I've seen a few times lately that right after an initial alignment it seems like we regularly have to go to check_mich_fringes.  I had a look at the ADS DOF 10 signals that are used for MICH_BRIGHT_ALIGN, and it looks like we're basically skipping doing much of anything.  In the attached screenshot, we begin MICH_BRIGHT_ALIGN at the first t cursor, and exit the state for offloading 9 seconds later at the second t cursor.  Clearly the ADS has not converged, since neither the inmons nor the outputs are around 0. 

On lines 1022-1023 of ALIGN_IFO the convergence threshold is set to 10, but none of these signals ever get that high.  We need to lower this value to something like 0.5, so that the DOF10_OUTPUTs (ex. pink on bottom row) have a chance to actually converge. 

I will do this on Tuesday, when I know that I have time to test it, unless we happen to be doing an initial alignment when I'm on site before then (so that I can test it).  For now, operators will likely find that they need to do check mich fringes after an initial alignment.