Any electronics experts here? microvolt comparator needed

[rurwin]

The accuracy of reasonably cheap resistance meters scares me. For example, the top-rated one of fasttech,

• It reads from 0.01 to 9.99 ohms
• Accurate to +-0.4%

0.4% of 10 ohms is 0.04 ohms. That's OK if I'm building a 2 ohm coil, which to be fair is what I'm likely to be doing, but no use at all if I'm building a 0.2 ohm coil and significantly imprecise if I'm building a 0.6 ohm coil.

A DMM on its 200 ohm range is likely to be less accurate still.

I've built a resistance measuring device based on a wheatstone bridge:

In use, one adjusts R3 (a ten turn potentiometer with a calibrated dial) until the voltage/current between A and B is zero.

If I measure the voltage/current with a DMM, it works very well. I'm getting close to three digit accuracy.

However I would like to have the whole thing in a neat box. If I put a panel meter in place of the DMM, then the needle is going to hit the endstops hard before the knob is adjusted. I thought I could just use an op-amp comparator with no feedback and a couple of LEDs, but the bias voltage (or something else) seems to be making it uselessly inaccurate.

Can anyone suggest a simple and cheap circuit which will compare two voltages down to around 100 uV and give a high/low indication? It needs to be calibratable with standard DIY equipment, ('scope, DMM, etc) and preferably not require any calibration.

(I'm not sure of the values of R5/6. They are only there to keep the power in R2 within spec and the measured potentials away from the power rails. Working the figures again, I think a total of 57R should be suitable.)

 

[Alexander Mundy]

I have been busy as a duck at a June bug eating contest, but if I get any extra time I will design something for you. You will need to use 2 sections of a multi section op amp or comparator that works off a single ended supply. Here is the basic circuit.

 

[rurwin]

Thanks Alexander. If you do have time that would be great. I think I see where you are going with that, but...

I think what you are saying is that the error is only on one side of zero, so by comparing in both directions we can null it out. But which direction that is, and whether I can run a LED off the outputs directly are still mysteries. Is it important that the LM339 is open collector, or is that just the first example you found?