i found the adc of stm32f103ret6 is nonlinearity

[eason87lee]

Posted on May 29, 2014 at 16:56

hi,

i am using stm32f103ret6 to test the voltage from the signal genarator(DG1022 from rigol). i found the adc of stm32 is nonlinearity. i found that,when the voltage from DG1022 is less than 1V,the voltage get from adc of stm32 is smaller the 1V,but the volmeter's result is acurrate. Next is the result: DG1022 (mv)       adc of stm32 3000                    3000.352 2000                    2000.654 1000                    996.415 500                      495.2234 250                     246.88 150                     146.877 60                       57.627 30                       27.982 10                       8.219 #stm32 #adc #stmf1

[waclawek.jan]

Posted on May 29, 2014 at 18:56

Is this a DC measurement, i.e. of a steady input voltage?

Do you switch the ADC's multiplexer during the measurement? Is the voltmeter connected directly at the ADC input pin and AVSS pin? Is AVDD/AVSS properly decoupled? Is the supply voltage (namely AVSS) rock stable? Can't this be some floating-point calculation/rounding issue? Isn't there any negative current flowing into some of the other pins (i.e. any pin connected to voltage below VSS)? Do you perform ADC calibration? JW

[eason87lee]

Posted on May 30, 2014 at 03:28

thank you

yes ,it's a DC measurement, a steady input voltage no,i don't switch the mux,when measuring, yes,directly connected yes, no,not the floating point problem,we sample 1000 times,and get the average voltage no negative current, all is positive and gnd ADC calibration must be performed. so, i guess the ADC of stm32 is nonlinear.

[chen]

Posted on May 30, 2014 at 10:00

Hi

The datasheet does specify the 'non linearity' of the ADC. From memory it is not as bad as you have measured. My colleagues (electronic engineers) tell me that if the measurement time is increased (rate decreased) the linearity improves. It is because the ADC is a successive approximation type, it dump the charge from capacitor to capacitor. I guess this means it has problems with low input drive.

[eason87lee]

Posted on May 30, 2014 at 10:30

my adc sample time is71.5,clock is 9M.i measured 1000 times,and get the average.

so,i don't know,,, ps:i modified the testing data(this time,it is true.last is from my memeory).

[chen]

Posted on May 30, 2014 at 11:01

''i am using stm32f103ret6 to test the voltage from the signal genarator(DG1022 from rigol).''

Can you try the test with a PSU instead (or a pot across the supply)?

[eason87lee]

Posted on June 03, 2014 at 03:26 OK,I'll try it. thank you