I thought it would be interesting to see if I could measure oscillations of the tuning fork without applying any drive signal. Sure enough, a small, noisy signal (noise from SPMS) from the preamplifier output is visible on the scope at 100 mV/div vertical scale and 20 us/div timescale:

The tuning fork oscillates at it's resonant frequency despite there being no drive voltage. This could be induced thermally, electrically by conducted or radiated noise, or acoustically by a nearby air-conditioner. The small oscillation signal is easily measured by the FPGA lock-in detector despite the high noise. Here's the result of a narrow frequency sweep:

Not bad. I then placed an SMPS power brick next to the preamp, burying the signal in switching noise. The peak-peak noise is ~40X larger than the signal. This scope trace is at 1 V/div and 20 us/div:

The resulting amplitude curve is essentially unchanged, since the signal and noise are at different frequencies. The lock-in measures only the signal in a very narrow band around the frequency of interest. Pretty cool!