Hi Snorlax,
I typically differentiate between AFE and external AFE.
anyway, I would like to apologies, I didn't see that your excitation frequency is 3kHz.
You have two problems,
Your frequency is too low, close to the DC and 50/60Hz bin.
Settling time = 0 means coherent sampling
If you need a detailed explanation about how affects both things to your measurement, just let me know.
I'd like to demonstrate my theory,
Just a simple example, may you change your excitation frequency little bit higher, lets say 5kHz, what about the error? Is reduced?
Now a little bit higher, lets say 10kHz...
Its the overall error reduced?
Regards,
Miguel