Report on GNIRS status meeting - 5/26/99

Dave Montgomery
 
 

With comments in response by the GNIRS design team.

Present: NOAO,USGPO: MT,NG,JE,BG,LG,GM,
Gemini: DM
 

This document is identical to Dave Montgomery’s original report, with the addition of comments by the GNIRS design team. All of our added comments are in italics.

Introduction.

This meeting was held in large part to communicate the status of the GNIRS project to Gemini and to involve Gemini in identifying and resolving the design and interface issues.

This document is a short commentary from my point of view and is intended for feedback to the GNIRS team and the Gemini instrumentation manager as well as a personal record. The comments are in the order I wrote them down at the meeting, not order of importance.

Summary.

The GNIRS team presented an instrument configuration significantly more advanced than the previous informal presentation that I attended 3/15/99. There has been a great deal of progress on the internal configuration and the new design is addressing the requirements and design issues.

The comments below are intended to be helpful. As this is a snapshot of work in progress, apologies in advance for stating the obvious here and there.

Comments on presentations.

Collimator adjustment.

The GNIRS team has identified the collimator as a component that will probably require adjustment. Some kind of jacking screws were suggested. I suggest that the team design in ‘thick shims’ to allow for adjustment in the collimator should it be required instead of screws. If the screws are retained, then some form of fiducial and position indicator would be very useful for backtracking during adjustment. By way of suggestion cleaned-for-vacuum micrometer heads can be used, they provide a fine pitch and a vernier scale (but I still think thick shims are better overall).

 The “thick shim” approach is what we have advocated in our “Adjustment Strategy," SDN0003.20

Adjustment screws may be used (with dial indicators) in an alignment jig, during room-temperature alignment. Final positioning will be established by shims. The team agrees that fancy (and delicate) adjustments should not be a part of the final instrument.

Light seal on optical support box.

The use of Indium and overbolting to light seal the box seems like an unnecessary design complication and operational overhead to me. Even if the intention is to have a partially light tight outer radiation shield, I believe an effective light tight seal can be made by a rebate in the lid or the box wall (the lid overlaps the wall or the wall overlaps the lid). Only a few retaining screws would be needed. More complicated labyrinth seals could also be considered.

This suggestion of ours was motivated by the positive experience we have had with indium o-ring seals in Phoenix. However, its supporters admit that there are simpler ways, and several members of the GNIRS team were convinced that it is a significant operational complication. We do advocate high stresses in the bolted joints (where they exist) for reasons of mechanical stability and thermal transport. The plates that will be joined will be part of the main thermal-mechanical backbone of the cryogenic structure. (A casting has not been ruled out.)

Outer radiation shield.

Nothing was presented on the design of the radiation shield. During the discussion on sealing the optics box it was stated that the radiation shield would not be fully effective from the point of view of photon noise but I am not aware of the design decisions that have led to this. Is it implicit in the design or an assumption that the radiation shield will have significant leaks despite efforts to make it light tight?

 Usually, the effectiveness of the radiation shield are compromised by the following things:

1. Leakage at necessary joints on lid or removable panel(s)
2. Leakage at necessary penetrations for coldheads, optical paths and the thermally insulating support structure.
3. Conductive leakage though wires that are not adequately heat sunk to the radiation shield.
4. Radiation from hot spots on the radiation shield.

My experience is that all of these problems can be dealt with adequately with careful design of the thermal isolation trusses, cabling colds sinks, optical baffles, radiation shield thickness, superinsulation etc.

It is unfortunate that the shielding philosophy was not discussed at the meeting. We agree completely with Dave’s concerns, but think he misunderstood the role of the perhaps misnamed “radiation shield” which is not a photon seal, but a structure to intercept unacceptably high and variable thermal loads on the cold structure.

Shielding for the spectrograph is at three levels. Innermost, the optics and detector are in a light sealed cold box. This is the optical support structure. The second level of shield is a shell around that structure which is cooled by the first stage of the coolers. This shield is made of multiple parts fitted to allow removal for access. Some motors and wires are thermally connected at this stage. This shield can be thought of as a thermal buffer between the instrument bench and the (changing) ambient environment, which will affect the temperatures of the outer dewar shell and of the radiation shield. The outermost shield is what we have called the radiation shield. Minor radiation leaks in this outer layer will not affect the background inside the optical support structure. The radiation shield is doing its job if it limits heat transport to amounts less than or similar to that from other sources (e.g., dewar window and motors), a few W.

Filter wheel.

It is noted that the intention is to have 10 positions in the filter wheel, a change from 12 previously.

The question of pre-mounting the filters (as presented) verses machining some of the retaining detail in the slit wheel came up. There are pros and cons both ways but if the filter wheel is not easily removable and the technique is to replace filters for re-configuration, then the slits (filters) should be pre-mounted in a ‘cell’ as presented.

I would recommend that normal stainless steel angular contact bearings be used for the filter wheels (cleaned for vacuum use and with some form of vacuum/cryo compatible lubrication). I also suggest that the bearings and preload structure be made in the form of a cartridge, or hub. ROE have had success with this approach. The bearings mounted back to back with a 5/10 thou shim preloading the inners.

I should have asked the question, but it was not clear to me how modular the filter assemblies were. Is it possible for instance to run a filter module independently of the main structure? I think it should be and this goes for all the other mechanism assemblies, where practical.

We agree with Dave’s concern. The filters should be accessible (for changing). But having individual cells may be unnecessarily complex and bulky. The filter wheels (as indeed all of the mechanisms) will be modular and capable of being removed and reinstalled without realignment. Thus we believe that the individual filter cells will probably be unnecessary.

We think Dave’s cautions about bearings are well taken. The design team is concerned about the risks of using non-standard bearing technology, which in addition to being risky may involve some development work.

Cryo motor testing.

It is noted that plan ‘A’ is to use the Phytron motors and there is an acknowledgment that some development work will probably be required. I suggest that it will be useful to make a device that can measure motor output torque when the motor is cold. This is a direct way to diagnose thermally related problems with bearing pre-load or internal motor clearance that are not great enough to stop the motor but reduce useful torque and cause premature failure of the motor or bearings. This is probably more critical if the motors deliberately isolated from the main structure.

ROE is having some problems with mechanism drives in the cold testing of Michelle. It has not been determined yet if the problem is mechanism sticking or motor torque reduction. Ian Bryson of ROE has promised to report the findings of an investigation into the problem.

Those test results have been supplied. Thank you ROE!

I believe it is important to determine a service torque rating for the motors and pick a service factor (suggest 4) to determine the maximum allowable torque for a mechanism that will be powered by that motor. Initially, the mechanism torque at the motor can be estimated and later verified with the real hardware warm and cold, for single and multiple cycles. It should be noted that initial step speed is also a factor in determining available torque and motors should be ramped up and down.

It would be nice to show working prototypes at the RR but I for one would be satisfied with a test plan for the motors/mechanisms, especially when the GNIRS team can point out the success of other groups, particularly NIRI using the Phytron motors.

Our plan is, for example, to run the motor shafts out of the test dewar through a ferrofluidic feedthrough into an adjustable load. This will enable us to characterize the motors and mechanisms in their working environment, with respect to delivered and required torques respectively. This will be an important product of the prototype testing.

Grating Turret.

I suggest that the team consider a brake instead of a tensator spring for the anti-backlash device. The brake will be stiffer, provide damping, and a more consistent resistance. Disc and caliper brakes are used in all the rotary mechanism of CGS4. The brake pads are Vespel, rotors are brass, Aluminum brake calipers are ‘floating’ with machined in flexures with adjustable pre-load springs. As all mechanisms are nominally balanced, the brake force is not large.

We think Dave’s suggestion is worth trying. This is something that will be convenient to do in the prototype testing.

WFS sub assembly.

Just a suggestion, but it would alleviate some integration problems if the WFS components were mounted and pre-aligned on a sub plate. Perhaps the IfA could do this, but even if they do not, it will still give scheduling flexibility if the WFS can be tested independently of the MOB.

This is certainly what we intend to do. What we discussed at the meeting, which may have misled Dave as to our intentions, is whether additional modularity is desirable (we think it is) which permits removing single OIWFS assemblies without having to remove the entire OIWFS.

MOB thermally isolating support.

I did not see anything presented on the current design for this (the schedule has 20% complete so maybe not enough to present). The implementation of this will influence the design of the MOB, radiation shield and vacuum vessel. It will also affect access and serviceability.

The meeting with Dave was scheduled before these matters had been thoroughly discussed. Since then all have been addressed.

MOB.

Having a central, contiguous structure with modules bolted to reference surfaces on it is definitely the way to go. The structure itself is still under development and Larry intends to stiffen the structure with webs. It still looks like an ‘I’ beam in section though (with the covers removed) and this is not good for torsional stiffness. How badly this affects alignment and flexure depends a lot on the load path from the thermally isolating support structure. I would prefer to see a box-like structure. And not rely on the covers for stiffness.

With the wavefront sensor installed on the top (installed with numerous screws) the structure is a box. The walls are covers but are heavy and integral to the structure.

I have no experience with constructing a MOB by bolting plates together, but I know Phoenix uses this technique. My fears would be reduced thermal conductivity and creep of the bolted joints on thermal cycling/transient heatloads causing alignment changes between cooldowns. Increased hysteresis under self weight loads is also a possibility. ROE uses castings and IfA have machined from solid in the case of NIRI to avoid these problems (at least in part).

Castings have not been ruled out. On the other hand the experience with bolted structures (e.g. Phoenix) has confirmed that they can be stable under thermal cycling. We understand that castings differ slightly in CTE from machined 6061 parts; since we have also raised the issue of 6063 for the bench we should look at the general issue of CTE mismatch.

Drawing tracking.

The intention is to assign and track drawings and part status using Microsoft Access and a sample report was shown. This can also be used to track the mass budget.  This looks like a neat way to handle book keeping of the drawings.

Thermal analysis.

The first order thermal analysis is thorough and well thought out.

I was surprised that a preliminary analysis in SDN007.01 indicated that transient thermal effects could be significant if the motors were bolted directly to the structure.  As suggested in this document the intent is to design motor drives that are thermally isolated from the structure, short drive shafts are employed and some form of wicking. It will be prudent to test the motors with a similar stand off, drive shaft and heat sink arrangement to ensure that the heat sinking is adequate from the motors point of view and acceptable in terms of the thermal background. The drive shaft and couplings will need to be very stiff to avoid wind up.

We are doing additional thermal analysis at about the same level as SDN007.1 to further the address the potential problems from thermal transients. We think that the reason why this has not been identified as a problem in previous instruments is primarily the much tighter tolerances in GNIRS.

This will be another subject which will be more easily addressed via prototype testing than thermal modeling, because of the complexity of the thermal paths and the uncertain loads.

I believe that in general the global stiffness of most structures and the transient temperature distribution will reduce the effects of motor operation sufficiently to allow motors to be bolted directly to the structure. I agree that this would have to be confirmed with a detailed analysis.

We believe it is possible to have the majority of the heat generated by a motor delivered to the cryocoolers directly without compromising the stiffness of the mechanical mounts. We see this as a conservative approach given substantial variation and uncertainty in the level of heat generation by the motors.

The number of cold heads and cooldown times were discussed, as was the possibility of LN pre-cool. It was stated that adequate cooling capacity was a pre-requisite for thermal control, but some basic calculations should be done to justify the four cold heads and to estimate cooldown times. This is planned in SDN0007.02.

It is stated that work done so far indicates that using two cold heads would give marginal performance at steady state.

Considerable work on these areas has been done since the meeting with Dave. See SDNs 0007.01-.03

Camera lens assemblies.

I liked the design for the barrel assemblies and I had one comment on the radial seats that should probably be milled flats rather than radiused. Assuring repeatability is just a question of getting the axial and radial pre-loads right to keep the lens in the ‘V’ and on the ‘three point support’. If the integrated expansion co-efficients show that the axial displacements are OK, then you are in. If it is marginal, you can deliberately displace the CaF or BaF lens to half the relative movement between them (you can do this if the ‘Vs’ are milled).

The radius was to diminish stress and increase the area of contact for thermal contact. Given the inflexibility of the materials in contact it is not clear that this helps. The thermal stress may depend more on the total force rather than the nominal area of contact, given the stiffness of the materials. The plan is to continue work on modeling the thermal contact, stress and geometrical definition issues with the lens mounts.

One other possibility that I have used to mount ZnSe lenses in Aluminum: You could define the radial position athermally by three ‘planetary’ nylon cylinders locating in a simple bore. The nominal diameters of the cylinders can be calculated to compensate between the lens material and the barrel material.

Athermalization at room and operating temperatures is attractive but it is hard to deal with stresses that can develop during thermal transients (cooldown and warm-up) when the system is not isothermal.

Slit slide mechanism.

I think the approach should be to count on balancing this mechanism actively. The use of ‘V’ type cam followers can cut the number of followers in two and the shaft mounts can be drilled horizontally or vertically.

The moving mass of this mechanism is fairly heavy. Given the precision we require, Dave’s concern is well placed. Inclusion of a counterbalance has certainly always been an option. We will look at the balance weight again before we are done with the design. This has also always been one of the prime targets of the planned prototype.

We have preferred not to use the V rollers because we think the cylinders are more accurate.

Optical status

The changes from CDR were noted, I believe these are on the GNIRS web site.

Just a reminder that an Acquisition mirror deployment was added, grating turret reduced to three positions, no extra drive required.

Actually, the net change is one motor – acquisition and focus mechanisms have been added, but the grating drive has gone from two axes to one.

Schedule and status.

The tracking information that was presented was detailed and up to date including the project plan to RR. The team acknowledge that the RR date is a given although there was some thought expressed that more time is needed to advance certain areas of the work.

I think the team should show at the RR that the functional requirements and ICDs and evolving design issues show every indication of being met. Issues that were stumbling blocks in the past review should be covered such as handling (0% on the plan at present, Ming scheduled to start soon), mass, serviceability.

At the time of the meeting with Dave, the engineering team had still not dealt with a number of major issues. It was clear then, that the goal of the RR would be to present a plausible path to a successful solution of all engineering challenges, but that it would not be late enough in the design and prototyping process to be considered a Critical Design Review.

It is also important to show on the programmatic side, adequate project tracking and appropriate response to reviews.

There is a projected deficiency in software effort through the year that is being addressed.

Issues of software planning have had to be postponed for lack of suitable manpower. There is still time to plan and execute the tasks but we are concerned that software not become a pace-setting item.
 
 

NOAO Intranet Services

NOAO Copyright  Statement