(Calibration for Attachment 5 corrected Aug 27, 2015)

Now the test procedure fo the unit is written in the document https://dcc.ligo.org/LIGO-T1500404

And the test result of the first unit (S1500117) has also been uploaded to DCC https://dcc.ligo.org/LIGO-S1500117

Here are some supplimental information with plots

Attachment 1: OLTF of the AM amplitude stabilization servo.

Attachment 2: CLTF/OLTF of the 2nd AM detector self bias adj servo

The secondary RF AM detector provides us the out-of-loop measurement. The secondary loop has an internal control loop to adjust the DC bias.
This loop supresses the RF AM error signal below the control bandwidth. This has been tested by injecting the random noise to the exc and taking
the transfer function between the primary RF AM detector error (MON1) and the secondary one (MON2).

Then the closed loop TF was converted to open loop TF to see where the UGF is. The UGF is 1Hz and the phase margin is 60deg.

Above 10Hz, the residual control gain is <3%. Therefore we practically don't need any compensation of MON2 output above 10Hz.

Attachment 3: Comparison between the power setting and the output power

Attachment 4: Raw power spectra of the monitor channels

Attachment 5: Calibrated in-loop and out-of-loop AM noise spectra

Attachment 6: TFs between BNC monitor ports and DAQ differential signals

BIAS2 and CTRL look just fine. BIAS2 has a gain of two due to the differential output. The TF for CTRL has a HPF shape, but in fact the DC gain is two.
This frequency response comesfro that the actual CTRLis taken after the final stage that has LPF feature while the CTRL DAQ was taken before this final stage.

MON1 and MON2 have some riddle. I could not justify why they have the gain of 10 instead of 20. I looked into the issue (next entry)

Attachment 7: TF between the signals for the CTRL monitor (main unit) and the CTRL monitor on the remote control test rig

The CTRL monitor for the test rig is taken from the CTRL SLOW signal. There fore there is a LPF feature together with the HPF feature described above.
This TF can be used as a reference.