[SAADC] how to measure battery with more precision digits

[Louis]

[CONDITION] 1. R1 = 1.5M, R2 = 430K 2. GAIN = 1/2 3. RESOLUTION = 10bits 4. Single End

[ISSUE DESCRIPTION] 1. If voltage is changed by 0.1V step (4.1V -> 4.0V), the outputs from ADC is ok to distinguish. 2. However, if voltage is changed by 0.01V step (e.g. 4.12V -> 4.11V), the outputs from ADC is overlapped for some range, resulting in ambiguous readings to tell 4.12V from 4.11V, or vise versa.

[QUESTION] How to make SAADC work to get battery voltage with more precision digits: e.g. 4.1V, 4.11V, 4.111V, ....?

[Roger Clark]

If you need more than 10 bit resolution, you would need to use an external ADC to measure the voltage, which has more than 10 bit resolution.

You could investigate oversampling the internal ADC e.g. See

www.embedded.com/.../Oversampling-with-averaging-to-increase-ADC-resolution

However, this is a far from optimal solution, hence why you can buy 20 bit ADC's

You also start to need to consider the stability of your voltage reference when you are measuring to these sorts of resolution.

[Roger Clark]

nRF52832 has a 12 bit ADC, and the spec says

"14-bit resolution with oversampling"

This is 2 bits better than the nRF51's ADC

I'm not sure however I fully understand your problem.

If your resistor divider is 1500k and 430k , you can measure up to voltages of 4.246V as that divides to 3.3V at the ADC

For for a 10 bit ADC, thats 1024 steps so 4.246 / 1024 = 0.004V resolution. Which is better than you now say you require.

However....

I think you do not fully understand the operation of an ADC. See devzone.nordicsemi.com/.../

So you need you either need to use much lower value resistors (which may be impractical as it would flatten the battery). Or you need to implement a circuit described here

devzone.nordicsemi.com/.../