Sunday 15 January 2017

Hacking LED voltmeters

These LED voltmeters are cheaper than the 7 segment modules!



They come scaled typically 0-100V. The linearity is awful below ~1.0 and accuracy is around 10% above 1 and -10% down to 0.5 when it's around -20% These particular ones have had the part number taken off the microcontroller, but it's a safe bet they have a 10 bit ADC to cover the wide voltage range.  One nice feature of these is that they autoscale below 10V, so you get 9.99V then above that 10.1V up to 99.9V. The resolution +/-0.03V (0.02 on 1 module I have), so below 10V don't expect too much. Some modules out there only have 1dp and right justify the display.


I've marked the signal path from the yellow input wire. Basically it goes through the 22k pot, through a 20k 1% (marked 30C), then a 3k9 (marked 58B) to the adc input and also to a 1k (marked 01B) to 0V.
Or to put it another way, there is a potential divider of 23.9k of fixed resistance and a 20k pot in series, and a 1k to 0V. Assuming the pot is mid-position, the ADC input is therefore Vin (1/1+11+20+3.9) or ~0.0278 or (1/35.9)
I needed a 0-5V input to read 0-15 ish. A 18k across the 20k & 3k9 achieved this.
The basic accuracy at low inputs can be improved at the expense of having the meter flicking from 0.00 to 0.02 with no input.  For the above mod, a 1M resistor from pin 5 to 8 (+3v3) means it reads with less than 1% error to at least 4.0 (didn't test above that)
For an unmodified meter, a 1M5 does the same to 12V (didn't test above that)