Am I completely off on this?
No, not completely off. Those things do contribute to the fatalities, of course.
I have to admit, as I do in my postings on this subject when it comes up yet again, that I do not really understand 100% of the problem. If I did understand it, I could do a lot better than saying "here be dragons."
My experience with reversed-ground pedals started from doing a newly-commissioned custom build of some two-transistor pedal that originally had a positive-side-ground. No problem, I'll just do a reversed power/ground and be done with it. Needless to say, it didn't work. Some days of debugging later, I had found that when it had reversed ground, it oscillated in the FM radio band. The only change needed to make it work perfectly again was to reverse what was considered signal ground compared to the power supply, with concomitant changes to the grounds on input, output, and controls. Note that the DC conditions inside the active circuit did not change, so bias was untouched. Neither were there beer soaked and corroded contacts and components.
Well, OK, I did try drowning it in beer in frustration at one point, but then I had to rebuild it, and the re-build did much the same.
I said to myself, surely this is a power supply bypassing problem or one of those RF black magic things that needs small ceramic capacitors sprinkled in at odd places to suppress/enhance/extend RF magnitude/phase response and/or suppress spurious feedback. Another few days convinced me that capacitor-sprinkling was far inferior to ground changes.
Ahah! (I said to myself) all I need to do it better bypass the power supply and this will go away, as it is one of Maxwell's Demons camped inside the electron paths and those are banished by good power supply decoupling.
No good. Make it positive grounded, it works. Make it negative grounded, it sings but refuses to work. I rewired it so often, including two new and different versions of hanging-garden star grounding, that I wore some leads out. Still, all that works is putting the ground back where it said to in the schematic. I ran out of tricks, both of mine and my grizzled ham-radio friends who also happened to be degreed and practicing EEs. Their collective wisdom was "don't do that, most especially if it works with the intended power supply wiring".
I still do not understand the phenomena completely. It gets better with good power supply bypassing and immaculate wiring, as those have fixed some other things I have done with reversed ground power wiring. But I have run into a few others that can't be fixed with any combination of tricks I've collected over the past 40 years.
I'm forced to conclude that indeed, sometimes there are dragons there. Maybe not often, but sometimes. The incidence is not zero.
But all of these things do not really apply to designing a new PMOS or PNP effect, which needs its own new bias anyway, do they?
No, your comments don't apply to new PMOS or PNP effects. But Nyquist and Mother Nature do. I don't think that it's always bias. Every hard case I actually had in my hands was RF oscillation, including oscillation at frequencies that a 20MHz scope thought didn't exist. PMOS, with its very high frequency response, has the potential to make that worse, which is why I was on the point of suggesting that you put 100R to 1K resistors in series with those MOS gates as close to the FET as you can get them, as some situations can have MOS devices oscillating from lead inductance leading into the gate, capacitive load on the source, or both. But I digress. I don't think that just getting the bias right and not changing is a slam-dunk answer. It's an answer to some degenerative cases, but then...
So what I am wondering is: Why no new designs?
I suspect that it's just become a positive-power world. Silicon's leakage superiority had it simply eat up the bipolar world from germanium, hole and electron mobility be d@mned. Silicon is more naturally an N-type, although advanced processing has recently (last decade or so) made PNPs as good as NPNs, mostly. I think it's more that you have to have a positive reason to pick a harder to find and more expensive part to go P-type.
But for PMOS or Ge-BJTs there is some benefit to be had, imho.
And that leaves us with opinion. Well and good. Many if not the vast majority of effects are set up using the side effects of devices and circuits. If you have discovered that you like some characteristic of PMOS, great. Go with it. There's no reason to do any of this at all unless we are discovering new shiny things that please us for some reason.
Gemanium BJTs - that's another matter. The Fuzz Face has had an - um, well, effect, I guess - all out of proportion to what it really does. It is a good idea to remember that simply rubbing germanium on an effect does not make it sound better. The purveyors of effects which are simply the 27,386th clone of the Fuzz Face with secret-sauce "mods" to make it sound like flying saucers and lasers would have you believe that germanium is magic, even if it's just used for an emitter follower in front of everything. The Fuzz Face itself was not universally good, and many of the real, no-fooling Fuzz Faces sound simply terrible compared to today's clones, as we have learned to measure the gains of the transistors, sort out the obviously bad ones, then compensate bias for the remainder. But simply using germanium isn't magic. Maybe someone can design a new circuit using a side effect of germanium if there are enough germaniums in the world left to use.
So don't take this as any kind of "don't do that" message. Go do what sounds good. But when and if you run into one of the funny situations, perhaps after increasing the gain a bit, or doing a new layout, remember the caveats. You'll almost certainly be able to work around the issues, having been forewarned.
