PDH error signal curve superimposed DC bias

[Future-li-cloud]

I use PDH frequency locking module experiment. A sawtooth wave is used to scan the laser cavity length to obtain the PDH error signal curve. Theoretically, the curve should be near zero, but the obtained error curve has DC bias voltage superimposed, and the bias voltage is not very stable. Why is that?

[michaelcroquette]

I'm not sure this is an issue with pyrpl. If you have residual dc voltage on the PDH signal, this can mean that you have some parasitic amplitude modulation. Perhaps when scanning the cavity length, you are causing some misalignment. My advice would be to try:
-with a smaller amplitude modulation
-at a different modulation frequency
-Have you tune the demodulating phase correctly ?
-to use a sinusoidal wave instead of sawtooth to avoid non-linearity
-using the network and spectrum analyzer to debug your setup