This morning while the interferometer was tied up with other work,  I took a quick look at the length to angle coupling of ETMX. 

The bottom lines:

• M0 seems OK
• the impulse response of L1 L2P seems bad. 
• In the configuration used for ALS DIFF, the DC coupling seems OK

I had a look at the impulse response by applying a million counts offset in L3 INSINF and watching the oplev time series. The first plot shows the reponse of the three stages used in ALS DIFF (Mo, L1, and L3) with the filters set as they would be for DIFF, with the other stages turned off.  You can see that L1 causes a large response in pitch, while M0 seems to be fine. 

The next plot shows the response of the L1 stage alone. With the L2P decoupling off, we get less of a response in pitch, while turning off the L2Y filter doesn't make much of a difference. Based on this I have turned off the L2P on L1 for now.

The last plot shows the immpulse response of all the stages, with everything as nominal for ALS DIFF.   In the red trace, the L1 L2P filter is off, in the blue trace it is on.