We continue to investigate the aliasing seen around 2.5 kHz in the Sc channels, as reported in entry 42467 and comments.
Today, we took data in free swinging Michelson, and injected two lines on NI mirror actuator, one at 200 Hz (8h35 UTC, purple) and one at 4800 Hz (8h26 UTC, blue). The figures compare the results, the second one is only a zoom around 200 Hz.
- pad1: channel generated in CALnoise and sent to the DSP at 10 kHz: only the injected line is visible (if we forget the lines below 1e-8)
- pad2: "same channel" but readout by Sc: in addition to the injected line f_0, we see the aliased line at 5000 - f_0 Hz. Note that the amplitude of the line at 200 Hz in Sc_NI_MIR_Z_CORR is similar in both injections.
- pad3: the reconstructed deltaL channel: we see the line at 200 Hz only we we indeed want to inject the line at 200 Hz (purple).
- pad4: readout of the coil voltage via the DSP: we see the same aliased lines as in Sc_MIR_Z_CORR
Here are the observed effets:
- When we inject the line at 200 Hz, we see it in Sc_MIR_Z_CORR with an amplitude of ~0.1 V/sqrt(Hz), and we also see it in DeltaL with amplitude 5e-13 m/sqrt(Hz)
- When we inject the line at 4800 Hz, we see a line at 200 Hz in Sc_MIR_Z_CORR with similar amplitude of ~0.3 V/sqrt(Hz), but we do not see it in DeltaL.
If the aliasing was in the driving part, the measured SC_MIR_Z_CORR would be sent to the coil actuator and the line at 200 Hz would have been seen in both case in DeltaL. Since it is not the case, we can conclude that the aliasing is not on the driving part, but only on the readout part, when the 40 kHz DSP channel is filtered and down-sampled to 10 kHz and sent to the DAQ.