SDN0002

SYSTEM DESIGN NOTE

SDN0006.03 - Use of Castings in GNIRS Structure

Prepared by	Date	Approved by	Date	Rev.	Rev Date
Jay Elias	11/1/99	N. Gaughan	11/1/99

1. Introduction

NOAO has not previously used castings in cryogenic instruments, but has used them successfully in other applications. Other large cryogenic instruments have used castings for their cold structures (e.g., CGS4, built at ROE).

The purpose of this note is to discuss the advantages and disadvantages of using castings in GNIRS, and to set out the criteria for deciding where to use them.

2. Applicable Requirements

As the discussion below (section 3) makes clear, the most appropriate use of castings in the instrument would be in large structures which do not require a lot of detailed finish machining. There are two areas roughly meeting these criteria: the cold bench structure and the dewar vacuum vessel. The requirements in these two areas are somewhat different.

2.1 Cold Bench Requirements

The cold bench structure must meet requirements in several areas:

1. Rigidity. There is a tight requirement on instrument internal flexure, of which bench flexure is (potentially) a significant component. This implies the use of reinforcements, ribs, etc. to maintain strength. Also, use of bolted joints should be minimized.

2. Light tight. The bench structure also serves as the primary seal against external radiation. This requires the use of light seals at all joints in the structure, as leaks through joints formed by two flat surfaces will be significant. Thus any joint involves more machining than just producing a pair of flat surfaces. Cracks and holes through which light can leak are also undesirable.

3. Good thermal conductivity. The overall thermal design is intended to avoid making the bench conductivity a critical factor in either cool-down time or thermal stability. Even so, good conductivity is desirable as it simplifies design of the pre-cool system and thermal distribution system. Thermal conductivity is reduced across joints and through thin cross-sections.

4. Low weight. The bench should meet its weight budget, and cannot be arbitrarily heavy.

5. Minimal differential contractions. The optics up to the cameras are most easily aligned if the instrument is athermalized, so that an alignment done warm will be maintained when it is cooled to 60K. Furthermore, significant CTE mismatches between the bench and assemblies mounted on the bench require much more complex interfaces.

2.2 Vacuum Vessel Requirements

The requirements for the vacuum vessel are somewhat different:

1. Vacuum tight. The dewar must hold a vacuum. One prefers to arrive at this without extensive, iterative leak-checking and leak repair (though some leak testing is essential, and some leak repair is probably inevitable). This is the most important requirement.

2. Rigidity. Although the flexure requirements are not as severe as for the internal bench, the instrument as a whole must remain aligned within its allowed tolerance to the optical axis. The dewar bulkhead is the key part in this regard, since all major assemblies are attached to it.

3. Weight. The bulkhead is possibly the single largest part, and must stay within its weight budget.

3. Discussion

There are both pros and cons to the use of castings instead of fully machined parts. These are listed below, as applicable to GNIRS (e.g., we ignore the advantages of casting for mass production).

3.1 Pros

1. Cast parts can be made that are not possible with conventional machining. In particular, enclosed structures like the optical bench can be made with fewer parts.

2. Shapes involving a lot of hogging out may involve less effort if produced as castings.

3.2 Cons

Castings are generally more expensive than a machined part if the machined part is feasible. This is because the casting requires making a model, then production of the casting, and then machining of the casting at all critical surfaces (all the interfaces for the case of GNIRS). The cost will inevitably be much more unless the model is much cheaper than a part produced directly and there is little post-machining.

1. Preferred casting alloys (e.g., A356) do not have a good CTE match to 6061, which is what will be used for the metal optics and mechanisms (overall differential contraction between room temperature and 60K is ~0.03%). While 6061 can be cast, it is not likely to work well for thin-walled structures or structures requiring vacuum integrity. The mismatch between alloys is small enough that using them for non-cryogenic parts of the instrument is not a problem.

3.3 Conclusions

The key point is that a casting will be more expensive than a part made by machining so long as the design permits production by either method. (Note that estimates from vendors for the structural modules for the “Old GNIRS” design suggested casting to straight machining cost ratios of at least 1.5:1). Our more recent conversations with casting vendors support this.

Therefore:

1. We should not plan to make any part for GNIRS using castings that can be produced in a straightforward way using conventional machining.

2. Parts should be designed to be produced by machining unless performance is impacted.

Are there any areas where the application of the “rules” given above still might lead to the use of castings?

For the cryogenic bench (optical bench) structure, it is possible that a nominally equivalent structure could be built up out of bolted plates – and might be cheaper. But its thermal and structural performance would undoubtedly be worse, and would be more difficult to model accurately. This makes it a higher-risk approach. Additional effort would also be required for assembly and optical alignment, and in dealing with light seals.

These considerations suggest that the critical parts of the optical bench should be designed as a small number of castings. However, in doing the design one should keep in mind the fact that straight machining is still the preferred method of production. Thus there might be as many as four castings or as few as one (the main spectrograph bench seems to be unavoidable as a casting). The castings must have a CTE compatible with the rest of the instrument, which implies use of 6061 aluminum as a casting material. Designs that will not accommodate this (or vendors unable to handle the alloy) are therefore excluded.

For the vacuum vessel, a cast bulkhead structure does not offer any performance advantages, and the issue is strictly one of cost and schedule. Thus, the time to make a decision is after a preliminary design has been taken to the point where accurate cost estimates can be made. In this case, too, the choice is not strictly between the two approaches (we would need to evaluate a welded structure with post-machining). The decision should be made at a point where it is still reasonable to adjust the bulkhead design to match the fabrication technology (but not to alter the overall dewar design).