Previously with Yuhang we measured the open loop transfer function of the SHG cavity control servo:

Spectrum analyzer source - Perturb IN
Channel 1 - EPS2 OUT
Channel 2 - EPS1 OUT

Spectrum analyzer frequency response = Ch2/Ch1

EPS1 is just before an adder which injects Perturb IN, and then EPS2 is right after the adder. Without the Perturb IN both give the error signal of the loop when in SCAN mode.

The OLTF for the SHG control should have a ~ 43 dB bump at 100 Hz and then UGF at about 2-3 kHz (Aritomi thesis). However, on the new spectrum analyzer, we saw just a flat response of 0 dB, up to the kHz range, while the old spectrum analyzer gives the correct transfer function. As it turns out, it was just a setting issue. On the old spectrum analyzer, when the signal goes out of the screen, the Y-axis is adjusted to fit the signal and says "Auto Range in Progress". On the new spectrum analyzer though, one of the features is that the signal acquisition and on-screen display are decoupled, so we can have different measurement span and range compared to what is shown on the screen. This is convenient for certain types of measurements that I haven't figured out yet. Anyway, pressing Auto Range on the new spectrum analyzer toggles between Automatic and Manual range limitation for the input signal. In this case, it was set to Manual Y-axis range with maximum 0 dBV, so the acquired signal doesn't go above 0 dBV. Turning Auto-Range ON fixed the issue.

When the signal overloads on the display you should instead press Auto Scale A/B.

Afterwards I get 4.5 kHz UGF on both analyzers. It increased a bit from last time when we set it to 2.2 kHz and 43 degrees phase margin. The best excitation level to send in this case was about 200 mV white noise, which gives the best signal coherence between input and output. Above 200 mV there seems to be a lot of noise.