Simulating Additional Leakage in Ge Transistors

[Big Monk]

Kind of an odd topic but one I've been thinking about lately. Some Germanium fuzz circuits (Tonebenders in particular) do very well with leaky transistors.

What would be the advantage of simulating additional leakage, ideally through the use of base to collector resistors, in low leakage units that have a good base tone? Is there a benefit, i.e. in the form of noise reduction, ease of biasing, etc.?

I have a nice little experimental Tonebender board going to manufacturer soon and I have spots left on the board to either dedicated to these faux leakage resistors.

My motivation is that I have some nice generic NPN transistors that nail the gain bucket I want but are lower leakage the typically used in the circuit.

[mozz]

If you have a way to measure leakage, you can add the resistor and then measure it. I've done the experiment but can't remember what values i used or if it changed the sound to my liking. There's also a capacitance crappy transistors have, you can measure and simulate that also.

[Rob Strand]

Kind of an odd topic but one I've been thinking about lately. Some Germanium fuzz circuits (Tonebenders in particular) do very well with leaky transistors.

What would be the advantage of simulating additional leakage, ideally through the use of base to collector resistors, in low leakage units that have a good base tone? Is there a benefit, i.e. in the form of noise reduction, ease of biasing, etc.?

IMHO, the sound of most pedals is strongly dependent on the collector voltages.   The problem with ge transistor circuits is you only get the "correct" collector voltages with transistors which have the magic combination of gain and leakage.

I'm much in favour of what you are suggesting.  Adding some tweak resistors, or changing part values,  which take up the slack so you can get the required collector voltages with virtually any transistor.   

There seems to be a reluctance to change part value to reach the desired goals because people like to see the original part values.   However, no one complains about acoustic and classical guitar builders tweaking the top-plate thicknesses and brace thickness to compensate for wood variations to achieve the final goal.  Mainly because they only see outside of the box.

There's a few ways to simulate leakage.  One is a BC resistor, the other is a resistor from the base to the +V rail.   For small amounts of leakage it doesn't matter but if you have to add a low BC resistor say 1MEG then the Miller effect would amplify the effect of that resistor.   For a stage gain of 50 the additional input impedance caused by the leakage resistor would be a parallel 1MEG / 50 = 20k; when looking into the base.    If instead we connect a resistor to the supply to simulate the base leakage current then the equivalent resistor value would be about 2MEG.   The additional input impedance looking into the base is a parallel 2MEG since there is no Miller effect.

The point here is if leakage is a *current* we only need to inject it into the base.    A high value resistor connecting to a "high value" voltage source is a way to emulate a current source.   While in reality the base leakage current is supplied though the collector resistor there's not real need to add that in since the collector leakage = gain * base leakage is somewhat greater than the base leakage.   So this is why an an added resistor to supply will have less impact on the circuit than a BC resistor.

You can emulate leakage with diodes but you have no control over the leakage current.  In order to stop forward conduction of the diode you would probably have to use two diodes connected in inverse series.    Transistor current sources are an option but the idea is we are tweaking the leakage so the high value resistor should be a good start.

[iainpunk]

i'd put in a Jfet based constant current diode, and experiment with the resistance, between 500k and 5M is where i'd start.



something similar to this. the down side is that the current with the same resistor will vary from unit to unit, and it needs quite a bit of headroom, so the miller connection is out of the question here.

cheers, Iain

[Big Monk]

i'd put in a Jfet based constant current diode, and experiment with the resistance, between 500k and 5M is where i'd start.



something similar to this. the down side is that the current with the same resistor will vary from unit to unit, and it needs quite a bit of headroom, so the miller connection is out of the question here.

cheers, Iain

The issue here would be board space. I have enough room for 3 more resistors that can be configured as either BC units or B to Vs units. Definitely no room for 3 resistors and 3 FETs.

In any case, I'm not sure something this sophisticated would even be warranted. My issue is that the great sounding NPN Ge transistors I have can be tough to bias, as Rob noted. The Tonebender MKII/Supafuzz is my go to circuit and that thing was pretty much designed around PNP transistors with substantial leakage.

[Big Monk]

I went through my layout for my TB board this morning and I believe I'll have the space for additional (and optional) base to power resistors. My prototype board may have configurable connections so I can also connect them base to collector, but the main board is designed around 3 pole term blocks anyway, so that may not even be necessary.

Thanks for all the advice!

[bms19]

I have a similar problem with Russian transistors like GT308s they have 0 to 1 leaking so in many cases they are completely gated. Been trying to find a way to put a leaking simulation on them but so far no luck.
Ben

[Big Monk]

I have a similar problem with Russian transistors like GT308s they have 0 to 1 leaking so in many cases they are completely gated. Been trying to find a way to put a leaking simulation on them but so far no luck.
Ben

Have you tried simply playing with the collector resistors? The low leakage means that once you dial in the proper voltages, they should stay put.