Matching transistors

[Dr.EM]

I will need to match some transistors for a project, so I googled for articles about how you do it. I got various results, most seemed to involve measuring currents and voltages for different models (though I never got a clear diagram of how to set up a test, closest was one for a MOSFET). Anyhow, i'm wondering, if I measure them on my multimeters transistor hfe setting, and a pair read the same, does that mean they are matched? I've heard of matched gain, but is that the same as matching to say 0.3v?

 

[Roff]

What are you going to use the matched transistors for? I have generally matched Vbe, which you can do (crudely) with an ohmmeter.

 

[Russlk]

Dual and multiple transistor arrays are available. These are matched by virtue of being made at the same time on the same silicon substrate. Check out Panasonic XN0553100L, and I believe that the old CA3045, 5 transistor array in a 14 pin DIP, is still being made.

 

[Dr.EM]

I'm making one of these up as a unit:

**broken link removed**

I have one in my synth which doesnt use matched transistors (the 3904 & 3906) and works, but i'm trying to make this one good quality. I assume its good to have all of them matched a bit? The ones parallel to each other on the ladder (bear in mind i'm not sure on how this works, but I know it does :lol: ). If anyone can show me a setup for doing it fairly accurately with just a multimeter on a breadboard, that would be great. I think it involves setting the transistor up with a fixed base current and measuring an output voltage into a fixed resistor, which would determine the gain :?: . If it only determines gain, then the hfe measure on my multiemeter is the correct thing to use?

 

[Wrighty]

Yes but there is many more considerations, like temperature co-efficiancy (sure ive spelt that wrong!), internal resistance, stray inductance and capacitance. I cant get on to that link that you have posted so dnt know what you are trying to achieve. If you go to RS, they sell matched transisters that were made at exactly the same time, this makes sure that they are matched (its the only way)

 

[Nigel Goodwin]

Have you considered that part of the sought after tone of the originals may be due to the use of non-matched transistors?.

 

[Dr.EM]

Yeah, I did try changing some transistors for lower noise models, and it did alter (and degrade) the tone. Thats why I will have to use the TIS97s, and not a matched transistor chip. Its just on the notes it specifies that the 3904 and 3906 should be matched to 3mv (that seems awfully close :? ) and I presume thats how it was done in the day? Is the hfe test the right thing to use?

 

[Nigel Goodwin]

Is the hfe test the right thing to use?

I wouldn't have thought so? - mV isn't an hfe unit.

I presume it's probably a measure of the Vbe voltage, aren't the transistors used pretty much as diodes?, and isn't there even a version that uses diodes?. It's really a pretty confusing circuit to understand!.

 

[Dr.EM]

Yeah, there are diode ladder filters too, and there is one where the filter uses transistors connected purely as diodes (the roland tb303 one I think does). This one does use them as transistors in the ladder, not purely diodes, but the transistors that need the matching arn't actually in the ladder. They appear to form part of the voltage control circuitry. Perhaps it assures it tracks at 1v per octave with the resonance on, otherwise the key tracking might go out as you went up the scale. For me, that isn't too important actually as I intend it as a stand alone unit, but I want to do it properly; is there any simple way to measure the vbe voltage? I only have a multimeter to use. Yeah, it is a confusing circuit, i'm not certain how it works myself, quite an ingenious design.

 

[Roff]

This should get you close enough. If you're really picky, you could match the resistors first. They don't have to be exactly 100k. They just need to be the same value, say +/- 1%. Any voltage between 9 and 12 should be fine.
It may not be easy, since they are not the same type transistors. You may have to experiment a bit with different part numbers to find two that will match, and even then, temperature coefficients will probably not be exactly the same.

 

[Dr.EM]

Thanks for that, I have some 100k 1% around. I'll just match them the best I can with what I have.

 

[Dr.EM]

Ok, I matched the resistors first, then tried the various 3904 and 3906 I had. I only currently have about 8 of each, but found a pair that read 0.601 and 0.600v respectively. The last digit there represents the mv, so they are within the +/-3mv of each other needed. The temperature certainly changes the readings I noticed, so I left them for some time to settle after touching them.

 

[Roff]

Ok, I matched the resistors first, then tried the various 3904 and 3906 I had. I only currently have about 8 of each, but found a pair that read 0.601 and 0.600v respectively. The last digit there represents the mv, so they are within the +/-3mv of each other needed. The temperature certainly changes the readings I noticed, so I left them for some time to settle after touching them.

That sounds good. On the schematic, I was indicating that you can also measure the matching differentially. Put one probe on one tranny, and the other probe on the other tranny.