Simon Morris (Durham University, Chair)
Klaus Hodapp (U. of Hawaii)
John Rayner (U. of Hawaii)
Robert Nolan (Gemini)
Mark Hunten (Gemini)
Corinne Boyer (Gemini) - telecon part time
Doug Simons (Gemini)
Don Hall (U of Hawaii)
Complete 3d modeling needed
don’t need all 2d mechanical drawings
complete optical design
complete electronics design
software - all ICDs completed?
See VG of changes since CoDR
See VG of RSAA identified issues
Relative performance comparisons? Claims discrepancy was due to optimistic predictions for NIRSPEC, so now thinks is accurate. Agree
I raised dark current reproducibility concerns. Klaus thinks temperature stability helps with dark current stability. Thinks pattern will not be visible after do dark subtraction. Don Hall also thinks dark is stable now. John said residual images is biggest problem. Klaus agrees is likely to be biggest concern. Doug suggests could get involved with CIRPASS engineering to get experience with dealing with residual images.
AO guide stars. Debate on NIRI OIWFS - h=15 currently OK.
I mentioned AO hole vignetting, need to include in sky coverage calculations.
BH demographics - need to push on M/L limits to improve science
NLR - Klaus mentioned possibility of shift and add for objects with bright continuum. I wonder about using NDR and shift and add is anything possible? Would need to have a model for the source brightness along with all the linear NDR information?
One night per object Ok by me.
Doug, 2003 - 2004 for laser with Altair. MCAO PDR docs describe laser technology studies - available in about a month. Are pushing this very hard. Altair gets first one.
Doug confirmation that without OIWFS will have a problem - NIRI tests with long set of integrations - is significant flexure between NIRI and PWFS - 10 arcsec in f6 over several hours.
Extrasolar planets good for poor transparency? Peter mentioned also maybe think of NIFS for bad seeing as image slicer?
Klaus wondered about putting in dichroic to allow OIWFS to use target?
Mechanical flexure error budget a concern - doesn't pass required limit?
OIWFS gimbal rotation biggest contrib.
Grating rotation next biggest - for both allocation in table is probably low by factor 3-5 relative to what will get. Will cover in later presentation. Peter unwilling to push on grating until understand OIWFS gimbal better as will dominate.
Need to discuss how get under budget?
Risk summary - most will get explored in detail during later presentations. Revisit during NIFS team discussion on last day.
See CDR docs on changes since CoDR. IFU geometry changes to facilitate diamond machining.
Doug wonders what being done about filter specs given the fact that the light bundle is small going through (4mm) Answer is nothing. Do they need to map out the filter? Can continuously move filter, so could choose sweet spot if knew where they were?
Doug asked about wedge on filters? They think neither wedge nor tilt are a problem?
John checked if will get water free fused silica - will do.
Still not clear on effect on science case of differing PSF in wavelength versus spatial direction - probably I'm just not understanding the issue? Consider case of cube from point source, zero velocity dispersion emission line - what will it look like? In x, y (and lambda) each have a PSF or MTF. In sliced direction have clean fairly sharp cut by slices, then blur information in that slice by PSF including any diffraction from fact slice is narrow. Along slice direction have PSF of spectrograph without such a big diffraction contribution, but don’t have the clean separation between slices. Is there ever a case for observing some target with 2 orientations - one rotated 90 degrees relative to the other?
Doug asked about gaps between slices - depends on masking. What about cross talk? Peter says will reduce field size - have to throw away ends of slits?
Now with grating rotation can cover K with one grating. Debate about whether need 2 K gratings with different blaze - probably not necessary but would like info on grating performance out to 2.5 microns..
Question from John about differing blaze function for polarization - Peter says wire grid means will only ever see one polarization. Will tune orientation of wire grid to optimize throughput.
I disagreed with statement in docs that wavefront budget unjustified, mostly due to slicing.
I asked about location of fly cutting geometry error in error budget - says is negligible and so doesn't appear.
Showed 2 versions of the pupil mirror array - first was meant to be final version. Had defect - had done roughing cut for final cut - had lines across. Spec <10nm roughness, got 13.8. Just got second example - is better, but not perfect? Maybe need different material? See tearouts. Doug suggests should talk with ATC about what they use? Have measurements showing pupil mirror arc very good - wanted radius 448, got 449.1+/- 0.5 mm. Easily good enough. Boundary positions within errors of measurement technique.
I asked about whether coating helps - he says doesn't change it.
Hall asked whether their machining may affect surface? He thinks not.
Will investigate other alloys - University of Bremen only now told them of their preferred substance.
Cryogenic compensation - refractive indices stuff, thermal strain stuff discussed.
Ghost images - spec based on assuming well depth 50000, brightest OH reaches that in hour.
Klaus says can calculate ghosts from detector, one sapphire to vacuum, other jump about the same. Also asked about whether should get better coatings for transmitting optics. They claim better coatings don’t gain you much.
Hall asked about Roland ghosts off gratings. Peter says asked for info from Richardson - says they didn't respond.
Peter makes good case that ghosts are OK if uniform, current detector dark was optimistic, assumption of 100 saturated OH lines is pessimistic. I'm still worried about structure - need to compare with section on point source ghosts. Will get image of detector, in some cases get focussed image of detector.
Surface roughness - I asked about measuring BRDF for hardware they now have in hand. Could get someone else to do this for them? Maybe something we could do at Durham?
John asked about bare metal versus protected? Will there be problems with grunge on surface, and how then clean? Klaus endorsed and mentioned vacuuming to get out grunge whenever remove component.
John mentioned black coating is (a) toxic and (b) rather thick. RSAA didn't have much experience with this. Klaus endorses using it on large surfaces where outgassing is an issue. He used simpler stuff on smaller surfaces.
John asked about ghosts/dust on cryostat window. Klaus does AR coat the window. Is a long way from the focus.
More rather random discussion on structure in ghosts. No clear conclusions. Gabe convincing that optical design optimal for getting rid of in focus ghosts.
John asked about feeding calibration sources before get to use GCAL. Peter not concerned because of small field, thinks can do flat field illumination.
Hall worrying about persistence of OH lines. Helps NOT to dither in wavelength. Want to have single wavelength campaigns?
Issues - grating wheel stiffness
- OIWFS gimbal mount stiffness
OIWFS - SH bit changed to remove focus option
- removed OIWFS filter wheel.
Doug not worried about reflective cold stop, Jim Oschmann (and I really mean ultimate) authority.
Deferred discussion about alignment to later time.
Machining for slicer - Doug happy as is close to the ATC design.
Doug asked about choice of disk brake approach. Peter says can't fit in geneva mechanism. RSAA experience otherwise has been with detented mechanisms, which wont apply. Peter says need this kind of mechanism for grating, then might as well reproduce.
Klaus asked about control - count stepper motor steps, can confirm at some locations with hall effect sensors. Friction disks means don’t need to rely on motor torque (backlash a worry). Friction disk also helps with cooling (particularly for grating).
Klaus asked about ball braking support - think is OK.
Issue raised about whether can roll NIRI over while at summit - answer is no.
Also asked about capturing screws - Klaus mentioned is always at least 1 loose screw in NIRI (!).
I asked about grating adjustment mechanism- they have experience and say works well cold. Also helps with thermal control of the grating.
Klaus asked why didn't use worm gear. Answer was that worm gears are 'horrible'. GNIRS has also switched. John's system also uses spur system. Same question about tapered rolling bearings. Answer - may not be able to get in the sizes needed.
Discussion on need for repeatability of the grating position - even if is not achieved then maybe can fix operationally by never moving between calibration and science.
Raynor mentioned need for heat treating grating substrate to minimise problems when cool. Peter commented these gratings smaller. Raynor also wondered about putting gratings into cassettes to avoid having to realign whenever remove grating. Answer is will not remove gratings.
Some discussion about how often can use the flip mirror without causing persistence. Answer is can use ND filter. That lead to discussion of how to get genuine ND filters - are never Neutral across J to K. Not clear NIFS team have an answer to this yet.
Klaus asked about soldering - answered with example of successful test.
Klaus asked about cold performance of fused silica. Wants to see plots.
John asked if array is centered in socket. Answer is no. Hence if change between MUX, detector etc have to do realignments. Klaus says could measure from bonds. Klaus also asked about detector tilt adjustment. Answer is will do it here with a microscope - also have 60 microns of focus depth.
Mild worry about Hilo at sea level and condensation on outer connectors and corrosion. Mark commented is OK if air conditioning will be on. Also recommends make all such connectors gold so don’t corrode. Don Hall also mentioned could tack cables to outside of the dewar to get to room temp before reach connector. Can untack at summit if desired.
Mark checked about safe handling - bare detector surface will be within 2mm of another surface so should be safe from loose screws.
John asked about baffling - is it to keep stuff out? Answer seems to be no. Will become more of a problem if go to 4 microns. John also wondered about astronomical light hitting the MUX? Klaus thinks IR filter will cure most of this.
Very nice description of alignment procedure. Certainly beats what Altair reported at CDR!
Klaus not sure whether can use MUX or not - silicon may saturate at slow frame rates? Klaus says he could try to test it. Peter says could still put a screen there and an eyepiece.
Mount distortion - two plates + towers => very rigid. For worst 15 deg swing, get 0.068 pixel shift.
Grating turret flexure - for 15 degree swing got 0.08 pixel shift (outside of budget).
Gimbal flexure - for 15 degree swing get 0.4 pixel shift. (outside of budget)
Sum up 0.132 allocated, 0.41 predicted pixel shift.
Doug commented on tests of flexure of OIWFS sensor versus NIRI focal plane mask. Doug says it met a spec.
Will get results for Gimbal but already have bits manufactured - thus will accept gimbal, wait for results and if then turn out to be a problem will have to request remanufacture.
Don’t worry about grating flexure, think fringing model is very reliable.
Lens strain. - convincing that can just machine out a bigger hole to avoid any stress for CaF.
Klaus also thinks at low temperature the apparent peak in deltaT is going to be killed by conduction.
Are going to test with a blank - we agreed this was crucial? Also helps to do slow cooling.
Can rotate in one direction. Could probably also pick up cryostat alone and put electronic on trolleys. Doug is happy with this, and can ignore other bits of flexure. 15 degree increments. Pinhole in mask wheel.
See page 145 for changes since CoDR. Klaus thinks Rockwell wont do AR for Pace. Hall agrees would introduce schedule risk. Hall also worried about putting temperature sensors on. Also thinks is also a schedule risk? On the chip carrier is OK, Rockwell said they could put on multiplexer. Hall and Hodapp worried about this. Need to clarify what Rockwell are planning on doing. Hall said HST for NICMOS used measurement of FETs? General consensus seemed to be that temperature sensor is not crucial. Bob will check with Rockwell. Also nice if could know which pins will be used for temp sensor.
Detector to cooling block connection is crucial. Klaus wondered if Rockwell could put onto chip carrier before hybridized? People think unlikely as select after hybridization. Conclude that proposed strategy is likely to be OK.
Hall asked if reference signal adds noise. Hodapp says no as aren't read out simultaneously.
I asked about teflon conducting light. Pretty sure this isn't a problem (Klaus and Peter). Jeff never did do test though.
Klaus mentioned about generating high capacitance from copper shielding? Claims was dealt with in CoDR docs.
Mark worried about clamping two teflon cables on top of each other - crosstalk. Only are parallel for 50mm and are 5mm apart. Hall mentioned crosstalk between 4 outputs. Capacitative cross coupling in the chip or alternatively could be in ribbon cable.
Mark asked about nearby mechanisms - is nothing close. Mentioned cryocooler is only 300 mm away, possible worry about magnetic coupling with SDSU? If is a problem would put magnetic shield over controller.
Hall asked if are electrically isolating cryocooler - answer is no.
I asked about words on page 165 - say with reference circuit can monitor drifts during readout. Also should be repeatable. Klaus is calm about it.
Need controller temp stable to 2 degrees for longest integration time? Think that reference will save them. Also Mark described how Gemini water doesn't really track dome temp. Has very slow variation.
Hall wondered about SDSU stability - Klaus thinks is nothing to worry about.
Klaus thinks 5 % of pixels affected by MUX glow. Those will have higher noise. Also discussed benefits of multiple ADC samples. Klaus agrees that 4 samples better than 1, but also thinks bigger number of samples wont help.
Bob asked about shielding for signal cables? Answer - only shield one end of all cables? Mark mentioned about A&G shielding of power cables was important - shielded at both ends.
Hall mentioned idea of checking - if see break in ramp should also check adjacent pixels as may have smaller break.
Mark asked about use of 4 channel coadder board. Klaus said is not really using it to its full capabilities, but do have code to read out.
John - how much of signal chain will be duplicated? Answer - Detector assembly and wiring. Wont have 2 controllers. Cant run test dewar at same time as real dewar. Klaus wondered if Gemini has spare version. Issue is extra boards - 15k US? Klaus mentioned was useful having spare - they swapped bits in and out a lot.
Mark will need to talk with them off line about grounding and electronic interface.
John is worried by lack of duplicated array controller. Doug might be willing to discuss this as a spare
John also worried about flexure test plan. Klaus agrees is labor intensive and also hard to make it safe. Klaus commented he spent a lot of time doing this and still missed the crucial orientation.
John wonders about tilt table.
Bob worried about alignment of detector on carrier. Can be miles off. Need to think about whether procedure for aligning can cope.
Klaus still worried about loss of edges of the detector. Thinks will definitely loose some, and so cut down in slit length and wavelength.
5 months to build and run NIRI software
3 months to modify cc/oiwfs/is for NIFS
1 week to add offset command
1 month (still underway) to add support for Lakeshore temp controller
not yet done grating fine control (best done with hardware) - probably only a week
Working software within 8 months of delivery from NIRI.
Issues: Does NIRI satisfy Gemini, also concern that have not tested software with the hardware.
Bob asked if had changed organization of engineering screen. Haven't actually done this yet.
Peter asked about changes to NIRI software since they copied over in December. Answer is probably yes? Change control system exists, can give them access to these. Will be a year of maintenance contract for NIRI software after acceptance tests.
See page 216 for changes since CoDR.
Bob asked about how can tell on fig 142 which arrows are data transfer. Not sure was answered in a way at least I understood.
Doug asked about why DHS inputs anything into control system. This may be an issue - will discuss offline.
Bob asking about development of VME interface code for SDSU - Answer is that they say they will copy from somewhere else.
Possible concern that NIFS can generate 3.3 MB per second, is faster than DHS can handle. They aren't worried about this.
Bob asked whether can use all of the processor RAM within VXWORKS. Answer was yes.
Showed software time taken and seemed the estimates were not bad, apart from CDR preparation.
Software issue - how much data processing should be on the IOC? They are doing some whether IGPO wants it or not. Doug checked whether can get all raw data. Answer seemed to be yes.
Doug mentioned that NIRI code is not OCS compliant - doesn't comply with ICD? When have an OCS will be an issue. He thinks can fix that pretty easily.
Doug suggests get an OCS simulator, and could use this to check if are OCS compliant.
Doug checked that had DHS simulator - answer was yes. Have built and is running but have not driven it yet.
Corinne - something about slide 33 - processing time for linear fitting. Could be blocking by data processor? Suggests investigate this more. Answer - could be a big problem. Might need second CPU. Have to try it and see if gets locked up.
Question about timing for data display after transfer to DHS? Corinne says timing includes display. Therefore is about 10 seconds from readout to display. Thinks this is reasonable.
Broke for short closed session - got 4 comments from Corinne. She will email version of these to Mark Hunten - will append these to this notes file.
Hunten - how do you handle when clutch engages - how do you know when is slipping? Answer monitor hall effect sensor to diagnose this? Have to then decide what to do about any error - have to do new calibration and rerecord positions.
Klaus asked how many positions need to encode - maximum number is 12, some are 4,6,8. 3 levels needed. Klaus mentions calibration of these voltage levels is critical, and need to recalibrate after cooldown.
Mark asking about grating motion if want to do series of close steps back and forth. Not clear if they had an answer.
Discussion about likelihood of loosing steps versus likelihood of changes in Hall effect signals. Calibration sequence takes about 10 minutes per mechanism.
John mentioned is a worry about which side of the Hall effect one epoxies. If epoxy on ceramic side, and cool rapidly may break?
Bob raised issue about running cryocoolers when computers are not going. Klaus is not likely to make these changes to NIRI. Need to find out what they are doing before will change anything on NIFS.
Mark says need changes to the cassegrain connector panel interconnections (fig 120) - will do offline.
Mark asked about what happens if Hall effect sensor went strange and need to change which one used during the night? Answer - would have to do soldering, or could use spare board. Klaus commented this is unlikely during the night, more likely after cooldown. Doug will discuss spares needed on Saturday.
Doug asked if they knew about cryocooler protection plans in event of sudden loss of gas flow. Might involve retrofit of a differential pressure switch which would trigger in event of loss of gas flow. Is a retrofit.
Mark asked about what happens if there is a loss of coolant to enclosures. Answer is that there are overtemperature switches. Mark commented this is rather a brutal approach.
Bob asked about whether temperature controllers are self tuning? Asnwer is that there are several options. Does have self tuning, but they don't expect this to be needed after they have done the tuning initially by hand.
Discussion about 1mK control of the detector. Hall comments that rely on thermal transient involved in readout has to be the same from one read to the next. Can't guarantee that detector doesn't change temperature just need it to always be the same.
Comment that need to remove paint from ISS mating surface. Is a precision surface.
If cryocooler slips too much then IGPO may have some spare.
RSAA general - are short 2 TOs for electronics, are also 2 vacancies in mechanical designers The last 2 might impact schedule if don’t get more help.
I asked about vacations. They say are already folded in. John raised concern about need to 2 controllers again. Jan says will indeed need 2.
Klaus asked to comment on cooldown plan. Agrees are success oriented. Suspect will need the contingency cooldowns. Jan confirmed that there is budget for this but will result in schedule slip. Peter raised issue about inserting and removing masks before acceptance tests. Doug says should discuss this - is likely could document stuff before AT and skip some tests then.
John asked if will want to check alignment after shipping. Peter said masks are really for focus check, can do without them.
Committee agrees takes 3 months for a cooldown cycle.
Debate on whether could actually shorten further. Peter thinks detector deliver is the ultimate driver., along with time to characterize detector.
Went through risk analysis. Raised again idea of using NIRI for PSF determination. Doug thinks this may be a problem.
Proposed to put a bit more time into grating flexures to get the think further below budget.
OIWFS setting accuracy still TBD.
Bob asked how will make the OIWFS prisms. Will be done here. Gabe thinks he can make them.
Bob asked about whether external JFETs are necessary. Klaus thinks is still TBD. Also says they don’t do any harm. Maybe include PC board option on bypassing them?
Peter raised issue about whether OIWFS scale is too coarse for AO corrected image? Doug mentioned detailed analysis done in the past by Jenkins.
Peter asked about sacrificing some of the 3 arcsec square FOV to get more distant sky measurements. Committee not particularly excited by this idea.
Doug mentioned IGPO will be generating IFU software, and would like this to be used eventually. Also says is happy with NIFS generating their own code. Ultimately will generate 2 quicklook streams - one with idle mode data, one with final data, do not want a stream of processed data.