Frank Melsheimer's Visit Notes March 14, 1995 Attending: Jack Osborne, Eric James, Sandy Faber, Garth Illingworth (part time), Neal Jern, Frank Melsheimer, David Cowley Agenda 1. Review of the instrument at the time of the PDR 2. Review of the recommended changes 3. The current situation 4. Structure, drive, and bearings 5. Slit mask handler 6. Grating slide David covered the first two agenda items in the opening remarks. The source of the information was the PDR report, and the PDR committee's report. Frank elaborated a little on the PDR committee's comment about settings for the grating angles, by saying it was his opinion that a continuously encoded grating position could be as cheap or cheaper than a system using fixed positions, or "detents". As the grating slide/handler and the number of grating positions possible was of particular concern, we addressed the final three items on the agenda in reverse order. Gratings In addition to considering the "continuous motion" version to setting the grating angles, Garth wants us to consider the possibility of having 5 positions (one 8*12 grating, three 6*8 gratings and a flat mirror). Four positions rather than three is more than desirable. Two major variations of the grating slide where discussed; one a slide with autonomous encoders and drives for each grating, and one more of a grating handler that would position cells into a single drive and encoding unit. Both have advantages and problems. The grating slide is appealing because it probably is the most compact and would allow different sized holes for the 8*12 and the 6*8 gratings and even a different sized hole for the mirror. As a majority of the gratings would not be normally changed out, the positioning and encoding of their angles would be more constant. It is possible that the grating slide will fit in the congested area between the drive disk and the slit mask handler better than the alternative. Problems with the grating slide approach are the cost of the 4 sets of angular drive/encoders, the number of cables that would have to move with the slide, and that we would have to have a fairly clever mechanism to protect the drive/encoder units during installation and extraction of the gratings. A grating cell handler would have only one grating angle drive and encoder. Each cell would be placed into the drive/encoder kinematic mount by a handler mechanism. Only one drive encoder unit would be required. Manual extraction and insertion of grating cells into the handler would be simplified as you would not be mating to a precision mechanism. The exterior of each of the cells would be identical, thus somewhat simplifying manufacture. It would be relatively simple to deal with custom gratings as we would have a "standard" grating cell in which to mount. Also it seems that the total number of kinematic mounts would be reduced. Problems include: the size in the axis direction of the spectrograph would likely be greater than for the slide version; the exterior of each cell would have to be identical, thus would need to be sized for the 8*12 grating; as the size is greater, the moving mass would likely be greater. We agreed to investigate both alternatives. Slit Mask Handler Jack has investigated several versions, but only the "mill" allows us to be within 2 to 3 inches from the nominal focus of the telescope. The problem with the mill is that it is fairly large, and potentially interferes with both the drive disk and the grating slide. Frank proposed a caterpillar track version that would reduce the size of the handler is one direction. A couple versions of the track were explored on the chalkboard. It seemed that this alternative might avoid interferences with the drive disk and/or the grating slide. It was undetermined if this alternative was more or less complicated than the "mill". We agreed to investigate the caterpillar alternative with a design study. Structure and Drive. Three major areas were identified and dealt with somewhat separately: structure, drive disk placement, and drive. Frank stated that we wanted to design a monocoque structure such as the tube we mathematically modelled, rather than a space frame. He stated that a monocoque structure would be much stiffer in our application. We discussed at some length the materials and construction methods. They were basically Riveted Aluminum, All Riveted Steel, and Welded Steel. The pros and cons of each method are listed below. Riveted Aluminum Riveted Steel Welded Steel V Low Distortion V Low Distortion High dist Low Skill level Low Skill level High Skill No Heat Treat No Heat Treat Heat Treat High Mat Cost Low Mat Cost Low Mat Cost High Local Stiff Low local Stiff Low local Stiff Max thick 3/16 Max thick 3/16 Min 1/8 Splice end & Circ Splice end Splice end can add bulkhead can add bulkhead have to rivet Blister Difficult Blister Difficult Blister easy no experience no experience experienced In the end it seemed that a welded steel structure was the way to go, largely because we are experienced making these types of structures. It also seemed unlikely that we would get the required stiffness out of the Aluminum. (In fact a quick analysis shows that the aluminum would flex about 3 times as much.) In a monocoque structure, it is very important to apply all the loads properly to the skin. If you apply a load out of plane, or at a concentrated point, local distortions will be high. All the significant loads will need to be applied to the structure through a bulkhead arrangement. One possible construction sequence would have the instrument structure built in two parts, one in front of the drive ring and one behind. These two parts need not be identical in diameter, or even in structure if that were to serve a purpose. As the TV cameras do not necessarily have to be as stiffly held as the major optical parts (due to the presence of a reticle) it may be advantageous to have a space frame or square structure on the telescope side of the drive disk. Drive Disk Placement It was apparent that with a large diameter tube, the stiffness of the structure likely exceeded the stiffness of the individual mounts. It was also apparent that within some bounds you could tune the structure so that the 0 slope point occurred at a particular place in the structure. Other points on the structure will experience a reversal in slope as the instrument turns over 180 degrees, thus deflections and slopes need to be regarded as "plus/minus" the value calculated. The maximum slope calculated for base design was 0.4 arc seconds. A little thought was given to the point in the structure that we would like to have no slope. This point is apparently the mounts for the gratings. We discussed having the drive disk inside the hub of the Nasmyth bearing, and decided because of the size it would need to shrink to, and the moments it would place in the sub structure, this was not a very workable idea. Rotational Drive Frank gave some advice about the drive disk. That is if we are going to use a thin disk and harden it we need to remove at least 1/8 or 3/16th from each surface or it will crack in the final hardening process. The reason we would harden the disk is to allow contact pressures of around 100,000 psi between the drive roller and the disk. An alternative is to use a combination larger roller and thicker disk and not put it through the final hardening process. In this case you might drop the contact pressure to 40 to 50,000 psi. Frank recommended 4140 alloy steel for this disk instead of 1044 steel. It costs more but has other benefits. The 4140 has a yield strength of 65,000 psi without heat treatment. After burning the shape it would be normalized, then blanchard ground on both sides, machined (by lathe) on the O.D. and I.D. and lastly, ground on the O.D. The contact force would remain the same, thus the drive force would not change. We will investigate and possibility of using a thicker drive disk. Frank also gave us some advise about designing the rear bearing and seals.