During commissioning, the "Dark" filter was defined to be a combination of the nb1061 and the Ks filters. These two filters are in two different filter wheels, and by crossing them, a "Dark" filter position prevented light from reaching the detector.

One of our observers noted the inefficiencies of changing to a Dark filter and thought that there should be a method to save some time. An example of where the inefficiency is prominent is during mask alignments. Following the completion of a fine alignment, we set the filter to dark and wait for it to complete before configuring the CSU for the science setup. A move from J+open to Ks+NB106 requires ~30 seconds. A move from J+open to J+NB1061 requires 15 sec. Over time these seconds will add up: 15 sec, 5 target mask changes per night, 5 alignment masks, 70 total nights per semester. A ball park number is then 7 hours of time lost per year.

Through discussions with MOSFIRE team members, we concluded that there are other combinations in the two filters that would work to set a dark pair. Keck staff modified the ObsMode java gui to take advantage of the various dark filters, and the table below shows the implemented two filter wheel values for a filter currently in beam. For broadband filters, only filter wheel 1 will move and the primary blocking filter used is the NB1061 because open+ J -> NB1061 + J requires moving filter wheel 1 one position. We tested these combinations by exposing for 30s with the Ne and Ar lamps both on, and little to no light is detected with a mean value of less than 10 counts. If any filter was configured in beam, the lamps would saturate the detector in the minimum integration time.

Filter	DarkCombo
	Wheel 1	Wheel 2
Y	NB1061	J
J	NB1061	J
H	NB1061	H
Ks	NB1061	Ks
K	NB1061	K
J2	J2	K
J3	J3	K
H1	H1	K
H2	H2	K
Default	NB1061	Ks

One of the keys to these new dark filters is that we don't move from one combination to another and accidentally expose the detector to light. This is one of the reasons why the Y filter has the same dark combination as the J filter. When it comes to filter moves, the following was empirically determined and later confirmed by Jason Weiss and a review of the low level C code.

• If one of the filters is in an open position, a coordinated move is made: one filter wheel at a time.
• If a filter wheel that is currently in an open position needs to move to a new position, this filter wheel is moved first.
• If moving to the standard dark position, both wheels are moved simultaneously. The idea is that one filter wheel is moving away from an open position while the other is moving to a different filter position (see next bullet as to why it does not move through open).
• For a single filter wheel, all motion between filter positions avoids moving through the open position unless the intended destination is to open. So if moving from 5 to 1, the motion is 5-4-3-2-1 and not 5-0-1 (0 = open).
• If moving to the open position, the software finds the shortest path to open.

Related ECRs: Softewar ECR 2658