From JCM:
this is something I too became interested in a few years ago and what I'm about to say is not by any means proof of anything - just food for thought, and something that me and Mark talked about Sunday for a bit ...
One of the things I wanted to try out by installing tripots all over the place was to see if I could find any higher amount of distortion sensitivity to component values and biasing - interestingly enough wide-range bias variation didn't seem to affect things that much for me !
... incidently I have a BLACK script BMP with the RED pi, I've heard it's supposed to be one of the better ones - anyway, mine kinda sucks in my opinion, but then again I tend to play with single coils more often than not and with a Paul it sounds quite a bit better but it's still sucky - so multiple caveats here ... hence my tripot clone exploration ...
anyway, there was mention on RG's site and also Jack Orman's I believe a long time ago (sorry if I'm wrong on that) about switching to larger signal cap values ... one thing I've found with bipolar transistor fuzz circuits is that if signal caps are too large then it seems to me that the clipping stages get overwhelmed and the circuits tend to choke instead of sustaining ... my brain tells me that with bigger signals caps follows larger charge and discharge currents which will tend to interfere with the base-bias current which will then cause the stage to shut off and therefore produce a chocking sound ... I don't haver data to back this up, it's just a heuristic argument on my part ... incidentally this base-bias component would be missing in a FET circuit like Doug's Meteor - conversely I surmise this base-shutoff effect is what causes much of the clipping in Joe Davisson'S Blackfire hi-Z base biasing rather than collector saturation ... hard stuff to demonstrate though ...
... anyway, this suggestes that using larger signals caps to get more signal feed-through may not necessarily be the way to go with Bipolar clipping circuits because of this operational "tradeoff" component ... in my BMP clone I've settled for the smaller signal cap values to produce a tighter and crisper sound (I think I ended up sticking with original signal cap values - these should be selected to taste of course) ... now, this is a little different to using a larger cap value in the VOX treble booster to produce a Rangemaster pedal, which becomes a mid-booster - this is because there's much more clipping in the BMP and leads more easily to shutoff ...
anyway, my tripot BMP, which made Mark laugh for 15 mintues because it's wired point-to-point on the fly and covered with a piece of Levi jeans material to prevent shorting against the box, was tuned over the course of a weekend and originally included three way switches on the signal caps to test out Jack and RG's recommendations ... with these trimpots the emitter resistors could go down to zero and collector resistors pretty much could go doubled - hence very wide bias range possibilities ... I found that these didn't make much of a difference at all - which was interesting ... and as I said, I ended up settling on lower signal cap values ...
... what the circuit did respond to mostly, and quite drammatically, was the 8.2k base-series resistors which were replaced by 10k trimmers and tuned to taste ... I seem to vaguely recall that 5k was where they ended up landing roughly (it's very sensitive to these two values so you do this by ear) to get a decent crispness with low output single coil prickups ... in essence it will affect the amount of distortion and compression sensitivity - which I think Mark noticed ... if they're turned down too low the circuits get overwhelmed for the same reason larger signal cap values don't seem to work too good ... anyway, that's the sum total of my experience with the BMP circuit ... incidently, I used 2n5210 low-noise high-gain transistors ...
... as an aside, Craig Anderton once mentioned sticking a Compressor before a fuzz to help define all this action - this may squeeze more performance out if the circuit ... also, today it dawned on me that modding my sucky Red-pi BMP by replacing the 8.2k resistors with 10k tripots might be a way to wake that sucker up ... I will try that and report on it ...