Pre-Fabrication
Review Committee Concerns
and GNIRS Responses
August
15, 2000
These concerns were within the previous Pre-Fabrication Review Report, and here have been pulled out in order to address them more specifically. The GNIRS response follows each concern, and is italicized.
1. Given the current state of the flexure analysis, the committee can only comment that GNIRS appears to be “headed in the right direction” regarding flexure and early indications are positive but it is premature to assume that the aggressive 0.1 pixel flexure per 15º attitude change specification will be met by this design. Too little analysis was presented to make a definitive conclusion about meeting this spec and we anticipate a lot of work required to fine tune the bench design and incrementally increase the model’s complexity by adding additional loads, optics, mechanisms building in the influence function for all of the optics, etc.
The current bench
analysis does include the mass of all mechanisms and optics attached to the
bench. The mechanisms themselves,
however, are currently modeled as being completely rigid.
Therefore, the bench analysis to date will predict image degradation due
solely to flexure of the pre and post slit bench structures.
It does not include image degradation due to flexure of the mechanisms.
Including the mechanisms in the model is possible, but represents a very significant effort; Ed Hileman estimates between 3 and 4 man-months and some computer upgrade would be required. It makes more sense to use the direct measurements derived from the prototypes and the camera turret development to include mechanism flexure in the error budget.
Ed is
currently reducing the output deflection data from NASTRAN into a format Ming
can use to determine the image degradation with his ray trace program.
This effort will be finished (by decree) August 15, 2000, which appears
to be realistic.
2. The GNIRS team should further develop the fall back plan of active flexure compensation, based upon LUTs, in the event the actual flexure of the system fails to meet spec by a large margin. Specifically, they should identify an optimal compensator in the system (e.g., XY detector stage, move the collimator, etc.) and determine if they have adequate space/mass budget available to retrofit a mechanism it proves necessary to provide flexure compensation.
A
fall back plan has been adopted, namely correction at the collimator (as
suggested by Tom O'Brien). We are not pursuing the implementation unless and
until evidence shows it is needed.
There is, in principle, enough space at this location. The design of the compensator has not been done, but it is worth emphasizing that it should be very simple (small travel at low fractional resolution). One could certainly write a straw-man specification for such a device if needed but it seems more appropriate to await the imminent completion of the flexure analysis.
3.
The GNIRS team should consider running ultimate lifetime tests using their
prototypes after they are done using them to support the design effort – in
effect running the mechanisms until they break to learn about the intrinsic
weaknesses in the designs that wouldn’t emerge from the planned tests.
We have decided not to run the
lifetime tests on the prototypes, on the ground that the prototypes are not
quite representative enough of the real mechanisms for the tests to be
particularly helpful. We have carried out inspections on the mechanisms
(disassembly) and have seen no wear problems for use that represents >10% of
a "lifetime". Our current test set-up doesn't lend it itself to
prolonged operation without significant manpower, so this is not a trivial
effort.
Specifically,
the homemade bearings for the rotary prototype, which were the main cause for
concern, are not being used in the final mechanisms, so wear there (though none
was seen) is not relevant.
On
the linear prototype, the individual rollers are replaced by multiple crossed
rollers, which should have better performance in several regards; again, no wear
was seen in the present set-up.
The
vespel/metal interfaces have not shown wear.
4.
We recommend instead going straight to prototype #2, as we believe it has a
higher probability of meeting performance and lifetime expectations and has been
proven to work in other cryogenic instruments.
We are now proceeding directly to
the camera turret design using roller bearings, with the idea that it is not a
prototype. We have chosen not to adopt the angular contact bearings because our
measurements on the rotary prototype have shown that ball bearings have poor
conduction, and because there are concerns about load capacity for the camera
turret.
In essence, we have enough
confidence in the design to adopt it, but we are accelerating the fabrication
and test of the camera hub in order to ensure that we have time to detect and
remedy any deficiencies.
5.
We do not see any need for additional prototype testing of the lens mount, given
that this design has been used successfully in many other instruments and the
lessons which would be learned from any more testing can no doubt be extracted
from the experiences of others. We recommend contacting Klaus Hodapp to
understand exactly what difficulties he has had with this design as a rapid
means of closing out this effort.
We
have finished these tests. Klaus's design is somewhat different.
6.
If NOAO wants to pursue the idea of IfA installing elements in this module and
testing it as a unit before it’s shipped, that should be explored quickly as
OIWFS parts are scheduled to be built this summer at the IfA. Alternatively,
Klaus and/or Jeff can come on-site and help integrate/test the parts within
GNIRS.
We are pursuing this. Klaus wants to
see how NIRI does at the telescope before proceeding much further. This appears
to be mainly because he is concerned that additional alignment steps may be
needed.
7. Consider the use of invar bushings on the bolts that “permanently” connect parts of the main bench as a means of providing increased cold preload on that critical joint.
We are following this suggestion.
Rick and Ed are currently incorporating partially compensating collars in the
bench design. Ed has looked closely
at how much preload is lost vs collar length and has come to a suitable
compromise.
8. We expect that the architecture described will be able to meet existing ICDs and support all key functionality from both standalone engineering and observatory operations perspectives. That said we remind the GNIRS team that they must conform to all principal system interfaces.
We
are reminded.
9.
The GNIRS team need to clarify what will be delivered in the engineering system,
as this is a deliverable and Gemini would like to know if it requires additional
expertise, software licenses, etc. beyond what we normally expect to be required
to support our facility instrumentation.
At this time, we do not envision
delivering a system that would require any special software licenses.
It may be desirable to use an improved version of the EPICS display
manager, or to incorporate some new TCL code. However, we will neither deliver
nor require any software that is not a part of the standard Gemini environment
without approval from Gemini.
10.
GNIRS needs to thoroughly document the S/W that operates below the proposed
“thin EPICS layer”, again to permit long-term support by Gemini’s software
staff of GNIRS. This documentation is normally expected to exist in the form of
the software maintenance manual, which is already a contractual deliverable.
This is correct.
11. We suggest rectifying the discrepancy between CAS (Central Accounting) and Neil’s costs, by reporting in real dollar values to avoid in the future the confusion evident during the review. Specifically, we recommend reconciling the difference between accounting for scientists’ time between the cost schedules maintained. This will yield actual final total costs for various components of the program, which are arguably some of the most important metrics for gauging GNIRS status.
We are struggling to do this. We
have subtracted out all scientist labor charges from our planning to better
reflect what the actual charges represent.
We are still working with CAS on a more permanent solution.
12.
We believe GNIRS should release some parts, which are not very expensive or
complex, for fabrication as quickly as possible to (1) test the entire off-site
fabrication process with a vendor and (2) stimulate a change in the psychology
of the team, from one of only designing to actually building the instrument.
The prototype fabrication was
deliberately intended to "test the process" (and has done so). We are
releasing other parts as ready and have completed the (admittedly not critical)
environmental cover up to steps preceding final assembly and test.
13.
We suggest the use of an engineer independent of the GNIRS team (e.g.,
Earl Pearson) to double check drawings that depict key interfaces between the
main bench and mechanisms and optical assemblies. A useful exercise may be to
have him “reverse engineer” the location of the chief ray throughout the
opto-mechanical assembly, and check if his analysis of its location matches what
was used in designing the instrument.
We have effectively done this using our “new” mechanical engineer. Also, one of the points of the 3-D modeling is to avoid the dependence on building up the ray-trace from individual drawings.
Home | Management | System Design Notes | Requirements | Configuration Images | Subsystems | Archives | Staff | FTP Site | Internal Information |
 NOAO
Intranet Services |
 |
 |
Statement |
National Optical Astronomy Observatories, 950 North Cherry Avenue, P.O.
Box 26732, Tucson, Arizona 85726,
Phone: (520) 318-8000, Fax: (520) 318-8360