Sheila pointed out that she and Craig had rephased the IMC (alog 41432) after the EOM swap, but hadn't rephased the WFS.  I calculated that the extra 7 ns delay that was added corresponds to about 60 degrees of extra phase delay for 21.4 MHz.  So, for each quadrant of the IMC WFS A and B I subtracted 60 deg.  Allowing the WFS to work after the phasing is noticeably improving the IMC alignment from where the WFS with the wrong phasing had pulled it. 

Note to self:

wavelength of RF = c/freq

length equivalent of extra delay = c*time

ratio of new delay to one full wavelength = freq*time    (c has divided out, so you don't need the precise velocity factor to know the speed of light in a cable)

degrees delay = 360 * freq * delayTime