We borrowed the PZT noise hunting buzzer wand from the 40m (it is a bronze or copper dowel attached to a PZT with a BNC input).
Koji was kind enough to loan us a HV amp from the OMC lab (Thorlabs model MDT694A) - we are using it in the 100V max output mode (gain = 10, +/- 10V input, +/- 100V output).
I played around with it a bit yesterday with Koji and even with just the function generator (DS345), we could pretty clearly feel the buzz from 20 Vpp at 100-300 Hz.
Since we want to do band limited white noise, we ended up hooking up the DAC from the CyMac to the HV amp, and using awggui to make noise.
We put the DFD output into an audio receiver and listened to it on headphones while poking things with the calibrated poke stick.
At a signal of ~100V rms into the PZT,
I turned down the noise level to the PZT wand until I could touch it to the table without hearing any audible difference in the beat noise on headphones. Excitation strength normalized between the different points by letting the weight of the PZT wand itself provide the force pushing the wand to the optic mount (or whatever I was touching).
The loudest couplings by were (in order of noisiness):
A couple steering mirrors (and the PDs) things in the PDH path and the readout path were audible as well, but these were both quieter than the butterfly mount and cryostat
Notes and thoughts:
Another thing I noticed while listening to the beat - when I was tuning the temperature of the laser via the knob on the front of the Thorlabs ITC510s in order to zero out the control signal to the laser, the noise would change a depending on what the control signal level was. The minimum noise (by ear) was not zero, and lock broke at different values on the + vs the - end (was asymmetric by somewhere between a factor of 1.5 and 2). Further investigation needed. I'm guessing it's either an error point offset, or us seeing some of the 7th order mode when we are slightly off one side of the cavity resonance
Acoustic coupling: the top noise sources and our plans for mitigation:
Butterfly mount package
We are putting on sidebands at the diode itself via SMA connector on a PCB which touches the pins of the LD's butterfly package. We may be seeing sideband frequency modulation, or we may be seeing direct laser frequency noise. If the effect is from the sidebands, going to an EOM may fix this. Making little boxes for each diode is probably necessary: we could lift it off the table (via a rubber base?) and seismically isolate it. We also can thermally/acoustically isolate it, in case the coupling is from window / air currents / etc.
The flanges are tired directly to the cold plate, which is tied directly to the suspension/frame/rad shield. Based on elog:787, I think that the stuff from ~200Hz -> 2kHz is the cavity support structure (it goes away when I let all the LN2 boil off). We are designing a v2 of the cavity shield support. The idea is to have ~four 6" springs hanging from the cold plate, suspending a 2nd plate, and mounting the cavities on top of that plate in something that looks very similar to their existing shields. There is a list of questions we need to answer in redesigning this.
Beat readout steering mirrors / beat PD / beamsplitter
The acoustic enclosure box we are making
PDH path optics
Ghost beams? Scattering? The first thing to try once we get to this is looking at the noise spectra and trying to dump the ghost beams to see if we can eliminate any scatter paths. This is in practice slightly difficult because it is so hard to see the ghost beams. Even with an IR card + IR camera, we expect not to be able to see most ghost beams.