I'm breadboarding a two-stage inverting amp/filter to amplify a radar signal, using the TLV2464 opamp. My first stage has a gain of 45dB, the second stage 35dB. The input will be a radar signal which I offset to 2.5V so I get a positive output signal after the second amp stage that I feed to a comparator after which it will be fed to an STM microcontroller.
There are two major problems that I can't figure out how to solve:
When I add 0.1uF decoupling caps to the PWR pins the output signal disappears - it just stops working and starts again when I remove the cap,
Without any input (or even when tied to GND), the signal after the first stage is quite clean and stable. The output of the second stage however looks like high frequency noise with an amplitude larger than 0.5V, so the signal swings from below 2V to almost 3.5V. There seems to be a ~62Hz component in there (16ms period).
I power and scope the breadboard with an AnalogDiscovery2 USB scope which of course isn't the best tool around but it should be plenty good for this kind of job. Anyway, the first stage works just fine.
The TLV2464 ICs I have are 4CH versions; I've tried using two channels of one IC for both stages and I've tried using a different IC/package for each stage: same result.
So there must be something very basic wrong with my schematics I guess.
Any advice is welcome
This is with the input signal floating. Blue is after stage 1, yellow after stage two:
This is with the radar module connected to the input line but there's no movement in front of the radar. Blue again is after the first stage, yellow after the second stage:
Adding a cap to the bias reference line like @DAS suggested indeed seems like a good idea, but when I tried a 0.1uF or a 1uF I'm getting a signal with a weird bliep almost every second. Not sure what I'm looking at here...:
After the detailed response by @analogsystemsrf I looked at it again. I don't really understand the entire reasoning but not wanting multiple return paths is clearly a good design guideline! How I changed my circuit: Instead of using one voltage divider to create the bias at both opamps, I took the output of the voltage divider and fed it into two followers/buffers, each one which will be providing its output to each of the amp stages. The intention is three-fold: (1) controlling the impedance, (2) taking out the multiple return paths (not sure if that happened!), and (3) not leaving the 3rd and 4th channels of the package (4 opamps) unused; they have to be properly terminated or used. This is what it looks like:
Anyway, these well-intended changes did not improve my amp/filter: after the second stage I still get crazy noise with a large amplitude (V). Another issue that hasn't been resolved: there's still a very consistent signal in the noise around 60Hz. I have no idea where this could originate from!