Today Dave and company did the model restarts for the isc end station models, with the new matrix and filters intended for an calibrated circulating power channel. 

I am loading a calibration based on 15431 with a couple of changes:

3 dB of whitening gain, which is what we currently use in full lock. 

20 volts/32768 counts from the ADC, 1/2 Volts/volt for differential output, 1000 ohms transimpedance, 0.8A/W responsivity of the QPDs (I think this was a typo in the initial alog and this is really the responsivity, not the QE)  means 2621440 counts/Watt on QP

Keita points out (3077) that the red beam leaving the ETM hits 4 silver mirrors on its way through TMS, each with a reflectivity of about 98%, then we have 5% transmission through M12 and a 50/50 splitter for the two QPDs, (T0900385), so we get 2.3% of the light leaving the ETMs on each QPD.  

From the galaxy page, I think that our two ETMs have 3.8 and 3.9 ppm transmission each, I'm using 3.85 ppm as an average. 

Besides the whitening gain, this gives 3.96 kW of arm circulating power per count on the QPDs.