I tuned the lockin gain to have a better resolution for little signals.

I acquired 3 minutes of noise for different modulation frequencies, in the following conditions:

 - Probe ON;

 - Pump OFF;

 - Front door closed;

 - Lockin time constant 100ms;

I made an histogram for every acquisition and I computed the standard deviation in order to find the best chopper frequency.

The lowet value of the noise standard deviation is at 430 Hz modulation frequency. The second best value is for 480 Hz, as used previously. The highest noises are at multiples of 50 Hz, as expected.