Waht is it about Ge transistors that helps them "clean up"?

[slashandburn]

Hey all,

Question that's been nagging me lately thats probably been discussed to death but i cant pin down with the search function. I'm hoping someone can enlighten me with an explantion or link to another thread that will probably go way over my head! Why is it that Germanium transistors "clean up" significantly when rolling off the guitar volume than the silicon counterparts?

My small-brained and basic understanding of the big differences (and reason Ge was phased out and deemed less reliable/consistent) are current leakage and temperature sensitivity. I cant imagine the temperature sensitivity is the magic here and in my own limited experience I cant say ive noticed Ge transitors with high or low leakage having a big impact on "guitar volume clean up" in say a simple fuzz face.

I know im probably opening up a massive can of worms here. Just curious. If Ge didnt do the roll-off clean-up so well I'd imagine we'd likely have abandoned it like the rest of the world. Whats going on and why cant we get silicon transitors to do the same thing?

Cheers
Iain

[Fancy Lime]

I'm no expert on this but I would imagine that the fact that He transistors are *usually* much lower in hFE, or beta, plays a big part here. The Ge BJT base eats a lot of current, but if you roll down the guitar volume, you limit how much current can come from the guitar. That means the transistors output swing becomes restricted and stays more within the clean range provided by the power supply.

This may be nonsense but this is how I always thought it is, without ever bothering to really find out.

Andy

[Bunkey]

I'm no expert on this but I would imagine that the fact that He transistors are *usually* much lower in hFE, or beta, plays a big part here.

This and the fact their clipping has a much rounder edge to it, so it's more 'clean sounding' even though it's still distorting the input waveform by definition (ie the output differs to the input). The actual clipping limit is determined by the gain of the stage, supply voltage and biasing exactly the same as a Si transistor. I think it's really just these soft edges that makes the perceived difference.

The higher leakage and junction capacitance of Ge's is going to appear like negative feedback which will further reduce the effective gain and roll off the high end content that contributes to a harsher more obviously distorted sound. This is Ge 'wool'.

I can certainly get just as good a clean up using low gain B108's as I can with Ge's of a similar gain - the clipping is just more defined on the Si for the reasons above and takes on a slightly different character.

That's my less-than-educated take on it.

[iainpunk]

I'm no expert on this but I would imagine that the fact that He transistors are *usually* much lower in hFE, or beta, plays a big part here. The Ge BJT base eats a lot of current, but if you roll down the guitar volume, you limit how much current can come from the guitar. That means the transistors output swing becomes restricted and stays more within the clean range provided by the power supply.

This may be nonsense but this is how I always thought it is, without ever bothering to really find out.

Andy

i think this is correct, Ge's need more current into its base for the same collector current. limiting the base current limits the collector current and thus the gain.
its not that Ge ''cleans up better'' but its that hey have less Hfe,
Si's with the same Hfe have really similar clean up feel to them, especially reverse Beta transistors, they clean up really nicely

Si diodes also have a harder clipping knee, which gives a more aggressive sound even when clipping to the same degree as a Ge transistor, the Ge sounds smoother due to its soft transition in to the distortion. the wave folder clipped even more agressive, but it had a low enough gain to clean up well.

i believe that gain and clean up are closely related, the best clean up i had was on a wave folder fuzz i had, it needs not that much gain to totally fuzz out because of the extreme harmonic content created by wave folding, this makes it easier to clean up with volume control. it had a max gain of 10x, which is less than a Tube Screamer, but it totally fuzzes out when the threshold of folding is reached, and turning down guitar volume to 7 cleaned it up nicely.

cheers, Iain

[slashandburn]

Ah of course. That makes a lot a sense. Thanks!
The lowest gain Si transistors I've used were around 90hfe and made a really good Meathead. I don't remember them cleaning up it's not something I was looking out for.

I'm sure I have some 2222's around 150 hfe. If I can find them I'll have a play around.

[teemuk]

You can reduce silicon transistor's hfe by bypassing its emitter with another diode and some series resistance. Simultaneously the hfe decreases the clipping characteristics also turn softer. Certain Vox amps used this technique to emulate germanium transistors, certain to emulate "tube sound".

Do note that input circuits of ancient fuzz effects also tend to interact a lot with the source impedance. Varying source Z in, for example, Fuzzface alters both loop gain of the effect and resonant characteristics of the pickup. In addition, the ancient circuits are rather sensitive to effects of device gain. So it's not just about what type of semiconductor material an active device features.

[mozz]

 I think, in a fuzz anyway, it is more the circuit than the actual transistor. I'm using germanium's with a gain of 230 and they do clean up nicely, just not on 8 or 9.

EDIT: also the interaction of the guitar impedance (& vol control) as was mentioned. Often the cleanup is still ragged around the edges so to speak. I could probably just use a nasty fuzz and the volume control on the guitar and just do about any songs i doodle with. Doing some (weird) experimenting right now and the signal generator (600 ohms) is making the fuzz distort when down under the 10 millivolt range. The leakage, even slight and softer knee make it sound more pleasing. Measuring distortion too, about 65%. 2 high gain germs brings about 80% distortion. I do not have the second scope channel hooked up to monitor the filtered output(you would see twice the freq or 3 times usually). Even cleaned up sounds are about 10-25 % distortion.

I did AC beta tests a while back and often some had higher gains there when their static gains were similar. I can see drastic differences of the clipping on the scope with even a 10 ohm emitter resistor never mind a 75 ohm. I'm also using stock resistors wired wide open no pots and adjusting the Q2 emitter resistor till i get near 4.5-5v on q2c. 560 ohms gets me 4.6v. 2k trim pots on order. Q1 high gain germanium 230/.090 . Q2 is being swapped to silicon 150-200 and almost at the point a germanium is a waste in the Q2 slot. Works down to 2s322 (30 hfe) but sound great with 2n6519 (150).

[iainpunk]

2n2222a transistors sound good in reverse Beta operation, just flip the C and E pins around, and you lose about 2/3rd of its Hfe.

cheers

[slashandburn]

Cheers again. That sounds like a worthwhile experiment, Iain! Knowing me I honestly wouldn't be surprised I've done that accidentally somewhere along the line.

Derailing my own thread a bit here, but the trick teemuk mentioned about reducing transistor gain just reminded of something else I'm sure I read somewhere: Simulating Ge leakage with Si transistors. I've probably got this back-asswards but I'm sure I recall seeing in a tonebender thread that a resistor across two of the pins does the same thing?

I think, in a fuzz anyway, it is more the circuit than the actual transistor. I'm using germanium's with a gain of 230 and they do clean up nicely, just not on 8 or 9.

230 hfe is a bit out there for Ge, no? I'm sure nof my modest stash are half of that at most! In the ones I've got that might creep above 130 are those enormous russian GT402.

[KarenColumbo]

You can reduce silicon transistor's hfe by bypassing its emitter with another diode and some series resistance. Simultaneously the hfe decreases the clipping characteristics also turn softer. Certain Vox amps used this technique to emulate germanium transistors, certain to emulate "tube sound".

Do you happen to have a schematic fragment about? I'd love to ,,read in" into that

[slashandburn]

You can reduce silicon transistor's hfe by bypassing its emitter with another diode and some series resistance. Simultaneously the hfe decreases the clipping characteristics also turn softer. Certain Vox amps used this technique to emulate germanium transistors, certain to emulate "tube sound".

Do you happen to have a schematic fragment about? I'd love to ,,read in" into that

Sadly not, couldn't even say for sure if I actually read it or just dreamt it to be honest. Just something I think Iremember reading when I was looking for something else and thought sounded interesting.

could it be it be simulated? Is leakage just current that gets lost? (Genuine question, I really don't know me arse from my.elbow) How do we "fake" that?

[Bunkey]

You can reduce silicon transistor's hfe by bypassing its emitter with another diode and some series resistance. Simultaneously the hfe decreases the clipping characteristics also turn softer. Certain Vox amps used this technique to emulate germanium transistors, certain to emulate "tube sound".

Do you happen to have a schematic fragment about? I‘d love to „read in“ into that

Sadly not, couldn't even say for sure if I actually read it or just dreamt it to be honest. Just something I think Iremember reading when I was looking for something else and thought sounded interesting.

could it be it be simulated? Is leakage just current that gets lost? (Genuine question, I really don't know me arse from my.elbow) How do we "fake" that?

The most common Ge emulation trick I know is just using a C-B capacitor (100-220pF) to roll off the highs by miller effect. It's literally just negative feedback of the high-end content and that does a halfway decent job of softening things up. This technique is used as bandwidth limiting in a whole lot of pedals and amps already.

To quote a post from the forum which shall not be named:

"leakage in Germanium Transistors has alot to do with the inherent electrical properties of Germanium itself, (this may sound a bit technical) basically it has to do with how tightly or loosely the electrons are held in one of the shells surrounding the nucleus of a Germanium atom, this particular shell we're interested in is what's called the "Valence Shell" and determines the electrical properties of the Germanium atom, as it happens, the electrons in the valence shell of a Germanium atom aren't held in it as tightly as they are in a Silicon atom, because of this the electrons need less energy to get them to leave their place in the Valence shell, the energy could be in the form of electrical, or, thermal energy, which is why Germanium is so temperature sensitive, my understanding of it anyway..."

I'm not a betting man but I'd be quite happy to concur that the inherent chemistry of the substrate used has a lot to do with its behaviour.

Leakage itself is just current that gets through the reverse-biased C-B diode junction and then contributes to the B-E current. It's a bit like having a resistor there, except it isn't because If you put a resistor in place; that doesn't then offer the same characteristics as a Ge.

[Bunkey]

  I wonder if operating the Si around its breakdown voltage would make it behaive a little more like a Ge in this respect 

I think this is getting into the realms of what Iain was talking about with reversing the C and E on the 2N2222.

If you take the BC108 for example, its CB breakdown is 30v but its BE breakdown is just 6v, so operating it in reverse would exceed that breakdown voltage. Maybe if you ran it right on the limit at 6v (as this is easier than 30v), you might start to get a little bit of breakdown leakage current similar to Ge leakage whilst still having the junction predominantly 'reverse biased'.


Going to have to experiment with this one. I'm just thinking out loud here and could well be chatting a bunch of grapes.