What's All This Matching Stuff About? - part 1

People talk about matching transistors, but what do they mean?

There are a few instances in pedal circuits where matching transistor parameters to each other or to circuit requirements is essential for proper performance.

1. FET-based Phase-Shifters
Many Phase Shifters used JFETs as voltage-controlled resistors. The Phase 90 is a prime example. This schematic is from electrosmash.com. They have a very good explanation of how Phase-Shifters work, so I won't repeat all that here. They link to an article by R. G. Keen on JFET matching that is well worth reading.

Phase 90 (electrosmash).png
The transistor parameter that requires matching is the JFETs' Vp, the pinch-off voltage. Vp varies widely for the same part number. The trimmer resistor is there to adjust for the JFET's Vp, but there are four JFETs and only one trimmer, so matching Vp between the JFETs is essential for the circuit to function. How closely should they be matched? The closer the better; if you get all four withing 0.1V you're going to have a great Phase-Shifter. Get beyond 0.5V or so and it will be crap. If there is one take-away here it is that unless you buy a matched set, you will need to buy at least 5x as many as you need and test them yourself to end up with a good matched set.

2. Differential-transistor octave circuits

There are five types of octave-up circuits on PedalPCB. One of those circuits uses a pair of transistors to perform frequency doubling. An example is the Propolis.

TMR Bumble Buzz [Propolis] sch.png
Q3 and Q4 act as a full-wave rectifier, switching between two copies of the signal that are 180° out of phase. Without getting into the finer details of how or why this works, suffice it to say the two copies need to be identical for the strongest octave-up content. for that to happen, Q3 & Q4 need to be matched for Vbe so that they turn on and off symmetrically. In addition, some other parts need to be matched, specifically R7 & R8, R9 & R14, R10 & R13. If you use 1% resistors, you don't need to match them by hand. If you use 5% or 10% carbon comp resistors, get out your DMM. Also, Q2 needs to have hFE >100 so that the signals at the emitter and collector are the same amplitude.
 
Last edited:
Chuck, you've got a real knack for explaining these things in a succinct way pedal builders can understand.

As far as selecting JFETs for phasers, I've seen a lot of hand-wringing over using specific part numbers. Is there any attribute other than |Vp| that matters? Is there a 'goldilocks zone' for |Vp| for phasers?
 
Specific part numbers are not important, getting Vp in a zone that's compatible with the LFO and bias circuit is important. That zone depends on the specific Phase-Shifter circuit. Keen's circuit measures the Vgs required to achieve a specific Rds, 10K in this example. Having that and a Vp that's not too small is good enough. Ideally, we want to match Rds vs Vgs along the entire range of Vgs. We only need two points to do that, so in theory you could mix part numbers if you matched Vgs at two different Rds values. At the two extreme ends of the Vgs range, we have Vp and Idss. If you matched those, then the Rds vs. Vgs curve matches along the entire range.
 
And as we poor buggers can't pay you royalties or so we can pay you tribute by naming you Sir Chuck.... sort of royalties...
Cheers for that again!

So as I'm looking to build a propolis what is a recommended and easy enough procedure to measure and match Vbe?
Also: what matching would you suggest for the eqd Tentacles / Squidward?
 
... and while I'm at it: which parts would be critical to get right and matched in the Foxx Tone Machine / Fuzzy Foxx?
 
You ask a lotta questions for someone from New Zealand ;)

Vbe matching: use a real transistor tester if you have one, otherwise measure base to emitter with the diode setting on your DMM. Get 'em as close as you can, preferably better than ±25mV.

Squidward & Tone Machine: match the octave diodes for Vf, same as above. Use 1% resistors.
 
You ask a lotta questions for someone from New Zealand ;)

Vbe matching: use a real transistor tester if you have one, otherwise measure base to emitter with the diode setting on your DMM. Get 'em as close as you can, preferably better than ±25mV.

Squidward & Tone Machine: match the octave diodes for Vf, same as above. Use 1% resistors.
... well I'm born in Germany if that explains things a bit more.... ;).

Thanks for reply, I don't have a real transistor tester, just a cheapy ebay component tester. Will try with DMM and the cheapie tester and see how far it gets me.

And I'm sure I will have a lot more questions for you as we go.... thanks for spreading the wisdom.
 
I use a $20 Chinese component tester with very good results. I have compared the numbers against a curve tracer and they are plenty accurate for what we do.

Cheap Chinese Transistor Tester.jpg

There's a PF5102 in the socket. These guys don't measure Idss or Vp, just Id at some arbitrary Vp. Good for health checking JFETs, rough sorting and verifying pinout, but not very useful for matching. They are very good for testing germanium transistors because they measure Iceo & HFE accurately.

With a nickname like Cooder, I woulda guessed y'all was from Alabama ;).
 
Last edited:
It won't measure pickup inductance because the resistance is so high, but it will measure 500mH wah inductors no problem.

I just measured the primary inductance of a Mouser 42TM013 transformer I bought from Small Bear. 1.96H end-to-end, 132mH either end to the center-tap. The 132mH reading is plausible, but 1.96H is about 4x too high. This thing has issues measuring some inductors.

It's fun to test 2-color LEDs with it. You get a mini light show.
 
Last edited:
For testing GE transistors I use the R. G, Keen method, do you gusy recommend that as well? Should be fairly good way I guess with his expertise?!
The Chinese component tester seems off for those and doesn't get the leakage properly if I'm not mistaken.
 
I have measured Ge transistors on my Chinese tester and a curve tracer and they agree, so do not be too quick to discount the Chinese tester's accuracy. Do you have a link to Keen's method?
 
I have measured Ge transistors on my Chinese tester and a curve tracer and they agree, so do not be too quick to discount the Chinese tester's accuracy. Do you have a link to Keen's method?
On the geofex page 'technology of the fuzz face' he describes his method of measuring with a DMM under the header 'picking transistor for the fuzz face. What does Sir Chuck think of it?
http://www.geofex.com/

Edit: here's a better link. http://www.geofex.com/Article_Folders/ffselect.htm
 
Seen that method before. Steve Daniels describes the same method on his Small Bear how-to pages. It makes a proper measurement of leakage (Iceo) and DC gain (HFE). I expect the Chinese tester does something similar.
 
Thanks Chuck for another killer installment of The Boneyard! The information you put in these is Top Notch!

As for matching Ge transistors, I have used Steve Daniels method, I matched 2 Chinese Ge transistors out of a lot of them that someone gave to me. I then put them on a breadboard and cherry picked the bias resistors to get the proper voltages on the transistors, soldered them up and boxed it. It is a darn fine fuzz face if I do say so. I have a fuzz face with a matched set I bought. I compared the two, no difference really between them. They both sound great and track each other throughout the level and gain range. So Steve's method works really well.
I've also done this with Si NPN, but didn't worry about the leakage current, and it works just as well. To me a matched set and cherry picked bias resistors sounds better than using the trimmer pot for the bias voltages, but that's just my opinion
 
Thanks Chuck for another killer installment of The Boneyard! The information you put in these is Top Notch!

As for matching Ge transistors, I have used Steve Daniels method, I matched 2 Chinese Ge transistors out of a lot of them that someone gave to me. I then put them on a breadboard and cherry picked the bias resistors to get the proper voltages on the transistors, soldered them up and boxed it. It is a darn fine fuzz face if I do say so. I have a fuzz face with a matched set I bought. I compared the two, no difference really between them. They both sound great and track each other throughout the level and gain range. So Steve's method works really well.
I've also done this with Si NPN, but didn't worry about the leakage current, and it works just as well. To me a matched set and cherry picked bias resistors sounds better than using the trimmer pot for the bias voltages, but that's just my opinion
Great thanks! So what is a good practical methodology to cherry pick the resistor values and what are we aiming for?

I know, more and more questions.... :)
 
Back
Top