That's exactly how it should be done, it has a name, it's called opposition method or something like that if the translation is good.
The proper way to do it is with two multimeters, one as a zero detector and the other at the pot decider so you don't insert a load into the pot as a new error source. The zero detector could be used as voltage or current, it's always done for high impedance sources, but you have to take into account how much current will mean how much voltage to know the definition you need as a zero current detector, if you use a voltage detector the impedance of the zero detector will still be loading badly the DUT, so for 1GΩ DUT and 10MΩ meter, you will have 1% of the voltage, so the error will be 100 times higher than the null error, usually accounted as one or half less significant digit of the meter, but could be lower if a 50% duty cycle on the sign detector is used. Note that the error in the zero detector is not critical as lone as it's properly zeroed before starting, so you can archive errors lower than the instrument errors, the error on the other meter, measuring the voltage across the pot is the one that counts, let's say you try to measure 50V with a 1%+1LSD DMM at the pot, 200V range, 0.2V as 1LSD plus 0.5V as 1% of the reading, you have 0.7V error, so you probably want to detect your zero to be less than a few mV, so you are still on the order of magnitude of error, too much definition and you won't get the zero by tweaking your source. Also you want your source to have a low noise so it doesn't affect the reading, using two or three stages RC filtering or even a regulator to get rid of as much noise as you can is not a bad idea.
12afael said:
mmm do you think the impedance of the multimeter is high enough to not load the capsule? a tipical impedance for volt meter is 10Mohms, it is very low compared to 1G. provably the error will be always less when the "null" voltage is near zero but there should be an error always isn't? the point if it is low enough.
I can be very wrong if so please enlight me.
cheerz
Rafael
The error will always be there, the thing is we can know how big it is and make it as small as we want (with certain limits of course but probably good enough for a few digits reading) The thing about a good measurement is always know how much error we are making, so we are not fooling ourselves, the error will still be there but as I mentioned, we will know how much it will be, even if we can't cancel it mathematically, we can know a range where our measurement is valid, the difference with this method is that we are measuring within about 2% with the example I quoted, rather than measuring 1/10th of the actual value (1000% error)
dfuruta, if you came out with this method by yourself you could be pretty proud, I've seen people who have seen this schematic scratching the head for quite some time before getting it, most indirect measuring methods are quite counterintuitive, and being able of using an instrument to get a measurement better than it's own error is quite good, since your zero detector could have 10% error and still be good enough to get 0.1% reading at the end of the day. Of course the other instrument must be better than 0.1% but it doesn't need to be high impedance at all, as long as you can feed it from your source, it may be getting few mA and being quite a load for the pot, but still being good enough, since you know, as your zero detector is saying, that the voltage at the pot is the same (within the margin) than at the DUT.
JS