Modeling germanium transistors

[mojotron]

I'm working on modeling germanium transistors using si parts and I'm wondering what our collective knowledge might be able to pull together.

So, far I've done a lot with caps - not really that great, and I'm going to try the piggy-back stuff... but I'm wondering if others have found a good way to model a Ge transistor to get that really warm sound from an Si part?

[R.G.]

As I originally suggested - piggy back transistors for lower gain.

Do the search in the archives for "piggybacking".

[mojotron]

As I originally suggested - piggy back transistors for lower gain.

Do the search in the archives for "piggybacking".

This does lower the gain - but is also doubles the parasitic loss - and that's what I was thinking about checking out. So on a Si transistor this would act kind of like a cap between the base and emitter. But, most of the current producers of Si parts don't publish specs for parasitic loss - so to what degree this is tunable - it would have to rely on someone's ear.

A 2N2369 is really nice for the gain needed for a FF, but I was really looking for the "tone" of the model being like a Ge part.

[brett]

Hi.  
Here's my 2c.  Reducing the hFE has a dramatic effect on most circuits (e.g. fuzzface).  Next issue, as you've identified, is reproducing stray losses (capacitances, etc ??).  I suspect that the gain-bandwidth product of Ge devices was very variable, and although for example AC128s are rated at 1 MHz, they probably ranged from around 0.2 to something above 1 MHz.  In a circuit like a fuzzface, individual AC128s therefore roll off high-end frequencies to varying degrees.  When using Si transistors in a fuzzface, I add a capacitor across the 330/470 ohm collector resistor to roll off high frequencies (fc~4kHz, 6dB per octave).  That such filtering gives pleasant "Ge" tones suggests an Ft of about 0.5 MHz for the mellow-sounding ("good") Ge devices for the fuzzface.

You can probably work out the internal capacitances, etc that would result in such a charactistic.

cheers

[Paul Perry (Frostwave)]

Making the transfer curve more 'rounded'? maybe by putting a Ge diode as part of the load? I think for a fuzz, the frequency response is the least of it, and the roundness of the overloaded transfer function the most of it.