STM32 ADC noise

[David Hess]

I read that the digital and analog grounds must be connected in just one point, it means separated ground planes but connected in 1 point. it seems ground separation does not help to reduce the noise. if you think an external ADC is more stable, then I gonna test an external one.

That is how it is suppose to be done for the reason I identified; a single point ground removes ground loop errors.  A good ADC should produce less noise than that but ADCs embedded with microcontrollers are often compromised by all of the digital switching noise no matter what is done.

[VanitarNordic]

That is how it is suppose to be done for the reason I identified; a single point ground removes ground loop errors.  A good ADC should produce less noise than that but ADCs embedded with microcontrollers are often compromised by all of the digital switching noise no matter what is done.

Do you have experience with an external ADC? for my case 10 or 12bits 1Msps is enough.

I think that's the reason why in many equipments we see external ADCs

[David Hess]

That is how it is suppose to be done for the reason I identified; a single point ground removes ground loop errors.  A good ADC should produce less noise than that but ADCs embedded with microcontrollers are often compromised by all of the digital switching noise no matter what is done.

Do you have experience with an external ADC? for my case 10 or 12bits 1Msps is enough.

I think that's the reason why in many equipments we see external ADCs.

That certainly is a factor.  It is very difficult, and sometimes impossible, to achieve low noise using an ADC built into a microcontroller.

There are lots of external 12 bit 1MS/s sampling ADCs which achieve close to their noise free resolution.

[VanitarNordic]

Look at this circuit - it amplifies 1mV (50us) pulse superimposed on 5V DC. It will do with "any DC" voltage ("DC" is "filtered out" by R1C1 time constant). Amplification of the diff amplifier set to 100x.
An example only, you have to modify it according to your reqs.

Hi imo

I came back to this topic because now I want to focus on it.

in my case, the small signal has the same frequency as the input signal. actually, the input signal itself might change. for example, if the input signal is 100Hz, then the variations are also 100Hz, something like phase shift with changes in the amplitude, but very very small in the range of uV which should be amplified before detection. for example, if it is a square pulse, then the falling edge of the pulse shifts a bit and starts falling sooner or maybe later. something like this.
if I amplify the whole signal, then the variations will be destroyed because the opamp amplifies everything and many signal points will be clipped by the opamp when they go near supply rails. therefore I will miss those points. Therefore if I could be able to just amplify the changes, it is perfect.