SDADC Tolerance issue with STM32F373CCT6 MCU.

[PMahe.1]

Hi,

We have a problem with SDADC on STM32F373CCT6. We are using two SDADC Analog pins to measure the voltage. Even through the voltage on the both the Analog MCU Pins SDADC-1 and SDADC-2 are same, the digital output voltage and ADC values from the MCU differs.

It is not same on all the MCUs, some MCUs are fine, some are showing this issue? Is it known issue with this part number STM32F373CCT6??

How can we go forward to resolve this issue?

Thank You,

Praveen

[waclawek.jan]

Is this on two channels of the same SDADC, or on two SDADCs? Is it outside the characteristics given in the DS? Do you perform calibration as outlined in Launching calibration and determining the offset values subchapter of SDADC chapter in RM?

JW

[PMahe.1]

Hi Jan,

Thank You for your response, here is the ans for your questions

1. Is this on two channels of the same SDADC, or on two SDADCs? Is it outside the characteristics given in the DS?
   1. its pin PB2 (SDADC1_AIN4P) and pin PB14 (SDADC3_AIN8P). I think its two different SDADCs (Pin PB2 is using SDADC-1 and Pin PB14 is using SDADC-3). Correct me, if i am wrong.
2.  Is it outside the characteristics given in the DS?
   1. Ya, The analog voltage we are trying to measure on both the ADCs are exactly same. If we measure directly on ADC pins with a precise multimeter, its same. But we have a variation of around 5 mV to 10mV between pin PB2 and pin PB14 in SW converted value, but there is no difference in measured value on MCU pins directly. I don't think 10mV difference is not acceptable as we are using all 16bit conversion and one LSB is equal to 50uV.
3. Do you perform calibration as outlined in Launching calibration and determining the offset values subchapter of SDADC chapter in RM?
   1. Not sure about this, but do you think the calibration would cause that much difference. I hope calibration is for uV differences.

It works, if we swap the MCU, but we cannot keep on doing that, and we don't know, weather the replacement MCU we will be swapping is precise or not.

Thank You,

Praveen

[Uwe Bonnes]

From the specs, the SDADC seem hardly usable. Gain is speced at     -2.4 -2.7 -3.1 % ( negative  gain error = result are greater  than ideal). So to compensate the gain error, you need to sample a second reference voltage lower that the SDADC voltageor about 0.7 % gain error between SDADCs is to expect.

[waclawek.jan]

Read carefully and thoroughly the specification in datasheet - if you use single-ended inputs, the offset error is in mV *after offset calibration*. And there's also a non-negligible gain error.

I know this may come as a disappointment, but SDADCs are principally prone to these sort of errors and they behave differently than SAR ADCs. They are not very well suited for absolute and DC measurements, they are better suited for AC signals like audio, where absolute amplitude and DC offset are less important than resolution (which is equivalent to dynamic range) and linearity (which boils down to distortion).

JW

[Uwe Bonnes]

There are SDADCs like the LTC24 series that are very well suited for low gain error DC measurements.