Here's my take on the project. Grounded input (scope noise floor 0.8mV):
I chose to use 400u MKP cap as the input capacitor. Settling time is excellent, 10V step takes only around a minute to stabilize. Overall pretty satisfying results, but the voltnut within demands more performance
There seems to be a problem with ADA4528 input current noise or my design of the input stage. Initially two channels were paralleled, but the resulting noise floor was noticeably larger than using just a single channel (
). I wonder if I got ICs from an underperforming batch or SMD metal film resistors I chose affect the performance so much.
Like described in the thread you may characterise your amplifier:
sweep across the frequency range of the LNA with a defined constant input amplitude to see the real amplification of the amp instead of assuming it,
make a high resolution FFT to see if 50Hz couples to strong in the LNA and that way artificially increases your noise when viewed with the oscilloscope,
test a few different noise sources to do a sanity-check: different voltage references like a LM399/REF102/ADR445/LTC6655, batteries, short
Only that way you will truly know your LNA and can actually reliably work with it. Otherwise you will doubt the results and it may have problems you dont know about.
Only a guess: did you use a dual channel ADA4528-2 or single channel ADA4528-1 and maybe now see intermodulation-related errors?
Folks, were is Andreas’ design for the LNA, mentioned several
times in this thread? Can someone share the link if available?
Thanks much folks. Andreas, do you have the gerbers somewhere?
I have the kicad project ... if you send me your email by pm, I can send you the zip with the prj.
I sent the project to 3 members, hope someone will build it and share here the results.