how to hand match resistor to tighter tolerances, <1%?

[ryan750]

how to hand match resistor to tighter tolerances, <1%?

[SoNic_real_one]

Wheatstone bridge. You will need two equal 0.1&#37; resistors for R1 and R3 and a voltmeter. R2 and Rx are the ones that you try to pair - mimimize the voltmeter indication.

[TerryO]

Originally Posted by ryan750

how to hand match resistor to tighter tolerances, <1%?

Ryan,
That's a good question. I don't know how it's done with anything but carbon based resistors, but with those we would measure the resistor and taking those that were over spec (a little too high) we would take a triangular file and start filing a groove in the body of the resistor a little at a time while measuring often to check the results. When it's dead on, take some fingernail polish and seal the "gap."
Some of the resistors may be under spec (value too low) and you either discard them (which I'm too cheap to do!) or save them for applications that don't need close tolerances.

BTW: get as good a meter as you can afford and a precision resistor of measured value for checking your meter's calibration from time to time.

Since you brought it up, I'm curious as to how others would do this...Good topic!


Best Regards,
TerryO

[Mooly]

For "normal" resistor values say 100 ohm to 1 meg try something like the old Wheatstone bridge. There are lots of variations, you can use 2 disimilar resistor values too.
Wheatstone bridge - Wikipedia, the free encyclopedia

[kevinkr]

Regardless of what technique you use to match them up you need to get a sufficiently large size sample to be able to do so. Sometimes it just makes sense to purchase more tightly matched resistors in the first place.

The wheatstone bridge is an excellent suggestion for matching resistor pairs to very tight tolerances, however if you have a sufficiently accurate meter this technique is not always necessary. I am fortunate to own a Keithley 2002 which is a 8.5 digit dmm with very high accuracy, but some Flukes having 4.5 digit displays may have sufficient resolution as well depending on the resistor value. Cheap meters should always be used with a precision wheatstone bridge.

I measure individual resistors and "bin" them according to value measured to the requisite accuracy - this does a few things for me:

1. It gives me the possibility of tightly matching resistors
2. It gives me a tight range of absolute values where the actual value matters a lot (like riaa phono equalizers)
3. It tells me something about the statistical distribution of values in my sample (you would be surprised to see in some cases that the distribution looks nothing like the bell curve you might expect indicating that the someone may have sorted for a specific range of values before you did..)

You can apply the same technique to LEDs, caps, and active components.

The wheatstone bridge works great for matching, but for determining the absolute value of a component the bridge needs a third resistor that has to be adjustable and precise. Wheatstone bridges are pretty passe these days and can be obtained cheaply in a lot of cases and are useful for all sorts of precision measurements of resistance, capacitance and inductance values. I find them a bit cumbersome, and have had at least 3 that I can remember - all of which eventually got ebayed.

Genrad made a very nice digital LCR bridge, I wouldn't mind having one of those..

[Redshift187]

Here's a detailed explanation on how to do it:
Tutorial: How to match resistors up to 0.01% using a cheap DVM - Head-Fi: Covering Headphones, Earphones and Portable Audio

[star882]

If you have an ohmmeter that can connect to a PC, you can write a program that automatically keeps track of resistances, then finds the best matched pairs. Manually trying to find the best matched pair very quickly becomes impractical for any significant number of resistors. (To pick the best matching pair from just 50 resistors would take more than 1000 tests!)

And resistors are very cheap so buy a lot and match them. Put the probes side by side such that you can just put one resistor on them, tell the program to record the number, then put the resistor aside and put on another one. Then sort the resistance values and calculate the differences.

[TerryO]

Originally Posted by star882

If you have an ohmmeter that can connect to a PC, you can write a program that automatically keeps track of resistances, then finds the best matched pairs. Manually trying to find the best matched pair very quickly becomes impractical for any significant number of resistors. (To pick the best matching pair from just 50 resistors would take more than 1000 tests!)

And resistors are very cheap so buy a lot and match them. Put the probes side by side such that you can just put one resistor on them, tell the program to record the number, then put the resistor aside and put on another one. Then sort the resistance values and calculate the differences.

I believe that Frank Van Alstine had his children measuring and matching resistors years ago, using a setup like described above.

It really doesn't have to be much work at all, besides you're providing the food and shelter!


Best Regards,
TerryO

[Conrad Hoffman]

Covered well above, but I'll add that I've often purchased bags of 200 1/4W MF resistors from Digikey (probably Yaego) and it didn't take long to come up with many pairs matched to better than 100 parts per million. I get a piece of cardboard from a cardboard box and stick the resistors in the little holes at the edge to sort them. After setting up your Wheatstone bridge you can also dunk the resistors in a beaker of warm/hot mineral oil and easily see the temperature coefficient. A little hint for super precise resistor matching- don't be fooled by thermocouple voltages generated by dissimilar metals. For example, plated alligator clips are terrible. Use bare copper clips crimped, not soldered, to bare, not tinned, copper wire. The wire pulled from old phone and network lines works great. Clean the oxide off the wire.

Hey, match 11 resistors well, get yourself some rotary switches or even pin headers, and you can build yourself a nice Kelvin-Varley divider. Do it 6 times and you can have a nice divider that can be set to surprising precision- better than most meters unless you can afford that 8.5 digit treasure!

[TerryO]

Originally Posted by Conrad Hoffman

Covered well above, but I'll add that I've often purchased bags of 200 1/4W MF resistors from Digikey (probably Yaego) and it didn't take long to come up with many pairs matched to better than 100 parts per million. I get a piece of cardboard from a cardboard box and stick the resistors in the little holes at the edge to sort them. After setting up your Wheatstone bridge you can also dunk the resistors in a beaker of warm/hot mineral oil and easily see the temperature coefficient. A little hint for super precise resistor matching- don't be fooled by thermocouple voltages generated by dissimilar metals. For example, plated alligator clips are terrible. Use bare copper clips crimped, not soldered, to bare, not tinned, copper wire. The wire pulled from old phone and network lines works great. Clean the oxide off the wire.

Hey, match 11 resistors well, get yourself some rotary switches or even pin headers, and you can build yourself a nice Kelvin-Varley divider. Do it 6 times and you can have a nice divider that can be set to surprising precision- better than most meters unless you can afford that 8.5 digit treasure!

Those are some excellent tips. Thanks.

Best Regards,
TerryO