A proposed solution to germanium temp drift...

[m_charles]

Hi again. Haven't been on in a while, family etc but wanted to throw another one out here to get some input.
I have a family friend across the pond who I consider an electronics wizard (he's actually a Physicist for real!) I keep in touch with via email.
Only thing is, he has no knowledge of electronics related to audio, music, effects, etc.
So just for the hell of it, I posed a question to him.
"Do you know any way to help deal with the temperature fluctuations of a germanium transistor in this simple circuit?"
I sent him a standard Fuzz Face schem with the 10k trimmer at the collector of Q2. I was curious if he could suggest something besides the ole' trimmer on Q2.
Here was his response:

"This is a normal thermal issue, especially for germanium transistor.
The voltage drop between base and emitter is temperature sensitive. When temperature changes, current flows through transistor base changes, and then collector current changes as well. In this case, the voltage at Q2 collector certainly changes. I am not sure how thermally stable you need.
First of all, you need to use low temperature coefficient resistor for the one at Q1 collector (33K). This resistor defines the working point of Q2. If this resistor is stable, the voltage at Q2 collector is relatively stable. Second, trim pot normally has bad temperature coefficient (200ppm/C), it is good for you to use a good resistor (low temp, coefficient) and a smaller value trim pot to replace the trim pot in order to have less resistance change when temperature changes. Third, if all the changes above still can not meet your requirement, you need to use thermal resistor (temperature sensitive resistor) to replace the 33K with opposite temperature coefficient to compensate the change made by transistor."

Having no experience with temp coefficient resistors, I was wondering if this seemed like a plausible solution to any of you?
Thoughts?

[DiscoVlad]

What about a reverse biased Ge diode across the B-E junction of the transistor?
Like the Mk III Tonebender uses: http://fuzzcentral.ssguitar.com/3knob.php

[PRR]

With odd-lot Germanium parts in a poor circuit, the Vbe is usually not the heavy drift. It is the Ico problem.

Germanium leaks. Older, low-price, or sorted-out Ge can leak a LOT. This leakage is sometimes enough that a transistor will "amplify" with NO designed-in bias. This leakage doubles every 10 degrees C.

I suspect this is much-much more effect than the worst pot tempco.

> "When temperature changes, current flows through transistor base changes"

He didn't finish this thought.

Use this example:  http://fuzzcentral.ssguitar.com/fuzzface/fuzzfacepnpschematic.gif

In hi-gain lo-leak Silicon, we would assume the voltage drop across the 100K base resistor is very small. In lo-gain hi-leak Germanium, it is NOT. if base current is 10 microAmps, *either way*, the drop across 100K is 1V, *either way*.

Either way because we know we want 260uA in Q1 emitter. For hFE of 20 to 70 we need 13uA to 3uA gross base current. But collector-base leakage may be negligible or >10uA at room temp, and doubling when hot. So the externally-supplied current may have to be 3uA IN to the base or 7uA OUT of the base.

So the drop across 100K could be +0.3V to -0.7V.

In simple design we would assume the top of the Fuzz pot is equal to Q1 Vbe, say 0.15V for Germanium. Adding drop in 100K, the voltage at the top of Fuzz pot could be 0.45 to "less than zero" (actually it quits at zero).

We can't reduce the 100K much because it loads the guitar.

This means that Q1 must have reasonable hFE and _LOW_ leakage. If leakage is low, voltage drop in 100K is low, and has little effect even if it rises some on a hot day.

The Vbe temperature drift is secondary. About 2mV per deg C, or 20mV for 10 deg C. That means 0.150V or 150mV will drift down to 130mV. About a 15% change in Q2 current. Not great but not awful. Perhaps enough to shift the extreme-fuzz tone, but not to quit working.

[m_charles]

Hey Paul. I'll admit, I'm a bit lost on his explanation, and your reply, ha, ha.
So you're saying he missed the point in his solution?

[Mark Hammer]

There was an automatic Fuzz Face bias circuit posted by Gary Mould in 2001, for a "Servofuzz", that uses feedback via an op-amp circuit, from the collector of Q2 back to the base of Q1.  I can't post it from work, but I'll try and post it once I get home.

[aron]

What happened to the BritFace??? I guess it went down with geocities???

[DougH]

GFR proposed a way of doing it with diodes too, IIRC, back in the day.

[slacker]

What happened to the BritFace??? I guess it went down with geocities???

It still exists at something called reocites http://www.reocities.com/SunsetStrip/Studio/2987/index.html **warning you may be bombarded with some NSFW advertising depending on your browser settings**

[GFR]

I just took the simplest temp compensation possible straight from the pages of "integrated electronics, Milman & Halkias.

Theres also a FF with servo from Joh Holis, the "rock face":

[pinkjimiphoton]

aron, i have the brit face archived as a pdf if anyone needs it.

basically, all you need is a reverse biased ge diode (only works with germanium transistors) in parallel with the b-e junction of the transistor.

works great, i use it often.

it works by ,...kinda, as is my limited understanding... by counteracting the leakage of the ge tranny with the opposite amount of leakage thru the ge diode....kinda equalizes it out.

all i know is, now i can use my fuzzface ALMOST every situation.

i'm uploading it to my box account, will post a download link when it's done.

EDIT: here's the britface, archived from plate to plate via the wayback machine..

https://www.box.com/s/79i9el477a6u4vcwf2ju

[Mark Hammer]

As promised...

[Mac Walker]

As promised...

Lol, isn't that essentially a closed loop proportional controller?  I think you need to add a derivative and proportional term, for faster response!  Just kidding....

[EBK]

There was an automatic Fuzz Face bias circuit posted by Gary Mould in 2001, for a "Servofuzz", that uses feedback via an op-amp circuit, from the collector of Q2 back to the base of Q1.  I can't post it from work, but I'll try and post it once I get home.

My first necrobump! Woohoo!  Anyone (Mark?) still have this?  I'm a throw-lots-of-extra-parts-at-a-problem kind of builder, so I naturally find this idea appealing.