Draining leakage

[mac]

It is no mistery that a leaky ge can be problematic if you are building a FF or TB.
Leakage can be seen as a permanent collector current. This is not exact but it is usefull to do some estimations.
For example, in a FF 100ua of leakage across the 33k means 3.3v of pure leakage voltage. Add some micro amperes into the base coming from the 100k and you have 33k*hfe*base_current volts more.
The circuit reacts sending less base current to Q1 as leakage increases, or temparature goes up. If not, leakage voltage plus the true collector voltage would be more than 9v. At the same time Q1 collector, Q2 base and emiter go down. Since there is less emiter or collector current you need a higher resistor to get 4.5v.
Imagine a 200ua or higher leakage ge...
This estimation can be extended to any circuit, for example TB MKII Q1 and TB 3 knob Q3.

I tried to figure out how to use leaky ge in a FF without altering it much, ie, keeping the 33k and biasing the thing with a value closer to 8k2. To be precise, I want to build a very dark sounding FF using some Matsushita 2SD352 ge that leak a lot and have a freq transition close to 20khz.
I remember my experiments with a TB MKII, in parfticular Q1. If you look at the schem you'll notice that leakage bias the transistor, but what does the 100k to gnd do? It drains some leakage to gnd so as to make it bias close to 8v or so. More leakage and you need a to decrease the 100k to shunt excesive leakage to gnd.
Maybe this is why the Marshall Supa Fuzz differs from the Sola Sound.¿?

I used this approach and added a 100k from the base of Q1 to gnd. It shunted enough leakage to let me bias it with a value close to 10k at Q2 collector. How much leakage I drained? Vbe*hfe/100k.
At the same time I increased the input cap since the 100k lowered Zin a bit.
I did the same thing in a TB 3knob Q3 and it worked without modding the 18k.

I have to test if it helps with temperature.

mac