Reports until 16:54, Friday 31 May 2024
H1 CAL
jeffrey.kissel@LIGO.ORG - posted 16:54, Friday 31 May 2024 (78165)
Post OMC-Swap DCPD Transimpedance Amplifier Update: Finally A Mystery Solved
J. Kissel, F. Clara

At my wits' end with measuring the analog response of the OMC DCPD transimpedance amplifiers via the OMC DCPD whitening chassis's preamp monitor interface (LHO:78090), the only thing left to question was the SR785 we'd been using that had been out by the SQZ and HAM6 racks -- a trusty race horse that had been in service out on the floor for a long time. 

Fil and I have discovered that indeed, it was time to put this race horse out to pasture.

We pulled the unit from the floor, grabbed a random spare electronics chassis we had lying around (Fil chose a D0902783-style anti-aliasing chassis), and measured it's transfer function using a suspected good, and the suspected bad SR785 from the floor.

Our suspicions were proved correct: the floor SR785 that we'd been using for months (if not years) is broken in a nasty, subtle way.

Check out the attached comparison of transfer functions of the same electronics channel, but between 
    - The "bad" floor SR785, where the CH1 and CH2 inputs are in differential, (A-B) mode :: the configuration that's needed for the measurements I was doing for months,
    - The "bad" floor SR785, where the CH1 and CH2 inputs are in singe-ended, (A only) mode :: totally fine
    - The "good" EE shop SR785, where the CH1 and CH2 inputs are in differential :: totally fine
    - The "good" EE shop SR785, where the CH1 and CH2 inputs are in singe-ended, (A only) mode :: totally fine

If you compare the ratio of the "bad" floor SR785 in differential mode to the "good" SR785 in differential mode against the "measurement setup" with pages 3,4, and 5 of the plots from LHO:78090, then you see exactly the same "gain is 2 [V/V], and has a frequency dependent hump above 1 [kHz]" that drove me insane this past Tuesday.

GUH. It's been a rough couple of months when you find out you've been using a calculator that computes 2 + 2 = 5...

The unit is now boxed up and off for "re-calibration," which (a) it hadn't had in a really long time, and (b) usually means that we'll get it back fully-functional.
Non-image files attached to this report