Accurate DAC Reference Data/Voltages from P2 Silicon

Tubular

Here's some accurate raw data taken of each code level 0..255 from each pin P0..31 of "Board2"

I've run out of time for now, but will come back to this for analysis down the track. I post it here in case its useful for others. My intention is to use this DAC data, captured from a calibrated multimeter, to in turn study the performance of the ADCs on each pad.

The column data is in volts. The multimeter was fixed on 10V range for the duration of these tests. For each pin there are 260 readings - two GIO readings (second is more settled/accurate), the 256 code steps of the DAC, then two VIO readings (second is again more settled/accurate).

I'll come back to measurement of the slow comparator DAC at some later time. Chip has some thoughts about how to do this, which should work.

evanh

Here's the 123 ohm DAC linearity table graphed. First graph is using min/max DAC voltages as the ideal span, while the second graph is using GIO/VIO pin voltages as the ideal span.

EDIT: Updated with axis labels
EDIT2: Fixed the image resolution
EDIT3: Change "scale" to "span"

Tubular

Thanks for plotting all those evanh

It'll be interesting to see how these vary from chip to chip. We can take some broader measurements across a batch of assembled boards

evanh

Here's the spreadsheet. It can be duplicated and the first table filled with new data for auto-graphing. There is a couple of lines below the first table for changing which span scaling to use.

EDIT: Update spreadsheet with span labelling

evanh

Huh, that's strange, the resolution of the PNGs are way too low. It's not the forum software doing it either, they're done that way by LibreOffice .... sorted Had to use an export option in the file menu instead of right-clicking. I've not had to do that previously. I can only guess it's a new "feature".

PropGuy2

In an un scientific trial, I programmed and tested several DAC and ADC Prop2 smart pin pairs (jumpered together) , and found the resolution and accuracy was within 3-decimal places in all cases (+/- one digit count ) I measured all the voltages with a decent 4-1/2 digit multimeter. The example code I wrote has been running for well over 2000 thousand hours and continues to be stable and accurate.
The test example - I set three different DACs to exactly 1.000 / 1.500 / 2.000 volts, and read the corresponding ADC voltages within +/- one digit count (again verified by my voltmeter ) Hope this helps.