January 2000 Commissioning of PFU

THE AAT GETS A NEW PRIME FOCUS UNIT

"True" colour B,V,R image of the spiral galaxy NGC1808 obtained with the MITLL3 2Kx4K CCD mounted on the PFU. Data obtained by S.Lee, D.Malin & C.Tinney. Colour imagery by S.Lee. Copyright AAO.

This page contains a report on the Jan 2000 PFU commissioning run. For further
information on WFI and PFU see the WFI/PFU Page

Introduction

Between January 28 and January 31 this year, the AAO began commissioning a new Prime Focus Unit (PFU) to support direct imaging with large format detectors on the AAT. The PFU offers observers a facility for mounting detectors up to 130mm x 130mm in size, with either the doublet or triplet corrector. Observations can be performed through up to 6 filters of 165mm x 165mm, mounted in a filter wheel. Exposures are controlled by a shutter of a two travelling blades, which are independtly controlled to produce identical linear excursion profiles, permitting exposures as short as 1s with 1% uniformity over the entire field of view.

AAT Prime Focus Imaging - A Short History

For those of you not familiar with the prime focus imaging facilities of the AAT (and given the small number of allocations for imaging over the years, there aren't many of you who are), imaging at prime focus has been offered using only a 1K detectors mounted on a slightly modified version of the photographic camera head, or using a 1K detector and some f/1 re-imaging optics. In both cases, the AAT has offered facilities un-competetive with better image quality sites ... when you have a 6' field of view, the difference between 1" and 1.5" median seeing is critical. As a result, there has been little incentive in the past to upgrade the existing camera head - in spite of its completely manual operation, which necessitated an observer riding in the cage.

The advent of large format detectors, and mosaics, however has meant that the AAT's wide field of view can give it a scientifically important niche in imaging again. By concentrating on large shallow surveys (the sorts of surveys other telescopes usually eschew), the AAT can carry out important imaging science even in 1.5" median seeing. This has been one of the primary drivers beheind the decision of the AAO, RSAA and University of Melbourne to construct an 8Kx8K Wide Field Imager (WFI, pronounced "wiffee") for the use on the AAT and the RSAA 40" telescope.

However, to mount the large (roughly 0.8m on a side) and massive (roughly 100kg) WFI and expose with it, it was obvious that a new camera head would be neccessary for the AAT, to replace the existing manually operated head. For the last 18 months, the AAO has been designing and constructing PFU for this purpose, so that we can have a mount ready for WFI when it gets commissioned on the AAT in July, this year.

The Prime Focus Unit (PFU)

This figure shows a plan drawing of the filter wheel and shutter system for PFU (an expanded view is available here), together with an image of the unit itself (larger version here), prior to installation on the telescope. For reference, the open jaws of the caliper sitting on top of PFU are 100mm apart. (Due to a shortage of time, we were unable to anodise the unit prior to the run, hence its silvery colour.) The other image shown is an image from the MITLL3 2Kx4K detector used on this run. The field shown is the H II region NGC 2014 in the LMC, and it is 8'x16', and was taken in 1.0" seeing in the V band during the commissioning run.

Once installed in the prime focus cage, PFU leaves very little room. Since there's no need for an observer, the seat in the cage has been removed. Indeed, any observer larger than the petite John Stevenson (who's shapely legs are pictured below) would not fit. The filter wheel itself has a diameter of 840mm, while the two travelling blade shutters are each roughly 130mm square. The filter wheel is driven by a belt around its edge. When a filter position is reached the belt is de-tensioned, and a pneumatic detent locks the wheel into position, with a satisfying 'clunk'.

The shutter is operated as two independent blades sliding on teflon surfaces, and driven by DC servo-motors via an extremely low-pitch lead screw. The low mass of the shutter blades permits them to be accelerated rapidly to their operating velocity, {\em before} they cross the edge of the field of view. Each blade is then driven at a constant velocity to produce a 1s open- or close-time, before being decelerated {\em after} crossing the other edge of the field of view. Since both blades are precisely controlled to the same velocity profile, the resulting system can produce highly uniform exposures at net exposure times down to 1s. Though not currently installed, sensors will eventually record the shutter blade crossing times, so that a precise record of each exposure's absolute {\em observed} time can be stored in file headers, as well as the requested time.

PFU Filter Access port All PFU filters will be used mounted in metal holders to reduce wear and tear on glass filters. A new feature for PFU is that the filter wheel also contains a barcode reader (a barcode containing the information "empty slot" can be seen at the right of the empty filter slot in this image). At initialisation, each filter position is read, so that bar-codes placed on the filter holder can be used to initialise the file header look-up tables. In this way, it is planned that observers will ALWAYS know exactly which piece of glass was used for their exposures, via a unique identification number from a master list of filters which we are compiling.. (The barcode reader is, of course, turned off during observing). Filters manufactured from Schott glass have been acquired to match the UBVRI passbands (GIF, Postscript), and SDSS g'r'I'z' filters are on order.

The January Run

Commissioning was carried out in January, using a single MITLL3 detector, giving a 8'x16' field of view.

Night 1 - Cynics (not me this time) had been heard to suggest prior to this run, that clear weather could only be expected when we couldn't use it. As the sun set on the first night, therefore, a week's worth of cloud, rain and fog cleared to deliver clear conditions and 1" seeing, when all we {\em really} wanted to do was test how the shutter worked. What we quickly found, was that having unblackened insides to a camera is a very bad idea. Because of time problems, PFU had not been blackened or fully baffled, and we soon found that enough light leaked onto the detector during read-out to pose a serious problem for flat-fielding. Nonethless, we proceeded to observe standards and focus the telescope through a variety of filters. We were able to obtain images of uniform quality over the 8'x16' field of a single MIT/LL CCD, down to 1". The standard star data also revealed the outstanding sensitivity of the MITLL CCDs in the red, with the MITLL3 delivering {\em twice} the sensitivity of the TEK in the I band. This data will also enable us to prepare a meaningful WFI S/N calculator, since the MIT/LL detectors will be the same or similar in WFI. We also found that all trace of the reflective ghost image, which had previously made I-band observing with the triplet impossible (see http://www.aao.gov.au/local/www/cgt/cgthome/uh_8k_i_ghost.gif), has now been eliminated by the improved AR coating of the last triplet element.

Night 2 - The second night saw us unable to acquire further afternoon test data, due to a shutter failure. After struggling to diagnose the problem with data alone, it was pointed out by night assistant extraordinary Steve Lee, that we could look at the shutter in action by climbing up on the telescope and peering through the triplet corrector. The next hour saw me propped up on the AAT top-end shining a torch at the triplet, and indeed we found one blade of the shutter was inoperable. Another hour's poking around inside the PFU with both myself and John Stevenson in the cage, revealed that the AAO's lack of "Go-Go-Gadget" arms would prevent the problem (a loose grub screw) being solved that night. At which point, software maestro Serge Ivanoff leaped into actaion and re-programmed the shutter to operate in single blade mode, instead of two blade mode, so we could at least spend the rest of the night acquiring long exposures. Fourty minutes later, we were on the air again, and acquiring data for a number of service observing programs.

Night 3 - At the crack of dawn the following morning, Gordon Schafer, Brendan Jones and John Stevenson turned up to remove the top end and PFU, in order to deal with that pesky grub screw. By that afternoon, we were back on the air. Once again night assistant extraordinary Steve Lee earned his pay, by pointing out to a sleep deprived commissioning astronomer that we {\em could} get dome flats (though not sky flats) for our data un-encumbered by leakage during readout, by turning the dome flat lamp on and off. Duh! Amazingly, we found that even B dome flats acquired in this way flattened our data very nicely (which makes me wounder if the "dome flats are no good in the blue because the lamps are too red" might not be folklore induced by leaky instruments, rather than sky colour). The following night saw us struggle with 3" seeing and cloud, resulting in only a little useful data. We were able to test one of the Taurus blocking filters (R8, 910nm/40nm), which produced very nice images. These intermediate pass band filters have very high throughput, and many are optimised to reject OH emission (eg. R8 counts 1/5th the photons of source in I, but 1/10th the sky counts), making them potentially very powerful for gravitational redshifts.

Night 4 - Flexure tests showed that dust doughnuts in flat fields at the zenith are not being corrected properly at 2 airmasses - though this might be filter movement more than flexure. A problem that is being looked into. Tests of the shutter unformity revealed exposures to be uniform at the 0.5% level for 1 second exposures - twice the specification! Absolute timing, however, must await the fitting of blade edge sensors. This night saw the best seeing of the run delivered - though also some of the worst clouds. Steve Lee and David Malin acquired a number of three colour images, so stay tuned for some solid-state AAT colour imagery, of which the images below are just a sample. All images are copyright Anglo-Australian Observatory 2000.

Finally, for the really terminally in need of a life, here are my handwritten logsheets from the observing run. Page1, Page2, Page3, Page4, Page5, Page6, Page7, Page8, and Page9. You can also make your own logsheets from the data stored in the AAT archive at http://www.aao.gov.au/local/www/cgt/wfi/commission1.html. And you can review my cryptic notes on how I reduced the data at README.

Conclusion

Like all good commissioning runs, this one delivered sucesses and problems. The problems we will be dealing with in the coming months before WFI arrives in July. For the sucesses I'd like to thank: the members of the PFU Team - Alan Lankshear, Rob Patterson, Chris McCowage, Serge Ivanoff, and Steve Lee; the guys who actually built it - Gordon Schafer and Brendan Jones; the Coona staff who made the run go so smotthly - Gordon, Brendan, and John Stevenson; and for their invaluable presence during the run - Steve Lee, Serge Ivanoff and David Malin. In particular, we sucessfully knocked off all the current service programmes accessible in January. Proposers will be contacted shortly, once I've worked out how to process the data.