Considerations for lock-in amplifier design for sub 50uV signals using AD630

[super7800]

I don't think one would need the last gain stage, as the noise for the ADC input should be well less then 2000 times the input stage noise.

you are correct in that I don't need the third gain stage (4 gain steps). Better to have three gain steps.

If mains hum is an issue, a notch filter would be more practical solution

I would think it would be better to filter this out in the digital domain, correct?

The MUX at the input would add some extra noise from the resistance (some 125 ohm each). One should consider a lower resistance MUX.

Good point. I will switch all the analog switches to relays. SIP-1A05.

[Kleinstein]

Reed relais are low resistance, but they also tend to produce thermal EMF and the power consumption can be an issue.
There are lower resistance switches if needed (e.g. ADG1409). The later mux to choose between the amplifier stages would not be critical with the noise / resistance.

For the lowest gain path one could consider a divider (e.g. 1:4 or so) after the INA that is still powered with +-12 V to than get a path to the +-2.5 V powered part. This would allow to have more gain at the front and less noise from the gain setting resistor / INA.

If the mains hum level is low, the ADC can handle this. This problem could be a relatively high level of mains hum, that could drive an amplifier to saturation if the gain is high. A notch filter is also not ideal, but could be an option if there is really a lot of background. If there is only a relatively small hum level, there is no need for a notch and digital filtering should work better.

[super7800]

For the lowest gain path one could consider a divider (e.g. 1:4 or so) after the INA that is still powered with +-12 V to than get a path to the +-2.5 V powered part. This would allow to have more gain at the front and less noise from the gain setting resistor / INA.

Attached. Also a preliminary layout.

If the mains hum level is low, the ADC can handle this. This problem could be a relatively high level of mains hum, that could drive an amplifier to saturation if the gain is high. A notch filter is also not ideal, but could be an option if there is really a lot of background. If there is only a relatively small hum level, there is no need for a notch and digital filtering should work better.

I expect the mains hum to be low, so i will proceed with the idea of digital filtering.

[Kleinstein]

I see no need for the extra buffer amplifier U21.1. There is a buffer already after the mux.
I would consider having the path via the divider DC coupled ( other side of C39). With so little gain this should work and could help detecting problems without an external measurement. It would also reduce the required coupling capacitor.
With 1K/10 K at the -10 amplifier stage the coupling capacitor should likely be larger anyway or one should use larger resistance.
A coupling capacitor in the 1 µF range would likely be PE film and not longer C0G.

The path to S2 at the mux should have a series resistor to limit to current incase the voltage goes beyound +-2.5 V.