Ge Xstr Leakage: Red Flag or Red Herring?

[Joe Kramer]

Hi Friends,

Does it make sense to waste a good sounding and somewhat rare germanium transistor based on a technicality that even the original manufacturers were able to live with?  Isn't it sort of like divorcing Cindy Crawford for eating crackers in bed?

I mean, germanium transistors went to the moon in the 60's, and yet some of them were probably leaky.  Millions of items, from radios and TVs to missiles and satellites, were produced over decades using germaniums, a few of which also probably leaked a little.  Why, exactly, do they have to meet a standard that they were never held up to in the first place just for a fuzz box?

Unless I'm misunderstanding the problem of leakage (mostly temperature instability leading to bias changes), it seems to me that with the reverse-diode remedy (see Fuzz Central), that problem is solved.  Anyway,  show me an authentic, great-sounding, original Fuzz Face and I'll wager I can show you a couple of damn leaky transistors.  But if it sounds good, who cares?

Granted I may be missing some critical point about the matter (please educate me further), but otherwise, I would like to see the obsession with Ge xstr leakage relegated to the academic nitpicking it seems to be.  Talk about funny mojo stuff. . . .

Note to leakage sticklers: I'll be happy to take all your leaky OC71s, OC44s, AC128s, NKT275s, etc., off your hands and make some great sounding Fuzz Faces with them. . . .

Regards,
Joe

[MartyMart]

Hmm, on some occasions, its "Red Flag" I'm afraid !!
I have built quite a few "FF" variants, and have taken all the great advice
from the guys here and Phillips site.
I have several PNP and NPN Ge circuits which work great, thats because:

A- I bought a few "matched" pairs from either Steve/Banzai
B- On the others, I did a LOT of checking for hfe/leakage to find the right
ones...
C- I just got "Lucky" on a couple, with almost no checking... as on my
new "Range Blaster" NPN build ....

I have a few dozen AC128's/127's that just "wont" work right in any
circumstances, they are just too "unstable/leaky"     :cry:
They can always be used as a diode mind ......

Regards,
Marty.

[Doug_H]

Try breadboarding your circuit first and see how well it works. If you leave it on for a while through temp changes, phase of the moon, sunspots, etc and it stays reasonably stable (i.e. stable enough to produce consistently good tone), don't worry about it.

Doug

[R.G.]

Granted I may be missing some critical point about the matter (please educate me further), but otherwise, I would like to see the obsession with Ge xstr leakage relegated to the academic nitpicking it seems to be. Talk about funny mojo stuff. . . .

OK, education is always good.

1. Can a Ge transistor have too much leakage?
Obviously, yes. Some leak a few uA and some leak so badly that they're almost unbiasable. Even with circuit tricks to compensate out the leaking bias, some devices will not work.

Then there's the question of what kind of leakage. Germanium has taken a bad rap in the industry, for a couple of reasons. One is that Ge is simply leakier than silicon. We can live with that thousand-fold difference because silicon is so darn good at not leaking that Ge's intrinsic leakage is an annoyance, not a killer.  

The problem is that mankind learned to make semiconductors *at all* on Ge. That's because it's easy to work with, much easier than silicon. We learned to process semiconductor intrinsic material on Ge, learned to dope semiconductors on Ge, and learned to package semiconductors on Ge. As a result, there are more forehead slapping moments to be had as we fumbled our collective way around finding the right recipes.

Early Ge manufacture was a case of getting some pure-ish germanium, cutting it into tiny pellets, stacking it in a sandwich with pure-ish indium pellets and baking the resulting tiny sandwiches in a diffusion oven until the indium pellets "soldered" themselves to the Ge and then diffused into the germanium from both sides. The qualities of the transistor, the gain, leakage, voltage withstanding ability, depended on how pure the Ge and Indium was and how far and how much it diffused into the Ge pellet. As you can imagine, the first setups and the first manufacturing setups were not all that different from baking cookies. Sometimes the indium didn't solder to the Ge. Sometimes it did not diffuse in well. The thickness of the base region is determined by how close the two indium diffusions came to meeting on the inside. What do you have to do to control that? Well, you control the Ge thickness, its dopant purity, the contact area of the indium pellets, the oven temperature, the surrounding gasses (!! Air will not do!!) the length of time you diffuse it, and also which direction you throw salt over your shoulder.

When you got the newly made transistor out of the oven, you tested them. Every single one. As you can imagine, since you hand-made each one, every one was different. Once they were tested, you had a bunch of people with clean-ish hands, then gloves, then bunny suits work under microscopes to bond gold wires to the indium/Ge/indium pellets, and glue them down inside a package. Then test again. Then close the package. Then test again.

Air inside the package would oxidize the outer layer of the pellets, leading to surface leakage. Ugh. Surface leakage is leakage all right, but ugly, intermittent, noise and hiss ridden leakage in addition to just DC leakage. It took years to pin down what the contaminants were and then go build ever-more-complex, clean and repeatable fabrication factories that got better and better at making identical, high gain, low leakage devices.

Surface leakage is ugly indeed when it's on the input transistor of your mega-gain distortion pedal...

I have some of the last and best germanium transistors ever made. They're essentially non-leaking, only the intrinsic leakage, high gain, high voltage devices. Dandy stuff. I bought every one I could find through fair means and foul. Personal stock, like tucking away a case of a good cabernet. These are some devices that Fairchild made after the discovery that you could diffuse in dopants from a gas, not a solid pellet of Indium on a monstrous big three inch diameter wafer of germanium which had been grown by the Chrowalski process (and I probably didn't spell that right - doesn't look like enough consonants) and zone refined  to high purity, then dice them up into identical devices.

The problem with this is that by the time we had learned to make good germanium, we had also learned to make good silicon transistors, and silicon offered a thousand-fold difference in performance in many areas, and the electronics world turned to silicon. Germanium processing in general quit improving, as the old equipment was surplused and sold off to anyone who would buy it.

So the quality of your new old stock germanium devices depend on who made them, and where on that learning curve.

And leakage. The quality of leakage, as well as the quantity matters. Some is OK, some is ugly. None is desirable.

2. How much leakage is too much?
So we've established that there is a point that can be too much. How much is that?

When I came up with the cutoffs I suggest, it was clearly a judgement call. A decision was called for, and in writing for an effects building audience that sometimes didn't know what a leakage was, there was a strong need for a guideline.

But where to call it? If you set a guideline of 300uA, is a germanium with 301uA of leakage useless? Of course not. Is a germanium with a leakage of 3milliamps useless? Probably so, even with circuit tricks to suck the nasty stuff out the base.

For beginners, trying to duplicate some vintage circuit that, as in the case of the Fuzz Face, only had one in ten with good sounding devices, you don't want the newbies wandering around in the tall grass with device selection. So you give them a cutoff. Is that an absolute cutoff? Clearly not. But it gets down to what can a beginner handle. Obviously, some beginners can't handle the complexity of getting an original FF together, let along compensation circuits. And the more complex the compensation the less an inexperienced beginner will believe that it sounds just like the original, because its NOT like the original. How can they believe that the other stuff in there is not somehow subtly poisoning the sound of the "original" whatever that was.

So beginners need cutoffs, and back when the leakage guidelines were written, there were only beginners in this pocket of the woods.

Fast forward to now. A few of us supplied years of instruction to the effects building community. That made the pool larger and it finally attracted more knowledgeable people. Now there are beginning to be more people interested in building effects that are not starting from the water analogy.

Can the big boys cope with higher leakages in Ge devices, and take their lumps when it doesn't work? Sure they can. And I do enjoy seeing them contend with the real world.

As a last thought, what's a magic mojo? As you've used it, it's something that's a myth, a fable that is hyped beyond its true underlying, real-world facts. That's clearly not so with leakage. Ge leakage is a solid, measurable fact.

There's a continuum of leakage, both in amount and in quality of leakage. Somewhere in the continuum, it becomes not worth the trouble to keep working harder and harder to cope with leakage, and everyone calls it quits. Like the song, you can run marathons, pump iron, eat only sprouts, do whatever you like, you're still gonna die. Hard to call that eventual giving up "mojo".

That's why the idea I came up with to piggyback silicon transistors is so intriguing - it gets darn close to germanium sound without the warts.


Which gets us to one of your other questions:

Why, exactly, do they have to meet a standard that they were never held up to in the first place just for a fuzz box?

Exactly because the circuits that they were used in were designed to make the leakage, gain, and hiss problems not matter. I had to sit through some of the last design-for-devices-not-mattering circuit design classes. Did they work in "from radios and TVs to missiles and satellites"? Dang sure did. That's because the circuit designers busted their chops to make the peculiarities not matter. It's rapidly becoming a lost art as we now have almost-perfect transistors for almost-free.

In musical effects, you have the opposite situation. Particularly for germanium fuzzes you *want* the devices to matter. You want dat ole germanium buzz to shine through. That means you can't use the stock tricks for making the device not matter. Other tricks exist, but none of them will cure surface noise, and surface noise is highly correlated with total leakage.

At least that's what I think...

[vanhansen]

Wow, nice write-up R.G.  That is great info.  I vote that this thread be added to the FAQ forum.  I definitely learned something reading this.  Thank you Professor Keen.  

[MartyMart]

Thanks RG, very useful and interesting read !! ( as usual )

Regards,
Marty.

[nelson]

Enlightening. Cheers RG.

[Joe Kramer]

Thank you RG for the great description of early semiconductor processing.  It's awe-inspiring and humbling to ponder the brilliance of the people who thought up this stuff.

1. Can a Ge transistor have too much leakage?
Obviously, yes. Some leak a few uA and some leak so badly that they're almost unbiasable. Even with circuit tricks to compensate out the leaking bias, some devices will not work.

I may just be lucky, but I have found only about one in a hundred xstrs to be absolutely unusable, and these were merely noisy.  I've never yet found one that wouldn't bias up (knock on wood).  So far, given 100 Ge xstrs, I could probably give you 49 Fuzz Faces and a Rangemaster.

For beginners, trying to duplicate some vintage circuit that, as in the case of the Fuzz Face, only had one in ten with good sounding devices, you don't want the newbies wandering around in the tall grass with device selection. So you give them a cutoff. Is that an absolute cutoff? Clearly not. But it gets down to what can a beginner handle.

Your consideration of beginners (a class from which I consider myself barely graduated) is generous and right-minded.  IMHO, your "Technology Of" papers helped lay the groundwork for the present DIY resurgence.  I enjoy and refer to them all the time.

But certain kinds or uses of information can have a bogging-down result.  I remember once reading an online description of flanging that was several pages long and thinking, "Wow, a beginner reading this is going to think he needs a degree from MIT before he can ever use a  flanging effect."  It's mind-broadening and sometimes necessary to understand the physics and mechanics of how to ride a bike, but sometimes it's better to just jump on and pedal away.

Consequently, I think it's missing the point when somebody gets a handful of bitchen Ge's, races straight to the leakage tester, and comes away sorely disappointed when they fall short of a largely esoteric standard.  I say, put the horse before the cart, reach for your breadboard and give them a listen first.   Measure leakage as an afterthought, if at all.

As a last thought, what's a magic mojo? As you've used it, it's something that's a myth, a fable that is hyped beyond its true underlying, real-world facts. That's clearly not so with leakage. Ge leakage is a solid, measurable fact.

I recognize germanium transistor leakage as an undebatable and concrete fact, and I never meant to imply that it wasn't.   By calling concerns about leakage "mojo" I mean to say that it is in danger of becoming a fetish or hang-up that conflicts with the more important value of Ge xstr sound.  I'm not a "mojo-killer" either--there are plenty of mojo factors I fully embrace.  But just like humor, mojo is good as long as it doesn't spoil something else.

(FWIW, I did second-think using that mojo phrase because it seemed a little inflammatory.  Suffice it to say, my overall intentions here are gentlemanly and in good sport only!)

In musical effects, you have the opposite situation. Particularly for germanium fuzzes you *want* the devices to matter. You want dat ole germanium buzz to shine through. That means you can't use the stock tricks for making the device not matter. Other tricks exist, but none of them will cure surface noise, and surface noise is highly correlated with total leakage.

This is a great point.  You DO want the device to matter because these particular devices have a very unique and musical sound.  All the more reason to focus attention on the SOUND of the device and not a technicality that may unnecessarily invalidate it.  A glass of Dom Perignon with a speck of cork floating in it is still a glass of Dom Perignon!

Regards,
Joe

[RCZ53]

Good thread. Thanks RG for a very educational and interesting post.
Really great!!!

[DiyFreaque]

Good thread. Thanks RG for a very educational and interesting post.

Indeed!  Great info, RG.  And thank you, Joe, for pointing out that one's ears are the final arbiter of whether one's empirical measurements stand up to how things actually turn out.

Before I joined this forum, I was of the opinion that to get distortion, you clipped a signal and moved on with your life.  But, I've grown to appreciate that there's a helluva lot more to it than that.  It's not black and white; there's a whole spectrum of timbre to explore.  My hat's off to all of you in search of the ultimate fuzz.....

Cheers,
Scott

[Joe Kramer]

And thank you, Joe, for pointing out that one's ears are the final arbiter of whether one's empirical measurements stand up to how things actually turn out.

Thanks Scott.  Wasn't sure my mic was on.  

Joe

[Doug_H]

This has to be the best example I've seen lately of someone asking for the time and being told how to build a watch. :wink:

Nice history lesson, R.G. :wink:  :wink:

Doug