Prescaling on ADC battery voltage measurement

[Eliot Stock]

I'm trying to measure a scaled down battery voltage level on an ADC pin. Given my resistor values the voltage on the pin should be about 0.279 times VBATT, to fit within 1.2V. It's a lithium ion battery that shouldn't go over about 4.1V or 4.2V.

Here's how my ADC pin is set up. Note the "NoPrescaling".

	NRF_ADC->CONFIG	= (ADC_CONFIG_EXTREFSEL_None << ADC_CONFIG_EXTREFSEL_Pos) // We don't have any external reference pins.
		| (ADC_CONFIG_PSEL_AnalogInput2 << ADC_CONFIG_PSEL_Pos) // Use analog input 2.
		| (ADC_CONFIG_REFSEL_VBG << ADC_CONFIG_REFSEL_Pos) // Use internal 1.2V bandgap voltage as reference for conversion.
		| (ADC_CONFIG_INPSEL_AnalogInputNoPrescaling << ADC_CONFIG_INPSEL_Pos) // No prescaling.
		| (ADC_CONFIG_RES_8bit << ADC_CONFIG_RES_Pos); // 8 bit resolution.

There's some discussion of the arithmetic to be done on the NRF_ADC->RESULT in this thread:

devzone.nordicsemi.com/.../what-s-wrong-with-measuring-lithium-battery-voltage

My question is, why is the code in that thread using an ADC_PRE_SCALING_COMPENSATION value when the ADC is set to no prescaling?

Multiplying by 4 would get me a value pretty close to what I'm measuring with a multimeter, as it happens.

[Eliot Stock]

Here are my constants:

#define BATTERY_ADC_FACTOR (0.279f)
#define BATTERY_REFERENCE_VOLTAGE (1.2f)
#define BATTERY_ADC_MAX (1024)

And here's my arithmetic, with the values I'm seeing:

	// This gets us a fractional number from 0 to 1, a proportion of the maximum.
	float adc_result_proportion = ((float)NRF_ADC->RESULT/(float)BATTERY_ADC_MAX);
	// eg. 234 / 1024 = 0.229

	// This gets us the actual voltage on the ADC pin.
	float pin_voltage = adc_result_proportion * BATTERY_REFERENCE_VOLTAGE;
	// eg. 0.229 * 1.2V = 0.274V <-- Too low considering I measure 4.2V across the battery directly.

	// This gets us the corresponding voltage across the battery terminals.
	m_battery_voltage = pin_voltage / BATTERY_ADC_FACTOR;
	// eg. 0.274V / 0.279 = 0.983

Let me know if you need my resistor values, but they're not too far off those posted in the other thread. Same orders of magnitude, anyway.