ADC measured voltage is 200-300mV over the real voltage

[jloliveira]

Dear Madam/Sir,

We integrated the ADC measurement code from the ble_proximity app example (nRF52840 Preview DevKit, SDK15.0.0, Softdevice6.0.0) to measure the battery from a non-rechargeable lithium battery, and realised it is measuring around 200-300mV over the real measurement with a multimeter.

We tested the ADC measurement from both the nRF52840 PrevDevKit and our custom board and both are measuring similar voltage with the same over 200-300mV to the real multimeter measurement.

We then used the PowerProfileKit with different voltage inputs from the voltage regulator and setup the nRF52840 PrevDevKit and our custom board as an external DUT. From different input voltages, which match what we measure on the multimeter, we realised the ADC measurement keeps the same +200-300mV deviation on both boards despite the input voltage we chose.

Is there anything wrong with the ADC formultation in the ble_proximity app example? Should we unconsider the DIODE_FWD_VOLT_DROP_MILLIVOLTS from the line batt_lvl_in_milli_volts = ADC_RESULT_IN_MILLI_VOLTS(adc_result) + DIODE_FWD_VOLT_DROP_MILLIVOLTS?

Looking forward to your reply.

Thank you,

João

[ecasado]

Hello,

If your design does not have a forward diode to protect the battery, you must not add that 270 mV from DIODE_FWD_VOLT_DROP_MILLIVOLTS to the result. As this answer says, that's in the example code just to show how it's done in that case.
Regards,

Enrique

[jloliveira]

Hi Enrique,

I understand your answer but doesn't the nRF52840 Preview DevKit integrate a forward diode to protect the battery?

I even thought the nRF52840 chipset itself included the forward diode battery protection although I see from the answer you shared that it doesn't.

Thank you,

João

[ecasado]

Yes, the DK has diodes on its voltage inputs (which are D80, D81, D82 and D83).

How do you have the battery connected? If you are connecting the battery output directly to the ADC input, then you are bypassing the diode and effectively adding those 270 mV to the voltage. The diode's drop should only be compensated if you are measuring the battery voltage after the diode, not before.

[jloliveira]

Hi Enrique,

The battery is connected to the DK through External Supply P21 pins. That means we are indeed bypassing the diode and should unconsider the +270mV?

If I consider the coin cell battery power input in the DK than I should account for the +270mV?

Thanks again,

João

[ecasado]

Hi again João,

I've reviewed the schematics for the nRF52840-DK again, and it seems the external supply pin input (P21) does not have a protection diode, it's managed by load switches unlike the nRF52-DK (nRF52832) which is simpler. Those diodes (D80 to D83) are there only for the 5V inputs.

So, ultimately, I believe you shouldn't add DIODE_FWD_VOLT_DROP_MILLIVOLTS to the final result because there is no such drop on your boards, and neither will be if you use the CR2032 cell. 

Best regards,

Enrique

[jloliveira]

Hi Enrique,

That explains it all.

Thank you again for your clarification,

João