MITLL2A has been withdrawn from service and is no longer available. The information on this page is provided for historical interest. 28 August 2002.

The MITLL2A Lumogen Coated 2Kx4K 15um Pixel Device

1. The Device
2. Differences with the TEK
3. Lumogen Coating
4. Use with AAO Instruments
1. Taurus
2. UCLES
3. UHRF
4. RGO
5. Others
5. Serial Links, NONASTRO Speed and Possible Problems with MITLL2 Observations 18 Oct 1997 - 14 Feb 1998, and ALL CCD Observations 20 Jan 1988 - 14 Feb 1998

MITLL2A is available for use on Taurus, UCLES, UHRF and RGO Spectrograph. With better read noise, smaller pixels and larger area, the MITLL2 should be the detector of choice for most programs performing observations longward of 5000A.

1.The Device

Format

2K x 4K edge buttable, with 15 um pixels. The pixels on the MITLL2A are read out flipped in X (where X is defined as the X-direction as seen on AAO XMEM images and FITS files) relative to the MITLL2. It therefore now produces images in the same orientation as the MITLL3.

Quantum Efficiency

The device is a set-up grade thinned 2Kx4K 15um chip from the MIT/LL/UH consortium. The SiO AR coatings used on the LL device are optimised for the red. A Lumogen yellow coating was applied in Dec 1997 in order to try and improve the blue quantum efficiency. The overall performance of the device both before and after coating is shown in the following figure in comparison with our current standard TEK 1K detector. A larger version is also available, as is a colour postscript file (NB: because this file was created by an Evil Empire Microsoft product, it will not preview with Ghostview version 1.5 or earlier. It will preview with Ghostscript version 4 or later, and it will print out).

This data was measured obtained with MITLL2, but will also apply to MITLL2A

Ratio of QE (compared to TEK at 200K*) integrated over Imaging Bandpasses
U (340nm cutoff)  U (no cutoff)  B  V  R  I  TEK (170K) 0.84  0.84  0.86  0.90  0.91  0.90  MITLL2 0.76  0.87  0.57  1.02  1.07  1.18  Blue Thomson 0.38  0.44  0.28  0.51  0.55  0.55  Red Thomson 0.006  0.006  0.22  0.52  0.59  0.52
* Recall that the TEK is used at 200K for imaging applications to obtain higher QE at the cost of higher dark current. It is used at 170K for spectroscopy.

 

A few words about the QE measurements are in order. The 'before coating' numbers come from Gerry Luppino at the University of Hawaii. They were not made with the CCD in the same operating conditions at which it is now being used. The 'after coating' measures were made by John Barton at the AAO. John had to struggle somewhat in order to derive sensible number from three different photo-diodes with out of date calibrations. The QE curves for the 'after coating' device are therefore uncertain by as much as +-5% in the extreme blue and extreme red.

Having said this, a QE of 35% short of 4500A is exactly what one would expect from the lumogen coating and the 'before coating' curve, so we beleive these results are robust at the +-5% level. No 'before coating' QE measurements are available in the blue for this device. However, based on a comparison with similar MIT/LL devices, the before coating QE at 4000A was ~ 25% and at 3500A was ~ 1%. So while the coated device is worse in the blue than the TEK, it is much better than it was.

The main conclusion to draw is that longward of 5000A the coated MITLL2A device has QE as good as, or better than, the TEK 1K.

Count rates

The QE numbers before coating numbers have been roughly checked by looking at Landolt standard observations in B and R made with Taurus. We find that the LDSS B filter + MITLL2A on Taurus produces count rates of 11 photon/s for a B=22.5 star, and the KPNO R + LL on Taurus gives count rates of 52 ph for a R=22.5 star.

 

Comparison of these numbers with the expected count rates of the TEK at prime (which requires some correction for extra optical surfaces, reflections and filter throughuts), shows that the LL device is about a factor of two down on the TEK (170K) at B - in line with expectations.

Lumogen Coating

See below for a report on the lumogen coating.

Cosmetics

There is a bad light emitting strip to the bottom right, which blats out about 10 columns on one edge of the chip, and a further charge trapping site. For a set-up grade device it generally looks very nice.

 

There is a 'brick-wall' pattern present on all these LL devices, caused by a flaw in the laser annealing process. The brick wall is wavalength dependent, being much stronger in the blue than in the red. Prior to lumogen coating it produced variations in the device QE of about 20% at U and 7% at B and even less at redder wavelengths.  After lumogen coating this was reduced to about 4% at U. This is not surprising since the photons the CCD detects at U are actually being re-radiated by the lumogen at longer wavelengths (about 5500A), where the brickwall is much smaller.

For an example of this pattern see http://gardiner.ucolick.org:80/~ccdev/lincoln/w66c2/w66c2.html which shows some test results carried out at Lick for other LL devices. Because this is a QE variation, and in the chip it flat-fields out. There are fewer cosmic rays than our  Tek1K, - probably due to a known problem in TEK manufacture which is not present in these LL devices.

Fringing

The LL chips are known to fringe in the red. Observations in dispersed light with the RGO spectrograph show this fringing pattern to be extremely regular, with an amplitude of 3% peak-to-peak at 7000A, 6.5% p-p at 8000A and 10% p-p at 9000A. The frining at 10000A is still about 10% p-p, making fringing considerably smaller with this device than for our existing TEK.

In dispersed light the fringing pattern appears as a very regular series of bands with peaks appearing at a wavelength spacing of about 30A.

Broad-band images obtained with Taurus seem to show no evidence for fringing.

The lumogen coating seems to have had no effect on fringing - again as expected.

Focus

The LL detector sits in a standard AAO dewar, and will mount on all the AAO instruments - however it has a preferred orientation because it is rectangular. This means, for example, that on the RGO it has to be mounted rotated relative to the TEK.

While care was taken to mount the chip in its MITLL2A incarnation with the same back-focal distance, it will almost certainly have slightly different focus from the MITLL2. When use on UCLES, shimming of the MITLL2A was required to bring it into focus, where before it was not. Experience will be gained with this on an instrument by instrument basis as MITLL2A moves back into operation.

Modifications have been made to the AAO's existing controllers to take advantage of the superb read-noise performance of these LL devices. This has allowed a device 8 time larger than our TEK1K to be read in about twice the time with similar or better noise performance.

 

SPEED	INT (us)	GAIN (e-/adu)	READNOISE (e-)	ALPHA (x1e-6)	SAT (Ke-)	READ RATE (us/pix)	READ TIME (s)
							1x1	2x2	5x5
NONASTRO	1+1	4.8	5.26	-1.48	125	6.5	60	24	9
FAST	2.5+2.5	1.8	3.2	-0.16	115	11.5	105	42	15
NORMAL	4.5+4 .5	0.9	2.4	0.02	59	19	170	59	19
SLOW	12+12	0.37	1.8	0.04	24	34	300	93	24
XTRASLOW	48+48	0.091	1.4	0.01	6	106	940	260	56

Full well seems to be about 115Ke-, but this limit is only reached in the FAST and NONASTRO  modes - all other modes are limited by the 16-bit A-D at 65535 counts. The next generation of AAO-2 controllers will decrease these read times for large format devices significantly.

 

The linearity term ('alpha') above corrects for an assumed linear increase in non-linearity with count-rate. When alpha < 0.1e-6, such an approximation is quite good. For larger values, however, a polynomial parametrisation is usually preferable, as the CCD's non-linearity is non linear (ie they get more non-linear at high countrates).  See the WFI linearity discussion for more information (WFI  has one MITLL Phase I CCD in it, like the MITLL2A). Observers for whom linearity at high countrates is critical must acquire their own linearity data.

 

To remind you of the TEK performance see Chapter 2 of the CCD imaging manual from which the following table comes

 

Table 2.3: Readout parameters of the TEK CCD (for a 1050x1024 window).
		Readout Speed (TEK)
	XTRASLOW	SLOW	NORMAL	FAST	NONASTRO
Readout time (s)	394	120	75	52	33
Readout noise (e-)	2.3	3.6	4.8	7.2	11
Gain (e-/ADU)	0.34	1.36	2.74	5.5	11
Saturation (ADU)	65535	65535	65535	65535	35000
Alpha  x 1.e-6 (see  Section 2.1)	negligible	negligible	-0.03	-0.07	-0.14

VERTICAL AND HORIZONTAL CHARGE TRANSFER

No direct measuremts have been made. Cosmic rays appear crisp with no smearing detected either in X or Y.

DARK CURRENT

MITLL2 - A 160K operating temperature has been adopted.  When tested at the AAO's Epping laboratories at 160K the dark current was small - a 4000 sec dark showed it was less than 0.3 e-/pix/4000sec.  Tests carried out in the coude room at the AAT, however, have showed significantly higher dark currents, which decrease with a time constant of about a day. UCLES and UHRF observers are strongly urged to read the following report.

MITLL2A - measurements have not yet been made.

TRAPPING SITES

A quick look for trapping sites revealed only one obvious candidate. There are possibly several smaller traps that produce no delayed charge into the vertical overscan region and so have remained undetected.

RESIDUAL IMAGES

MITLL2 - There are no signs of residuals in pinhole exposures of 10 times chip saturation, i.e. at 1.6Me-/pix.

CLOCK INDUCED DARK CURRENTS

MITLL2 - were found to be about 0.007e/pix. Thus binning up 5*5 pixels should be little affected. However, a 10*10 bin would accumulate 0.7 e-/pix which is easily seen and may need to be taken onto account.

SATURATION STRIPES

MITLL2 - in the row direction are found on bright saturating pinhole images but only in readouts with binned rows. These come from the CCD itself and can be eliminated at rather odd V and H clock levels. Needs further work but since they do not occur on unbinned images or in images binned only in the row direction the problem was put to one side.

POWER-ON CLEANOUT

MITLL2A - Measurements have not been made

MITLL2 - The CCD cleanout rate following power off and on, with a cold CCD, as measured in 100 sec dark frames are:

2. Differences with Tek1K operation.

• Data must be saved in USHORT, to save on disk space and time.
• The MITLL2 has a preferred orientation - it may be mounted on your instrument differently to the TEK 1K
• NONASTRO modes can only be used at all with the optical fibre serial link.

3. Lumogen Coating

As of 23 Dec 1997 only cosmetic tests of the flat-fields of the coated device have been carried out. A thorough report can be found here. However, to summarise -- while the coating applied was not of the ideal uniformity we would have liked, this non-uniformity only affects the performance at wavelengths longer than about 8000A, and then at a non-serious level.

The lumogen coating was successful, however not as uniform as one would like. The coated CCD has a pastel green appearance which is  notably darker in the central rectanglar region of the image area extending to within a few mm of the edges. This "rectangle" is slightly wider towards the readout end of the CCD and narrowes slightly towards the far end of the CCD where it shows quite rounded corners. Surrounding the central rectangular region are three ring patterns, the outer one close to the edges of the image area.

The coated CCD looks clean, the original spots and deposits on the surface were still evident and the only new features were about 8 smudgy spots where the coating appeared to be thinner.

The CCD was illuminated by various lamps with two 35mm diameter diffusers interposed about 50 and 80mm above the lamp. In between the lamp and the diffusers various filters could be placed. The lamp and diffuser assembly was held about 300mm away from the CCD so that it was equivalent to about an f/10 beam. This eliminated the effect of dust particles and defects on the window and provided a fairly even illumination over the 60x30mm CCD image area.
 

General Characteristics

All flats exhibited a domed structure -- the intensity in the centre being from 3 to 13% higher than that at the mid-edge and a few % extra above that at the corners. Profiles of a central column and row were quickly analysed for the very large scale peak-to-peak amplitude of the dome (called "Vdome" and "Hdome") expressed as a % of the flat-field intensity.

In all flats except for two, the brick wall ("BW") structure could be seen and in these profiles the peak-to-peak amplitude was estimated as a % of the flat-field intensity.

An image of the CCD unbinned (UV curing lamp+RG1000 filter) shows a circular pattern which is almost certainly the polishing marks produced by the wafer fabrication process.

IR LED illumination (880nm)

10% Vdome 14% Hdome 3% BW. Central rectangle and surrounding rings dominant. BW faint and in background.

RED LED illumination (660nm)

7% Vdome, 6% Hdome, 5% BW. BW dominates with a slight darkening of the edges of the chip

UV+SL1 (350-400nm)

8% Vdome, 6% Hdome, 4% BW, a good flat flat, BW dominates

UV+SL3 (300-350nm)

7% Vdome, 6% Hdome, 3% BW, a good flat flat, BW dominates

Quartz Halogen (QH) + I band filter

13% Vdome, 14% Hdome, 3% BW, the central rectangle dominates with no sign of the fringes seen with the UV lamp which has strong IR lines, BW pattern is present.

QH+R band filter

10% Vdome, 9% Hdome, 4% BW, the BW pattern dominates with doming into the corners and only a very faint central rectangle detected.

QH+B band filter

3% Vdome, 6% Hdome, 6% BW, a very flat flat, the BW pattern dominates

Curiously, only the I band seems to be affected by the uneven coat with the central rectangular region being clearly more visible in the flats. The fringing above 700nm (not seen with the I band filter on QH lamps, but seen only with strong emission line sources) probably has nothing to do with the coating. The flats taken below 600nm down to about 300nm are very flat, the brick wall pattern dominating in all of these.

A critical examination of the brick wall pattern in the QH + B filter flat showed that at worst the peak-to-peak pattern was about 9%. Overall, a histogram of this flat showed that 99% of all pixels fell within an intensity range that spanned the average +/- 6% and this includes the fall-off in illumination at the edges and the corners of the CCD.

4. Use of the Device on AAO Instruments

4.1 Taurus

The MIT/LLs are excellent devices from the point of view of sampling.

The TTF at f/8 now gives 0.37"/pix over the full 9.87' field and in 1" seeing, this areal advantage is really paying off for the wide field surveys. There is not much need for f/15 now, and observers who do wish to use f/15 with the LL chip should think carefully about why they want to. Also the detector allows the full 10' diameter field to be charge shuffled between two frequencies, and almost the entire field to be charge shuffled between three.

For countrate calculations, it may help to know that the LDSS B filter + MITLL2A on Taurus produces count rates of 11 photon/s for a B=22.5 star, and the KPNO R + LL on Taurus gives count rates of 52 ph for a R=22.5 star.

There is little reason to prefer the TEK over the LL device for TTF use.

For Taurus use in the blue, however, the TEK device may be preferable.

4.2 UCLES

CGT/ March 2000 - MITLL2A has also been successfully used by C.Tinney in March 2000. No significant differences in perforamance were noted (relative to MITLL2) except for a pronounced pick-up noise at SPEED=FAST. This had 1adu p-p amplitude (irrelevant for our program, but potentially not for others), with a roughly 30 pixel period. At speed NORMAL the same pick-up was compressed greatly in Y, but just visible at the same amplitude. It is though this is a problem with the  extended data lines at Coude, rather than the MITLL2A per se.

UCLES users will get smaller pixels (15um) improving spectral and spatial resolution, a somewhat increased wavelength coverage, and complete sampling (with no inter-order gaps) much further into the red.

The fringing numbers found with the RGO ( a regular pattern with an amplitude of 3% peak-to-peak at 7000A, 6.5% p-p at 8000A, 10% p-p at 9000A, and 10% p-p at 10000A)  can be expected to produce similar effects with UCLES.

When tested at the AAO's Epping laboratories the MITLL2's dark current was small - a 4000 sec dark showed it was less than 0.3 e-/pix/4000sec.  Tests carried out in the coude room at the AAT, however, have showed significantly higher dark currents, which decrease with a time constant of about a day. This can have a significant impact on UCLES observations of very faint targets. UCLES observers are strongly urged to read the following report.

An important point to note is that the current UCLES camera optics cannot illuminate the entire area of the LL detector (which is 60 x 30mm in size). In fact the unvignetted region which can be observed is more like 38.5 x 18.8 mm (for less than 10% vignetting). The region covered at 50% vignetting is 60 x 34 mm, which is approximately the entire LL chip, however unless you are working in the very red, the echellogram will not put any light on much of the chip.

The following  sample GIF images from ECHWIND the region of the LL chip illuminated and the effects of vignetting.

Suitable windows should therefore enable the read time to be cut by a factor of 2 over the 'full chip' times given above. If binning in the spatial direction is used, a further factor of about 2 should be obtained, which should make use of the XTRASLOW speed tractable (4 minutes).

(If the fact that we can't actually illuminate the whole chip seems insane to you, then I suggest you contact your ACIAAT representative and start lobbying for the UCLES Camera upgrade as soon as possible!)

Note of course that the 'boxes' shown are for one wavelength set-up. You can move the echellogram anywhere on this field - but the relative locations of the boxes will stay the same. As with Taurus, the main reason for preferring the TEK over the LL device is its superior QE in the blue.

4.3 UHRF

The MITLL2 device has been  used on UHRF. MITLL2A has not been used yet.

When tested at the AAO's Epping laboratories the MITLL2's dark current was small - a 4000 sec dark showed it was less than 0.3 e-/pix/4000sec.  Tests carried out in the coude room at the AAT, however, have showed significantly higher dark currents, which decrease with a time constant of about a day. This can have a significant impact on UHRF observers, where data is often binned over up to 1200 pixels. UCLES and UHRF observers are strongly urged to read the following report.

UHRF users get smaller pixels (15um vs 24um), and a significantly increased wavelength coverage - though not the full 4096 pixels, as the shutter currently limits the clear range to ~3300 pixels. Resoutions of 900,000 have been obtained with the MITLL2. The main reason for preferring the TEK over the LL device is its superior QE in the blue.

4.4 RGO Spectrograph

The MITLL2 device has been used on RGO.  MITLL2A has not been used yet.

RGO users get smaller pixels (15um vs 24um), together with a much increased wavelength coverage. Observers can either use smaller slits, or get better sampling of sky lines. Once again vignetting stops the whole chip from being used, with only about 3000 pixels being illuminated by the RGO - which is still more than double the wavelength range covered by the TEK1K. You can play with the possible options using the WWW version of RGOANG (with user a specified detector of 3000 x 15um pixels).

Once again only observers looking at wavelengths longer than 5000A should consider using the LL device, while those observing in the UV may prefer the TEK. Observers of single objects with the LL can achieve quite short exposure times by windowing out the largely useless spatial direction.

4.5 Other Instruments

The use of the MITLL2A on other instruments will be attempted if required by observers on a shared risks basis.

5. Serial Link Problem : 18 Oct 1997 - 14 Feb 1998