I have bought ZB2L3 battery tester. It measures the discharge current by a 20 mOhm shunt resistor and amplify the voltage about 30 times using LM321. The op amp has typical offset 2 mV; in this setup each mV means about 50 mA error. This is unacceptable and the device measures about 80 mA less. How to fix it?
Simple solution would be to add a resistor to fix the offset. But what about drift? If I measure the offset now and use resistor to compensate is it reasonable to expect the offset stays low after a weak? And a year?
You should perhaps have used an opamp with much lower offset. Or even "zero" offset - they do exist.
The LM321 is not "precision" at all - its offset is actually as much as 7mV, not 2mV - you always use the
worst-case values in the datasheet, not the "typical" values, as that's asking for disappointment...
Precision opamps offer down to 250µV to 50µV offset voltages, although tend to be more expensive.
An example general purpose 5V rail-to-rail precision opamp is the AD8656 (250µV offset worst case),
but you'll find many more if you scan manufacturers' selection guides.
For zero offset the search terms are "chopper stabilized" and "auto-zero", for instance the TC913, though
that needs more than 5V supply, has 15µV offset worst case, and 0.15µV/C temp drift...
welcome to the analog world. It appears to be operating inside its specification. What is the tolerance of the shunt times 30. The posted circuit is designed with a DC bias to improve voltage measurement accuracy. When the terminals are not connected, the unit will display a small voltage. (~0.06v). This does not affect the actual measurement. If you short the input terminals, the unit will display 0v (mine does).
Thanks for the input. I did not design the circuit, only bough it in China. It is not my fault the cheapest op amp is used. What I did as a quick fix (I don't have a precision op amp to replace it) if anyone is interested:
The 510k resistor on the picture is between 3V3 from regulator and IN+. Since there is 330R between IN+ and GND this resistor simply increases the voltage on IN+ by about 2 mV. But despite this my device showed about 100 mA less. After some calculations I decided the right value is 262k, I used 270k standard value. This still left about -15 mA offset so I soldered a 3M6 resistor over the previous one (they are now parallel) and now the offset is +2 mA which is acceptable.
After fixing offset I have found the gain is also way off - about +5%. The gain is controlled by the lower 10k resistor. I have soldered a 200k resistor parallel with it and the gain is now OK.
Now I can wait and see how long it will take for the opamp to drift into unacceptable values.
