Fluke 8502A expectations?

This is really just a guess, but with a meter of that accuracy, the last digits will fluctuate due to the fact that you are measuring a very small magnitude with those digits. This is one of the reasons that the meter has a 4 wire resistance measurement. By having separate force and sense lines for resistance measurements, you get a more accurate and stable measurement.


I just bought a used 8502A and when measuring resistance, the least 2 significant digits won't settle down. Is this something that calibration will fix or is this just inherent in a 6 1/2 digit device?

You're complaining you can ONLY get 4 1/2 digits of accuracy? You can always average out the reading but if you took physics, you'll know how hard it is to get beyond 3 digits reliably.

I'll dip my oar in here. The 8502 will have erratic ohms readings, especially at high resistance values of 1M or more, if the current source is noisy. It's a fairly common problem. You should expect some last digit bobble, say +/- 2 digits in these instruments -- and more at very high resistance measurements where noise and stray capacitance are issues.. See if the 4-wire measurement, say at 10k ohm, is any better. I had an 8505 that I never could get to work right on resistance, even though everything else cal'ed well. I sold it before I got to the real problem, then read a note somewhere about the current source issue.

Hope this helps.
Think about the precision (not accuracy) of a 6.5 digit readout. That's +/- 2,000,000 count. Or a range of 4,000,000 count. So the least significant digit is 1/4000000th of the full scale, or 0.000025% (twenty-five millionths of a percent). To put it another way, it's -132 dBFS (132 dB below full scale). To put it yet another way, it's 22 bits of precision. This is around the noise limit of modern electronics. You will never likely see the LSD of a 6.5 digit meter settle down, simply due to noise. It's also why you don't see a 7.5 digit meter out there. It might be possible to measure with more precision if the electronics were cryogenically frozen, but otherwise you just can't get away from the noise.

Keep in mind too that resistance changes with temperature. You can very easily see this on even a 4.5 digit meter with a carbon resistor. Just clip it on, and note the reading, then grab it tightly in your fingers to heat it up (don't touch both leads) and you will see the reading creep up. Let go and the value slowly creeps back down. With metal resistors, the effect is much less (you can test for fakes this way). Even the small amount of current used by the meter when measuring resistance will cause some self-heating of the resistor and cause the value to creep, especially at 6.5 digits.

Don't forget too, that you can't measure anything with 6.5 digits of accuracy, even if the precision is there. The accuracy, even freshly calibrated, is orders of magnitude worse than the 0.000025% precision. The last few digits are only for comparing any two readings taken around the same time. Taking the same measurement at a different time can and will give different readings at 6.5 digits due to temperature of the meter and of the component under test, humidity, drift, etc.
True, with enough filtering, you can knock out the noise on any steady state signal. Is that your bench at work or at home? Not many here will have a nearly $10k voltmeter on their bench. I wonder though why would anyone need a 8.5 digit meter? Who really needs to measure to a few parts per billion? (and what is the accuracy vs. resolution?)

I think my second point, the temperature of the resistor, is probably spot on though. Using my old 6.5 digit meter, the reading is constantly in motion when measuring a carbon resistor. If you even breathe on the resistor, the measured value can shoot 1000 counts one way or the other. On a steady voltage (like a AA cell) the final digit does settle to within +/- 1 count.
The 3458A is on my home bench -- I'm long retired from active work, and never had one at work. For a while, I was deeply involved in precision standards as a hobbyist. Yes, resolution is not accuracy, but the engineering in the 3458A shows just how good such things can be, accounting for many issues of stability and drift. The 3458A will hold accuracy of a few parts per million over a period of a year for DC voltage and resistance, not so much for AC voltage or any current measurement. Still, it's the best, and my unit is now nearly 20 years old, and still meets factory specs....

You are indeed spot-on, not many folks need such precision or resolution, certainly not for general audio or many other kinds of lower-precision measurement use, where a good 3-1/2 or 4-1/2 digit unit will serve quite well. That said, there's some comfort in knowing that a measurement is as close to accurate as is possible outside of a standards lab. For me, that means being able to trust the instrument and not be wondering about it, and instead concentrate on the work in hand. These are the same reasons I have a very good Tektronix 'scope, a high-stability counter-timer, a GPS-DO frequency standard, and some other precise and very useful pieces of gear.

Not a bad idea to have a nice 6-1/2 digit meter, either, especially when they are so easy to get on the auction sites...