Why round to 2 decimal places?

[shortbloke]

More of a question before I look to make changes and submit a PR.

The Automation HAT features a 12-bit ADC, thus there are 4096 values:

• 24V ADC, max voltage 25.85 a single bit is 0.0063 per bit
• 3.3V ADC, max voltage 3.3 a single bit is 0.000806 per bit

In order to show the resolution represented by each bit shouldn't the output be to:

• 24V ADC: minimum of 3 Decimal places
• 3.3V ADC: minimum of 4 Decimal places

Surely part of the purpose of ADC4 is to provide a higher resolution for lower voltage measurements?

Anyone wanting to handler a lower resolution can round the result themselves. Right?

[Gadgetoid]

The ADC is 12-bit, but the 12th bit is the sign-bit so positive voltages between 0 and 24v (actually 25.85v because of how the resistor divider is set up) so it's closer to 0.012v granularity.

Except the full-scale range of the ADC when reading a 0-25.85v value is actually set to 4.096v (not 3.3v) giving only ~1649 possible usable values, which brings us to a granularity of 0.015v (25.85 / 1649) for 24v and 0.002v for the 3.3v inputs.

In this instance we could probably still deal with 3 decimal places on the output voltage and let the end-user oversample the ADC and round as they see fit. So you're still right. A PR would be appreciated.

This is documented pretty exhaustively here - https://forums.pimoroni.com/t/automation-hat-accuracy/7252/3 (I should put a link to this post somewhere in the Automation HAT docs!)