Help with Germanium Leakage to Compute Fuzz Face Bias

[mac]

I want to update a little public program I wrote to compute the bias resistors of a Fuzz Face circuit, and Big Muff and Voltage Divider stages:

http://geocities.com/guitarfxs/exes/BiasCalculatorX.sit
http://geocities.com/guitarfxs/exes/BiasCalculator.exe.zip
(bug: change decimal point if you get strange results)

I want the program to include the effect of leakage. My starting point is RG test to measure Ge transistors real hFE and leakage. If I am not wrong ic* = ic + leakage = hFE*ib + leakage.
See this:
http://www.diystompboxes.com/smfforum/index.php?topic=45481.0

Is this correct? Can I consider the collector current as the sum of the 'real ic' and a small amount of current that finds its way to the emitter through the base?

Any suggestions and/or corrections are welcome.


mac

[brett]

Hi
That's correct -mostly. 

Of course, "leakage" refers to a single symptom (increased Ic) that is the result of a number of illnesses.  So leakage might be constant in some devices, and vary a lot in others. 

One factor that many people recognise is the way temperature drives leakage and Ic, and thereby Hfe and probably other things (and therefore tone).  Hfe also affects input impedance, which may be why Q1 in a fuzzface requires special selection.  From memory, Q1 sits on the edge of the Hfe - Ic "knee" with the standard Rc of 33k.

cheers

[mac]

Thanks Brett.


mac

[bioroids]

Hfe also affects input impedance, which may be why Q1 in a fuzzface requires special selection.

I think Q1 requires more attention because the change in leakage/hfe gets multiplied by the gain of Q2.

So if the leakage current goes to the emitter through the base, shouldn't it be amplified too?

Luck!

Miguel

Mac no habia visto tu programa hasta ahora. Buena onda!

[brett]

Hi

So if the leakage current goes to the emitter through the base, shouldn't it be amplified too?

Interesting question.  There probably is a component of Collector leakage that has been amplified through the base.
But what we measure is Collector leakage, so we already take into account any amplification.
cheers

[mac]

Just guessing...
From a quantum stat view, isn't the base a kind of finite barrier, a square potencial separating the collector and the emitter, and as there is a probability that an electron in the collector region can pass to the emitter region, leakage current can be seen as those electron waves that dig a hole in the base?

Miguel, are you saying that leakage could be amplified in the same transistor, or that a part of Q1 leakage goes to Q2 and gets amplified by Q2?

BTW, I've done the math and make a first build. The program behaves as expected, raising Q1 hFE or leakage slightly increases Q2 collector voltage fast. I am going to buil a real FF and compare the results. But the equations are very sensible to small changes so I do not expect great precision. As soon as I build the interface (thanks realbasic!) I'll post the links.


mac

[bioroids]

Miguel, are you saying that leakage could be amplified in the same transistor, or that a part of Q1 leakage goes to Q2 and gets amplified by Q2?

I think that part of the leakage current can be the result of the same transistor amplifying a smaller amount of curent that goes from the collector to the emmiter through the base. It's just a theory, but in any case I don't think it's important for these calculations.

I've run some tests on several Ge transistors at different temperatures, measuring leakage and Hfe. I can send you the data if you want to compare with the results of your equations.

Luck!

Miguel

[mac]

Miguel, por favor enviame esos datos cuando puedas. Gracias.

If eqs.predict results closer to the real thing I will post the eqs and the code I wrote using realbasic.



mac

[brett]

Hi.
I think RG's site (geofex.com) has some information on causes of leakage (or maybe it was in the forum here?).

A large change in Q2 bias with a modest change in Q1 leakage sounds right.  Leakage is a high-order function (exponential or cube or something) of temperature.  This effect is shown for normal, healthy devices in datasheets such as the 1N4148 datasheet. 

My limited experience indicates that devices with high leakage rates due to physical problems (cracks etc) also have leakage that rises rapidly with temperature.  This makes RG's recommendations about leakage for Ge devices somewhat difficult to interpret.  Where I live, daytime temps are 100 F in summer and 50F in winter, and the same device might leak 500mA in summer, but only 250mA in winter.

FWIW, I think the bias calculator is a good idea, even if it's only a first approximation.

cheers

[Joe]

I think the problem may be that the effective leakage changes depending on the biasing. So it's hard to predict beforehand. I've gotten closer by using 0.2 for Vbe and measuring leakage with the Geofex circuit, but with a 100k (or whatever the bias value is) from base to ground instead of open. I haven't tested this theory extensively yet.

[mac]

FWIW, I think the bias calculator is a good idea, even if it's only a first approximation.

That's the idea, mainly because calculation of Vc2 is sensible to small changes, in particular variations of Q1 properties, hfe, vbe and leakage. On the other hand, Vc2 is not so sensible to Q2 values.
The input variables are the resistances, vcc, hfe, vbe and leakage.

An example.
Q1:(hfe=70, Vbe=0.2V, iL=50uA)
Q2:(hfe=90, Vbe=0.2V, iL=50uA)
Vc2=4.84V

changing one variable of the above,
if iL1=100uA (all others unchanged) Vc2=5.42V
if iL2=100uA (all others unchanged) Vc2=4.83V
if Vbe1=0.1V (all others unchanged) Vc2=5.64V
if Vbe2=0.1V (all others unchanged) Vc2=4.80V
if hfe1=90 (all others unchanged) Vc2=5.35V
if hfe2=120 (all others unchanged) Vc2=4.81V



mac