Of Germanium transistors, hFE and leakage

[splice]

I've built up a Fuzz Face clone some time ago and I'm about to build a Tonebender MK II (and possibly another FF, Bonamassa version) and I have some questions regarding Ge transistors, more specifically about the desired gain and leakage ranges.

There are many references to desired ranges for hFE and leakage. For example, a TB MKII has recommended hFE of 60-70 for Q1 and Q2, Q1 with high leakage and Q2 with low-medium leakage, Q3 at hFE 100-120 and medium leakage. Simple enough.

I have a transistor tester on my DMM but without leakage measurement. Geofex's handy circuit can test both, so that's the way to go for me.

The one thing that confuses me a bit though is that Ge transistors do not have a constant gain across varying current. Some recommended transistors do not attain the desirable gain range until a certain current. They may read very low hFE in the geofex circuit, but increasing the current to what the datasheet says make them attain the desired gain range.

Which brings me to my question: how were the desirable gain ranges measured in the first place? Were they measured in-circuit? Were they measured using a circuit like geofex's at the same current? Were they measured at the current specified in the datasheet for the transistor? If I want to build a circuit using the recommended ranges, should I just use geofex's circuit and disregard all other issues of datasheet specs and transistor type?

I've also read of some builder who worked for dunlop who said they didn't measure gain or leakage at all, but rather used a scope until the waveforms converged or something like that. Any idea what procedure that is? If I wanted to get a scope and match transistors like that, how would I do it?

Just a couple random questions I've been thinking about...

[R.G.]

There are many references to desired ranges for hFE and leakage. For example, a TB MKII has recommended hFE of 60-70 for Q1 and Q2, Q1 with high leakage and Q2 with low-medium leakage, Q3 at hFE 100-120 and medium leakage. Simple enough.

I have a transistor tester on my DMM but without leakage measurement. Geofex's handy circuit can test both, so that's the way to go for me.

Once the pedal hacker community had the geofex gain/leakage tester, they went wild with specifying gains and leakages on new nuggets of germanium they found. As with so many things in the pedal hacker mythos, the "right" values became the subjects of legend, and worse - advertising.

The one thing that confuses me a bit though is that Ge transistors do not have a constant gain across varying current. Some recommended transistors do not attain the desirable gain range until a certain current. They may read very low hFE in the geofex circuit, but increasing the current to what the datasheet says make them attain the desired gain range.

It's not complicated - all bipolar transistors have a gain that is a function of the collector current. It's quite low at low currents, rises at increasing currents, then starts to fall off after some peak when the secondary effects inside the transistor start becoming too significant. The rise of small signal gain with increasing collector current is in fact the basis of most VCAs and multipliers. One term is the signal across the bases, the other term is the emitter current for the pair and the more emitter current, the higher the diffamp's AC gain.

A side light to this is that there is no one transistor gain. As you've found, it's variable. I used to post this advice all the time until the world pedal hacker community got so big and so diverse that it wasn't useful to type it.

I designed the gain/leakage tester back then because people had no clue that germanium leakage showed up as "gain" on their meters, which tested at some unspecified current and other conditions. The point of the geofex tester was to remove one layer of mysticism and misdirection. It's a handy device, but very limited, and intentionally so.

Actually doing the kind of testing you're thinking about requires more testing that most people have the ability to (1) do and (2) understand if they did it. Testing transistor gain across current ranges and temperature variations AND leakages gets into a lot of data about every transistor. The supply of germanium in the world is limited, and frankly, most people are better served getting moderate-gain (i.e. 80-150 or so) devices with low leakage and then listening.

Which brings me to my question: how were the desirable gain ranges measured in the first place? Were they measured in-circuit? Were they measured using a circuit like geofex's at the same current? Were they measured at the current specified in the datasheet for the transistor? If I want to build a circuit using the recommended ranges, should I just use geofex's circuit and disregard all other issues of datasheet specs and transistor type?

Like so much of the legends and lore of pedals, they were posted to the internet as Absolute Truth by whomever happened to be posting. If I were you, I would consider all of the "Absolute Truths" about transistor gain to be someone's opinion gleaned from their one, two or ten builds, and measured by some method, which varied. Semiconductor testing is a complicated subject, as you're finding. 

I've also read of some builder who worked for dunlop who said they didn't measure gain or leakage at all, but rather used a scope until the waveforms converged or something like that. Any idea what procedure that is? If I wanted to get a scope and match transistors like that, how would I do it?

I would be very, very surprised if in a production environment the did not first pre-select transistors in to buckets of gain and leakage, and then build from the "best" buckets, discarding the obvious junk, of which there's a lot. The "using a scope and making the waveforms converge" is highly suggestive that they they had a "golden fuzz face" that someone in authority believed was what they wanted them all to sound like, and set up a test jig with the golden unit and the unit under test, then tweaked resistor values or subbed in transistors to get a scope trace to look close.

It's possible, but then you have to have a "golden unit" as a reference to start with, and you have to have the time (and a scope!) to do the tests.

[smallbearelec]

How were the desirable gain ranges measured in the first place?

In many of the early pedals, the devices were not measured for gain at all, so the pedals sounded awful. There are stories, probably true, of Jimi Hendrix trying one FF after another to find one that sounded "right". There's math that dictates the proper biasing for the voltage-feedback amplifier circuit, which someone had to know to define the values that you usually see in the generic circuit, but those resistances presume specific gain buckets for the devices. For more detail on this than I can give you, see "The Technology Of The Fuzz Face" at GEOFEX.

Ge transistors do not have a constant gain across varying current

True, and the current in the FF circuit is at the very low end of the datasheet figures.

should I just use geofex's circuit and disregard all other issues of datasheet specs and transistor type?

Yep. When we sort germanium devices, we measure gain with a Collector current of about 1 ma. Only a relatively small percentage hit the right buckets at that spec with low-enough leakage. Here's my FAQ:

http://diy.smallbearelec.com/HowTos/FuzzFaceFAQ/FFFAQ.htm

Note that the FF circuit is very tolerant, so you Can get happy results within a fairly wide range of gain buckets. Invest in a breadboard and have at it.

Regards
SD

[splice]

Thanks for the answers to both of you, you're confirming a lot of my suspicions regarding the desirable values. I'll probably use my retail Bonamassa FF as a golden master for a clone or two... Breadboarding ahead.