I won't get into an arguement with you Erik, I respect you too much. But, be assured, this design HAS passed Weights and Measures, and is approved to 1:10,000, which it comfortably exceeds. And, as you know, that includes all of the environmental tests you mention. Would that all electronic products (apart from the military) went through such a test regime. Trying to read repeatable sub-uV signals in the presence of up to 10v/m RF from DC to 2Ghz can be challenging... (and before the pedants leap in - it's 3v/m and 1GHz - I know - but our tests went further...)

As Stefan points out, we are ONLY talking about the electronic part here, so my simulation is perfectly relevant - as was my point.

Stefan, the 5532 is at max gain (64x), so 640mV into the modulator with, of course, 5v reference.

I've never tried, but Crystal (I should say Cirrus Logic, as that is what they are now) suggest that amplifier noise dominates at high gains, modulator noise at low; as you would expect. I should do the maths and see if reducing PGIA gain and upping modulator gain would improve the situation. I suspect not and, as I keep saying, commercially, it's fine: "if it ain't broke, don't fix it". This is just an academic exercise - and I learnt something (as I always do) from Kai - I researched a little more into lock-in amps ;-)

Interestingly, I just answered an email from my buddy who is currently vacationing in Varna, where he has bought an apartment and some building land. I was surprised he had a broadband connection. He replied that he hadn't - he took his laptop with him from England and found 3 unsecured broadband connections in his apartment block ;-)

Dave