I made a noise check of the 2 probes, with the pump OFF.

the attached image shows 6 plots in 2 rows.
For each plot, I acquired 5 minutes of the AC signal from the lockin at 100ms of sampling period (3000points).
The first row is the 633nm probe, the second row is the 1310nm probe.
The first column is with the probe OFF, the second and the third columns are with the probe ON.
The first and second columns are with the chopper ON, the third column is with the chopper OFF.

The circle is centered in the mean value of the 3000 points, and the radius is the standard deviation.
The blue line connects the center of the circle with the zero of the axis.

I converted the size of the circle in ppm*W (ppm valid for 1W of pump power) using the calibration I measured for both the probes.
R=19.4 1/W for the 633nm probe (from entry 1033)
R=5.25 1/W for the 1310nm probe (from entry 1045)

The noise with the 633nm probe looks to quite small, 0.3ppm*W, but the specs of the original setup said better than 0.25ppm*W.
The noise with the 1310nm probe is far too high, 13ppm*W not enough low to measure the crystalline coatings (<1ppm).

The enclosures of the setup are all open, so the noise could reduce a bit after I cover everything from the wind.

The chopper doesn't seem to contribute much to the noise.

All these measurements where don with the pump OFF, so the stray light from the pump may increase a bit the noise.

I don't know the polarization orientation after the HWP, so I'm going to check the noise for different rotation angles of the HWP.