step size error (ADC accurate voltage measurement)

[DANXIA]

Hii... i am using STM8s003f3 controller,which have 10 bit ADC.
i am trying to measure +-1000v DC with 0.1 accuracy.
i know normally its not possible because of step size error,that is 1.955..
i want ask is there any way to achieve this goal without changing adc rsolution.
if there is any way please share some reference .
Thanks!!!

 

[BradtheRad]

Consider creating more than one range for your voltmeter similar to store-bought meters.
For a range 0-100V you'd have 0.1V accuracy or 0.1 percent accuracy, whichever you want.

Or by subtracting a certain amount you get an expanded scale meter. Therefore to measure 900 to 1000 V, first subtract 900 using a series zener diode. (In practice you'll scale it down with resistive divider.) This effectively results in 0-100 V to be measured, with 0.1V accuracy or 0.1 percent accuracy.

 

[DANXIA]

Consider creating more than one range for your voltmeter similar to store-bought meters.
For a range 0-100V you'd have 0.1V accuracy or 0.1 percent accuracy, whichever you want.

Or by subtracting a certain amount you get an expanded scale meter. Therefore to measure 900 to 1000 V, first subtract 900 using a series zener diode. (In practice you'll scale it down with resistive divider.) This effectively results in 0-100 V to be measured, with 0.1V accuracy or 0.1 percent accuracy.

so according to you there is no other way to get 0.1 accuracy with 10 bit adc in the range of +-1000v

 

[KlausST]

Hi,

Accuracy and resolution are different things. Precision is another one.

Usually the ADC accuracy is worse than one LSB.
For the accuracy you need to take into account: Vref_accuracy and noise. Voltage divider accuracy, ADC offset, ADC full scale error, ADC noise, ADC linearity...and all their drifts ... but not resolution.
Thus read the ADC datasheet on accuracy specification.

Theoretically you may have an ADC with better accuracy_error than one LSB. If so: To get use of this high accuracy you need to add analog dither and a digital low pass filter.

*****
Generally microcontroller ADCs are not very accurate. This is because Reference inaccuracy and drift, offset ... and ground bounce.
Mind that every microcontroller GND pin current will shift the ADC GND reference and thus causes errors. Microcontroller internally will cause GND current, but also periferals connected to a port may cause GND current. One of the worst may be a (blinking) LED connected to a microcontroller port.

Thus - for high accuracy - I recommend to use an external ADC.

Please check whether you really need "accuracy" or just "precision". In doubt do an internet search.

Klaus