The troubled binary IO code (https://alog.ligo-wa.caltech.edu/aLOG/index.php?callRep=2420) was fixed by JoeB. Here is the not-too-accurate recap of things as I understand.

TMS was working fine in an old binary I/O regime.

Several days ago, there was a change in the binary I/O code. Unfortunately some typos were introduced in the TMS model (different typo from what Jeff wrote in the earlier alog) when it was copied from other model, creating some funny name channels that nobody could have guessed. This caused the first problem (i.e. not being able to turn the coil driver DAC input on), and was fixed by Joe by fixing the typo and then making an MEDM screen to control binary IO.

The same code change had other side effect. The whitening/dewhitening filters were in FM9 and FM10 in the coil output filter module in the old days, but in the new code they are standardized to FM2, FM6 and FM7 (FM9 corresponded to FM6, FM10 to FM7, and we used to turn FM10 on and off instead of turning FM2, that is an inverse filter of FM7, off and on). Because of this we couldn't match analog and digital, therefore we couldn't damp the mass after the coil output was enabled. This was fixed by Joe by running a matlab script to copy standard set of filters (Joe had to fix the script first).

Now that things are working, I asked Dave to make a safe.snap file so the model can automatically burt restore after restarts.