I doubled checked that the noise eater monitor is not usefull-  the readback we have is an accurate reading of the voltage out of pin 12 on the laser diagnositcs cable, which the manual says is a noise eater readback.  However, this voltage is around 3.1V when the noise eater is on and when it is off. If we are going to use the noise eater we should have a way of remotely switching it on and off, and it would be nice to have a way to tell if it is oscillating. 

Also, just as a check I tried changing the input polarization of the beam leaving the ISC end table.  The arm cavity was misalinged, and I watched refl_B_LF as I inserted a half wave plate and rotated it. I could reduce the power in the reflected beam by diong this, the maximum power I could get back with the waveplate was the same as we get back when we don't have a waveplate.  So it seems as though the low TMS efficiency is not due to polarization.  This makes sense because we already knew that the input polarization is right (s-pol at the bottom of the periscope).

So the question of why the TMS efficiency is low remains.