From: Jay Elias Date: Tue, Apr 2, 2013 at 12:11 PM Subject: Re: iSHELL review meeting To: Alan Tokunaga Alan, Here are some further comments - 1) My fellow review panelists don't seem to agree, but I still think you may be able to save time/money overall if you revisit some of the mechanisms. In particular, if you adopt a standard simplified design for the filter/slit/order sorter wheel (means fewer filters in the SV channel), it may be that the engineering effort is repaid in reduced detailing and fabrication time. This requires there to be enough space for a common design. The K-mirror is also over-designed, I think, but there is much less payback. The current grating selector design and dekker designs seem about right. The requirements on the XD mechanism are demanding, but if you could eliminate the need for the tilt adjustment and just have a selection mechanism, you could keep that portion of the design and save on space and detailing/fabrication effort. 2) In order to eliminate the need for the tilt adjustment, you need to reduce the number of configurations you support. This would not be easily reversible, but here are some thoughts: (a) If you reduce the maximum slit length, then the number of discrete configurations goes down proportionately. Thus if you go for a maximum slit length at JHK of 5 arcsec and at L of 10 arcsec I think you end up with 3 configurations for JHK, 3 for L, and 2 for M'. This is less than the 12 in the wheel. The different L configurations have only modest overlap. I don't know how many different rulings are needed for this, and how many would be custom. (b) It's also worth looking at using XD gratings in both first and second order. In particular the efficiency of ruled gratings falls off at both high and low frequencies, so it's possible that the second order efficiency of a 400 or 500 l/mm grating is pretty comparable to the 1st order efficiency of the 800 or 1000 l/mm grating. This seems to work best going from L to H, although you might be able to use an L or K ruling at J with a different tilt - only worthwhile if it saves on a custom ruling. If you find custom ruling costs really drive matters you can work in still higher orders, but I think this likely past the point of diminishing returns, plus the rulings become coarser and not necessarily more efficient. (c) If you have extra spots in the wheel, you could consider installing PI XD gratings to enable special programs - there are obvious logistical issues but it enables specialized programs without a more complex mechanism or a larger investment in gratings on your part. You would keep the baseline gratings installed at all times, but the number of open spots would probably accomodate a semester's worth of PIs with that kind of money.