Hi,
I am implementing a lock-in amplifier for a milli-ohmeter. I am following this project; http://cappels.org/dproj/dlmom/dlmom.html

I intend to use better components like a FET op amp for the high pass filter. I simulated the first 'two' stages which I attached.

Current source;
Simulation produced sufficient results although I had to add a 100nF cap across the switching mosfet in order to attenuate massive spikes due to fast switching time of the pulse generator.

It was set to 100ns. I would be driving the mosfet gate by a microcontroller. Do you think this solution is practical enough?

Integrator

If I understood the operation of the lock-in correctly, the unwanted noise would be attenuated in the rectification stage which is the last stage in my schematic. Is an integrator the only solution before reading with an ADC for example? Or is it the most practical one? I will be using an external ADC attached to my micro controller which will be chosen accordingly.

 

I'm not crazy about your current source. When the FET is off the 317 will go to maximum voltage trying to drive the 50 ma.. When the fet turns on the current will be very high until the 317 goes back into regulation.

 

thats what i thought initially .. I left it that way after the simulation gave valid results. Obviously simulation is not reality and i guess the spice model of the 317 has its limitations.

I will go for the 220Ω in series with the test resistance since my test resistance will only vary from 10mΩ to 100mΩ. Therefore my max error would only be (220.1-220)*220/100 = 0.045%

Is there any other feasible method ?

 

You have to check skip the initial operating point to see it. It doesn't last long. If you look at the transient response curves 5 or 10 usec. is about it. (The simulation shows less but I'm not sure I believe it) So I think you could filter it out at the op amp.
The big cap you have (220Ufd.) will mess up the sample time a bit and the ESR will be to high to hide it anyway.