SYSTEM DESIGN NOTE

SDN0003.20 - Adjustment Strategy (Preliminary) - Ver. 2

 
 
 
 

Prepared by	Date	Approved by	Date	Rev.	Rev Date
B. Gregory	6/2/99	N. Gaughan	6/8/99

 

Introduction

In what follows, we have tried to describe an overall adjustment scheme with the chief aim of identifying where we do and where we do not require some kind of alignment adjustability built into the instrument. There are several levels of adjustability: manufacture only (no adjustability); shim; and adjuster. An adjuster is a permanently installed mechanism that permits adjustment without disassembly. Adjusters are both expensive and points of vulnerability (in flexure, motion in thermal cycling, and in potential mechanical failure) so reducing the number of adjusters to a minimum is an important goal. We don’t mean to suggest that a shim is an afterthought; where indicated the design should provide a means for doing it reliably and without major disassembly or re-machining of major components,

Inevitably the instrument will be misaligned at some level. The point is to define the level of alignment required. Most of the critical alignments are internal to modules (Offner, cameras). With the possible exception of the collimator, most of the other misalignments results in shifts of the image on the field, or shifts of the beam on the cameras which, while not ideal, do not affect the scientific usability of the instrument.

We circulate this in the interest of eliciting comments about the strategy, or to elicit alternative strategies and to check the logic in setting levels of tolerance. Ming will want to pay particular attention to these numbers.

Focal Planes

We take the focal planes to be determined ultimately by the plane of the slit edges. That Focal Plane (FP), the Slit FP, will be reflected back through the Offner onto the pickoff mirror which becomes the Offner Input FP. The Offner Input FP must coincide with the Telescope FP. The OIWFS will be focussed at integration so that its focus coincides with that of the Offner Input FP. To bring the Telescope FP and the Offner Input FP into coincidence, the entire instrument will be shimmed in the z direction. The BFL from the ISS to the focal plane is 300 mm, +/- 1mm, with the guts cold, obviously. Jay mentions that the tight tolerance, coupled with the fact that the laboratory test temperature is high compared to the operating range of temperatures on Mauna Kea, means that we should take ambient temperature into account when doing the adjustment. The strategy is to leave the mounting pad thickness TBD until after the entire instrument is assembled and fully tested, cold, before adjusting  it to produce the desired BFL.

In the other direction from the fiducial slit FP is the science detector FP, re-imaged by the collimator and camera. In this case a focus adjustment is available to bring them into coincidence.

To make the BFL measurement, and many others, it will be useful to have a simple telescope simulator on an x-y stage mounted on the ISS simulator. (See Appendix A)

Optical axis position

The position of the optical axis will be defined by the center of the slit. It is a mechanism whose position is awkward to adjust, post facto, and it is associated naturally with the fiducial beam direction defined as follows:

Beam direction

Some long baseline mechanical line, (whose position is presumably relatively difficult to correct by moving it), should define the beam direction of the instrument. The obvious choice is the line between the slit center and the mechanical center of the collimator mirror (not the vertex of the parent paraboloid). The practical problem then is to work back from this fiducial beam direction to the input of the instrument. The principle is to direct a beam into the instrument, through the center of the cold stop and adjust the foreoptics so that beam illuminates the collimator symmetrically. This can be done to about 0.5mm precision (at a mask located on the collimator). This corresponds to a precision of  3.3e-04 radian = 69 arcsec. An oversized beam (f/10, say) from the stellar projector, focussed at the input focal plane of the Offner, will be trimmed to f/16 by the stop at the Offner secondary to f/16 and continue to the collimator, where the beam position can be checked.  (Note that the beam is not being defined by the alignment of the projector but by the position of the Offner cold-stop). If the beam placement is within the 0.5mm tolerance; the Offner is consistent with our fiducial and indeed, it (together with the slit center) now become the beam-defining elements.

If the beam is not within that tolerance, the Offner should be shimmed in angle so that it is.

Window

The window is unique in being the only optical component which is not directly mounted on the cold bench. It is a lens (though a weak one) so its mounting has optical consequences. Given the large build up of tolerances there will be between the cold structure and the cold mass it is important to consider two matters:
 

• Flexure of the cold mass with respect to the window. This will result in a deviation of the beam and vignetting on the cold stop.
• Alignment of the window. This is required, or at least desirable, to ensure that the instrument beam emerges parallel to the nominal instrument axis. This is relevant to the present document.

It appears however, that the optical power of the lens/window is so weak as to make the effect of window decenter irrelevant. A decenter of 1mm appears to be tolerable, though this is inconsistent with Ming’s tolerance of 0.1mm. This should be resolved.)

Were it important, tt is a simple matter to have the window mounted in a cell that can be translated laterally small amounts (<1mm). (The design should be such that the adjustment is not disturbed if the window is removed and replaced.) At final assembly, the window would be adjusted so that the beam direction emerges parallel to the original beam direction.

It should be pointed out that this is a redundant adjustment, in effect, to that provided by the proposed shimming of the ISS mounting pads. An alternative approach (of the tolerance buildup were modest) would be to have a completely fixed window and correct for the beam angle by shimming the ISS. This is a less ideal solution: it leaves the principle ray in a bent state. It is correcting one misalignment with another. However, the effect appears to be negligible.

The window should be adjustable only if it is required.

Lateral Centering of the Instrument

Gemini requires that the center of the instrument input focal plane coincide with that of the telescope to within 2 mm (Estimate by Jay Elias; this should be checked). Again because of the buildup of tolerances in the dewar, it will be difficult to fabricate and mount the cold mass so that its optical axis coincides in position with the optical axis of the telescope to less than about 1mm. One could shim the point of support of the cold mass with respect to the interior of the dewar bulkhead, but we propose, again, to use the ISS support points as the place at which the final adustment of lateral placement (~<1mm) will be made.

The adjustments of the beam direction, lateral position of the instrument, and window centering, would all use the telescope beam simulator on the x-y-z stage on the ISS simulator. (Appendix A).

Foreoptics

We take the Offner module to be a  module with no internal adjustments. It will have been assembled, adjusted and tested as a module, as described in Appendix B. The pick-off mirror will be part of Offner module, rigidly fixed to it. (No adjustments). The module will be fixed in the dewar, but can be shimmed to put the Offner Output FP at the slit edges. As described above, the Offner beam axis can now be taken to be the beam direction of the instrument.

The effect of the fold mirrors associated with the Offner module should be carefully taken into account. The assumption here is the errors produced by those fold mirrors, by design and precision fabrication, will be small enough to be corrected by the alignment step at the ISS described above. That is the mirrors must produce no greater beam deviation than 0.4mrad. (Engineers: tell us if this is un-reasonable.)

Angular Alignment of Offner beam with Telescope beam

The tilt adjustment can be accomplished in the same step as the BFL adjustment, by adjusting the mounting pads to the ISS.  We propose to use this method for the following reasons, which are basically the same as for making the final focus of the instrument at the ISS interface:
 

• Tolerance build up in the several subassemblies between the Offner and the entire instrument mounted on the ISS is such that it will be very uncertain to determine the required beam adjustment until the entire instrument is assembled. At that point, access to the pickoff mirror adjustment requires disassembly of the instrument.
• Eliminating the tilt adjustment at the pickoff mirror simplifies its design and makes the unit mechanically more robust.

One assumption implicit here is that the fold mirrors in the Offner assembly are assembled to sufficient precision in angle that the input beam to the Offner is parallel to the output beam to a precision that makes it practical to do this. Let us set the maximum angular adjustment to be 0.5mm at one of the ISS mounting pads. This corresponds to an angle of 4.2e-04 radians. (Assuming 1.2m width of the ISS.) This, then, is the required precision of the parallelism between Offner input and output beams.

So far:
 

• Offner is shimmable, in z for focus (telescope focal plane (OIWFS) to Offner input focal plane)
• The pickoff mirror has no adjustments.
• TSS mounting pads have to be adjusted to tilt instrument beam to adjust the BFL to match the specification and to match telescope beam direction.
• The OIWFS is assumed to be shimmable or otherwise focusable to have its input FP coincident with the Offner Input FP.

Having made these adjustments, the Slit FP is reflected back onto the pickoff mirror. The mapping of the slit center to the center of the pickoff mirror is not a high precision requirement, but the pickoff mirror is only slightly oversize (how much ???) so it can’t be sloppy either.

The Offner input and output beam angles (including the effect of the flats) have been measured to be parallel to each other in independent testing. (See Appendix B.)

Slit Mechanism

The slit mechanism is assumed to have no adjustments nor need it be shimmed in z; it determines the focal planes of the rest of the instrument. Tilt of the slit plane around the local x-axis (the axis normal both to the optical axis and the slit) will cause a tilt of the focal plane at the science detector. It should be small, by fabrication  (0.5mm error from one end of the slit to the other  = ~ 8 mrad. Is it a problem at this level? If it is a problem, the slit assembly should be shimmed.)

Collimator

The collimator is heavy and difficult to support rigidly with adjusters. We avoid adjustment mechanisms and provide for shims as it is likely they will be required. We propose that shims be used to:

a.) Move the mirror axially to place the unique focal point of the paraboloidal collimator mirror on the slit plane (focus).

And both

b.) Tilt the collimator mirror
c.) Rotate the collimator about its mechanical (not optical) center.

to bring the unique focal point of the collimator to the center of the slit. The rotation angle is likely to be adequately defined by mechanical assembly alone. (What tolerance on angle (roll; about z-axis) is required?)

These adjustments involve adjustments in the thickness of three defining points behind the mirror (or something mechanically equivalent). The mechanical center of the mirror is not moved laterally as it defines the internal optical axis of the instrument. Let us suppose until proven otherwise that a shim in rotation is not required.

The adustments should be done by using a flat mirror (>100mm diameter; of excellent quality), placed in the vicinity of the prism turret, to autocollimate the beam from the slit and collimator. The image of the slit reflected back to the slit can be inspected (using, for example, a beam-splitter) to determine focus and elimination of coma (tilt).

Alternatively, an interferometer could be used to examine the wavefront from the slit and collimator. To gain access to the beam, the interferometer could look in through the camera turret, via the flat acquisition mirror.

Thus: No adjusters desired. Some warm shimming almost certainly required.

Grating, Prism modules

These actually should not require great precision in placement. The optics in them act like mirrors. Say +/-1mm in lateral placement would be tolerable. Angular displacment is another matter but still not very stringent: grating:

 tla – affects nominal height of spectrum on chip; 10 pixels ok; this translates to about 0.1 mrad.
 tlb – affects zero of spectrum; completely compensated by rotation of grating.
 tlc – affects orientation of dispersion direction on chip. Also affects motion of spectrum on chip as wavelength is changed. Say 10 pixels tilt across spectrum; 0.5 degrees produces this across the chip, but the center wavelength can be changed by tilting the grating ~10 times the spectral width represented by the chip, so the requirement on alignment is about 10x stronger, say .05 degrees ~ 0.1 mrad.

No adjustments screws permitted! Whatever shimming is required can be achieved by adjusting the height of hard point supports. Provision for doing that should be designed in.

Cameras, on turret.

Treat as a module. We should take as a goal that the module not require alignment in angle or in displacement. Displacement results in slight shift of beam on optics. Beam is ~100 mm in diam. 1mm shift will have acceptable effects (??? check this).

Angular mis-alignment of cameras shifts the nominal center of the field on the chip. This can be compensated by shifting the chip. Again, a 1% field shift is almost certainly tolerable. This is 10 pixels or 0.3mm.

Tilt of the camera axis results in a tilted focal plane. The defocus must be zero to within about 12 microns (worst case; from Ming). This implies a FP tilt of no more than 0.012/[(1.4*1024*.027)/2] = 0.6 mrad

No adjustments permitted!  Some shimming required? Probably not.

Detector Module

It may be convenient to shim the detector module in x-y (rather than the camera turret) to bring the field center to the center of the chip. Focus is, of course not a problem. Angular adjustment both pitch and yaw as well as roll will very likely be required (by shimming). I propose that the detector be shimmed in angle rather than the camera turret, for the obvious mechanical reason that the detector module is smaller and lighter.

In summary:

In the above all adjusters have been eliminated!! Shimming may be required internal to the modules:

Offner
Camera
Gratings and prisms on turrets

Assuming this is done so that the modules are internally well aligned, the modules themselves will require some shimming, as follows:

a.) TSS mounts (in focus, possibly in both x,y and tilt
b.) Offner, in focus only, possibly in tilt
c.) Collimator: tilt, defocus. No x-y: collimator center defines beam.
d.) Prism turret, grating turret. Possibly some shimming in angle but very likely not. (Beam translation of less than 0.5 mm is ok.
    Probably nothing else.
e.) Detector mount.  Shim in yaw, pitch and roll (sorry about that!)

Appendix A  - Tools
 

• ISS simulator – This would permit mounting the GNIRS on it for flexure measurements and for beam projection measurements.
• Flexure rig – Apparently a plausible candidate has been identified in a heavy duty welding positioner.
• CCD with moderately good tools for image inspection (e.g. IRAF). The interface between the CCD and the display/analysis software should be more-or-less real time. The CCD should be capable of being mounted on the focus stage of the GNIRS.
• Star projector – A simple projector which projects a very small field image with am f/16 beam . This would be carried by an:
• x-y-z stage –  This would be mounted on the ISS simulator. The x-y motion is important for precise centering. The z-stage is important for precise setting of the BFL which is used to measure and adjust the BFL of the instrument.

Not written yet

Appendix C – Internal Camera alignment – incomplete!!

Cameras, internal: presumably require high internal alignment precision. This will be achieved and tested warm. No adjusters! Provision for shims? This requires more analysis. Ideally, the optical elements will have external features (edges) which enable them to be centered, with zero tilt to the required tolerance.

Camera-to-camera alignment:

Focus: not required (detector focus  available)
Field center: 10 pixels ???
Focal plane tilt: worst differential tilt 0.6 mrad. ???
 

Brooke Gregory
27-May-99
8:52 AM
 
 
 
 

---

---

If you have any questions or suggestions regarding this website, please contact Melissa Bowersock.

NOAO Intranet Services

NOAO Copyright  Statement