I do know you voltnuts are going to shoot me hard on this!
After a few days fighting to get working an STM environment I decided to have some fun. Just testing an ADS1115 (16b ADC, with internal reference, MUX and diferential PGA) as it seemed a nice toy, Arduino nano, adafruit example, see what it throws. Out of the box, no calibration, 50mV test gives results within 0.5% of the BM235!!!
Ok, let's take out the noise, some averaging... 0.1% WTF!!!
Of course, take it as it is, single sample with single 3.5 digits DMM. I'm not saying this will be my volt reference, but I guess there is quite some value in it! Could be useful for some projects.
JS