
Technology

Current News

Status, February 1st, 2010.

A complete Review of the Work undertaken with the Focal Plane Imager,
Controller Hardware and System optimisation which proceeded from
September 2008 until the end of 2009.

1. Initial Work undertaken in late 2008/early 2009.

Much has happened since the rather now out-dated information was presented in the
links below, the last being November 20th, 2007. At this date, the Focal Plane had
been fully populated and Imaged, 8 devices at a time, through one of the FIB interface
boards. We subsequently found, however that the design of these left something to be
desired and the channels were oscillating and causing excess noise. This was subsequently
fixed but the overall design anyway came under closer scrutiny and it was abandoned
later (mid-2008) when it was found it would not meet its performance spec., see below
This is where this review begins, though at that time I was in the UK and didn't return
until Sept. 2008.

After almost 9 months of persisting with the Focal plane Interface Boards (FIB),
we took the decision to re-design the connectivity between the STARGRASP Controllers
and the Focal plane.

This decision was taken, as it became clear that the design of the FIB, although
working to 1st order, would not allow us to produce low noise image data from the
CCDs. In addition, the boards were rather over-complicated, designed this way, to
provide a high level of independant operation of all 64 amplifiers and to allow us
to select amps using links on the boards themselves. None of this was a bad thing,
but the confines of space and the design we had put together resulted in more than
an acceptable amount of heat being generated, just in the wrong place.

So, we had discussions with the suppliers of our controller hardware, the University
of Hawaii, and it was decided that they should provide us with interface boards
which would allow us to connect direct to the Pre-amplifiers in the STARGRASP
controllers, via a series of Ribbon cables, terminated in 2x20 way IDC Samtec connectors.

This design was pushed through, and when I got back to Australia in late September
2008, this was the-state-of-play. We had 1 STARGRASP Chassis, with 1 controller
board installed (4 CCDs) and a Test Chassis with a similar setup, located in the
Detector Lab.

With the benefit of hind-sight and knowing what I do now, this was far from adequate
to measure proper system performance as we demonstrated time and again over the next
year, that the system operated satisfactorily (read noise about 6e) when there was
only ever 1 board set in the controller. We needed at least 2 sets and preferably 4
(the full complement) operating the 16 detectors on one side of the focal plane. The
following is an image of the Eng. Setup showing all the components in the system,
without the Imager Vacuum Jacket, all parts sat on the bench for easy inspection..

What can be seen here is the STARGRASP Chassis in the background, the ribbon cables
connected to the Ribbon Interface Boards (RIFs) at the controller end and these then
connected to what would be the outside of the Imager Feed-through panel, at the other end.

In the foreground, which would be inside the Imager vacuum jacketnormally, are 4 Flex
cables connected to the Feed-through connectors at one end and to the CCDs at the other.
2 of the CCDs are actually 'dummy detectors' which emulate the electrical characteristics
of the real devices, for testing purposes. The other 2 are real CCDs, 2 of the Engineering
parts which are sat in their transit boxes, supported by the metal carrier with the cut-outs,
the CCDs themselves facing downwards, so that the silicon surface is out of harms way!
See the next picture.

We ultimately received another SAC (STARGRASP Array Controller) board for the
STARGRASP#1 Chassis. This meant we could then operate 8 CCDs and at this stage we
started to see problems with what looked like cross-talk pattern noise. The following
3 pictures show the connectivity using the Ribbon cables (8 in all) for the 2 SACs and
the 3rd picture shows one of the channels connected to the Test Dewar.

3. 4. 5.

The arrangement with the Test Dewar allowed us to confirm that it was possible to
obtain low noise performance form one of the E2V detectors when mounted in an
independent system. We consistently achieved 5-6e rms from the Test Dewar when
connected either to the standard ARC (Astronomical Research Cameras) Test Controller
or in the way shown above. Even when the SG#1/SACs were producing higher noise from
the focal plane CCDs.

In addition it was also possible to verify that the Focal Plane CCDs themselves were>
performing correctly by connecting a pair of these directly to the ARC based Test system,
see next picture..

We spent a considerable amount of time in attempting to optimise this system to produce
good low-noise read-out performance. Although it was apparent that the individual devices
themselves could produce the sought for low-noise, and even 4 devices together on a
single SAC controller was producing acceptable results, as soon as we moved to 2 or
more SACs in one Chassis, the noise performance with the hardware described above, was
atrocious - see next images.

7. 8.

Even the noise on these data frames was no where near as bad as it had been during the
course of the optimisation... I have been unable to find an image of that data though..

2. Remedial Action taken with the Ribbon Cables.

It became increasingly obvious during the course of the optimisation process that the
cause of the pattern noise seen above, was due to a problem with the connectivity, in
other words, how the signals were being conveyed/retrieved from the front of the Pre-amp
boards to/from the feed-through connectors on the Focal plane. The next picture shows the
complete board set which constitutes one controller (SAC). On the left is the DACQ
Motherboard with the FPGA daughter board on the other side. The FPGA is the Fully
Programmable Gate Array which is programmed with down-loadable firmware, which then
generates the clocks, biases etc for the CCDs and is responsible for data transfer to
the host Sun. The DACQ mother board hosts the circuits to provide the analogue to digital
conversion of the video signal from the Pre-amp board, a 32-channel part seen in the
middle of the picture. This board has links to permit data to be retrieved from the CCD
output in Left, Right or Split Serial mode. These boards are programmable so that offsets
and gain can be set along with the biases and clock levels for the CCDs. The board on the
right is the Ribbon Interface Board (RIF) which permits us to connect, via the Ribbon cables,
to the CCDs in the Focal Plane.

The clocking waveform applied to the detectors is obtained by editing the serial, analogue
processing clock waveform in an application called CestlaVie (Clocking Engine Serial Time Line
And Video Instruction Editor). This permits the all important serial transition phases to be
adjusted, the number of pixel multi-samples and the timing of the various signals to achieve
the best performance. See next picture for the clock waveform pattern we are currently using,
the so-called 2500ns4+1 pattern.

10.

As is maybe evident, at this stage we were in a position of having a variety of possible
sources for the pattern-noise problem. What was it about the connectivity and/or the clocking
pattern which would give rise to the excess noise, which at the outset was 300-1000adu above
nominal. In addition, was there any other possible source for this inter-activity, eg. could
the way we had connected the CCDs to the controllers be causing us a problem. At the outset
the ribbon cables were connected so the each horizontal pair of CCDs were connected to one
device channel (dev0 or dev1) on each of the RIF boards, was this the correct way to do it?.
There are 4 CCDs connected to each RIF boards, 2 CCDs are in parallel, so this is our 'quantum'
for control. There appears to be no performance hit in operating 2 CCDs in parallel, in terms
of clocks and biases. One dev channel (0 or 1) therefore controls 2 CCDs. It is up to us how
we sequence these to read from 1 amplifier (left or right) or 2 amplifiers (split) - see later
for the change that had to be made to the read-out mode in order to expedite the total read-out
time from the focal plane.

At this stage we were also reading out the CCDs through one amplifier only, as this had been
chosen to be the preferred way of doing things. This was partly due to the fact that this
permitted some redundancy if an amplifier failed, we could then switch to its partner at
at the other end of the register and this would allow us some contingency in case of problems.
Unlikely with the robust E2V devices, but this was thought at the time to be the best way to go.

What follows now is a summary of the work undertaken, with various pictures taken at each
stage, in an attempt to reduce the excess pattern noise. Many things were tried, using
Aluminium foil in the 1st instance, then to Aluminium metal with insulation wrapped around etc.
All this was also tried with shortened Ribbon cables. The ones with which we had been supplied
with were 500mm long and I considered these to be way to long for our requirement in any case,
but this length of multi-conductor ribbon (very similar to the sort of stuff used to connect
hard disk drives to motherboards in PCs etc) may also have been the cause of the problem. So
these were shortened, see the next 2 pictures for the original length parts (500mm) and a
shortened version of the Ribbons which were, 150mm long.

11. 12.

The length of the ribbon cables made no difference to the noise performance. What was noticed
was that moving the cables with respect to each other changed the pattern of the high noise.
This was not good. It appeared that any movement with the cables caused the noise to change in
un-predictable way... even if it was predictable - it still wasn't good!. So...
The ribbon cables were cut to a defined length based on the physical position of the FP and
controller hardware location, and this new length cable was used in all further tests; this
length was 290mm. From the somewhat hotchpotch arrangement with Aluminium foil and insulating
sleeving, seen above, we moved to a more tidy arrangement with Sticky-backed Copper tape which
could be applied to the ribbon cables, a strip on each side, the tape was exactly the correct
width for the ribbon cables and provided a setup much more reliable in an attempt at finding a
solution, See next picture.

13. 14. 15.

Here, it can be seen we have 2 RIF boards connecting to 8 CCDs and this time the CCDs are connected
so that each horizontal pair are connected to the RIF such that they mate to separate devs.
This did not produce any change in the noise performance and so we appeared to have at least
one parameter 'on the ground'. In this picture it is also more evident that the ribbons had to
have a twist in them to accommodate the connector orientation on the RIF boards. This twist
was subsequently removed when we turned the STARGRASP chassis over, so that the connector came
up to the RIFs from underneath, see the 2nd picture above for this detail.
Finally, for this arrangement, the 3rd picture shows the Ribbons in a test mode where the
conductors carrying the low level video signal (OSL and OSR) where removed and 2 pairs of low
impedance video coax. cables substituted; they are the black conductors lying either side of
the Ribbon cables. This was to test whether the pattern noise was being induced into the low
level signals being conveyed on the ribbons to the input of the pre-amplifier boards from the
CCD output nodes.

This solution too produced absolutely no change in the noise performance - it stayed persistently
high, now in the 50-100adu region

3. Final Solution - The RIF board problem.

During the course of our weekly Telecon meeting with our colleagues at the University of Hawaii
it was pointed out that there did appear to be one possible cause for the pattern noise being
induced into the video signal chain. Inspection of the RIF boards revealed that the large edge
connector which plugs into the Pre-amp board, had been assembled WITHOUT a shield across the
connector mating surface. See picture 16 below.

16. 17. 18.

As a quick fix and to test the idea, I installed copper tape over the connector mating surfaces,
ensuring that the copper was grounded at each side by connection to a pin close to the copper
tape. See picture 17.
These 4 boards were modified in this way and installed in the controller, see picture 18.

This produced an immediate and exceptional improvement in the noise performance.

It appears that in production, the RIFs couldn't be assembled with the correct mating, SHIELDED
connectors as they weren't available and so parts without the shielding (see picture 16) were
substituted.

3.1 An explanation of the Real source of the Problem.

Further discussion during ther next Telecon revealed some interesting facts.
The node sensitivity (i.e. the conversion factor in uV/e) of the E2V CCDs is of
the order of 6uV/e. This sensitivity can be changed by modifying the voltage on the
2nd output gate, OG2, which by default is set to a level of -4V (see picture 39 below
for a detailed scematic of the output section of the CCD). The sensitivity changes
with OG2 voltage and at +10V the node sensitivity is 3uV/e. This effectively reduces
the gain of the system and so can be used say, to select the sensitivity for observing
bright objects.
However, this feature was not what was causing the problem, but the fact that compared to
the Lincoln Labs (CCID45 & 58) OTA CCDs, the node sensitivity was quite low. The Lincoln Labs
OTA CCDs have a node responsivity of something like 18uV/e, three times that of the E2V
devices. The consequence of this reduced responsivity meant that when the input stages
and video processing electronics were set up on the pre-amps, the overall system gain had
to be increased to accomodate the lower node responsivity of the E2V devices. The
up-shot of this is immediately apparent, we have very high gain pre-amps in our system!
We all know what high gain amplifiers are good at doing, amongst other things, picking up
stray signals, if they are about, and wreaking havoc with what you are trying to measure.

To make sure that there was no possibility of pattern noise injection anywhere else,
I went to inordinate lengths to ensure everything else was shielded and grounded to the
common ground on the RIFs. The next series of pictures shows where this took place and
how the connections were made. Clearly we needed the proper boards supplying which would
do this job correctly but at last we had made the first big step in gaining the low noise
performance required from the SkyMapper Focal Plane.
The following pictures show the progress being made in attaining a low-noise performance
by adopting wide ranging shielding options on the front-end hardware-

19. The new EMI shielding being installed on the 4 RIF boards.
this had insulating material a sealing strip running along the length so that
it could be properly secured in place.
20. shows 2 of the RIF boards installed in the Focal plane with the above mode
21. shows the connection made form the Ribbon shields to the ground plane on
the connector on the RIF boards to ensure a continuous single ground connection.

19. 20. 21.

The 2nd set of 3 pictures related to the above show the addition of an extra ground planes-

22. attaching the copper tape to secure..
23. a copper plate across the pre-amp boards, common ground with connector shield
24. additional Aluminium plate across the RIF boards, as a test only

22. 23. 24.

Finally the last 2 pictures show-

25. all the above mods made and the 4 RIF boards installed in the Chassis and in operation.
26. the New RIF boards delivered from the UofH with the correct connector shielding installed

25. 26.

This assembly, along with some optimisation of the clocking pattern finally resulted in stable,
low noise operation of this side of the Focal Plane. At this stage we still only had 4 controller
card sets and it remained to be seen how the performance would change when we took delivery
of the remaining controllers (SACs) for the other STARGRASP chassis, and then these we operated
in parallel with the current hardware.

The remaining task would be to produce the lowest noise by carefully tuning what are
known as phase delays between the controllers, so that any pattern noise or
transients which take place during the read-out process, can be kept away from the
portion of the signal processing most sensitive to noise. This is described in the
next section on phase delay optimisation.

4. Optimisation of the detector read out timing.

Along with all the work which went on at the time as the changes and investigations into
the performance of the hardware, there was time also devoted to the optimisation of the
pixel sequencing and probably more importantly, the phasing of the controller
read-out process.

The phasing is a means of delaying, by any time between 10-2000ns, the commencement of the
read-out of each of the 4 pairs of controller channels. As each controller (SAC) channel has 2
independent devices (dev0 and dev1), each themselves controlling a pair of CCDs, we had 8
numbers with which to play. In general, the delays between dev0 and dev1 remained fixed
but the delay between each of the SACs was varied in an attempt to find the lowest noise (a
minimum if you will) in the read out process. Clearly as the numbers could be changed in say
10ns steps, anywhere between 10 and 2000ns, (Note the pixel time is 2500ns so there is little
point in extending the delay more than say half a pixel time), this represented a lot of variants
in the phasing. Taking 10ns steps from 10 to 1500ns is a lot of steps and considering there were
4 pairs of these at this stage (there would be 8 pairs once we got the full complement of
controller hardware) this looked like the job for an automated script which would 'probe'
the phase space to find one or more minima.

The attendant difficulties here were-

which controller (SAC1->4 and SAC5->8) is delayed with respect to which,
are dev0 and dev1 on each controller delayed separately to the controllers,
which chassis 'goes first', STARGRASP#1 or STARGRASP#2,
can some phase delays be the same.
How is the 'triggering of the delays between the chassis's to be achieved.

Although seeking a minima was tried at various stages, in an automated (scripted) way,
manually selecting phase values and trying them produced reasonable results. These would
have been much more reasonable were it not for the fact that at the same time the work
was continuing with the hardware to ensure IT was not causing us any grief in terms of
cross-talk etc and, as we have
seen from above, this was not the case.

So the work took on a double edged approach, attempting to sort out the hardware so that it
did not degrade noise performance, whilst at the same time attempting to run optimisations on
the pixel and phasing in software to attain the lowest noise we could, given the hardware was
operating in an optimal fashion.

5. Final Controller Hardware Configuration.

In the middle of all of the above, at the beginning of August 2009, we received the other
4 boardsets for STARGRASP#2, these would be SAC5->SAC8; SAC1->SAC4 being housed in STARGRASP#1.
This immediately meant we could start optimising for the lowest noise using all 8 controllers
8 pairs of phasing values being required. It also of course meant that at last we were able to
read out the whiole Focal Plane array in one go and look at the complete data.

At this time we also considered a fall-back option. As the UofH group had used a Rigiflex
solution for their inter-connectivity, we decided that it may be prudent for us to have a
similar option in case we found later that the Ribbon cables, even in their current configuration
may turn out to be problematic when we started moving things round. So we set to and designed
a Rigiflex alternative to the Ribbons. I decided we may as well use a similar design to the
flexes we already had inside the focal plane, but it would be prudent to add some extra shielding
with options to have the shields grounded or not at either end, i.e. at Focal Plane feed-through
and/or RIF. This design was quickly put in hand and whilst we waited for the delivery of 40 of
these - 32 Science and 8 spare, we continued with the Focal Plane noise reduction/optimisation.

During this time we also achieved what looked to be our optimum read-out time for the whole
focal plane, into our top end application, Cicada. This time was around 35 seconds and was
too long for the Skymapper Science requirements. As we had moved from an 1800ns pixel to
the quieter (relatively at present) 2500ns pixel this had had an immediate impact in the
total read-out time. Due mainly to Annino's efforts in code optimisation and efficient data
retrieval from the controllers, we managed to get this to around 25 seconds; still too long.

The only solution was to move from single port read-out of the detectors, to split read-out.
We had easily enough hardware to cope with the extra 32 channels (2 per CCD instead of 1),
and with only link changes required on the RIF boards we could accomplish the split read out
mode fairly quickly. In addition, it was considered that some of the pattern noise problem we
were seeing was being exacerbated by the relatively long row (2048 pixels) read-out time on
our devices, (compared to the UofH GPC 1 camera which utilises Lincoln labs OTA CCDS with only
600 pixels per line). This longer line read-out time for our devices resulted in a higher pattern
noise effect during this time. Configuring for split read-out would cut down the row length to
1024 and remove the excess pattern noise. It would also mean we would get the data out twice as
fast, which is what we wanted. This would result in a total focal plane read and disk dump of
12-15seconds and very close to the Skymapper science requirement.

Although we now had a much better noise performance from the hardware - it was obvious we
were still not seeing optimum noise performance due to incorrect phasing and from not having
an optimum pixel timing sequence. We were of course now seeing much better noise performance
than earlier on in this period of work.
The following image is a Bias frame taken on 16th September and shows a largely respectable
performance with several 'outliers', see the associated text file showing the stats.
Also clearly visible is the now much subdued pattern noise on some detector amplifiers,
which still require some attention either in the phasing or pixel timing.

27. 28.

5.1 Gain and Linearity.

Throughout this period, as well as measuring the noise (in adu, analogue-digital-units) an
almost constant assessment was made of the system gain, this was required in order to know
what the real noise of the CCDs were in electrons so we could compare what we had with what
we expected. It was always found that this varied with the signal level, being higher at
lower (few thousand adu) signal levels than at higher (20k adu) signal levels.

Here is the gain table from July 20th, 2009 showing the effect I mean.
The pre-flash level is the length of time the calibration LEDs are on. These LEDs are
situated around the periphery (there are 12 of them, 3 along each edge) of the imager
and they can illuminate the CCDs for a pre-determined length of time. Using these and
the well known Photon Transfer method, it is relatively easy to obtain the gain without
recourse to any other optical or external aids.

Preflash Level
(ms)

Net Signal(de-biased)
(k adu)

Gain
(e/adu)

5 5 1.3

5 6 1.6

5 8 1.5

5 15 1.1

10 8 1.1

10 15 0.6

10 20 0.7

10 24 0.73

The change in gain with signal level is a completely unacceptable situation. The gain
must be constant across the whole dynamic range of the CCD and electronics, from a few
10s of adu up to 65,000 adu, the limit of the 16-bit A/D. By the system gain I mean
the number of electrons that each adu measured on the display, i.e. in the data,
represents. Our goal was to have a system gain of around 1e/adu, with an option
that this could be switched to a lower gain (more e/adu) of between 1.3-2e/adu if
required for brighter objects. This can be easily achieved by changing the voltage
on the 2nd output gate (OG2) on these devices, which is the last gate before the
output node, at both ends of the serial transfer register.

The gain problem stayed with us throughout the whole of this period, up until the
end of the work in October 2009, whenm the issue was finally addressed.

What made matters worse, the few times I measured the system linearity, was that
it showed non-linear behaviour at low (<10,000 adu) counts. A plot of the incremental
signal measured versus the input signal from the Pre-flash LEDs, produced a plot with
the line breaking sharply at the 10k count signal level. It then continued to low signal
levels with a completely different slope. Both these problems (the gain and linearity),
when we arrived at a situation where we could be sure we had a stable, low enough
system noise, I felt were related. They both indicated a non-linear system and again an
un-acceptable situation for serious scientific work.

The following picture shows a 'quick-Lin' measurement taken in August and the problem is
easily seen.

29.

These measurements were repeated several times on different CCDs - all with the same
result, clearly something, somewhere was wrong!

5.2 Rigiflex Replacement for the Ribbons.

Whilst the above problems with the non-Linearity were being dealt with, the Rigiflexes
arrived and these had to have their connectors assembled, be cleaned and then tested
on our Room Temperature Imaging system. See next 2 pictures.

30. 31.

These show one of the Rigiflexes under test and connected to image one of the detectors
through amplifier B(Right), the other input to the ARC controller is underneath. On the
right is a picture of the inside of the so called 'Poor Man's Focal Plane'. This black
Pelican< box houses our 4 Eng. CCDs, 3 on the original Invar carrier, and the 4th on a
'roving' CCD mount block which allows us to test each of the Feed-through connectors as
and when required. All the CCDs are mounted with their active Silicon surfaces facing down.
The Room Temp. imaging rig provides data as a useful guide as to the state of all the
connectivity. If the wiring passes all DC tests, it is installed here and a Room Temp.
image obtained from the detector(s).

Although these images are subject to high dark current (due to the Temperature) for which
the signal ranges from about 5000adu to 65000adu, the signal sampling is fast enough for
us to obtain valid data, where features in the images can be seen. If these are nominal,
compared to those taken when the device is cold (these images having being earlier archived
for comparison, then we can be fairly confident that all is working correctly and the
wiring can be installed in the real Focal plane.

The following 2 pictures show first, 4 pairs of Rigiflex installed, so that for noise
comparisons can be made with the other 4 pairs of EMI-shielded Ribbon cables. Picture
33 shows the assembly with all 8 pairs of (version 0) Rigiflex installed.

32. 33.

At this stage, as can be seen from picture 33, the Rigiflex were connected to the Focal
Plane in such a way that pairs of detectors in a vertical plane were connected to the
device (0) on the RIF board; the next vertical pair to the right being connected to dev1.
Although this arrangement keeps neighbouring CCDs together on the same RIF board, it does
mean that some Rigiflex have to 'stretch' somewhat to reach the appropriate position on the
RIF board from the Feed though connector on the Focal Plane. The 2nd one in from the left,
lower bank, having to stretch the furthest. This was felt to be both unwanted and made the
whole Rigiflex assembly somewhat less flexible, an undesired consequence of connecting
the CCDs up this way.

There were one or 2 features of the version 0 Rigiflex which became apparent after we
connected them up, that were thought not to be advantageous. We therefore set about a slight
re-design to solve these problems and deemed these would be the final install parts, for
fitting just before Xmas. This re-design was done quickly and we had the final parts in
early November.
Again these had to be assembled, DC tested, cleaned and baked and then imaged on the
Room Temperature Rig in the Detector lab, before they could be fitted to the Focal Plane.
It was decided at this stage that if there was no impact on the noise performance of
the system, we would re-configure the way they were connected such that all the 8 left-hand
CCDs would connect to the 8 left-hand device channels (dev0) on the RIF boards and all
8 right-hand CCDs would connect to the 8 right-hand device channels (dev1). This would
mean a more logical way of connecting up the devices and would over-come the problem
that one or two of the flexes at the extremities in the previous assembly, which were
stretched a little too much, would be relieved.

The next 4 pictures show the more natural way the Rigiflexes now connect the RIFs to
The Focal Plane Feed-through connectors. These pictures, unlike the earlier ones, were
taken with the imager now mounted on the simulator with the detectors facing the sky.
The wiring assembly and controllers have therefore been inverted w.r.t the earlier pictures.

34. 35.

36. 37.

A consequence of connecting up the Rigiflexes in this way is that the dev0 and dev1
channels on each board now connect to CCDs which are separated by 3 other pairs of
devices. This was all checked out by comparing the system performance to both the
original way the Rigiflexes had been connected and to the Ribbon cable performance.
There was no performance hit in connecting up the CCDs this way and so this is now
the default connection assembly layout.

6. Attaining Low Noise, Linear operation.

Before the assembly could be finally moved to the simulator for final AIT before
deployment to SSO, we had to locate the source of the excess noise, the changing
gain with signal level and the system non-linearity. Though I deemed the last 2
were related and would be solved together.

During the last months of 2009, we (i.e. myself, Annino Vaccarella, Mike Ellis and
Mike Petkovic) had been having weekly Telecons with our colleagues, Peter Onaka and
Sidik Isani at the University of Hawaii. They are members of the group which supplied
us with the STRAGRASP hardware, see -
STARGRASP for much more information about STARGRASP, OTA's and the GPCs.

Several suggestions were made at these Telecons as to the possible sources for the
problem with the system non-Linearity. It was obvious we now had to address this
problem as a matter of some urgency.

There were several possibilities for the cause of the non-linearity and one of
the most obvious is the signal sampling, in the analogue processing electronics,
within each channel on the Pre-amp boards. The sample processing is controlled
via analogue switches and these themselves are sequenced through the pixel routine
which was mentioned at the start of this review.

Having operated, edited and used a few dozen pixel sampling sequences during this
period (the pixel sequences for STARGRASP aren't much different to the ones
we use for the ARC DSP based controllers at SSO), it didn't appear likely that
there was anything wrong with this aspect. Picture 38 is a screen shot of the pixel
sampling sequence we had been using up until the end of October 2009.

38.

In almost all respects this was identical to the DSP analogue processor sampling
we use on the Imager, DBS, Test and WiFeS ARC CCD controllers. The most notable
difference and what gives this system the great edge over the ARC based systems is
the ability to do fast pixel multi-samples This can be seen as the 1+4 pulses (the
first sample is discarded) on the bottom lines marked SAMPCLK and ADCCLK. The total
pixel time can be seen to be 2500ns and the 3-phase serial triplet, moving charge
to the Summing Well (SW), ready for measurement at the node, can also be clearly seen.
The PEDESTAL or DC level of the system is measured by the first 1+4 multi-samples,
the charge dump occurs and then the PEDESTAL+SIGNAL SAMPLE measured by the 2nd
set of 1+4 multi-samples. Subtraction of these 2 samples gives the analogue level
of the signal charge at the node. All this looks perfectly correct...

As I recall now, almost at the same time that we were looking at this and discussing
the issue, I happen to put together a mosaic picture of 3 bits of circuit related to
the output amplifier section of the SkyMapper E2V CCDs and pulled out the front-end
block diagram of their pre-amp board. See next pictures..

39. 40.

The first picture shows the end of the serial register with its clock connections, the
output FET (this is the component which is static sensitive and the pair of them, one
at each end of the register, the reason why we take inordinately careful anti-static
precautions when we go ANYWHERE near the devices!) and the connections to
the CCD on its mating Tactics connector. The 2nd picture shows a simplified form of
the input to one of the 32 analogue channels on the Pre-amp card.

What these diagrams did, during the course of the Telecon discussions, and knowing the
form of the pixel routine illustrated above, was raise a concern that there could be
an issue in having the CCD RESET clock and the amplifier Clamp (VCLAMP, which effectively
provides a zero reference for the input) co-incident in the pixel routine.

This was something new, there had been plenty of instances where the co-incidence of
these clocks had been used, and there hadn't been a problem. It is usually expedient
to try and put as many clock transitions in parallel, as possible so this does not
slow down the pixel sampling. Care has to be taken in doing this as some clock transitions
can't occur at just any time and some can't be co-incident with others, for performance
reasons. One of the main rules-of thumb is to have nothing 'noisy' going on in the
pixel, when the pixel signal sample is being taken, particularly in the gap between
the 2 (1+4) multi-samples.

So, almost the 1st thing which was done was to move the RESET and VCLAMP clock pulses
away from each other. The following picture is a repeat of the one shown above and shows
clearly that the 2 clocks have been separated by about 100ns. compare this with the
picture above, 38.

41.

Again this had an immediate effect on the CCD noise and pattern noise. It looked as if
the co-incidence of VCLAMP and RESET were having a detrimental effect on system
performance. On our ARC based systems, doing the reseting and clamping in this way
causes no adverse effects but here, attempting to reset the CCD node whilst 'black
clamping' (as it is called) the input to the video processor were causing us grief.
It was a simple matter, using CestLaVie, to move the RESET pulse, earlier in the pixel
routine and this was all done in such a way so as not to lengthen the pixel time and
so maintain its 2500ns duration.

The effect of having the 2 clocks co-incident, we believe, was that the CCD was only
partially reset when the clamp was removed. This would leave a signal at the input to
the pre-amp which would be more dominant at higher signal levels. This residual
signal having more of a detrimental effect as the real signal level increased - exactly
what we were seeing in the Linearity data.

6.1 The Linearity Problem.

Having demonstrated that the problem with residual excess pattern noise was attributable
to un-shileded connectors on the RIF boards, which had now been fixed - it was also
evident that reseting and clamping the CCD and video processor incorrectly, could
be responsible for signal not being measured correclty at the higher levels.
In other words the problem with the non-Linearity lay not at low signal levels
(<10,000adu) but at high signal levels. This made much more sense than what had
been thought earlier, missing signal at low signal levels. So, naturally, the Linearity
was measured all over again. The result is shown in the following picture...
it speaks for itself I guess!!

42.
(The vertical blue line is a scanning artifact.)

We appeared to be 'home and dry'... !

After this, noise and gain were again measured at various settings of ADCGAIN, this is
a software settable feature to allow us to modify the overall system gain of the video
processor so we could 'tune up' to the actual gain we wanted, namely, 1e/adu.
Maintaining OG2, the clock phase which can be used to change the node sensitivity,
at OG2=-4V, we obtained the following results for System Gain.

ADC Gain

System Gain
(e/adu)

1.0 1.3

1.25 (the DEFAULT) 1.12

1.5 0.95

In addition, with ADCGAIN fixed at the default value of 1.25, the voltage on OG2 was
changed and similar measurements taken by varying OG2 this time and keeping ADCGAIN
fixed at 1.25. OG2 is the clock phase which can be used to alter the node sensitivty,
and hence the gain. The results obtained in this case are again shown in the table below.

OG2 Voltage
(V)

System Gain
(e/adu)

Noise
(e)

-4 (the DEFAULT) 1.12 5-6

0 1.3 8

+4 2.2 12

All now perfectly consistent and operating as expected!
The change to OG2 can be effected from the front panel of the Cicada Application,
there being a gain option for the observer to use, with the above characteristics.

The noise measured for the whole focal plane after all this work was complete, towards
the end of November 2009, is shown in the follwing picture. This is a statistics table
produced by 'Director', which is the Engineering interface we use to do all the low
level quantification and testing.

43.

A quite remarkable set of numbers considering the mess we started with over a year ago!

These numbers show the noise statistics across various areas of the detector, time is
on the left followed by the controller number. Each SAC has 2 devices (0 and 1) and
each device has 2 CCDs connected in parallel - in split serial read-out mode this
results in data beeing generated for 4 amplifiers, 2 per CCD, hence the ch0-ch3 entries.
There are therefore, 64 of these amplifier data sets. The SAC stats. are split into dev0
first followed by dev1. The meaning of the various statistics parameters are explained below.

ped: If a readout was performed with "bpp=32" and the FITS file was saved with
"bpp=32" then the mean pedestal level will be reported in the FITS header. Usually, the
pedestal image is subtracted from the data before saving but this can be a good diagnostic.
Pedestal levels should be within a few 100 ADU of 5000 or whatever value is set in the scripts.
Bias: is the normal bias level in the overscan, and well known.
back: measures the mean level (above bias) in a portion of the image area.
rmsped: If you have a 32bpp STARGRASP FITS file (see ped= above) then this
statistic reports the RMS in ADU of the pedestal. Expect this to be quite high in many
cases. This is the uncorrelated sampling noise. By default we only use 16-bit settings
for data acquisition and retrieval. The upper 16 bits should be zero.
We can retrieve the full 32 bits but this slows the system up somewhat and doesn't
give us any more information, in a science sense, than the 16-bit data. It is useful
however, in 'diagnostic mode'.
rmsov: This is the RMS of the subtraction of (vid-ped) which is the 16-bit
value normally delivered as the image, for pixels in the serial overscan region..
rmsmask: The RMS in ADU of part of the image after masking out any bright
chip defects.
region in the image area.
rmsrow: The RMS in ADU of the same region, after applying a "row-by-row"
bias correction to each pixel row individually.
rmsplane: The RMS in ADU after the above treatment by rmsrow, but after
fitting a plane to the image to take out any overall horizontal slope.
rmstilt: The RMS in ADU after fitting a sloped bias correction to each
individual row of pixels.

7. Some Pictures taken with the Imager on The Simulator.

The CCD images in picture 47 and 48 were taken using Cicada to control the full
system as described above. A small pin-hole had been installed at the front of the
tube seen in pictures 45 and 46.

The pictures are-

44. An image of the Focal Plane showing all 32 CCDs during the time in which
the Imager was rotated so that the focal plane faced the sky and moved to the
Simulator. It being moved from its Test Bench where it had been mounted upside-down
for ease of use in Testing, and onto the Simulator.
45. An image of the complete system on the Simulator showing the 2 STARGRASP
chassis SG#1 at the top and SG#2 at the bottom.
46. An image showing the focal plane, taken down the Simulator tube, with the
shutter and filters withdrawn.
47. A pin-hole image of a group of noteworthies - taken as soon as we had all the
image formatting correct in Cicada.
48. Zoomed in portion of the above showing more detail of the folk present. The
ghostly image of me in the lower right hand corner is due to my having to get up and
down during the 20 second exposure, to open and then close the shutter.!

44. 45.

46. 47. 48.

8. Some Reference Data Frames taken in Cicada.

The 4 data frames which follow will be useful in assesing the Focal Plane state and
verifying the operability of the 2 STARGRASP controllers and associated hardware
once the instrument is at Siding Spring.
The Imager and its associated hardware is currenlty set for transportation
on the 16th February and the Imager should be in operation on the Telescope,
2 weeks after that..

49. A standard zero second exposure Bias Frame
50. A standard 5ms LED Pre-flash frame
51. A 2 minute exposure on the Test Pattern

52. A Room Temperature (309K) image taken just before the Imager was
was removed from the Telescope Simulator (on Monday 8th February) after warm
up on Saturday February 6th.

9. Some On-Sky data taken in early March 2010 ('1st Light').

The Focal Plane was re-connected to the controllers on Wednesday February 24th.
The process, in a step-by-step, considered way, took 4 hours. There were no
issues and the process went well. By mid-day the whole mosaic array was connected
and the power turned on just after that. The first image out, a Room Temp frame
to confirm the FP's operability before cooling, looked good - all amplifiers were
working and the frame looked similar to that shown above in picture 52, this RT
reference frame taken BEFORE dis-assembly so that comparison could be made once
all had been re-connected at SSO.

After this, it was decided that the cable drape should be configured correctly
and this work took most of the following (Thursday) morning. It was decided that
it was better to get this sorted out before we commenced the cool-down process.
The cable drape work took all Thursday morning and was complete just before lunch.
Once we had all agreed that the sub-systems were ready to go, the cool-down
process was enabled from Cicada and the cryo-pumps started.

Detector System Assembly Pictures.

A full set of pictures showing the STARGRASP controllers being assembled onto
the Vacuum Jacket, the installation of the Rigiflex Interface boards and
the Rigiflex connections to all 32 CCDs being made, can be found *here*

The focal plane was imaged during the cool-down process and bias frames taken
to monitor performance.

The following images were taken on the Friday morning, early, just before I
left SSO. They represent comparison data frames taken for reference and are
used to confirm the status of the system w.r.t. that when the imager was
here at R.S.A.A.

53. A standard zero second exposure Bias Frame at operating Temp.

*MEXT 64 Channel Science CCD R/O FITS data (546Mby)
54. A standard 5ms LED Pre-flash frame.

*MEXT 64 Channel Science CCD R/O FITS data (546Mby)
55. A 600s Dark exposure, showing CRE's

*MEXT 64 Channel Science CCD R/O FITS data (546Mby)
56. A 120s exposure, taken on the 8th March, of the Tarantula Nebulae,
NGC 2070
Notes: this frame has not been de-biased or had any flat-field correction.
The narrow and 2 wide gaps are the spaces between the rows of detectors
the wide ones being caused by the space required to allow for the bond-wire
connections to the CCDs. The dark vertical strips are the CCD under-scan
strips, each 50 pixels wide. These are real pixels in the serial shift register
which do not receive light and are the 1st pixels out at each side of each
detector. The full x-width of each CCD panel being 50+1024+1024+50 pixels.
The variations in the bias signals for each of the 64 amplifiers (2 per CCD)
are caused by slight differences in the video processing chains in the read-out
electronics. The Bias levels are between about 1200 and 1800 adu.
In addition, this frame was taken with the STARGRASP X-trigger> function
disabled. This is a signal which is conveyed from one of the Chassis (#1)
to the other (#2) and enables correct timing readout of the array between the
2 controllers.
The result of (accidentally in this case) not enabling this feature (there is a
check-box in Cicada to enable this function) is that the data 'slips' when
retrieved and the frames look both noisy and have vertical features in them
where the horizontal lines have been displaced with respect to each other,
in a simliar way to a TV picture frame slips due to the horizontal line scan
lock being disrupted. Something which rarely happens these days..

*MEXT 64 Channel Science CCD R/O FITS data (546Mby)
57. A 30s exposure, taken on the 9th March, of the Lagoon Nebulae,
NGC 6523

*MEXT 64 Channel Science CCD R/O FITS data (546Mby)

58. A 60s exposure, taken on 11th March, of the Jewel Box
or Diamond Cluster, NGC2516-R

*MEXT 64 Channel Science CCD R/O FITS data (546Mby)

And that, as they say, is that. Done Job.
Paddy Oates. March 11th 2010.

* End Latest 2010 Update *

STOP ..........

Documentation below refers up to the end
of 2007, Some FITS data has now been removed from
the docs below.

Status, November 20th, 2007.

All 32 Science CCDs Imaging thru FIB into SAC.

I am further pleased to announce that today, we finally imaged
all 32 Science CCDs into the SAC#1 through our 8-channel FIB
inter-connect and signal routing board.

As only 1 of the FIB has been built and tested, we imaged the Science
CCDs, 8 at a time through the 8-channel FIB and into the SAC#1
(STARGRASP Array Controller).

The 4 images below show all 32 CCDs being imaged at Room Temperature,
our next Cooldown (#4-) is on hold until we reach a suitable pressure in
the SkyMapper Vacuum Jacket, this is now unlikely to be before next week.

The following data was obtained yesterday afternoon (Mon 20th) and the
data is oriented with the S/M Vacuum Jacket upside-down, CCDs facing
down. So what is termed UPR/LWR bank in the following text is 'as it
looked when reading out' and this is opposite to what the designation
would be were the instrument facing the sky!
(Don't want to cause any confusion!)

4, 8-channel Science CCD Mosaics read out
through FIB and into SAC#1.

SAC #2 side UPR bank of connectors

Tif format image

*MEXT 8 Channel Science CCD R/O FITS data (141Mby)

SAC #2 side LWR bank of connectors

Tif format image

*MEXT 8 Channel Science CCD R/O FITS data (141Mby)

SAC #1 side UPR bank of connectors

Tif format image

*MEXT 8 Channel Science CCD R/O FITS data (141Mby)

SAC #1 side LWR bank of connectors

Tif format image

*MEXT 8 Channel Science CCD R/O FITS data (141Mby)

Close inspection of the last image will reveal that devices 1st and 4th from
the right do not appear to be imaging. The frames are just bias frames.

A test this morning with the ARC 2Chip configuration assembly revealed
the CCDs ARE actually functioning through both of their output ports.
So we appear to have a small anomaly - we know the external hardware
works, it has been rigorously tested and is the same hardware used for
each bank of 8 CCDs. And we know the CCDs are imaging correctly...
so, a small puzzle remains..

Paddy Oates, November 20th. 2007.

Status, November 16th, 2007

All 4 Engineering CCDs Imaging thru FIB into SAC.

I am now pleased to announce that today, we finally imaged the
4 Engineering CCDs, located in the PMFP (see below), through one of
the new Focal plane Interface Board (FIBs) and into the STRAGRASP
Array Controller (SAC#1).

4-channel Eng. CCD R/O through FIB and into SAC.

*MEXT Eng CCD R/O FITS data (70Mby)

This marks yet another very significant milestone in the operation of
the full Science focal plane, which is now awaiting its next cooldown,
(#4-) and connectivity to the 2 SAC 16 channel controllers.

It is planned to interface the current 8-channel system we have been
testing, to the Science Focal Plane, on Monday. This will enble the
science array to be read out, 8 CCDs at a time, into the SAC and under
control of cicada.

Following on from this, as soon as we have the remaining 3 board sets
tested and the Focal plane is at operating temperature, a full 32 CCD
read out can then be undertaken and the system noise performance measured.

Status, November 15th, 2007

First Image retrieved from new FP FIB.

Thursday afternoon, November 15th. The first image was obtained from
one of our E2V MS#02 CCDs (a fully operational 'mechanical sample' CCD)
- taken at Room Temp using the new FIB FP inter-connectivity boards.

This means we are now on track for connecting to our Room Temp. 'Poor
Mans Focal Plane' (this is a 4 Eng CCD rig to allow us to perform
Engineering work, see pictures below) and retrieve images to confirm
the operability of the new hardware.

From there it will be possible to extend the R/O configuration to 8, 16
and 32 devices and so image the true focal plane at Room Temp before
Cooldown#4- proceeds next week.

We are now on the way to having a fully operational focal plane, interfaced
through our new connectivity hardware and for which, subsequently, the
electronics can be optimised to provide the low noise optical
detector focal plane, we have all been working so hard to achieve.

Status, November 9th, 2007

Focal Plane Connected and Working.

Subsequent to the Full population of the Focal Plane, an extensive connectivity
test was performed yesterday (Thurs) which was successfully completed.
Having obtained the external shorting harness yesterday lunch-time,
all 32 detectors were interfaced to their mating Tactics connectors and
then a full Room Temperture Imaging test performed, 2 detectors at a time.
Data being obtained from both the CCDs output amplifiers to verify all was
working.

The RT imaging confirmed the full operation of all devices in the Focal plane
and so around 4pm, the Vacuum pump was connected and pumping of the
instrument commenced. We will now move to Cooldown#3 (the 4th), first thing
Monday morning.

This will be the first thermal cycle with a full complement of Science CCDs
and so is an important step in the Focal Plane characterisation process.

Well done all concerned is what I say!

Status, November 7th, 2007

Focal Plane Populated with E2V Science CCDs.

Cooldown #2C was terminated on Friday November the 2nd in
preparation for this weeks activity....

The 32 Science CCDs were installed during a 9 hour
period yesterday, November 6th.

All the installation work went well and the Focal plane looks
stunning. we have a great instrument.

Two shots of Fully populated Focal Plane.

A full set of pictures can be found
*here*

My thanks to Mike Petkovic who sat and provided support all day and
did the documentation on the installation process so there is now
a formal written record of what I did! Both for the installation
and removal of a device.

In no particular order...
Thanks should also go to Peter Conroy and Andrew Granlund for taking
the E2V CCD installation design and implementing it so well in SkyMapper
and providing a very safe system to work on.
Andre de Gans assembled the Flexes, did all the labelling and cleaning and
the internal wiring harnesses for temp. sensing etc.
Errol Kowald laid out the flex cable design & Samtec panel feedthrough
connectors.
Mike Ellis and Marian Szczepkowski have been and are reposonsible for all
the low noise cabling and focal plane interconnectivity respectively, the
latter of which will be characterised tomorrow and in the next few days.

In the workshop, Col, Robbie and Ross have provided excellent mech.
Eng support and last but not least all those 'hidden folk' in Admin who
make all the big decisions and are after all the most important
folk of all :-8)))

Status, October 3rd, 2007

SkyMapper Focal Plane at Cooldown 2B.

We have pushed along the refit of the focal plane with the modified Invar
plate carrier and some up-dates to the internal mechanical and electrical
arrangements. All this was carried forward over the last few weeks with
the efforts of the Mech., Electronics and Computer Teams here.

Namely Peter Conroy and Andrew Granlund on the Mech. side
Andre de Gans, Marian Szczepkowski and Mike Ellis in Electronics
and Annino Vaccarella in the Computing section.
Overseen and 'encouraged' by our Project Manager, Mike Petkovic.

The Focal plane is now operating once more at temperature (-120C)
to allow me to do some more noise characterisation. We will then be
setting up to interface our new SAC hardware via the FIBs (Focal Plane
Interface Boards) which allow us to inter-connect the CCDs to the SACs
and to then do an end-to-end test with the 2 SAC controllers. This will
now be operated directly with Cicada in either 8, 16 or 32-channel mode,
(see details below for an account of the progress with this aspect).

We are now also in a position to interface 3 devices (1 Mech Sample
and 2 Eng CCDs) via the new FIBs to the 2 SACs at Room Temp.
All this hardware is being proofed at Room Temperature using our
newly contructed 'Poor Mans Focal Plane'. This is the old Invar
plate carrier from the VJ, mounted in a black Pelican box with an
aperture to allow us to mount one of the spare Samtec Connector
carrier plates and connect, via 3 spare flexes, the Mech. Sample #02
and Engineering CCDs #03 & #04 CCDs mounted on the Invar plate,
see the following pictures.

Some pictures of the PMFP!! Assembly.

The following pictures illustrate the installation of the Mechanical
Sample CCD and the 2 Enginnering CCDs into the Invar Carrier plate.

1. Draw bar pulling the 2nd CCD into place,
locating pin just behind the draw bar.

2. 3 CCDs now installed onto the Invar carrier, showing
shorting pads still in place and all ready to go in the Box.

3. Invar Carrier and 3 Detectors in the light tight Pelican Box,
and ready to be brought outside!.

These E2V CCDs image sufficiently well at Room temperature,
using fast pixel sampling (0.2+0.2us), to retrieve images to
confirm the SAC/FIB hardware is working as expected.

This will then give us confidence that we can safely connect to
the $2.5million Science CCDs when they are installed in the Focal plane.

The population of the Focal Plane with the 32 Science CCDs is set to
take place before the end of this month (October 2007).

Status, September 18th, 2007

SkyMapper Focal Plane Imaging Data taken
from STARGRASP in 32-channel mode.

We have just acquired the 1st 32-channel R/O from a single
32-channel SAC and reading direct into Cicada.

Again - we only have 2 CCDs connected and they are in positions
6 and 7 as indicated in the small diagrams below.

Due to the relatively slow pixel sampling (2.5 or 4us) being used at
present, the effects of dark current are swamping the 2 images. We are
attempting to get a faster read-out implemented (0.5-1.0us sampling)
and this should result in us being able to see more of the detector
defects, evem at Room Temp.
The image below shows a 32-channel R/O and this data was taken from
a SINGLE SAC unit with 2, 16 channel Pre-Amp and DACQ cards mounted
in the box.

This setup is different from what we plan to do when the system is
deployed, where we will have 2 SAC boxes, one at either side of Focal
Plane and each housing a single set of 16-channel hardware consisting of
one P/A and one DACQ/FPGA card.

There will be an identical system residing at RSAA acting as a Test bed
and as a 'hot-spare' in case of problems at SSO.

32-channel R/O from a single SAC.

Room Temp. Bias Frame data JPG image
Room Temp. Bias Frame data TIF image
*MEXT Data Cube or Mosaic IRAF FITS data (564Mby)

Status, September 6th, 2007

SkyMapper Focal Plane Imaging Data taken
from STARGRASP in 16-channel mode - WORKING.

We just achieved 16-channel read out from the SAC, see the
image below, which again shows an LED Pre-flash frame,
of 100ms duration with the 2 real detectors in the positions
indicated below in the small table.

Annino Vacarella deserves credit for his persistence with the
Cicada Integration and I would like to thank John Tonry, Peter
Onaka & Sidik Isani at the University of Hawaii for their
excellent design and continued support of our
STARGRASP Controller.

Bank:-DEV#0------------||-DEV#1-----------
--------------------------------------------------
CCD# 01 - 03 - 05 - 07 || 09 - 11 - 13 - 15
CCD# 00 - 02 - 04 - 06 || 08 - 10 - 12 - 14
-------------------------^------------------------active detectors
-------------------------v------------------------

100ms LED Preflash Frame data JPG image
*MEXT Data Cube or Mosaic IRAF FITS data (282Mby)

The mapping of devices has been set so that the images on the
display match the physical location of the devices when we have
the full complement of Science CCDs populating the Focal plane.

The pixel values in the device (Eng. CCD#03) at position #06 range
from a few thousand adu to about 18,000 adu and the pixel values
in the device (Eng CCD#04) at position #07 range from about
18,000 adu to 52,000 adu. These values are consistent with those
found on the ARC Test system and from the fact that device #07
is vertically above device #06 in the focal plane. It is also
nearer the source of the LED light. See following picture:-

Arrangement of Eng. CCDs in Focal plane.

There is also evidence once more of a small amount of X-talk into the
other channels but given that none of the remaing 14 channels are
connected to anything or that the wiring and inter-connectivity hardware
has been finalised, this is not un-expected.

This afternoon we will be procedding with some noise tests and early next
operating both the 16 channel controllers to finally achieve 32-CCD read out.
This work should now procedd without event as the Cicada environment to
operate 32 CCDs is already in place.

Status, September 5th, 2007

SkyMapper Focal Plane Imaging Data taken
from STARGRASP in 8 & 16-channel mode.

We have just achieved Read out of the 2 of the Focal plane Eng CCDs
into the SAC (STARGRASP Array Controller) under control from CICADA.

The 2 images below show the 8-channel readout out, only 2 CCDs are
connected at the present as we have yet to take delivery of the
2 sets of 2x8 channel SAC to VJ inter-connection boards,
FIB, Focal Plane Interface Boards.
These will allow us to finally read both sets of 16 Science CCDs
when they have been tested and installed.

Block Diagram of the System.

The following image shows a Block diagram of the system illustrating all
the relevant parts of the hardware: SAC #1 & #2, FIBs 1,2,3 & 4 and the
Vacuum Jacket Invar plate.

Pictoral Diagram of the System.

The following image shows a mosaic of pictures fitted together illustrating
how we expect the final system to look. This reflects the components
identified in the block diagram above: The Vacuum Jacket with 4 Engineering
CCDs mounted, the two SAC#1 FIB and the SAC#1 with its temporary front
plane interconnection board which allows connection to any 2 of the 4
Engineering CCDs.

The installation of the 32 Science CCDs is now set for late
October and this operation will take place in the AITC Clean Room.
We may by that stage have set up a Web-cam - so the procedure
can be watched by interested parties... at my discretion!..

In the meantime we are imaging with any 2 of the 4 available Engineering
devices which have now been in the focal plane since the 5th May.
We have therefore maintained the detectors at operating temperature
for some 4.5 months and have seen no adverse effects with the operation
of the focal plane.

1st Image: an 8-channel R/O into Cicada

As can be seen, this is an LED pre-flash (100ms duration) of the CCDs
in the focal plane. We have an ring of LED light sources mounted around
the periphery of the focal plane just on the outside of the window. These
can be pulsed, in the same way I do with the internal LEDs on teh 2 test
dewars.
They are useful as they permit a quick and easy exposure to be taken to
confirm the operation of the complete system.
We can read out only 2 of the 4 available Engineering CCDs at present and
we are currently connected to Eng#03/#04.

The image shows all 8 channels read-out but only the top right-hand and
bottom right-hand image data is real. The other channels contain a small
amount (~0.1%) of X-talk from the 2 live channels.

Of note aew the detector defects - easily seen in the FITS data, and the
X-talk from the defects xcan be seen in the un-loaded (i.e. 'floating')
remaining video channels.

In the JPG image I have pointed the cursor at one of the artefacts on the
CCD surface, to confirm that we really do have the correct data. I have
seen this 'feature' in pre-flash frames taken using the ARC (SDSUIII)
controller when connected to the Focal plane. So all this is real CCD data.
Scroll over to the right to see the 'feature' in the magnified part of the image.

8-Channel, 100ms LED Pre-flash frame

Pre-flash images from from Eng#3 and Eng#4 CCDs at top and bottom right.

100ms LED Preflash Frame data JPG image
*MEXT Data Cube or Mosaic IRAF FITS data (141Mby)

2nd Image: a 16-channel R/O into Cicada

16-Channel, 100ms LED Pre-flash frame

Pre-flash images from from Eng#3 and Eng#4 CCDs, Data is missing!

You will note that the LED pre-flash light, which should now be
located in detector 'panels' 6 and 7, is missing - the data in
these panels looks like Bias.

The data is retrieved in 2 x 8 chip banks, and the arrangement
on the display has been organised so that the appearance matches
how the detectors would appear - looking into the front of the
vacuum jacket. The image data is therefore organised as follows:-

100ms LED Preflash Frame data JPG image
*MEXT Data Cube or Mosaic IRAF FITS data (282Mby)

We are actively seeking the solution to this problem but we
feel we are now very close to having the 16 channel controller
reading out the data correctly.

It is planned that today or tomorrow we will connect the fibre to
the 2nd SAC and we should then be able to retrieve 32 Channel data
straight into Cicada. This _should_ be straight forward as this has
been tested already under simulation. The hardware exists to do all
this - all we have to do is connect it up.

On Friday (7th September) the Focal Plane will be allowed to warm
up over the ensuing weekend, after 4 and a half months at temperature
(155K). This is to permit us to refit the Vacuum Jacket with modified
Invar and Copper plates and do some other engineering tasks.

We hope this work, refittment of the 4 Engineering CCDs and then
pumping and cooling again will be finished by the end of September.

There will then be a final Engineering characterisation phase
before the system is allowed to warm up for the installation of
the 32 E2V Science CCDs towards the 3rd or 4th week in October.

Status, June 21st, 2007

SkyMapper Focal Plane Imaging Data taken
from STARGRASP & E2V Eng#01 & Eng#02 CCDs.

I have just taken images from the Engineering #1 & #2 CCDs in the
SkyMapper focal plane, and read them out into the SAC (STARGRASP
controller) simultaneously - in similar fashion to the way the ARC
reads out the devices in 2Chip mode, as described below.

The imaging capability of the SAC and its and interface to the SkyMapper
Vacuum Jacket has been progressed with the hard work undertaken by
Bernie Keys in the electronics section here at RSAA. He has
continued to provide valuable technical support during this period,
to achieve the 2 CCD R/O from the SAC, into the Test System.

In addition Annino Vaccarella from the computing section
has made a commendable step foreward in providing the read out
environment under control of our front-end detector application,
Cicada. This was achieved today (Thurs 21st) and we can now
read any 2 of the 4 Eng CCDs in the Focal Plane, directly into Cicada
via the SAC controller.

From here we will need to extend this control to 16 channels, by
management of the hardware and software already in place to
achieve this. This is 'work in progress'...

The hard part was to get 1 CCD and then 2, simultaneously
operating; the move to N (16) should pass without event - this
is now being vigorously pursued.

Pictures of the experimental setup are shown below.

1. 2.

1. General view of the S/M Instrument showing all the components.
2. Detail of the Detector to SAC interconnectivity.

There are a selection of images below, taken using the SAC and connecting
to 2 of the 4 Engineering CCDs currently available in the focal plane.

All images were taken at the operating temperature, T=-110C.

LED Pre-flash frames from 2 CCDs.

10ms LED Pre-flash images from from Eng#1 and Eng#2 CCDs.

10ms LED Preflash Frame data JPG image
*MEXT FITS data (35Mby)

50ms LED Pre-flash images from from Eng#1 and Eng#2 CCDs.

50ms LED Preflash Frame data JPG image
*MEXT FITS data (35Mby)

200ms LED Pre-flash images from from Eng#1 and Eng#2 CCDs.

200ms LED Preflash Frame data JPG image
*MEXT FITS data (35Mby)

Dark Frame Data from 2 CCDs.

3600s Dark frame images from Eng#1 and Eng#2 CCDs. This is an approximate time only,
as we have not implemented an integration timer on the Test system at the present time.

T=-110C, 3600s Dark frame JPG image
*MEXT Fits data (35Mby)

Bias frame images from 2 CCDs.

Zero second Bias images having a sigma (noise) of xxadu from Eng#1 and Eng#2 CCDs.

0sec. Bias frame JPG image
*MEXT FITS data (35Mby)

Pictures of the Pin Hole Test Exposures.

Pin hole image of Colour photograph, stuck to the ceiling of the Pump Room,
imaged on to the ENG#01 & #02 CCDs at the same time. This photograph shows
the origianl members of the SkyMapper team in the Mech. Design demountable
just before the 2005 CDR in August.

15s Pin-Hole image of original Skymapper team JPG image
*MEXT FITS data (35Mby)

15s Pin-Hole image showing affect of water accumulation on
front window JPG image
*MEXT FITS data (35Mby)

Status, May 17th, 2007

SkyMapper Focal Plane Imaging Data taken
from ARC & E2V Eng#01 & Eng#04 CCD.

I have just taken some images from the Engineering #1 & #4 CCD in the
SkyMapper focal plane.

These images were obtained by exposing the detector to light passed through
1 of 4 small pin-holes, sat just above the detectors on the VJ window. The
availabilty of the 4 pin holes means any 2 of the 4 detectors may be so imaged.

To make things easy - our small iris shutter is sat above the appropriate
pin-hole so that timed exposures may be undertaken.

All images were taken at the operating temperature, T=-100C.

The location of the shutter was moved slightly, to more centrally position
over the pin-hole for the 2nd pin hole image. There is some evidence that
the 1st image was vignetetted slightly by the shutter's off-centre position.

Pictures of the Pin Hole Test Exposures.

1. Pin hole image of data sheet stuck to the ceiling of the Pump Room,
imaged on to the ENG#01 CCD detector.
2. Pin hole image of a photograph of the early SkyMapper Team, again
stuck to the ceiling of the pump room and imaged onto the ENG#01
CCD detector.
3. Central section of pin hole image showing the team members,
ENG#01 CCD.
4. Pin hole image of a photograph of the early SkyMapper Team, again
stuck to the ceiling of the pump room and imaged onto the ENG#04
CCD detector. The left hand side of the image is Eng#3 CCD which is
a bias frame as there was no light imaged onto this device.

1. 2. 3. 4.
2a. Pin Hole image of early SkyMapper team members Fits data, ENG#01 CCD
4a. Pin Hole image of early SkyMapper team members Fits data, ENG#04 CCD

Status, May 10th, 2007

Data Retrieval from SAC (STARGRASP) to Cicada.

We are now in a position to retrieve CCD data directly from the SAC to Cicada.

At present this is using our Test Dewar#1 with the E2V Mechanical Sample CCD#2,
this a fully operational device and so is useful for this type of Engineering work.

Annino obtained the Test Pattern data yesterday (9th) on the RSAA Test system
using external Temp. servo and manual shutter control, (i.e a switch, Mr. Rimmer)

The image below is a fairly standard looking Test Pattern frame, similar to
the data I have already obtained from this Test Dewar and CCD using the ARC
controller - see details towards the middle of these pages for this data.

The data below shows a 5s Test Pattern frame at T=-110C taken using
our Test Dewar#1 and the SAC controller.
This was all achieved using Cicada and communicating directly with
the SAC (STARGRASP) control hardware.

Test Pattern data JPG image
T=-110C Test Pattern Fits data

SkyMapper Vacuum Jacket Eng. CCD Test Phase

The work to characterise the Engineering detector's performance is underway
in the AITC detector lab. (LG1.12), with the Vacuum Jacket itself located in
the Pump Room (LG1.14), next door.

We have had the focal plane cold, at operating temperature (currently T=-100C)
and I am attempting to measure some formal system parameters, such as
System Gain, Noise, Dark Current, Light Leaks etc.

The two pictures below show the SkyMapper Vacuum Jacket in the pump room,
being cooled ready for detector characterisation.

The second image is a 'Flat-Field' unit we knocked together to do an
independent test of the system Gain. Normally we use a series of LEDs built
into the top of the instrument, above the window and outside the vacuum
environment, to perform this test (these LEDs are located just under the
dark slide seen on top of the instrument in the previous picture (with a
DVM sat on top of it).

As a check, this Flat-Field Unit was used to perform the same measurement as
the standard LED array. The FFU has a small LED at the top of the unit, behind
a Mylar diffusing screen and illuminates the 4 detectors in the focal plane,
which we image, 2 at-a-time.

1. 2.

Status, April 19th, 2007

Dual CCD Operation in Vacuum Jacket.

I am happy to report that since moving the Vacuum Jacket into the
AITC clean room (Room LG1.13), we have been able to progress with
the next stage in the CCD detector work.

This work is being undertaken in stages, and as can be seen below, we
have already proofed out operating 2 CCDs from our controller at any one
time in the 2 independent RSAA Test Dewars. These had a single E2V CCD
mounted and interfaced to the ARC controller, being read-out through a single
amplifier each and the image combined to make it look like a single, 4kx4k
'Super-CCD'.

This was undertaken to allow us to migrate the arrangement to the SkyMapper
Vacuum Jacket and install the 4 E2V Engineering CCDs and repeat the process,
this time at room temperature (see images below) and at operating temperature.

This work will give us a first indication that the internal arrangement of
thermal paths and electronics permit CCD operation in a low-noise environment.

To facilitate this we are able to select any 2 devices from the 4 available,
these now located in the focal plane, resulting in a total combination
of any 2 devices from 6; each pair of which may therefore be formed into
a single, 4kx4k, 'Super-CCD'

Pictures of the Engineering CCD Installation.

1. 4 Flex Connectors being tested for signal levels.
2. 1st CCD being raised into Focal plane.
3. 2 CCDs in Focal Plane
4. 4 Flexes attached to 4 Engineering CCDs.
5. External Connectivity for 2 CCDs, other 2 grounded
6. 4 CCDs in Focal Plane, ready for Room Temperature Imaging

1. 2.

3. 4.

5. 6.

Detector Lab. Pictures.

A full set of pictures showing the SkyMapper Vacuum Jacket being operated in the
AITC Detector Lab. (Room LG1.12 for the un-initiated!) at Room Temperature,
can be found *1.here*

Clean Room Installation Pictures.

The full set of pictures relating to the installation & integration of the
Eng. CCDs into the SkyMapper Vacuum Jacket can be found *2.here*

The Engineering CCDs - Room Temperature Data

1. Engineering #1 & #2 - 'Super CCD' JPG image
Eng#1 & #2 Room Temp Fits data

2. Engineering #1 & #3 - 'Super CCD' JPG image
Eng#1 & #3 Room Temp Fits data

3. Engineering #1 & #4 - 'Super CCD' JPG image
Eng#1 & #4 Room Temp Fits data

4. Engineering #2 & #3 - 'Super CCD' JPG image
Eng#2 & #3 Room Temp Fits data

5. Engineering #2 & #4 - 'Super CCD' JPG image
Eng#2 & #4 Room Temp Fits data

6. Engineering #3 & #4 - 'Super CCD' JPG image
Eng#3 & #4 Room Temp Fits data

Status, April 10th, 2007

STARGRASP Read noise and dark frame data.

The latest run with the E2V Mech.Sample CCD#1 running on the SAC was used
to obtain the following data from the CCD at an operating temperature of T=-110C

ARC Controller

1+1us sampling, system gain 0.97e/adu & 0.97e/adu.
7e & 4.5e rms for the left & right-hand amplifiers respectively.
Dark Current = 3.4e/pix/hour.
Read out time, ~40 secs, single port.
SAC Controller
1x4 us signal sampling, system gain 0.84e & 1.14e/adu
24e & 6.5e rms for the Right amplifier only.
Dark Current = 5.7e/pix/hour.
Read out time, ~20secs, single port.

Status, April 2nd, 2007

STARGRASP Imaging E2V CCD Test Pattern Data.

An Image of our standard Test Pattern frame has just been obtained with
the SAC and our Test Dewar#2.
The CCD is almost at operating Temp. (T=-110C)
and the 5 sec Test Pattern frame was obtained by the following sequence of
commands on the controllers specified.
(a bit of a hotch potch of a way to do it - but it works!)

clean ; clean ; clean (on SAC)
Open Shutter (on ARC)
Wait 5 seconds
Close Shutter (on ARC)
readout (on SAC)
transfer data frame to host (SAC to Sun)

where 'clean' clears out the ccd of residual charge and
'readout' reads the charge from the detector into memory.

The data fame below show a 5s Test Pattern frame at T=-110C taken by
the SAC controller using our Test Dewar#2 and the ARC controller to
servo temperature and operate the shutter.

Test Pattern data JPG image
T=-110C Test Pattern Fits data

The following pictures, again show the experimental arrangement.

1. 2.

3. 4.

STARGRASP Imaging E2V CCD in RSAA Test Dewar at RT.

A Room Temp Read-out was achieved on Friday from our E2V
Mech. Sample #2 CCD mounted in our own RSAA Test Dewar
(#2, the dewar normally used for WiFeS CCD evaluation).

This now means we can proceed to a cold test and evaluate the system
noise based on a dewar and wiring of our own building. This will be
the first step in us confirming we can operate the SAC with a known
low-noise system and achieve low read-noise and other science performance
parameters which are demanded by the SkyMapper Science goals.

The Templeton dewar, which belongs to the University of Hawaii (UofH)
can now therefore be returned to them, where further development work
will proceed to finish off the final design of the hardware.

The following pictures show the experimental arrangement.

1. 2.

3. 4.

The data fames below show room temperature & T=0C read-out of our
E2V MS#2 CCD, mounted in the RSAA Test Dewar#2 and operating on
the STARGRASP Array Controller (SAC).

Click Image for JPG data...

1.T=RT Fits data 2.T=0C Fits data

The data below shows a room temperature read-out of our
E2V MS#2 CCD, mounted in the RSAA Test Dewar#2 and
operating on Bob Leach's CCD Controller (ARC)

MS#2, O/P amp(L) on ARC, JPG image of RT readout.

MS#2, O/P amp(L), on ARC, RT readout, Fits data. (17Mby).

Status, March 22nd, 2007

ARC Controller Imaging 2 E2V CCDs at RT.

A Room Temp Read-out, from 2 independent E2V CCDs, each one
mounted in a separate Test dewar, has just been achieved in
the detector lab.

The detectors are interfaced via their flex cables, hanging outside the front
of the dewar in a 'heath-robinson' arrangement, through a custom controller
cable, constructed by Bernie Keys, to a dual channel ARC controller.

The following pictures show the experimental arrangement.

1. 2. 3.

4. 5. 6.

Two RSAA Test Dewars used in experimental setup to perform Dual CCD R/O

The ARC controller, in this mode, is acting as if there was a single CCD,
split R/O in progress. The images are at Room Temp., but identifiable
features can be seen on both the Eng#1 CCD on the Left amp (TD#1) and
the Mech. Sample#2 CCD on the Right amp. (TD#2).

The final image shows the Room Temperture readout which, at 0.5us signal
sampling, takes 30secs to readout the pseudo 4kx4k pixel detector!!

Eng#1, O/P amp(L) & MS#2, O/P amp(R) JPG image of RT readout.

Eng#1, O/P amp(L) & MS#2, O/P amp(R), RT readout, Fits data. (33Mby).

Another Significant Milestone.

So we now have a viable means of testing the vacuum jacket detector
connectivity - albeit only through 2 channels at a time, at present.
We will however get sensible results at Room Temperature, if the devices
are in the dark, and this will hence mean cutting an awful lot of time
out of doing the work if we had to cold cycle the system each time
we re-configured the CCDs for evaluation purposes.

Status, March 14th, 2007

STARGRASP Array Controller Dark Frame & Noise.

I now have some noise measurements from SAC and a 1537 second
dark frame.
The dark frame data was obtained at a temperature of T=-128C and the
temperature was measured by temporarily switching back to the ARC controller!

Measured Dark Current for this device during the 1537s exposure

System Gain = 1.046e/adu
Mean Dark Signal = 137 adu
Mean Bias in Y-overscan = 135 adu
Net dark signal ~2adu in 1537s => 5e/pix/hour
This is a little high, but there IS evidence that the dewar is leaking
light, maybe through the glue used to hold the Beryllium window in place.

Mech. Sample CCD#1Output Amp(L) JPG image of 1537s Dark frame.

Output (L) R/O Dark frame Fits data at T=-128C (17Mby).

Amplifier Noise measurements taken for Mech. Sample CCD#1

Data was taken for both the ARC and SAC controllers.

ARC Controller

1us sampling (system gain 0.85e/adu & 0.925e/adu)
4.2e & 4.7e rms for the left & right-hand amplifiers respectively.
SAC Controller
1+1us signal sampling (system gain 1.05e/adu)
5.6e rms for the Right amplifier only.

Status, March 13th, 2007

STARGRASP Array Controller Imaging CCD Cold.

I have the detector going cold now and have just taken the LED
pre-flash image shown in the link below.

Although I am not sure yet how to tell what the Temperature is, it's still
quite a bit away from operating temperature.

LED Pre-flash Data

I already have a pre-flash frame using the internal dewar LEDs in Templeton,
from our own test system, the data from which is available below. This data
was taken when the Templeton Dewar was here last August and was operating
on our ARC Controller based Test system. See the link:-
*Templeton/ARC PF frame*

And here is the current data - taken 10 minutes ago at a Temp=-85C (measured
using the ARC controller).

Mech. Sample CCD#1Output Amp(L) JPG image of Pre-flash at T=-85C.

Output (L) R/O Pre-flash FITS data at T=-85C (17Mby).

And a 2nd frame at T=-100C.
The defects - seen in the Frame taken last August at T=-120C, on our own
Test system, can now be clearly seen.

Mech. Sample CCD#1Output Amp(L) JPG image of Pre-flash at T=-100C.

Output (L) R/O Pre-flash FITS data at T=-100C (17Mby).

CDD Bias Frame Data

And here, a Bias frame, taken at a temperature of -90C. The split in the serial
register can be clearly seen, as again, can the X under and over-scans.

Mech. Sample CCD#1Output Amp(L) JPG image of Intermediate Temp. Bias frame.

Output (L) R/O Intermediate Temp. Bias frame FITS data (17Mby).

STARGRASP Array Controller Imaging CCD.

I have just taken the first two images at Room Temperature from the
new STARGRASP Array Controller (SAC).

The 2 CCD images I have look nominal - you can see the X-under and
X-overscans clearly and there is a lot of dark signal on the frame -
as expected. This data can be compared with similar images from our
own Test System, lower down these pages..

Mech. Sample CCD#1Output Amp(L) JPG image of Room Temp Bias.

Output (L) R/O Room Temp Bias frame FITS data (17Mby).

As can be seen - this image is assymetric - the frame size having been set to
dx=2200 by dy=4400. This is slightly different to the one I use here, which is
dx=2148 by dy=4200. The assymetry results from the serial shift register
being over-scanned by 2200-2148 = 52, which are hence virtual pixels. In
similar fashion to the vertical over scan which has 4400-4096 = 304 virtual
pixels (but not seen here due to the elevated dark current arising from the
Temperature at which this frame was obtained.

The Templeton Dewar was pumped last night and today I will obtain some images
with the detector cold. As the Beryllium window is still fitted - no external
imaging is possible but I plan to remove this tomorrow and replace with the
standard glass window so that I can image some test pattern data etc...

Status, March 8th, 2007

STARGRASP Array Controller now at RSAA.

The University of Hawaii's STARGRASP Array Controller (SAC) was delivered to
RSAA on Monday March 4th. See
STARGRASP Page
for details of this new generation of Array Controller.

This hardware is to drive the 32 CCD Focal Plane Array in SkyMapper and we
have now taken full delivery of the complete system.

The heart of this system consists of 3, 16-channel CCD (DACQ/PA/FPGA) CCD
read-out boards, 2 of which will be used as the Array Controllers for the
32 E2V CCD Mosaic for SkyMapper.
The other 16-channel system will remain at RSAA as a Test system and hot spare.

The following pictures show the SAC hardware on one of the benches in the
detector lab.

1. STARGRASP - General View of Kit

2. STARGRASP- Network Switch & Templeton Dewar

3. STARGRASP - Pixel Server and the Dual Agilent PSUs

4. STARGRASP- Compaq Laptop, SAC (side) & Templeton showing Beryllium Window

Looking clockwise from the front right side-

the large (& heavy) dual agilent power supplies (for the Test System only)
A linux based 'pixel server' (281.5 million pixels - to be read out ~20s)
Compaq Notebook to run the Test system 'demo version'
Chassis containing the 3, 16 channel DACQ/PA and FPGA boards.
A Dell PowerConnect, 24-channel Ethernet Switch
A fibre to Ethernet media converter
at the rear - the UofH "Templeton' CCD dewar, housing one of our Operating Mech. Sample CCDs

I have today (8th), finally got the Australain version of DC supplies and AC
cabling to be able to power up the system in 'demo' mode. This mode is to allow us
to operate the Templeton Dewar (see details of this, lower down the page) with
one of our Mechanical Sample CCDs.

I should be able to get, at least, room temperature images form the detector by
either late this afternoon or early next week.

Work with this new system at the moment revolves round (carefully) following the
'idiots' guide to cabling and powering up the hardware. It is hoped that either
today (Thurs) or early next week, a room temperature image can be obtained from
the Mechanical Sample E2V CCD housed in the Templeton Dewar.

This dewar, sent to us from the UofH last August, was equipped with MS E2V CCD and
operated on our ARC Test System here before being returned to the UofH, for them
to integrate the dewar into the new STARGRASP system.

The next step after this will be to operate the Templeton dewar cold, so that some
noise measurements can be taken at our operating temperature (now 170K). We will
then interface the SAC to one of our SkyMapper Test dewar, so that the Templeton
dewar, minus our CCD and assembly, can be returned to the UofH.

Some additional steps will be needed to get us from this position, to operating
the whole 32 CCD focal plane, namely, the interface hardware from the SAC to the
Focal Plane. This work is in an advanced stage of development, but will need testing.
The initial testing phase will be undertaken with 2 of the Eng CCDs, on the bench,
and then mounted in the vacuum jacket, with our 2-channel ARC controller. This will
be followed, it is hoped, by a similar configuration with the SAC hardware.

If all this goes well, we will then be on track for interface of the SAC to the
formal 32 CCD SkyMapper Focal plane and the real fun will begin!.

Status, December 5th, 2006

Summary of Science CCD Work to date.

The set of links below are pointers to the relevant parts of this page which
report the characterisation data for each of the 4 science CCDs tested and the
order in which they were characterised, the most recent at the end of the list.

1 - 12.04484-10-02, Batch #4 *CCD#12*
2 - 23.05191-02-01, Batch #1 *CCD#23*
3 - 21.05163-16-02, Batch #3 *CCD#21*
4 - 25.05255-02-02, Batch #5 *CCD#25*
Other links to important material in this page:-
E2V Detector Directory: *E2V Test Data*
E2V Detector Spectral Response Curves: *E2V QE Data*
STARGRASP: Controller images E2V CCD, *Stargrasp Image*
E2V CCD imaging in Templeton Dewar at RSAA: *ARC Templeton Images*
Re-furbishment of Templeton dewar pics: *Templeton Pics*

Status, December 5th, 2006
Characterisation Data for 4th E2V Science CCD
(05255-02-02, BATCH#5)

This is the last of the 4 SkyMapper CCDs selected for characterisation
and this work was completed today (Tuesday 5th Dec).
and therefore means the Science CCD quantification is now complete.

There are however one or 2 more measurements I would like to make,
which can be scheduled at any time - with one of the Engineering CCDs.

I would like to obtain some indication of what the red fringing is like -
this should be low, 0.1-0.2%.
I would also like to assess the H & VCTE from the frames I have available,
for one or 2 of the devices already tested.

Data for the SkyMapper Science CCD#25 was obtained this week,
along with a complete spectral response measurement.

This data consists of a measure of the 2 amplifier noises, a long exposure dark frame,
a QE curve and a pre-flash frame showing some pixel defects, and a standard test pattern
frame.

In addition I have also archived Test Pattern, Pin-Hole and Preflash data for:-

Binned - 1x2, 2x1 & 2x2. Left & Right amps
Windowed dx=1500xdy=1500 @ x=1074, y=2000. Left & Right amps

This data has also been archived for the other 2 Science CCDs,
described below on these pages.

All detector data for SkyMapper & WiFeS,
all 3 SII CCD systems - Imager, DBS Blue and Red,
& and all other miscellaneous systems (WFI, Tek)
is available on one of the system shares:-
/priv/samba2/ccd_data/

This device has ~500 dark pixel defects (reported in E2V's test data),
and this time are located throughout the array.

The device specs are again within the contract defect spec. for dark pixels.

This device was selected as it was the only part delivered form BAtch #5. Again this device exhibits characteristics either meeting or exceeding the contract
spec. we had in place with E2V for all the science CCDs.

10,000s Dark Exposure.

An 10000s dark frame was obtained at the the SkyMapper set point operating
temperature of T=-120C for the Science CCD #25.

The dark frame shows a wealth of CREs (Cosmic Ray Events) and, as usual, some of
them have very long tails.

The frame was taken with the blanking cap on the front of the Test dewar (#1).
This ensures that no extraneous light can enter the window and affect
the measurement of the dark current. This frame was taken using output amplifier
B (op(R))

Measured Dark Current for this device during the 10,000s exposure

System Gain = 0.88e/adu
Mean Dark Signal = 1667.39 adu
Mean Bias in Y-overscan = 1666.13 adu
Net dark signal ~1.26adu in 10,000s => 0.4e/pix/hour

Sci. CCD #25 Ouput Amp(R) JPG image of 10,000s dark frame.

Output (R) R/O 10,000s Dark frame FITS data (17Mby).

Artefacts seen in dark exposures and Flat Field data for this CCD

This device has no column defects and only ~500 dark pixel defects which this time
are spread over the whole image area.

Again all these characterstics are well within the contract spec. for these devices.

10ms LED preflash frame showing CCD pixel defects.

Sci. CCD #25Right-hand R/O JPG image of 10ms LED Pre-flash frame.

Right-hand R/O 10ms LED Pre-flash frame FITS data (17Mby).

1s Test Pattern exposure on Test Box.

The image below shows a standard test pattern exposure from the Science CCD,
mounted on the Test System. Vignetting, due to the shutter can be clearly seen.

Sci. CCD #25 Right-hand R/O JPG image of 1s Test Pattern image.

Right-Hand R/O 1s Test Pattern image FITS data (17Mby).

Spectral Response (QE) data taken for Science CCD #25

The curve below illustrates the spectral response for the Science CCD #25,
measured on the Test system on the 4th of December.

As can be seen, the response again appears very good right across the optical band
from 350nm out to 1050nm. There do however appear to be 3 discrepancies between
the RSAA measuremts and E2V's at 350nm, 500nm and 650nm.

Spectral Response of E2V Science CCD#25.

Amplifier Noise measurements taken for Science CCD #25

The following data confirms the read noise meets the noise spec. for the
SkyMapper Science requirements and is similar for both output amplifiers.
Data was taken for 1us and 2us signal sampling with what is estimated to be
~1e rms system noise in these figures.

1us sampling (system gain 0.92e/adu & 0.88e/adu)
3.75e & 4.8e rms for the left & right-hand amplifiers respectively.
2us sampling (system gain 0.464e/adu & 0.46e/adu)
3e & 3.4e rms for the left & right-hand amplifiers respectively.

There is also ~1e rms of system noise included in these figures,
so the devices are performing well and to spec. at the read-out rates used.

Status, November 28th, 2006

Characterisation Data for 3rd E2V Science CCD
(05163-16-02, BATCH#3)

Data for the SkyMapper Science CCD#21 was obtained last week,
followed this week by a complete spectral response measurement.

This data consists of a measure of the 2 amplifier noises, a long exposure dark frame,
a QE curve and a pre-flash frame showing some pixel defects, and a standard test pattern
frame.

In addition I have also archived Test Pattern, Pin-Hole and Preflash data for:-

Binned - 1x2, 2x1 & 2x2. Left & Right amps
Windowed dx=1500xdy=1500 @ x=1074, y=2000. Left & Right amps

This data has also been archived for the other 2 Science CCDs,
described below on these pages.

For this device many (~600) of the 1376 dark pixel defects (reported in E2V's test
data, are located in one position - see the low level preflash frame and a magnified
image of the area below.

The device specs are however within the contract defect spec. for dark pixels and it
will be up to us to decide where, in the focal plane, this device will reside.

This device was selected as it exhibited this large (relatively speaking,
of course) number of defects amongst the 32 science devices we have now received,
the E2V Contract now being essentially complete.
Again this device exhibits characteristics either meeting or exceeding the contract
spec. we had in place with E2V for all the science CCDs.

It is now planned to continue the work and characterise 1 more device from the last
batch (#5). I intend to complete this in the next 2 weeks so all the characterisation
will be complete before Xmas.
The Batch numbers of the CCDs are indicated in the small 'colour' table below and
shown in the CCD directory spreadsheet below.

E2V Test Data for all E2V SkyMapper CCDs

& RSAA Test Data for devices #12, #23 & #21

The link *E2V Test Data* references the *latest* E2V Detector Directory -
a table listing all E2V data and the sampled RSAA data.

8,000s Dark Exposure.

An 8000s dark frame was obtained at the the SkyMapper set point operating
temperature of T=-120C for the Science CCD #21.

The dark frame shows a wealth of CREs (Cosmic Ray Events) and, as usual, some of
them have very long tails.

Measured Dark Current for this device during the 8,000s exposure

System Gain = 0.9 e/adu
Mean Dark Signal=1455.7 adu
Mean Bias in Y-overscan =1454.4 adu
Net dark signal ~1.3adu in 8,000s => 0.53e/pix/hour

Sci. CCD #21 Ouput Amp(L) JPG image of 8,000s dark frame.

Output (L) R/O 8,000s Dark frame FITS data (17Mby).

Artefacts seen in dark exposures and Flat Field data for this CCD

This device has no column defects but a multitude (~600) dark pixel defects over
on the left hand side of the image.

All these characterstics are well within the contract spec. for these devices.

10ms LED preflash frame showing CCD pixel defects.

Sci. CCD #21Left-hand R/O JPG image of 10ms LED Pre-flash frame.

Left-hand R/O 10ms LED Pre-flash frame FITS data (17Mby).

This is a magnified part of the FITS image showing the cluster of ~600 dark pixels
over on the left-hand side of the image.

Sci. CCD #21Split R/O - part of the image.
Image magnified to show the dark pixel cluster on the left hand side better.

1s Test Pattern exposure on Test Box.

The image below shows a standard test pattern exposure from the Science CCD,
mounted on the Test System. Vignetting, due to the shutter can be clearly seen.

Sci. CCD #21 Split R/O JPG image of 1s Test Pattern image.

Left-Hand R/O 1s Test Pattern image FITS data (17Mby).

Spectral Response (QE) data taken for Science CCD #21

The curve below illustrates the spectral response for the Science CCD #21,
measured on the Test system yesterday (Nov. 27th)

As can be seen, the response again appears very good right across the optical band
from 350nm out to 1050nm.

Spectral Response of E2V Science CCD#21.

Amplifier Noise measurements taken for Science CCD #21

1us sampling (system gain 0.975e/adu & 0.88e/adu)
5.9e & 4.2e rms for the left & right-hand amplifiers respectively.
2us sampling (system gain 0.474e/adu & 0.45e/adu)
4e & 3.6e rms for the left & right-hand amplifiers respectively.

There is also ~1e rms of system noise included in these figures,
so the devices are performing well and to spec. at the read-out rates used.

Status, November 14th, 2006
SkyMapper E2V Science Device #21 installed in Test System
(05163-16-02, BATCH#3)

All data reported by E2V, along with a complete QE data set
for the 32 science devices is now at the link below.

Today I have removed device #23 from the Test Dewar and installed the 3rd of
the E2V SkyMapper Science CCDs, #21. 05163-16-02, a device from
BATCH #3.

This was selected as
(1) it is a device from a dfferent batch to the 1st 2 tested,
(2) it has a high mid-band QE and
(3) it has the highest number of single dark pixel defects.

Status, November 13th, 2006
Final E2V CCD Delivery #10,
Final 6 Science CCDs arrive at RSAA.

I have just received the 10th delivery from E2V - 6 devices and this marks
the end of the contract with E2V for Science CCDs for SkyMapper.

The data from E2V for all the suite of devices is
in the accompanying spread sheet.

E2V and RSAA Test Data for all E2V SkyMapper CCDs

The link *E2V Test Data* references the *latest* E2V Detector Directory -
a table listing all the devices for the SkyMapper Focal Plane.

Spectral Response Measurements for all 32 E2V SkyMapper CCDs

The following plot illustrates the Spectral response of all the Science CCDs,
shown together for comaprison.

Spectral Response of all SkyMapper CCDs.

Status, November 8th, 2006
E2V CCD Delivery #9, 05163-07-02
Last but one batch, 1 Science CCD arrives at RSAA.

I have just received the 9th delivery from E2V - a single Science CCD.
The data from E2V for this device and both science devices so far characterised
at RSAA are in the accompanying spread sheet.

Again another very commendable device from E2V

E2V and RSAA Test Data for all E2V SkyMapper CCDs

The link *E2V Test Data* references the *latest* E2V Detector Directory -
a table listing all the devices we have to date, i.e. data
for the 26 science devices, and for all 4 of the Engineering
CCDs, the latter data having been obtained on the RSAA Test System.

Spectral Response Measurements for all current E2V SkyMapper CCDs

The following plot illustrates the Spectral response of all 26 Science CCDs,
shown together for comaprison.

Spectral Response of all 26 E2V CCDs currently(08/11/2006) at RSAA.

Status, November 2nd, 2006

Characterisation Data for 2nd E2V Science CCD
(05191-02-01, BATCH#1)

The first set of data for the SkyMapper Science CCD#23 has been obtained this week.

This data consists of a measure of the amplifier noise, some long exposure dark frames,
a QE curve and a pre-flash frame showing some pixel defects.

This device was selected as it exhibited the next worse (relatively speaking, of course)
defects amongst the 25 science devices we have received to date.
This device exhibits characteristics either meeting or exceeding the contract spec. we have
in place with E2V for all the science CCDs.

It is now planned to continue the work and characterise 1 device from each of the
4 batches we have to date. I have colour coded these - for my own convenience only
as follows:-

BATCH#1
BATCH#2 - none
BATCH#3
BATCH#4
BATCH#5

10,000s Dark Exposure.

An 10000s dark frame was obtained at the the SkyMapper set point operating
temperature of T=-120C for the Science CCD #23.

The dark frame shows a wealth of CREs (Cosmic Ray Events) and, as usual, some of
them have very long tails. As was mentioned before, the long-tail events are
due to the deeper depleted silicon material used for the SkyMapper E2V CCDs.

Due to the increased size of the depleted region, this material is able to trap more red
photons and hence provide a higher red QE.
The down-side is the presence of many long-tailed CRE events.

The frame below is a 10000s dark frame taken with the blanking cap on the front of the
Test dewar (#1). This ensures that no extraneous light can enter the window and affect
the measurement of the dark current.

This frame was taken using output amplifier A (op(L)) and so there is no bias shift which
occurs when using split serial mode.

Measured Dark Current for this device during the 10,000s exposure

System Gain = 0.88 e/adu
Mean Dark Signal=1412.6 adu
Mean Bias in Y-overscan =1411.0 adu
Net dark signal ~1.6adu in 10,000s => 0.5e/pix/hour

Sci. CCD #23 Ouput Amp(L) JPG image of 10,000s dark frame.

Output (L) R/O 10,000s Dark frame FITS data (17Mby).

Artefacts seen in dark exposures and Flat Field data for this CCD

This device has no column defects and there are 2 pixel traps.
In addition there are 702 white and 255 dark pixel defects.

All these characterstics are well within the contract spec. for these devices.

The pre-flash frames are taken by utilising the internal dewar LEDs, these LEDs can be
pulsed after clearing the detector and just prior to read-out. They therefore provide
a means of calibrating the detector and can also be used to provide signal to inspect
any defects which may be present on the detector. The illumination in this instance,
ought really to be flat, but for calibration purposes it is _more_ useful to have a
non-uniform illumination.

10ms LED preflash frame showing CCD artefacts.

Sci. CCD #23Left-hand R/O JPG image of 10ms LED Pre-flash frame.

Left-hand R/O 10ms LED Pre-flash frame FITS data (17Mby).

3s Test Pattern exposure on Test Box.

The image below shows a standard test pattern exposure from the Science CCD,
mounted on the Test System. Vignetting, due to the shutter can be clearly seen.

Sci. CCD #23 Left-Hand R/O JPG image of 2s Test Pattern image.

Left-Hand R/O 2s Test Pattern image FITS data (17Mby).

Spectral Response (QE) data taken for Science CCD #23

The curve below illustrates the spectral response for the Science CCD #23,
measured on the Test system yesterday (Nov. 1st)

As can be seen, the response again appears very good right across the optical band
from 350nm out to 1050nm and approaches 100% QE in mid-band.

Spectral Response of E2V Science CCD#23.

Amplifier Noise measurements taken for Science CCD #23

1us sampling (system gain 0.88e/adu & 0.804e/adu)
4.9e & 3.68e rms for the left & right-hand amplifiers respectively.
2us sampling (system gain 0.44e/adu & 0.435e/adu)
3.3e & 3e rms for the left & right-hand amplifiers respectively.

There is also ~1e rms of system noise included in these figures,
so the devices are performing well and to spec. at the read-out rates used.

Status, October 30th, 2006
SkyMapper E2V Science Device #23 installed in Test System

A week or so has been spent attempting to measure night sky emission lines to see if
I could detect fringing on the Mech. Sample#2 CCD. As these are deep depletion devices
we do not expect anywhere near the level of fringing, seen on undepleted devices. On
these devices the fringes can be as much as 30% in the Red. See the fringing picture for
the 2.3m Imager on the Imager web-page.

On deep depletion devices the fringing is expected to be only ~0.1-0.2%

This experiment was unsuccessful due to to some vagaries of the experimental setup.
I plan to repeat this later, possibly with a Science CCD, now I understand what was
going on.

So, today I have installed the 2nd of the E2V SkyMapper Science CCDs,
#23. 05191-02-01

This was selected as
(1) it is a device from a different batch to the 1st device tested
(2) It has the highest mid-band QE of all the devices we have to date
(3) it shows some cosmetic defects including 2 traps but no column defects.

At this stage it is therefore planned to test one device from each of the 4 batches
(1,3,4 & 5) we have available. The Batch number represents a 'wafer run' on the E2V
Fab. (fabrication) line and hence represesnts what might be considered different
families of devices which just may exhibit slightly differing characteristics.
Though from the E2V data - these differences look to be minimal.

Status, October 19th, 2006

STARGRASP controller Images E2V CCD

I am very happy to report that the Pan-STARRS (STARGRASP) controller,
developed and built by John Tonry and Peter Onaka at the University of
Hawaii, has just produced its first test images from the detector we supplied
to them in one of their own dewars.
The link
*Templton Test Dewar*
illustrates the work we did to enable John's group to interface our device
with the STARGRASP controllers. This has entailed the Pan-STARRS group
constructing a test version of the controller, fabricatiing the custom cables
and generating suitable read-out code for this device.
THIS IS A VERY SIGNIFICANT STEP FORWARD.
See
STARGRASP Page
for the first of 2 test images taken with the new controller and the E2V CCD.

JPG image of the 1st data from the STARGRASP controller

Mech. Sample#1 CCD & STARGRASP

A test pattern image just obtained on the STARGRASP controller.

Image data of a similar pattern from our own Test system

The image below shows a similar test pattern exposure from the
Engineering CCD#4, mounted on our own Test System.
Vignetting, due to the shutter can be clearly seen.

Eng. CCD #4 Right-hand R/O JPG image of 3s Test Pattern image.

Right-hand R/O 3s Test Pattern image FITS data (17Mby).

Status, October 11th, 2006
E2V CCD Delivery #8,
Latest batch of 3 Science CCDs arrive at RSAA.

I have just taken delivery of the 8th batch of Science CCDs for the
SkyMapper Focal plane. Again - some of the charactersitics are superb,
one device achieving almost 100% QE at 500nm!

Again another very commendable batch of devices from E2V

E2V and RSAA Test Data for all E2V SkyMapper CCDs

The link *E2V Test Data* references the *latest* E2V Detector Directory -
a table listing all the devices we have to date, i.e. data
for the 25 science devices, and for all 4 of the Engineering
CCDs, the latter data having been obtained on the RSAA Test System.

As can be seen, the data from the latest science devices shows some
very respectable peak and one quite high UV response.

Spectral Response Measurements for all current E2V SkyMapper CCDs

The following plot illustrates the Spectral response of all 25 Science CCDs,
shown together for comaprison. Data between the points has been interpolated
by the Excel spreadsheet program.

Spectral Response of all 25 E2V CCDs currently(11/10/2006) at RSAA.

2nd Science device characterisation to start next week

Science device #15, the next in line exhibiting the highest level of defects of the
science complement will be installed in the Test system next week.

The formal characterisation will then proceed and a futher 2 devices investigated.
Currently it is planned to look at 4 in total of the 32 science devices.

Status, October 3rd, 2006

Characterisation Data for first E2V Science CCD
(04484-10-02, BATCH#4)

The first set of data for the SkyMapper Science CCD#12 was obtained last week.

This data consists of a measure of the amplifier noise, some long exposure dark frames,
a QE curve and a pre-flash frame showing some pixel defects.

This device was selected as it exhibited the worse (relatively speaking, of course) defects
amongst the 22 science devices we have recived to date.
This device exhibits characteristics either meeting or exceeding the contract spec. we have
in place with E2V for all the science CCDs

10,000s Dark Exposure.

An 10000s dark frame was obtained at the the SkyMapper set point operating
temperature of T=-120C for the Science CCD #12.

Due to the increased size of the depleted region, this material is able to trap more red
photons and hence provide a higher red QE.
The down-side is the presence of many long-tailed CRE events.

The frame below is a 10,000s dark frame taken with the blanking cap on the front of the
Test dewar (#1). This ensures that no extraneous light can enter the window and affect
the measurement of the dark current.

This frame was taken using output amplifier B (op(R)) and so there is no bias shift which
occurs when using split serial mode.

Measured Dark Current for this device during the 10,000s exposure

System Gain = 0.9 e/adu
Mean Dark Signal=1769 adu
Mean Bias in Y-overscan =1765 adu
Net dark signal ~4 adu = 1.4adu/hour = 1.3e/pix in 3600s

Sci. CCD #12 Ouput Amp(R) JPG image of 10,000s dark frame.

Output (R) R/O 10,000s Dark frame FITS data (17Mby).

Artefacts seen in dark exposures and Flat Field data for this CCD

This device has a column defect which can be seen in the Dark frame.
There are also 3 traps - though I can only find 2! and about 4000 white pixels.
The column defect is affecting the CTE at this point to such an extent that there
is charge trailing into the Vertical overscan. These artefacts can also be seen in pre-flash or flat-field images.

All these characterstics are well within the contract spec. for these devices.

The pre-flash frames are taken by utilisoing the internal dewar LEDs, these LEDs can be
pulsed after clearing the detector and just prior to read-out. They therefore provide
a means of calibrating the detector and can also be used to provide signal to inspect
any defects which may be present on the detector. The illumination in this instance,
ought really to be flat, but for calibration purposes it is _more_ useful to have a
non-uniform illumination.

10ms LED preflash frame showing CCD artefacts.

Sci. CCD #12 Left-hand R/O JPG image of 10ms LED Pre-flash frame.

Left-hand R/O 10ms LED Pre-flash frame FITS data (17Mby).

3s Test Pattern exposure on Test Box.

The image below shows a standard test pattern exposure from the Science CCD,
mounted on the Test System. Vignetting, due to the shutter can be clearly seen.

Sci. CCD #12 Split R/O JPG image of 3s Test Pattern image.

Split R/O 3s Test Pattern image FITS data (17Mby).

Spectral Response (QE) data taken for Science CCD #12

The curve below illustrates the spectral response for the Science CCD #12,
just (Sept. 2nd) measured on the Test system.

As can be seen, the response again appears very good right across the optical band
from 350nm out to 1050nm. The data measured here appears consistent with the figures
by E2V in the data sheet for this device.

Spectral Response of E2V Science CCD#12.

E2V and RSAA Test Data for all E2V SkyMapper CCDs

The link *E2V Test Data* references the *latest* E2V Detector Directory
- a table listing all the devices we have to date, i.e. data for the 22 science
devices, and for All 4 Engineering CCDs, the latter data all obtained on
the RSAA Test System.

Amplifier Noise measurements taken for Science CCD #12

The following data confirms the read noise meets the noise spec. for the
SkyMapper Science requirements and is similar for both output amplifiers:-

4.9e rms for the left/right-hand amplifiers and

This data was taken with system gain of 0.9e/adu. There is also
~1e rms of system noise included in these figures, so the devices
are performing well at the read-out rate used, approx. 250kHz.

Spectral Response Measurements for all current E2V SkyMapper CCDs

The following plot illustrates the Spectral response of all 22 Science CCDs,
shown together for comparison. Data between the points has been interpolated
by the Excel spreadsheet program.

Spectral Response of all 22 E2V CCDs currently(03/10/2006) at RSAA.

Status, September 4th, 2006

First E2V Science CCD (04484-10-02, BATCH#4)
Installed in Test System

The Test Dewar for the SkyMapper E2V device characterisation has now be returned
to its standard mode of operating, after the Vacuum Jacket Flex tests over the
last 2 weeks.

Science device #12, Serial No. 04484-10-02, has been installed and this will be
formally characterised and the data compared with that we have from E2V, see-
*E2V Test Data*
for all the data so far received for the SkyMapper CCDS.

It is expected that formal characterisation will tak ~2 weeks per device, and we are
currently planning to look at a selection of 4 devices.

As this work is going on in tandem with the detector work for the WiFeS FI CCD, and
with the work required to operate 2 of the SkyMapper CCDs (Eng.#4 & Mech.Samp.#2)
in the Vacuum Jacket using our ARC Test Controller, in about 4 weeks, it is expected
that the elapsed time to perform all the characterisation work may well be of the
order of 3 months.

Noise measurements

This data has been obtained from the Eng. devices and the 1st Sci. CCD.
Timing data at the end of the tables is for a single port read,
these times may be halved for dual port read with the same noise spec.

A read rate of 450kps, total read time of 20s, can be obtained by some
tweaking of the clock timing - this is in progress..

Reset+Sample (us) Analogue Gain/Sample Time(us) Sys.Gain
(L/R)(e/adu) Read Noise
(L/R)(e) Time/Pixel (us) Read-Rate (Kp/s) Full-Frame R/O Time(s)
1+1 2/F 0.93/0.95 3.8/3.8 4.0 250 36

0.5+0.5 2/F 2.05/1.85 8/6.8 3.0 333 27

Engineering Phase of SkyMapper Detector work is Concluded

Status, August 31st, 2006

E2V Vacuum Jacket Flex Connectors - Status

We have just operated the E2V Engineering CCD #4 in the Test System with the
1st of the 32 Flexible PCB harnesses. These Flex's allow interconnectivity
between the CCD and the outside world via a multi-connector panel at each side
of the SkyMaper Vacuum Jacket.

Each detector has its own flex PCB - a tactics connector on one end to connect to
The CCD and a 'Samtec', 2x20 way part at the other end. This samtec connector
interfaces to a mating plug inside the vacuum jacket and the pins of this are
conveyed through the panel - which is vacuum tight, to the outside world.

It is here where we will have out interface hardware before finally connecting to
each of the 2 Pan-STARRS, 16-channel controllers.

A few pictures of the internal arrangment of the test setup, operating one of the
Flex PCB's is shown in the following images:-

E2V Eng. CCD#4 (1st 6 images) & Mech.Sample#2 (last one)
being operated with the VJ Flex

1. Flex looking towards Detector Tactics connector.

2. Samtec to ribbon inter-connect. Jury rigged!

3. View from side showing Flex and Samtec inter-connect and
securing assembly! CCD Installed

4. Close-up of above also showing one of the calibration LEDs.
CCD Installed

5. Lower part of flex - CCD installed.

6. Detail of flex, Samtec and 50-way interconnect.

7. Mechanical Sample CCD, protective cover removed and Flex arrangement.

Future Plans

It is planned to use a modified hermetic to flex connection to trial operating the 2 JFETs
on the detector as a means of switching the video signal as part of the development of the
new Focal Plane Interface Boards, FIBs!.

These boards will permit us to have indepenedent control of the 2 output stages of the
detectors and to switch between these - should the need ever arise.

Status, August 11th, 2006

A near miss sort of a time for the UK travelling public!!

See *News Story* for all the details

E2V Mech. Sample CCD#1 now fully operational
in the (UofH) Templeton Dewar

All the assembly and detector integration pictures can be found at the 2
links further down these pages **Here**

This device has now been operated at -120C in the Templeton Dewar.
Data obtained, some of it presented here, shows the detector is operating
nominally and compares very favourably with data from these devices
on our own Test System.

Data from the device has been obtained in terms of read-noise, Low level Pre-flash
& a 3600s dark frame in Output Left, Right and Split serial mode.

The Dewar and CCD will therefore be shipped back to UofH for integration with
the SkyMapper Pan-STARRS 16-channel Test controller, before coming back to
RSAA, hopefully in a few weeks in preparation for integration of the 2,
16-channel science controllers, into the 32-CCD SkyMapper focal plane.

Templeton Test Dewar & Mech Sample CCD (02393-10-02) data.

The data presented here was obtained from the Templeton Dewar yesterday(10th).
The 3 frames show a standard split-serial Bias frame, a 'low-level' LED
Pre-flash frame a 3600s dark frame. Results and Conclusion follow the
data below.

Split R/O Bias Frame JPG image.

Split R/O FITS data (17Mby).

Split R/O Pre-flash JPG image.

Split R/O FITS data (17Mby).

Split R/O Dark frame JPG image.

Split R/O FITS data (17Mby).

Results

Detector: Mechanical Sample #1, Serial # 02393-10-02
Operating Temperature = -115C
Dark Current = 1.5e/hour (measured from a 1000s dark frame - a 3600s
is nearly done). V.good
Noise = 3.7e (including 1e system noise). V.good
Charge transfer on the Left hand side - appalling, caused by real bad
column defects towards left-hand edge. Hence no use as a science device.
Clear evidence of this can be seen in the pre-flash frames but more
clearly in the long dark exposure - which shows long horizontal tails on
all the CREs on the left hand side of the image, matching the position
of the column defects seen in the pre-flash frame.
No other problems or read-out effects

Conclusion.

A good device with which to verify operational performance with
the Pan-STARRS test controller.

Operation of this device in this manner represents little risk to
our Project.

Status, August 8th, 2006
E2V CCD Delivery #7,
Latest batch of 4 Science CCDs arrive at RSAA.

E2V and RSAA Test Data for all E2V SkyMapper CCDs

The link *E2V Test Data* references the *latest* E2V Detector Directory -
a table listing all the devices we have to date, i.e. data
for the 22 science devices, and for all 4 of the Engineering
CCDs, the latter data having been obtained on the RSAA Test System.

As can be seen, the data from the latest science devices shows some
very respectable peak and one quite high UV response.

Spectral Response Measurements for all current E2V SkyMapper CCDs

The following plot illustrates the Spectral response of all 22 Science CCDs,
shown together for comaprison. Data between the points has been interpolated
by the Excel spreadsheet program.

Spectral Response of all 22 E2V CCDs currently(08/08/2006) at RSAA.

1st Science device to be characterised next week

Science device #12, the one exhibiting the highest level of defects of the
science complement will be installed in the Test system next week.

The formal characterisation will then proceed and futher devices investigated.
Currently it is planned to look at 4 - the 2nd being Science Device #14

E2V Mech. Sample CCD#1 now installed in the (UofH) Templeton Dewar

This dewar - supplied by the UofH for testing purposes will be operated
today with the E2V Mech. Sample#1 CCD, it is hoped to have a room temperature
R/O of this device by early this afternoon.

If this proves succesfull, the dewar will then be pumped and then cooled tomorrow
so that some data (noise, Low level Pre-flash, dark current) can be obtained in
this system before it is shipped back to UofH for integration with the SkyMapper
Pan-STARRS 16-channel Test controller.

This whole rig will then come back here for us to commence integration of the
2, 16-channel science controllers, into the 32-CCD SkyMapper focal plane.

This work has been undertaken so that we have a system which works with one of
our E2V SkyMapper detectors here, and which can then be made to operate with
the new controllers before coming here.

It is hoped in this way that we can get 'on the air' quickly with the formal
focal plane array and the new controllers.

Templeton Test Dewar Assembly Pictures - CCD Installation.

Details relating to the installation of the Mech. Sample CCD
(02393-10-02) into the UofH Dewar. Can be found
*here*

Templeton Test Dewar Assembly Pictures - Preparation.

Details relating to the present state of the UofH
Dewar we are assembling to provide a working
E2V Mech. Sample (02393-10-02) detector for the
Pann-STARRS group, may be found
*here*

Status, July 28th, 2006
Characterisation of the E2V Eng. CCDs Complete

All four of the E2V SkyMapper Engineering CCDs have now been
operated and characterised.

In addition, we also have 2 working Mechanical Sample CCDs - one of
which is shortly to be used in the 'Templeton' (UofH) Dewar and will
be operated here before being shipped to UofH for integration with the
Pan-STARRS Test Controller.

We therefore now have a suite of 18 Science CCDs (with 4 more due in
the next week), 4 fully functional Engineering CCDs and 2 Mechanical
Sample parts - which also image.

The next stage will see the first of the Science devices installed in
the Test System and it is currently planned that characterisation will
continue for 4 of these devices before bringing this part of the testing
and evaluation process to a close - around the end of September,
the move to the AITC permitting!

Status, July 26th, 2006
Characterisation of the E2V Eng. CCD #4(04484-09-02)

Dark Frame, Spectral Response (QE) and Noise data
taken for Eng. CCD #4 CCD

The link *E2V Test Data* references the *latest* E2V Detector Directory
- a table listing all the devices we have to date, i.e. data for the 18 science
devices, and for All 4 Engineering CCDs, the latter data all obtained on
the RSAA Test System.

8000s Dark Exposure.

An 8000s dark frame was obtained this week at the the SkyMapper set point operating
temperature of T=-120C for the Eng CCD #4.

The dark frame shows a wealth of CREs (Cosmic Ray Events) as usual some of which have
very long tails towards the lower half of the detector. This is believed to be due to the
inclination of the detector with the incoming particles. The CCD is at a shallow angle to
the vertical and was oriented North-South. As was mentioned before, the long-tail events are
due to the deeper depleted silicon material used for the SkyMapper E2V CCDs.

Due to the increased size of the depleted region, this material is able to trap more red
photons and hence provide a higher red QE.
The down-side is the presence of many long-tailed CRE events.

The frame below is an 8000s dark frame taken with the blanking cap on the front of the
Test dewar (#1). This ensures that no extraneous light can enter the window and affect
the measurement of the dark current.

This frame was taken using output amplifier A (op(L)) and so there is no bias shift which
occurs when using split serial mode.

Measured Dark Current for this device during the 8000s exposure

System Gain = 0.9 e/adu
Mean Dark Signal=1705 adu
Mean Bias in Y-overscan =1705 adu
Net dark signal ~ 0 adu = 0e/pix in 8000s!!

Eng. CCD #4 Ouput Amp(L) JPG image of 8000s dark frame.

Output (L) R/O 8000s Dark frame FITS data (17Mby).

Artefacts seen in dark exposures and Flat Field data for this CCD

Nothing obvious on this device, though a large cluster of pixel defects can bee seen
which are affecting a group of colums associated with the defects and are degrading the
charge transfer down the rest of the array in those colums. The defects are affecting the
CTE at this point to such an extant that there is charge trailing into the Vertical overscan. these can be seen in pre-flash or flat-field images.

As this is an Enginering device - these sorts of effects are expected. These artefacts
may however not prevent use of these devices in the focal plane if this is required!
The following image was taken using the calibration LEDs, which are mounted in all the
latest SII & Test dewars and are also being fitted to the periphery of the SkyMapper Vacuum
Jacket and to the 2 WiFeS CCD cameras.

These LEDs can be pulsed after clearing the detector and just prior to
read-out. They therefore provide a means of calibrating the detector and can also be
used to provide signal to inspect any defects which may be present on the detector.
The illumination in this instance, ought really to be flat, but for calibration purposes
it is _more_ useful to have a non-uniform illumination.

10ms LED preflash frame showing CCD artefacts.

Eng. CCD #4 Left-hand R/O JPG image of 10ms LED Pre-flash frame.

Left-hand R/O 10ms LED Pre-flash frame FITS data (17Mby).

3s Test Pattern exposure on Test Box.

The image below shows a standard test pattern exposure form the Engineering CCD,
mounted on the Test System. Vignetting, due to the shutter can be clearly seen.

Eng. CCD #4 Right-hand R/O JPG image of 3s Test Pattern image.

Right-hand R/O 3s Test Pattern image FITS data (17Mby).

10s Pin-Hole exposure on Test Box.

The image below shows a standard pin-hole exposure from the Engineering CCD,
mounted on the Test System. Only a small array of holes are available in the
centre of the image.

The pixel defects mentioned above can be seen to be affecting the charge transfer
in the columns in which they appear and are causing charge streaking down the frame
from the point of origin of the defects on the Left-hand side.

The change in Bias level can also be seen due to the Split Read mode method of
reading out the detector in this case

Eng. CCD #4 Split R/O JPG image of 10s Pin holes image.

Split R/O 10s Pin Holes image FITS data (17Mby).

Spectral Response (QE) data taken for Eng. CCD #4

The curve below illustrates the spectral response of the Engineering #4 CCD,
just (26/07/06) measured on the Test system.

As can be seen, the response now again appears very good right across the optical band
from 350nm out to 1050nm. The data measured here appears better than the nominal data
taken for the E2V Science devices. However as can be seem from the family of QE curves
shown below, there is a variation, particulalrly in the blue, of the spectral response
from device-to-device. This reflects variations in manufacturing. All devices currently
meet or exceed the contract spec. right across the optical band.

E2V do not routinely measure the QE of their Engineering CCD parts but all 4 of the
SkyMapper Engineering CCDs have been quantified and we are therefore now in a position
to determine (taking into account other characterisation data), if they are useful for
the focal plane, if the need arises.

Spectral Response of Eng #4 CCD.

Amplifier Noise measurements taken for Eng. CCD #4

The following data, just obtained with the E2V Eng#4 CCD (04484-09-02), confirms the
read noise meets the noise spec. for the SkyMapper Science requirements and is
similar for both output amplifiers:-

4.3e rms for the left-hand amplifier and
4.3e rms for the right-hand amplifier

This data was taken with system gain of 0.93e/adu. There is also
~1e rms of system noise included in these figures, so the devices
are performing well at the read-out rate used, approx. 250kHz.

Status, July 11-19th, 2006
Characterisation of the E2V Eng. CCD #3(04484-06-02)

Dark Frame & Spectral Response (QE) data taken for Eng. CCD #3 CCD

The link *E2V Test Data* references the *latest* E2V Detector Directory -
a table listing all the devices we have to date, i.e. data
for the 18 science devices, and for the First 3 Engineering
CCDs, the latter data obtained on the RSAA Test System.

Templeton Test Dewar Assembly Pictures.

Details relating to the present state of the UofH
Dewar we are assembling to provide a working
E2V Mech. Sample (02393-10-02) detector for the
Pann-STARRS group, may be found
*here*

3600s Dark Exposure.

A 1 hour dark frame has just been obtained at the the SkyMapper set point operating
temperature of T=-120C.

The dark frame shows a wealth of CREs (Cosmic Ray Events) as usual some of which have
very long tails and some show curved structure. This is due to the deeper depleted
silicon material used for the SkyMapper E2V CCDs.

Due to the increased size of the depleted region, this material is able to trap more
red photons and hence provide a higher red QE.
The down-side is the presence of many long-tailed CRE events.

The frame below is a 3600s dark frame taken with the blanking cap on the front of the
Test dewar (#1). This ensures that no extraneous light can enter the window and affect
the measurement of the dark current.

This frame was taken using output amplifier A (op(L)) and so there is no bias shift which
occurs when using split serial mode as in, for example, the pre-flash frame below.

Measured Dark Current for this device

System Gain = 0.98 e/adu
Dark Signal=1719.3 adu
Background =1718.5 adu
Net dark signal ~ 0.8 adu = 0.72e/pix in 1 hour

Eng. CCD #3 Ouput Amp(L) JPG image of 3600s dark frame.

Output (L) R/O 3600s Dark frame FITS data (17Mby).

Artefacts seen in dark exposures and Flat Field data for this CCD

Close inspection of the FITS data will reveal a structure in the dark current, centered
in the middle of the frame about two thirds the way down the image, at the top of the
detector in fact (this structure can also be seen on the high resolution JPG image).
It is expected that this is due to manufacturing faults which has caused an
elevation in the dark current in this area. The mean dark current in this triangular
shaped area is between
2-9e/pixel/hour,
compared to that found in the rest of the image area, see figure above.

There are also some features in the image related to structural artefacts in the AR coating
these can be seen as low-levl streaks, easily seen in pre-flash or flat-field images.

As this IS an Enginering device - these sorts of effects are expected. These artefacts
may however not prevent use of these devices in the focal plane if this is required!
The following image was taken using the calibration LEDs, which are mounted in all the
latest SII & Test dewars. The can be pulsed after clearing the detector and just prior to
read-out. They therefore provide a means of calibrating the detector and can also be
used to provide signal to inspect any defects which may be present on the detector.
The illumination in this instance, ought really to be flat, but for calibration purposes
it is _more_ useful to have a non-uniform illumination.

This frame also illustrates the region mentioned above where there is an elevated dark
current signal. The edges of this region can be seen in the frame as a pair of diverging
lines, about 3/5th the way down the frame and starting just to the left of the central
read-out split in the image. There are a group of pixel defects, up, just to the left
of the origin of the lines.

200ms LED preflash frame showing CCD artefacts.

Eng. CCD #3 Split R/O JPG image of 200ms LED Pre-flash frame.

Split R/O 300ms LED Pre-flash frame FITS data (17Mby).

Spectral Response (QE) data taken for Eng. CCD #3

The curve below illustrates the spectral response of the Engineering #3 CCD,
just (19/07/06) measured on the Test system.

E2V do not routinely measure the QE of their Engineering CCD parts but I plan to quantify
all four Engineering CCDs before moving on to a selection of the Science devices.

However, the Astro-B AR coating which was specified for the SkyMapper CCDs is deposited on
all our devices hence these measurements are a good indication of the spectral response
performance for the SkyMapper focal plane CCDs.

Spectral Response of Eng #3 CCD.

Amplifier Noise measurements taken for Eng. CCD #3

The following data, just obtained with the E2V Eng#3 CCD (04484-06-02), confirms the
read noise meets the noise spec. for the SkyMapper Science requirements and is
similar for both output amplifiers:-

4.2e rms for the left-hand amplifier and
3.9e rms for the right-hand amplifier

This data was taken with system gain of 0.98e/adu. There is also
~1e rms of system noise included in these figures, so the devices
are performing well at the read-out rate used, approx. 250kHz.

Vertical CTE (Charge Transfer Efficiency) measurements
for Eng. CCD #3

A fairly rough-and-ready CTE measurement has just been made on the Engineering
#3 CCD. This was undertaken using the extended pixel edge response technique which
relies on measuring the net signal (In) above bias (Nb) in the last
few rows of the image area and comparing this with the signal (In+1) in the
1st row in the vertical overscan after n, 4096 transfers into the serial shift
register. A Low level pre-flash frame was used to do the measurements.

After 4096 transfers, the number of pixels in the image area, the following signals
were measured:-

In = 1839.756 average signal in the last 10 rows in image area,
In+1 = 1715.682 average signal in the 1s row in vertical overscan,
Nb = 1715.508 average Bias signal measured in 4 places in vertical overscan.

CTE = 1 - In+1 / (In x n ) = 0.999999658

A comendable figure!

Status, July 3rd, 2006
E2V Engineering CCD #3(04484-06-02)
now in Test System

The 3rd of the SkyMapper Enginneering CCDs was loaded into the Test System yesterday -
Sunday July 2nd, having removed Eng#2 and stored this away in its container.

The Eng#3 CCD will be characterised over the next 2 weeks after which the final Engineering
device will be installed.

Status, June 27-29th, 2006
Additional Data from the E2V Eng. CCD #2(04484-18-01)

Dark Frame & Spectral Response (QE) data taken for Eng. CCD #2

A 1 hour dark frame was taken last night (26th) at the SkyMapper set point operating
temperature of T=-120C.

The dark frame shows a wealth of CREs (Cosmic Ray Events) some of which have very long
tails and some show curved structure. This is due to the deeper depleted silicon
material used for the SkyMapper E2V CCDs. Due to the increased size of the depleted region,
this material is able to trap red photons more efficiently and hence provide a higher red QE.
The down-side is the presence of many long-tailed CRE events.

This frame was taken using output amplifier A (op(L)) and so there is no bias shift which
occurs when using split serial mode.

Measured Dark Current for this device

System Gain = 0.9 e/adu
Dark Signal=2779.6 adu
Background =2780.0 adu
Net dark signal ~ 0 adu = 0e in 1 hour!

3600s Dark Exposure.

Eng. CCD #2 Ouput Amp(L) JPG image of 3600s dark frame.

Output (L) R/O 3600s Dark frame FITS data (17Mby).

200ms LED preflash frame showing CCD defects.

This frame shows relatively few pixel defects, though there are a cluster
towards the centre near the top of the array, the effects of which are to
take out the columns immediately 'down-stream' of the charge transfer.

This Engineering device may therefore be suitable as a spare for the focal
plane - other characteristics of the device being met.

Eng. CCD #2 Split R/O JPG image of 200ms LED Pre-flash frame.

Split R/O 200ms LED Pre-flash frame FITS data (17Mby).

Spectral Response (QE) data taken for Eng. CCD #2

The curve below illustrates the spectral response of the Engineering #2 CCD,
just (29/6/06) measured on the Test system.

As can be seen, the response now appears very good right across the optical band
from 350nm out to 1050nm. The data measured here appears better than the nominal data
taken for the E2V Science devices. However as can be seem from the family of QE curves
shown below, there is a variation, particulalrly in the blue, of the spectral response
from device-to-device. This reflects variations in manufacturing. All devices currently
meet or exceed the contract spec. right across the optical band.

E2V do not routinely measure the QE of their Engineering CCD parts but I plan to quantify
all four Engineering CCDs before moving on to a selection of the Science devices.

Spectral Response of Eng #2 CCD.

The following data, just obtained with the E2V Eng#2 CCD (04484-18-01), confirms the
read noise now meets the noise spec. for the SkyMapper Science requirements and is
similar for both output amplifiers:-

3.9e rms for the left-hand amplifier and
2.97e rms for the right-hand amplifier

Status, June 26th, 2006
E2V CCD Delivery #6,
Latest batch of 4 Science CCDs arrive at RSAA.

E2V and RSAA Test Data for all E2V SkyMapper CCDs

The link *E2V Test Data* references the *latest* E2V Detector Directory -
a table listing all the devices we have to date, i.e. data
for the 18 science devices, and for the First 2 Engineering
CCDs, which has been obtained on the RSAA Test System.

As can be seen, the response appears very high right across the optical band from
350nm out to 1050nm. The data measured here is again consistent with that taken by
E2V for the 1st of the Science devices delivered to RSAA some while ago.

E2V do not routinely measure the QE of their Engineering CCD parts but I plan to
quantify all four Engineering CCDs before moving on to a selection of the Science devices.

Spectral Response Measurements for all current E2V SkyMapper CCDs

The following plot illustrates the Spectral response of all 18 Science CCDs,
shown together for comaprison. Data between the points has been interpolated by
the Excel spreadsheet program.

Spectral Response of all 18 E2V CCDs currently(26/6/2006) at RSAA.

Status, May 15th, 2006
First Data from the E2V Eng. CCD #2(04484-18-01)

Dark Frame & Spectral Response (QE) data taken for Eng. CCD #2

The curve below illustrates the (preliminary) spectral response of the Engineering #2 CCD,
measured last week on the Test system.

As can be seen, the response appears very high right across the optical band from 350nm out to 1050nm.
The data measured here is again consistent with that taken by E2V for the 1st of the Science devices
delivered to RSAA some while ago.

E2V do not routinely measure the QE of their Engineering CCD parts but I plan to quantify all four
Engineering CCDs before moving on to a selection of the Science devices.

However, the Astro-B AR coating which was specified for the SkyMapper CCDs is deposited on all
our devices hence these measurements are a good indication of the spectral response performance
for the SkyMapper focal plane CCDs.

Spectral Response of Eng #2 CCD.

The following data, just obtained with the E2V Eng#2 CCD (04484-18-01), confirms the read noise
is now similar for both output amplifiers:-

4.5e rms for the left-hand amplifier and
4.7e rms for the right-hand amplifier

These are taken with system gains of 0.906 & 0.95e/adu. There is also
~1e rms of system noise included in these figures, so the devices
are performing well at the read-out rate used, approx. 250kHz.

A 1 hour dark frame will be taken this week to confirm the dark current and an inspection
made of the defects for this device

In the meantime the data below is a 1000s dark frame taken last week on the Test System
using ouput amplifier B (op(R)) and at a temperature of ~-105C

1000s Dark Exposure.

Eng. CCD #2 Ouput Amp(R) JPG image of 1000s dark frame.

Output (R) R/O 1000s Dark frame FITS data (17Mby).

Status, May 4th, 2006
E2V CCD Delivery #5 and
Eng. CCD#2(04484-18-01) in Test System

The 5th batch of 3 Science devices was delivered to RSAA yesterday.
This represents 3 out of the 4 scheduled for delivery at this time.
The fourth device will be delivered with the next (6th) batch of science
devices scheduled for delivery towards the end of June.

We now have a total of 14 science devices out of a total of 32.

These devices again exhibit superb performance and have all passed
provisional acceptance for the SkyMapper Focal Plane Array,
they continue the trend for a very good science performance
for the SkyMapper FPA.

E2V and RSAA Test Data for all E2V SkyMapper CCDs

The link *E2V Test Data* references the E2V Detector Directory - a table listing the
devices we have, i.e. data for the 14 science devices we now have, and the
data presented below for the First Engineering CCD, obtained on the RSAA Test System.

E2V and RSAA Spectral Response Measurements for all current E2V SkyMapper CCDs

The following plot illustrates the Spectral response of all 14 Science and the
first Engineering CCD, all shown together for comaprison. Data between the points
has been interpolated by the Excel spreadsheet programe.

Spectral Response of all E2V CCDs currently(4/5/2006) at RSAA.

E2V and RSAA Test Data for all E2V SkyMapper CCDs

The first Engineering CCD (04484-07-01) was removed from the Test Dewar (#1) on
Monday (1st May).

This dewar has just had a modification made to the internal PCB - a new Version 2
unit was installed and this was checked out yesterday and the dewar is in the
process of being re-assembled.

The 2nd of the Engineering CCDs (04484-18-01) will be installed this morning
(Thurs. 4th May) and a Room Temp. read-out performed to ensure its operability
before pumping the system ready for next week.

Status, April 26th, 2006
Spectral Response (QE) of Eng. CCD #1(04484-07-01)

The curve below illustrates the spectral response of the Engineering #1 CCD, just measured
on the Test system.

As can be seen, the response is very good right across the optical band from 350nm out to 1050nm.
The data measured here is consistent with that taken by E2V for the 1st of the Science devices
delivered to RSAA some while ago.

E2V do not routinely measure the QE of their Engineering CCD parts.

However, the Astro-B AR coating which was specified for the SkyMapper CCDs is deposited on all
our devices hence these measurements are a good indication of the spectral response performance
for the SkyMapper focal plane CCDs.

Spectral Response of Eng #1 CCD.

Status, April 24th, 2006
Eng. CCD #1(04484-07-01)
Long Exposure Dark Frame Data

The following data, just taken with the E2V Eng#1 CCD (04484-07-01), confirms the dark current
performance for these type of devices.

The 1 hour integration frame, taken at -120C in Split Serial Read mode shows

0.5e/pix/hour on the right-hand side and
0.3e/pix/hour on the left-hand side.

Both these figures hence easily meet our contract spec. for <1e/pix/hour for these devices.

You will now clearly see many column and pixel defects in the image, including the rather bad (hot)
pixel defect which causes signal injection and then charge trailing up the whole image area. It is
thought that it is this defect which is affecting the signal processing and resulting in what apears
to be a higher read-noise from one amplifier compared to the other.

The read-noise measurements have so far revealed a figure of 3.7e/pix rms from the right hand
amplifier and 8.4e rms from the left-hand amplifier. It is estimated that there is approx. 1e rms
of system noise in this figure - so it looks as if these devices are also meeting the read-noise
requirement of the contract spec. which was for <4e/pix rms at our standard read-out rate.

It is assumed that the annomolous read-out noise figure is being caused by the bad defect - this will
be investigated with the other Eng. devices to confirm that the output amplifiers do have comparable
performance, which is what is expected.

Finally - 3 spot Spectral Response (QE) measurements taken last week at 380, 400 and 450nm
have shown that the devices UV response to be very respectible at:-

380nm, 79% (E2V data, 62%),
400nm, 75% (E2V data, 74.7%),

450nm, 83% (E2V data, 78%).

It is planned to make some more measurements this week, before moving on to the 2nd Engineering
device next week.

The poor cosmetic quality of this device, probably makes it unsuitable for
use in SkyMapper Focal Plane

3600s Dark Exposure.

Eng. CCD #1 Split R/O JPG image of 3600s dark frame.

Split R/O 3600s Dark frame FITS data (17Mby).

The data below is a short pre-flash exposure which also illustrates some of the many bad
(dark) pixels in the image area. In this instance, these pixels are causing loss of signal
as the charge is moved up the array, and so through them, to the read-out ports.

LED Pre-flash Exposure.

Eng. CCD #1 Split R/O JPG image of short LED Pre-flash exposure.

Split R/O Pre-flash exposure FITS data (17Mby).

Status, April 12th, 2006
Eng. CCD #1 Test Pattern Data in all 3 Read Modes

The 3 data frames taken, shown below, illustrate the E2V Engineering CCD #1
(04484-07-01) operating on the Test Box with a test pattern exposure in all three
read-out modes.

You will see many column and pixel defects in the images - sharp vertical and horizontal
overscan regions and the same sharply defined pixel and column defects.
This implies that the serial and parallel transfer are working correctly and
hence this is a stable operating environment for the E2V 4482 detector.

The DSP code to drive this device is a 'cloned' version of that used to drive the
2.3m Imager and DBS CCDs. A modified version (again in terms of clock and bias
settings only) was also used to drive the Lincoln Labs CCID20 devices, during the
phase of WFI Mosaic testing and remedial action, undertaken last year.

Again the generic Imager DSP code has been changed, at present, only in terms of the
Clock and Bias settings applied to the detector. It is now planned to start tuning up this
code to enable the best detector performance to be achieved for the 4482 devices for
the SkyMapper focal plane.

Eng. CCD #1 Split R/O JPG image.

Split R/O FITS data (17Mby).

Eng. CCD #1 Output Left JPG image.

Output Left FITS data (17Mby).

Eng. CCD #1 Output Right JPG image.

Output Right FITS data (17Mby).

Status, April 6th, 2006 - Eng. CCD #1 (04484-07-01)
now correctly Imaging Data

After the data taken below (23/3/06) was inspected, it was found that in fact there was a
minor problem with the internal Dewar PCB which resulted in only half the data ever
being read-out, i.e. that through the left-hand output amplifier.

The last 10 days has seen vigorous attention paid to the DSP code but in the event the
fault above was the cause of the problem.

We now have a stable system for both SkyMapper nad WiFeS (see the WiFeS CCD pages for
details about the latest frames from the Fairchild CCD).

The image below has just been taken on the Test System. It shows a Room temperature R/O
of the Eng. CCD and now correctly shows the data from the Right-hand side of the frame.
This is a split serial R/O but the single port op(L) and op(R) show exactly the same data
- but without the slight difference in Bias level seen in the split read-out mode frame.

I am currently cooling the detector and may have the 1st preliminary read noise and a Blue
QE check early next week.

You will be able to see the horizontal x-under and over-scans at the left and right-hand edges
and the y (vertical) overscan along the top edge and some column defects which may 'freeze out'
as the Detector becomes cold. The gradient in the frame is produced by an increase in dark
current as the detector is read out - the read process taking about 12secs in split serial mode.

E2V SkyMapper Engineering CCD #1 JPG image.

E2V SkyMapper Engineering CCD #1 FITS data (17Mby).

Status, March 23rd, 2006 - Mech Sample CCDs
(02393-10-02 & 05094-01-01)
in Dummy Focal PLane

The images below show the 2 Mechanical Sample CCDs mounted on the Dummy focal plane plate
which we have had manufactured to test the alignemnt of the CCDs in both X & Y

This is to ensure that the process of installing each device in turn does not compromise
the safety of the device's immediate neighbours.
The draw bar mechanism for lifting these devices into the focal plane, and the locating pin
ensure that when the CCDs are drawn up into the focal plane, it is not possible to hit or
touch any of the neighbouring devices as this operation is performed.
The whole focal plane may therefore be assembled with no risk to neighbouring devices as each
device in turn is loaded into the Focal Plane.

This operation will have to be performed carefully 32 times - each CCD having 3 securing
nuts and washers to install. In addtion each CCD has to have its shorting pad removed and
a 40-pin Tactics connector attached to enable electrical connectivity to the outside world.

E2V Mech. Sample CCDs being Aligned on Dummy Focal Plane

1. E2V Mech Sample CCDs mounted next to one another.

2. E2V Mech. Sample CCDs mounted next to one another in
Dummy Focal plane

3. View from connector end showing bond wires and shorting pad.

4. View from side showing shorting pads etc

5. Underside view showing securing nuts, drawbar and locating pin holes

6. View from top showing dark surface of CCDs

7. View along Dummy Plate showing reflection of Flow bench grill in CCDs

8. Close up of bond wire pads on both CCDs.
Also proximity of one CCD to the other

9. Rear shot of Dummy plate showing details of machined cut outs

10. Final view of assembly from under-side showing
CCDs mounted upside-down

Status, March 16th, 2006 - Mech. Sample (02393-10-02)
Imaging Data

The image below has just been taken on the Test System. It shows a 1 second exposure of
a regular Test grid pattern.

The CCD is again being read-out in split serial mode. This frame clearly indicates the quality
of this device as there is obviously a fault on the right and if you inpect the FITS data
a multitude of column or pixel defects can bee seen on the left-hand side amongst the test
image.

You can also see the horizontal x-under and over-scans at the left and right-hand edges
and the y (vertical) overscan along the top edge. These devices are 2048 x 4096 pixels and
the frame size is 2148 x 4200 pixels.

The quality of this device therefore lives up to its name - a Grade 6 Mechanical Sample!

E2V SkyMapper Mech Sample JPG image.

E2V SkyMapper Mech Sample Test Pattern FITS data (17Mby).

Status, March 15th, 2006 - Mech Sample Reads Out!

The pictures below illustrate the grade 6 mechanical sample being installed into the Test System.
This was deemed by E2V to probably not have working Silicon mounted, but for a first run with
the Test system, it was worth trying for a 'look see'...

The following image show the Room Temperature Read out of the CCD in 'Split Serial' Mode
As can be seen there is dark current build up on the LHS and on the original data a column
defect can be seen near the top. There is very little signal on the Right-hand side.
This almost certainly means the right hand ouput amplifier is not working and may be due
to the predicted DC shorts for these type of Grade 6, mechanical sample devices.

The X under and over scans can be seen on the left and right hand edges, the y-overscan can
just be seen at the top of the array (bottom of the picture here) and the serial split resulting
from the dual read-out mode of operation.

It has been instructive and useful to install this part as this device represents little risk to the
system operation and has confirmed the operational integrity of the DSP code (which is a modified
version of the E2V 2.3m Imager DSP code) and the internal and external wiring design.

The only changes which were made to the Imager DSP code were the Bias and clock settings.
The DSP code may however need optimising for this type of device to obtain the best characteristics
eg. the predicted noise level of 2e rms for these devices. This will be trialled on the
Engineering CCD, the 1st of which will be installed next week.

First Room Temp. Split R/O of E2V Mech. Sample CCD

1. E2V Detector Mount Block removed from Test Dewar

2. Detector Mount block and E2V CCD handling rig

3. Out-of-focus shot of DMB

4. Detector mounted on DMB and cover plate installed

5. Underside of DMB showing Flex connector

6. CCD mounted on DMB fron front with cover
plate installed.

7. CCD on DMB with cover, showing flex connector

8. Assembly mounted in Dewar with perspex protective conver in place

9. Another shot of assembly mounted in Dewar with perspex
protective conver in place

10. Assembly from side

11. Assembly from top left with cover plate removed

12. Close up of CCD with cover removed.

13. Close up of bond pad strip at bottom of CCD

14. Close up of CCD surface

15. CCD suspended on draw bar with locating pin installed
and shorting strip removed

Status, March 6th, 2006 - Delivery #4

5 more Science devices arrive at RSAA.

The 3rd batch of 5 Science devices was delivered to RSAA 2 weeks ago.

These devices again exhibit superb performance and have all passed
provisional acceptance for the SkyMapper Focal Plane Array,
and continues the trend for a good science performance for the SkyMapper FPA.

We are now also in possession of a second Mechanical Sample CCD for use in alignment
and testing positional accuracy in the focal plane before installing the 4 Eng.
and finally the full 32-element, science CCD focal plane.

Status, January 30th, 2006 - Detector Assembly

Mechanical Sample is loaded into Detector Mount Block(DMB).

The E2V Detector Handler has been built to ensure we mount the Science and Eng. devices
into the focal plane with the minimum of risk, to permit ease in fitting to the focal surface,
to attatch the (PGA) Tactics connector and align all the devices safely.

The link *here* Illustrates the process of mounting the mechnical sample CCD onto the DMB,
which was undertaken on 19th January and these pictures clearly show the
elements of the mounting process including mating the Tactics connector
and other parts of this unit for safely handling the SkyMapper CCDs

Status, January 10th, 2006 - Delivery #3

4 more Science devices arrive at RSAA.

The 2nd batch of Science devices was delivered to RSAA this afternoon.

These devices again exhibit superb performance and have all passed
provisional acceptance for the SkyMapper Focal Plane Array.
This clearly continues the trend for a good science performance for the SkyMapper FPA.

Status, November 11th, 2005 - Delivery #2

Last 2 Engineering devices have arrived at RSAA.

The last pair of the suite of 4 engineering CCDs have now arrived at RSAA.

Status, October 18th, 2005 - Delivery #1

2 Eng. & 2 E2V Science devices have arrived at RSAA.

The 2 Engineering and 1st of the 2 Science devices have now arrived at RSAA.

This somewhat unusual situation (getting 2 science devices before the full
complement of 4 Enginerering CCDs) has been brought about by the fact that
E2V are finding it difficult to find Eng. CCDs out of the most recent Fab.
run. The devices are so good that they are all being classified as
Science devices!!.

This clearly bodes very well for the SkyMapper FPA.

The remaining Eng. CCDs will follow and be delivered before the end of
November along with the next 2 science devices.

Status, September 28th, 2005

E2V CCDs due to arrive at RSAA in a week.

This from Steve Darby, SkyMapper Project Manager at E2V

I am just about to despatch two engineering devices and two science grade devices as agreed.

We have taken some photos of the devices just prior to them being fitted into the transit boxes.

From the quality of the photography I'm not sure of what value they will be to you as the images
are not that sharp though some show a good reflection of the camera and the hands of the
photographer!
(The numbering system of the photos should be self explanatary being notated by the device
serial number)

PS. You'll have to believe me that the photos are not one device taken eleven times !

I have also been informed by Paul Jorden that the SkyMapper fab run has produced such
good devices that they are having problems selecting grade 5 (Engineering) devices!!
All the CCDs are coming out as science grade!

This is the reason for Steve's comment above - we are getting 2 science devices before the
full delivery of the 4 Engineering CCDs.

This is very good news and bodes well for a quality fill of the SkyMapper focal plane.

Status, August 4th, 2005

E2V CCD Procurement status,
pictures of grade 6 CCD

This from Steve Darby, SkyMapper Project Manager at E2V

'...Just a short update of where we are on your project:
1) All the planned batches have commenced through the FAB area.

2) Fully processed wafers (backthinned) have been probed and suitable
devices selected and have been packaged for the grade 5 (engineering) devices.
These devices are now awaiting the camera test slot planned for this month.

3) Further wafers from the first three batches are now being back-thinned.

4) We have despatched the ZIF sockets - I hope that you have received these!
(YES we have - yesterday, 40 off at A$1700 each!)

1. PGA 'ZIF' socket for the E2V SkyMapper CCDs

2. PGA 'ZIF' socket showing opening mechanism

We see no problems in meeting the grade 5 device delivery in Sept.
unless of course all the devices selected test as science grade...'

This last statement may not turn out to be as unlikely as it may appear!! (Paddy)

We now have the Test system in an advanced stage of preparation to mount the grade 6
and the Engineering CCDs when they arrive in September. This test system will also
be used for the WiFeS Fairchild Imager CCD testing. We plan to utilise the
existing test dewar and controller and operate this in parallel with a duplicate
Test dewar which will be a re-fitted DBS Red (the old camera) dewar and both suitable
for mounting either the E2V 4482 CCD for SkyMapper or the Fairchild Imager
CCD486 for WiFeS

The pictures below illustrate the grade 6 mechanical sample - but as it turns out,
a working CCD in this instance, for the SkyMapper Focal Plane

1. Grade-6, Upside down in box, lid removed

2. Grade-6, Upside down in box, lid removed, close up

3. Grade-6, Upside down in box, lid removed, close up at an angle

4. Grade-6, Detector out of box and 'right way up'

5. Grade-6, 'Right way up' showing connector detail

6. Grade-6, Detector surface showing bond wire pads and mounting legs.

Status, July 18th, 2005

We are planning to operate the Grade 6 CCD in our test system in the next
few weeks.
A Grade 6 E2V CCD is usually just a mechanical sample but in this case this
device has a working, single channel detector mounted on the package and so will
be useful in setting up our test system before the Engineering CCDs arrive in September.

This will enable us to quantify the detector operation and investigate its performance
in terms of the SkyMapper Science Goals. DSP code to operate this device will be optimised
for this purpose, though it now looks likely that we will be operating the 32, E2V
Focal Plane CCDs with the GPC (Giga Pixel Camera or Tonry) Controllers.

These have been shown to offer superior performance for this instrument and at a cost
which is very competitive with the ARC hardware.

After successful testing with the Grade 6 CCD we will install the first of the
Engineering CCDs, which are set to arrive in September, in the test system and
perform some additional quantification and testing on these devices.

After this we will be in a position to operate the Science devices in the test system,
these are set to arrive early next year at the rate of 3-4/month. These devices will be
operated with the ARC Test controller in the first instance, although as stated above,
it is hoped we will make the transition to the new GPC Controllers during the latter part
of this year.

First E2V SkyMapper CCD arrives at R.S.A.A.
9th May, 2005

The first of the SkyMapper CCDs has just arrived at RSAA.

This is a 'grade-6', usually defined as a 'space grade', i.e. it can
be used in the procedure to design and set up the mechanical mounting
and occupies the space which the real science CCD will eventually occupy.
It ... doesn't go into space!
This device however does have Silicon on it & apparently one working
amplifier. It is also AR coated for broadband use.

We should therefore be in a position to operate this device well
in advance of the delivery of the 1st of the Engineerig CCDs, set to
take place at the beginning of October.

1. Grade-6, Mechanical Sample CCD, 02393-10-02

E2V SkyMapper CCDs publicity, 2nd May 2005

This was sent to me last night by Paul Jorden at E2V, requesting approval
to include this in their next Newsletter. This is a documnent which is
available and circulated publically and should see the ANU and RSAA in
particular given high profile publicity.

Quote..

E2V Newsletter, May 2005

36 CCDs for Australian ‘SkyMapper’ telescope

The Australian National University (ANU) Mount Stromlo Observatory has
placed a major CCD contract with e2v this year. The supply of 36 large area,
deep depletion CCD44-82s, for enhanced red sensitivity, signifies the largest order
to date for this type of e2v sensor. The devices provide low noise, very high
spectral sensitivity, together with close butting capability for maximum fill factor
within their large mosaic camera. They are destined for the ANU’s new 1.3m ‘SkyMapper’
telescope, which will digitally survey the sky at night with an 8-sq degree field
of view.

e2v has worked with ANU on previous smaller CCD orders and the Company is
delighted to have been selected for this major new instrument,
based on past performance.

Contract with E2V

Status, February 22nd, 2005

The Final contract for the E2V suite of Focal Plane Detectors for
Sky Mapper is now in its final stage. All technical and financial apects
of the contract are in place. A copy of the final contract is on the local
network share - for RSAA and MSO staff only, at this time, See -
\\skymapper\E2V Contract\SkyMapper E2V contract-Final.pdf
We are attempting to get the contract signed and in place by the end of
February.
This should see the 1st Engineering devices starting to arrive 6 months
after that date and the first science devices 3 months after that... :-|

We are currently in the advanced stages of placing a contract with E2V Ltd, Chelmsford,
UK for a suite of 32, 2kx4k 44-82 CCDs for the new SkyMapper telescope being built
at Siding Spring. This will utilise these detectors in a Mosaic array resulting in a focal plane of 16x16k pixels.

E2V 32-chip FPA and mating connectors

The simple looking connectors shown on the back of the FPA... are almost $2000 each!
$68,000 in total.:-(

RSAA is involved in the detailed contractual negotiations to specify these devices and their subsequent quantification and acceptance
by RSAA Detector personnel. See the SkyMapper pages for more information on this new project.

Impression of the Site and Detector Array

Documentation Available

E2V Detector for SkyMapper.

Request for Contract for Mosaic.

Detector CoDR Power Point Presentation.

Enquiries about the SkyMapper Contract and Detectors?
detman@mso.anu.edu.au

	Skip Navigation \| ANU Home \| Search ANU \| Directories
	Research School of Astronomy and Astrophysics Mount Stromlo and Siding Spring Observatories ANU College of Physical and Mathematical Sciences

Current News

Status, February 1st, 2010.

A complete Review of the Work undertaken with the Focal Plane Imager, Controller Hardware and System optimisation which proceeded from September 2008 until the end of 2009.

1. Initial Work undertaken in late 2008/early 2009.

2. Remedial Action taken with the Ribbon Cables.

3. Final Solution - The RIF board problem.

3.1 An explanation of the Real source of the Problem.

4. Optimisation of the detector read out timing.

5. Final Controller Hardware Configuration.

5.1 Gain and Linearity.

Preflash Level(ms)

Net Signal(de-biased) (k adu)

Gain (e/adu)

5.2 Rigiflex Replacement for the Ribbons.

6. Attaining Low Noise, Linear operation.

6.1 The Linearity Problem.

ADC Gain

System Gain (e/adu)

OG2 Voltage(V)

System Gain (e/adu)

Noise(e)

7. Some Pictures taken with the Imager on The Simulator.

8. Some Reference Data Frames taken in Cicada.

49. Bias Frame.

50. Pre-Flash Frame.

51. Test Pattern Frame.

52. Room Temp. Reference Frame.

9. Some On-Sky data taken in early March 2010 ('1st Light').

Detector System Assembly Pictures.

And that, as they say, is that. Done Job. Paddy Oates. March 11th 2010.

***** End Latest 2010 Update *****

STOP ..........

** Documentation below refers up to the end of 2007, Some FITS data has now been removed from the docs below. **

Status, November 20th, 2007.

All 32 Science CCDs Imaging thru FIB into SAC.

4, 8-channel Science CCD Mosaics read outthrough FIB and into SAC#1.

SAC #2 side UPR bank of connectors

SAC #2 side LWR bank of connectors

SAC #1 side UPR bank of connectors

SAC #1 side LWR bank of connectors

Status, November 16th, 2007

All 4 Engineering CCDs Imaging thru FIB into SAC.

4-channel Eng. CCD R/O through FIB and into SAC.

Status, November 15th, 2007

First Image retrieved from new FP FIB.

Status, November 9th, 2007

Focal Plane Connected and Working.

Well done all concerned is what I say!

Status, November 7th, 2007

Focal Plane Populated with E2V Science CCDs.

The 32 Science CCDs were installed during a 9 hour period yesterday, November 6th.

Status, October 3rd, 2007

SkyMapper Focal Plane at Cooldown 2B.

Some pictures of the PMFP!! Assembly.

Status, September 18th, 2007

SkyMapper Focal Plane Imaging Data taken from STARGRASP in 32-channel mode.

32-channel R/O from a single SAC.

Status, September 6th, 2007

SkyMapper Focal Plane Imaging Data taken from STARGRASP in 16-channel mode - WORKING.

Status, September 5th, 2007

SkyMapper Focal Plane Imaging Data taken from STARGRASP in 8 & 16-channel mode.

Block Diagram of the System.

Pictoral Diagram of the System.

1st Image: an 8-channel R/O into Cicada

8-Channel, 100ms LED Pre-flash frame

2nd Image: a 16-channel R/O into Cicada

16-Channel, 100ms LED Pre-flash frame

Status, June 21st, 2007

SkyMapper Focal Plane Imaging Data taken from STARGRASP & E2V Eng#01 & Eng#02 CCDs.

Pictures of the experimental setup are shown below.

LED Pre-flash frames from 2 CCDs.

Dark Frame Data from 2 CCDs.

Bias frame images from 2 CCDs.

Pictures of the Pin Hole Test Exposures.

Status, May 17th, 2007

SkyMapper Focal Plane Imaging Data taken from ARC & E2V Eng#01 & Eng#04 CCD.

Pictures of the Pin Hole Test Exposures.

Status, May 10th, 2007

Data Retrieval from SAC (STARGRASP) to Cicada.

SkyMapper Vacuum Jacket Eng. CCD Test Phase

A complete Review of the Work undertaken with the Focal Plane Imager,
Controller Hardware and System optimisation which proceeded from
September 2008 until the end of 2009.

Preflash Level
(ms)

Net Signal(de-biased)
(k adu)

Gain
(e/adu)

System Gain
(e/adu)

OG2 Voltage
(V)

System Gain
(e/adu)

Noise
(e)

And that, as they say, is that. Done Job.
Paddy Oates. March 11th 2010.

* End Latest 2010 Update *

Documentation below refers up to the end
of 2007, Some FITS data has now been removed from
the docs below.

4, 8-channel Science CCD Mosaics read out
through FIB and into SAC#1.

The 32 Science CCDs were installed during a 9 hour
period yesterday, November 6th.

SkyMapper Focal Plane Imaging Data taken
from STARGRASP in 32-channel mode.

SkyMapper Focal Plane Imaging Data taken
from STARGRASP in 16-channel mode - WORKING.

SkyMapper Focal Plane Imaging Data taken
from STARGRASP in 8 & 16-channel mode.

SkyMapper Focal Plane Imaging Data taken
from STARGRASP & E2V Eng#01 & Eng#02 CCDs.

SkyMapper Focal Plane Imaging Data taken
from ARC & E2V Eng#01 & Eng#04 CCD.

Status, December 5th, 2006
Characterisation Data for 4th E2V Science CCD
(05255-02-02, BATCH#5)

Characterisation Data for 3rd E2V Science CCD
(05163-16-02, BATCH#3)

Status, November 14th, 2006
SkyMapper E2V Science Device #21 installed in Test System
(05163-16-02, BATCH#3)

All data reported by E2V, along with a complete QE data set
for the 32 science devices is now at the link below.

Status, November 13th, 2006
Final E2V CCD Delivery #10,
Final 6 Science CCDs arrive at RSAA.

Status, November 8th, 2006
E2V CCD Delivery #9, 05163-07-02
Last but one batch, 1 Science CCD arrives at RSAA.

Characterisation Data for 2nd E2V Science CCD
(05191-02-01, BATCH#1)

Status, October 30th, 2006
SkyMapper E2V Science Device #23 installed in Test System