Well, as usual, I've spent much more time than I anticipated or wanted on the revision 1.4. So in this log I'd like to show where I got in the last two weeks, because rev 1.4 is still not finished.

If you've been following this project for at least a little while, you know that I switched the microcontroller from STM32F103 to STM32F373, because the latter is optimized for mixed signal processing and has much better ADCs. This new sigma-delta ADC is what prolonged the development so much - it is actually not easy to use it at all, there is very little documentation available and honestly, some of the claims made by ST are just wrong. But yesterday I finally managed to get acceptable results:

And I think those are more than acceptable results - the orange lines are 0,1 % tolerance. In other words, on the ±6 V range, the largest error we got was 1.6 mV. This almost makes it a true 6000 count meter, since this corresponds to the least significant digit. With this new ADC, we also have better resolution (16 bits), true differential inputs, dedicated REF pin and higher bandwidth (50 ksps).

And now is the time for bad news: this ADC needs calibration (albeit very simple one). This is generally true for most sigma-delta based converters, even though I was hoping I'll be able to avoid it. But the results are unusable without it. It is however true that most commercial multimeters are calibrated (since most of them use sigma-delta ADCs) and also the calibration is very simple - you just measure ADC values at the top positive and top negative value and input this into a special excel spreadsheet I made.

Another bad new is that this version won't be final - I made a few silly mistakes (as you can see from the photos) and the hardware will need to be modified a little.

And lastly, two photos from development: