STM32 OpAmp Differential Strain Gauge Bridge Input

[Rob Ashworth]

Does anyone know if the PGA OpAmp in the STM32 has the ability to interface with a differential input - like a Strain Gauge Wheatstone bridge?  I'm not convinced its going to work because my signal is going to swing +/-50mV, and the OpAmp does no look like it can handle a differential input.  There's probably some jiggly pokery that can be done to condition it externally, just wondered if anyone has any experience before I go down that route myself.

[AScha.3]

See in rm opamp :

So you could add all circuit you need for biasing etc external and it should work as expected.

But use up 3 pins to get an opamp with some 3mV offset and about 1000nV/(sqrt Hz ) - no big player for low level signals. :)

For low level like Strain Gauge Wheatstone bridge i would add a typical chip for such use, like HX711:

24bit ADC , with bridge PGA input , about 50nV total noise (!) ->

+ about 2 $ investment...

[MasterT]

If OPA's all 3 pins are accessible , than combine two OPA.

Circuits successfully implemented in one of my projects with nucleo-G474re.

*resistors 10k - internal PGA matrix.

[Rob Ashworth]

Thanks Ascha, I haven't come across that ADC before.  Looks good except too big for my application.

[Rob Ashworth · Accepted solution]

Ah ok, so I could do it, except I have the problem that I have 2 sensors, so would need 4 op-amps.  The G4 that handles this is too large, so looks like I'm back to using and external ADC.  Will probably go for something like the ADS1292.  Build in PGA and differential inputs.  Hoping ST bring out a controller with an Instrumentation AMP that can handle Differential signals some time soon.

[MasterT]

If G4 is too large, than ads1292 is small? Have you compare footprint?

BTW, STM has stm32f373 IIRC, with SD adc.

[Rob Ashworth]

The STM32F373 has the necessary Analog configuration & Resolution, unfortunately too big and no CAN-FD.

G4 Series, ok, possible, maybe... but the ADC resolution is only 12-bit.  

[MasterT]

Well, 12-bits + PGA x2 x64 => 7-bits = 19-bits total.

If power consumption isn't an obstacle, I'd run adc at 2.5 msps and "crunch" 5+ bits more, getting close to 24 total.

Running adc at max speed provides one more advantage, reducing noise of the OPA over wide band, so 250 nV (datasheet G4) mostly nulled-out,

[Rob Ashworth]

The ADC native resolution is still 12-bits no matter how much you amplify the signal, or use oversampling techniques.

I think the G4 solution is worth a test though as @MasterT  describes above

I have a dev board so will see if configuring 2 Op-Amps into a single Instrumentation Amp, then using 2 ADC channels in Diff mode provides an adequate solution.  If it saves me the space and hassle of adding an External ADC.

It may take me a few days to test..!