Matching transistors for a AceTone fuzzmaster

[knutolai]

I've decided to have another go at the Ace Tone Fuzzmaster fy-2 and want to try matching the transistors for the long-tailed pair to improve the octave-up. From what I've read it's the Base-Emitter Voltage Vbe thats the characteristic to be matched.
I found this method http://www.dragonflyalley.com/images/TransistorMatching/ianFritz-transmat0011_144.pdf (second page) but I was wondering if a setup with a simple 9v supply could yield as precise measurements?

would this setup be any good or are these unfavorable testing-conditions?

[kaycee]

I've built a lot of these superfuzz type circuits over the last 6months or so. I have found that using the hfe function on a standard multi meter is perfectly sufficient to get matched pairs that produce a good octave effect for these circuits. I just go through a pile and note their reading and then pair the closest ones up. I would suggest that lower gain pairs are better, say sub 200, but that's only my ears telling me, no scope or anything so YMMV. I usually use 2N3094 or 2N4401, but I'm sure many others are just as good.

[knutolai]

ofc my DMM does not have the Hfe function  
but I found this thread with a solution: http://forum.arduino.cc/index.php?topic=4842.0

thanks for the insight!

I'll try this:


Hfe = Lc/Lb

[PRR]

> would this setup be any good

Yes. It reads Vbe, which is what you are asking?

For best balance, read two transistors at once, so they get the exact-same Vb.

hFE *may* also be part of the balance. Remove R2, change R1 to 10Meg, and read transistors one at a time. Emitter voltage of half-supply is about hFE=100; lower emitter voltage is lower hFE and vice-versa. A very rough (couple Volt) "match" is very good here.

> a setup with a simple 9v supply

This _is_ a 9V supply, a meter, and some resistors. (The caps shown are not needed.) You need some resistance or it would just explode the transistors. You need a meter because you can't see the voltage that you want to know. You could avoid the two resistors R1 R2 with an added battery, but that's not really simpler.

[knutolai]

This _is_ a 9V supply, a meter, and some resistors.

I know. I drew it up   I was just wondering if that circuit would work the same way as the setup in the link I posted. As always thanks a bunch!

EDIT: R3 and R4 would be matched ofc

[PRR]

> I drew it up 

Apologies for not knowing one from another.

Yes, your drawing is much simpler, and functionally equivalent within practical limits.

If you can find a diode, _I_ would use that as the upper dropper and run it at a pretty good current. That can give a "Vb" node impedance near 7 Ohms, essentially zero drop for any purpose (even if you get a transistor with zero current gain).

Your caps reduce audio impedance but won't hold-voltage while a human reads a meter. If the circuit were liable to oscillate-- but it isn't. If your power is trash, a cap is small help (a resistor+cap is better), but better power is better. Power demand is well into battery zone.



If you can find an op-amp.... you could boost-up the difference to LED level, have a no-read no-thought Vbe-diff reader. But unless you have to produce a crate of matched pairs (or just love LEDs), this is probably more trouble than it is worth.

[knutolai]

Great Ill try your configuration with the diode. Would any general purpose diode do the job? The smaller forward voltage the better right? I have some 1N5819 handy..

[PRR]

> The smaller forward voltage the better right?

Why?

You need "some" collector-emitter voltage. Vbe is often specified at 5V, which in this case suggests a 4.4V diode drop from the 9V line.

Your 2k2+22k divider put about 1.4V collector-emitter. My diode makes it about 1.2V.

Does V(c-e) affect V(b-e) much?

Actually, if you use post-1967 devices, hardly at all. V(ce)/V(be) is similar to a triode's Amplification Factor "Mu". On clean Silicon devices Mu is around 5,000, and this is true all the way down to V(c-e)=0.2V. Changing V(c-e) from 0.2V to 5.0V will cause about a 1mV shift of V(b-e) at the same current.

So you "could" just nail the bases to 9V and let the V(c-e) be 0.6V, the V(b-e).

I recall some older devices where this was less-true, where as V(c-e) came below 1V they didn't conduct so good. That may be why I threw in a diode, to get a not-low V(c-e). But I now see this was over-cautious.

[knutolai]

How close is close enough for Vbe matching? I have two devices with 4mV difference. Good enough?

[PRR]

> 4mV difference. Good enough?

Your 100K+15K resistors will be far more offset, even if 1%.

If it is really important, use a plan with an offset trim resistor.



BTW, 4mV from any large run of BJTs from the same lot seems large.

[knutolai]

If it is really important, use a plan with an offset trim resistor.

Ah like in the superfuzz. Nice

BTW, 4mV from any large run of BJTs from the same lot seems large.

huh my DMM cant measure voltages below 1mV. Sure you don't mean close and not large? Other pairs i tested had differences of up to 15mV.