Why PNP?

[jim68000]

Hi

This is puzzling me: the three pedals I'd regard as being the definitive 60s gain/distortion boxes are the Dallas Rangemaster, the Fuzz Face and the Tonebender MkII. The connection is that all use PNP transistors and are therefore positive ground. The question I have is why? NPN Germanium transistors were available at the time and in most cases the NPN version of a circuit is exactly the same, with just the battery terminals reversed.

Is it just a coincidence? Were PNPs significantly cheaper than NPN? What advantages did positive ground offer?

Jim

[Buzz]

Hi mate.

Beginner here too... but I've been down this track.

When these pedals were made originally, the PNP transistors were way superior to their NPN equivalents, in ways like leakage and consistency.

The PNPs were easier to make to a higher quality due to manufacturing and design limitations imposed by the newness of the technology and the peculiarities of the element Germanium.

And if you want to build one now, with NOS parts, PNP is still the way to go.

Hope that sheds a bit of light, that's how I understand it.

Someone more experienced will probably give you a more technical ( and correct ) reply but....

Welcome!

[smallbearelec]

if you want to build one now, with NOS parts, PNP is still the way to go.

This can be true, depending on the compromise you want to make between cost and convenience. Yea, back in the day it was much harder to make NPN germanium with decent gain and low leakage, so there is less of that left today as NOS and it is much more scarce and expensive. But going PNP means either using a battery or a separate power supply. Don't try to go PNP negative ground; too many reported problems with oscillation. If you want to do a Rangemaster, check out the my Oh My Darlington:

https://www.smallbearelec.com/HowTos/BreadboardGeDarlingtons/BreadboardGeDarlingtons.htm

https://www.smallbearelec.com/Projects/OhMyDarling/OhMyDarling.htm

I worked out a way to use two low-gain, NPN germanium devices to simulate a much more expensive single part. The result is a classic pedal at very reasonable cost, with all the conveniences you expect of something modern. You can begin with a thoroughly-vetted breadboard build with lots of help and go all the way to the commercial version shown with the pretty BOSS-style enclosure. Welcome, and Happy Construction!

[jim68000]

That was the next question! I've got an expanding stash of Ge PNP low gain transistors (white dot NKTs and a variety of AC128s) and was thinking that there must be a Darlington solution to bulk up the gain but I don't have the experience to design such a thing. I'll breadboard that this evening. Thank you.

[jim68000]

@Buzz

Thanks. I thought there must be some technical reason behind it. As all the circuits require a healthy gain (and certainly healthier than I tend to see on the parts I've collected) I guess that must have been the thing.

[smallbearelec]

I've got an expanding stash of Ge PNP low gain transistors

Yer welcome! Yea, you can do a "perfect" PNP RM with these. But the absolute best use for them, IMO, is as Q2 in an NPN Sziklai pair. You get all the benefits of an NPN germanium device for very little $$. The only other thing you need is a low-gain piece of NPN silicon for Q1, and that isn't usually too hard to find. Let us know how it goes.

[pinkjimiphoton]

OR you can build a pnp fuzz and add a charge pump... if ya don't mind the power supply being more complex than the circuit.

pnp fuzzes just plain sound different than their npn counterparts imho...

[R.G.]

It's worth noting just for background knowledge: germanium makes better PNPs than NPNs. NPN germanium transistors have poorer performance than PNPs in terms of gain, voltage, power, and frequency, when made at equivalent levels of care and accuracy. Silicon however makes better NPNs than PNPs. For the same manufacturing precision, accuracy, techniques, an NPN silicon will be better, faster, yada, yada than a PNP.

This is for reasons to do with the semiconductor physics of the materials. It is difficult to make accurate mirror-image complementary devices (i.e. NPN and PNP) in EITHER germanium or silicon, and of course it's not difficult to make equal-complements of one germanium, one silicon: it's impossible.

PNP germanium was humanity's learner semiconductor material. It's much easier to make transistors out of it than silicon. But you can get higher performance out of silicon.

I won't even talk about SiGe - doped silicon with germanium atoms implanted in the crystal lattice that causes the charge carrier mobility to increase by stressing but not breaking the regularity of the crystal lattice.   

[DIMstompboxes]

I won't even talk about SiGe - doped silicon with germanium atoms implanted in the crystal lattice that causes the charge carrier mobility to increase by stressing but not breaking the regularity of the crystal lattice.   

You see... R.G. and even Mark never really share their secrets with us...hahahhaaa
Seriously now, can we make discrete leaky PNP using GE diode? like that discrete OPAMP I read before.
Thanks R.G.

[wavley]

I won't even talk about SiGe - doped silicon with germanium atoms implanted in the crystal lattice that causes the charge carrier mobility to increase by stressing but not breaking the regularity of the crystal lattice.  

We've been experimenting with SiGe devices here at the observatory quite a bit with pretty promising results.  Here is a paper from current Caltech/JPL PhD/engineer, the former director of our laboratory, and mentor of my current boss you might find interesting.  www.weinreb.org/sandy/LectureOnLowNoiseLna_May2010.pp Interesting side note, the Pospieszalski to which he refers is my first boss when I started here, who I still work with quite a bit.  And this bass player is his little brother, small world...

[R.G.]

I had a brief run-in with the fellows that invented (or maybe reduced to practice) SiGe devices back in the 90s before they were common, perhaps before it was public. At least I was sworn to secrecy. The semiconductor folks at TIC were off into trying to make this the Next Big Thing in semiconductors. It's not there yet, but it's got some real advantages at frequencies where the added cost of the material is worth the added performance.

[wavley]

I had a brief run-in with the fellows that invented (or maybe reduced to practice) SiGe devices back in the 90s before they were common, perhaps before it was public. At least I was sworn to secrecy. The semiconductor folks at TIC were off into trying to make this the Next Big Thing in semiconductors. It's not there yet, but it's got some real advantages at frequencies where the added cost of the material is worth the added performance.

The problem for us is that really only un-packaged ones are of any use to us because of resonances in the packaging at frequencies that don't matter to most folks and the package deforming on cooling to 15 Kelvin, so in many ways MMIC is the future, but discrete chip and wire has so many advantages at frequencies under 100 GHz.  It REALLY narrows down the field of devices we can use, right now we still rely on joint wafer runs we made with JPL in 1998, it's a good thing that a wafer has tens of thousands of transistors on it, but not all of them are good, sometimes I go to pick devices off the wafer and have to skip hundreds of them because they didn't pass the probe test.

That said, Mouser has 15 or so surface mount SiGe transistors for under $2.00.  I've thought about trying a fuzz circuit with them, but always forget to order them when I'm working up my parts list.

One of these days...

[Thecomedian]

Germanium has electron mobility of 3900cm2/V-sec, compared to 1350 cm2/V-sec for silicon, while Ge's hole mobility of 1900cm2/V-sec is the highest of any known semiconductor.

PNP is the movement of "holes", and as such, is less noisy. The processes back in the day which made germanium transistors noisy was due to the creation technology.

[R.G.]

I'd have to go look up electron and hole mobility and so forth, but the reason for germanium is that it's much easier to make bipolar transistors in germanium than in silicon. It's almost something that can be done in your garage. Almost.

Germanium was humanity's learning/training wheels semiconductor material. Early diffused germanium parts were literally made by making a sandwich of a slice of indium on both sides of a "base" of N-doped germanium and baking on a high-tech cookie sheet until the indium diffused into the germanium to make two P-type regions.

It may be that germanium has super powers, but its ease of fabrication is what made it first, and it's natural propensity for PNP types is what made PNPs most common. At one time, bipolars were hand crafted, tested to weed out the big % that didn't survive the cookie-baking, then hand-installed in transistor cases, so they were EXPENSIVE. One of germanium's not-so-super powers is that it's a slob about leakage. This is what drove the development of silicon devices. Later, when people got better about how to diffuse things in ovens, they figured out how to diffuse all three regions into the wafer from the top side only. This "planar" technology let humans make zillions of transistors on one wafer and then cut it up into dice and eventually package them automatically. Then they got cheap.

Can you point me to a reference on holes being quieter? My memory is that a lot of the intrinsic noise in bipolars is due to the formation and recombination of hole-electron pairs in a random fashion, and that seems to indicate they're about the same. Both NPN and PNP bipolars have current flow composed of majority and minority carriers, so both holes and electrons flow in both NPN and PNP transistors.

You're right - the crude processing techniques in early germaniums caused most of the noise; especially the lack of surface cleaning and passivation, which allowed oxides to coat the surfaces and offer paths for noise currents.