What Russian Germanium Transistor to Stock up on?

[Thecomedian]

http://www.learnabout-electronics.org/bipolar_junction_transistors_06.php

http://www.fairchildsemi.com/ds/PN/PN2222A.pdf

Let's take this example. at Ic = 150mA and Vce (voltage difference from collector to emitter) of 10V, gain is 100-300. At Ic = 500mA, gain is 40 minimum. That's a huge difference. Conversely, gain at 1.0mA is only around 50 minimum, which could mean that it's actually in the right gain pockets for the fuzz face circuit, provided the biasing keeps the gains around 70-120 or whatever the "good" numbers are, through amp "starving". When using most Ge parts for the fuzz face, such as  NKT175, they bias best at 1mA, which is approximately where Q2 lands anyway in the original circuit. AC128, on the other hand, is a 50mA bias http://html.alldatasheet.com/html-pdf/131441/ETC1/AC128/94/1/AC128.html so its a lot starved, and the highest gain versions of AC128 will probably be within the pocket when putting it in, if all transistors have as sharp a drop off in gain for "undervolt/undercurrent" biasing, but some graph help will prove or disprove that. http://alltransistors.com/pdfview.php?doc=ac128.pdf&dire=_valvo

This is what all those graphs on data charts help visualize: The way it's gain will react to different voltages and currents. That's why there are so many different types of transistor out there, they all have different reactions. Then again, different datasheets and different mfg and internet help will give you different answers. Some say hFE bias means the current going into the base as a function of how much to apply to get the right gain, some say its Ie, and some say its Ic, as per the AC128 and PN2222 datasheets. I'd probably assume Ie unless it specifically states Ic or Ib as the function, since that'd be almost what Ic would want anyway if it turned out Ic was the intended function to bias around by the mfg.

[tjdracz]

http://www.learnabout-electronics.org/bipolar_junction_transistors_06.php

http://www.fairchildsemi.com/ds/PN/PN2222A.pdf

Let's take this example. at Ic = 150mA and Vce (voltage difference from collector to emitter) of 10V, gain is 100-300. At Ic = 500mA, gain is 40 minimum. That's a huge difference. Conversely, gain at 1.0mA is only around 50 minimum, which could mean that it's actually in the right gain pockets for the fuzz face circuit, provided the biasing keeps the gains around 70-120 or whatever the "good" numbers are, through amp "starving". When using most Ge parts for the fuzz face, such as  NKT175, they bias best at 1mA, which is approximately where Q2 lands anyway in the original circuit. AC128, on the other hand, is a 50mA bias http://html.alldatasheet.com/html-pdf/131441/ETC1/AC128/94/1/AC128.html so its a lot starved, and the highest gain versions of AC128 will probably be within the pocket when putting it in, if all transistors have as sharp a drop off in gain for "undervolt/undercurrent" biasing, but some graph help will prove or disprove that. http://alltransistors.com/pdfview.php?doc=ac128.pdf&dire=_valvo

This is what all those graphs on data charts help visualize: The way it's gain will react to different voltages and currents. That's why there are so many different types of transistor out there, they all have different reactions. Then again, different datasheets and different mfg and internet help will give you different answers. Some say hFE bias means the current going into the base as a function of how much to apply to get the right gain, some say its Ie, and some say its Ic, as per the AC128 and PN2222 datasheets. I'd probably assume Ie unless it specifically states Ic or Ib as the function, since that'd be almost what Ic would want anyway if it turned out Ic was the intended function to bias around by the mfg.

Thank you very much! Definitely sheds some light on the subject. Much associated

[Thecomedian]

no problem. Throw into that the monkey wrench of different B-C junction capacitances, and different transistors for different fuzzes may sound very different.

Here's some bits from http://www.amazon.com/Designing-Audio-Power-Amplifiers-Cordell/dp/007164024X

Junction Capacitance
All BJTs have base-emitter capacitance (Cbe) and collector-base capacitance (Ccb). This
limits the high-frequency response, but also can introduce distortion because these
junction capacitances are a function of voltage.
The base, emitter, and collector regions of a transistor can be thought of as plates of
a capacitor separated by nonconducting regions. The base is separated from the emitter
by the base-emitter junction, and it is separated from the collector by the base-collector
junction. Each of these junctions has capacitance, whether it is forward biased or reverse
biased. Indeed, these junctions store charge, and that is a characteristic of capacitance.
A reverse-biased junction has a so-called depletion region. The depletion region can
be thought of roughly as the spacing of the plates of the capacitor. With greater reverse
bias of the junction, the depletion region becomes larger. The spacing of the capacitor
plates is then larger, and the capacitance decreases. The junction capacitance is thus a
function of the voltage across the junction, decreasing as the reverse bias increases.
This behavior is mainly of interest for the collector-base capacitance Ccb, since in
normal operation the collector-base junction is reverse biased while the base-emitter
junction is forward biased. It will be shown that the effective capacitance of the forwardbiased
base-emitter junction is quite high.

This may be another key to FF "mojo", or it may be a red herring. It will suffice to say that a FF goes into some form of cutoff when the Vbc  junction voltage difference drops below the value required, 0.2v for Ge and 0.7v for Si, which means that as the signal plays around this, the junction capacitance will be at it's largest, and so transistors with naturally larger capacitances will color tone a bit more. This is just my theory based on what I've read, however.

Current Gain
If a small current is sourced into the base of an NPN transistor, a much larger current
flows in the collector. The ratio of these two currents is the current gain, commonly
called beta (β) or hfe. Similarly, if one sinks a small current from the base of a PNP transistor,
a much larger current flows in its collector.
The current gain for a typical small-signal transistor often lies between 50 and 200.
For an output transistor, β typically lies between 20 and 100. Beta can vary quite a bit
from transistor to transistor and is also a mild function of the transistor current and collector
voltage.
Because transistor β can vary quite a bit, circuits are usually designed so that their
operation does not depend heavily on the particular value of β for its transistors. Rather,
the circuit is designed so that it operates well for a minimum value of β and better for
very high β. Because β can sometimes be very high, it is usually bad practice to design
a circuit that would misbehave if β became very high. The transconductance (gm) of the
transistor is actually the more predictable and important design parameter (as long as
β is high enough not to matter much). For those unfamiliar with the term, transconductance
of a transistor is the change in collector current in response to a given change in
base-emitter voltage, in units of siemens (S; amps per volt).
gm = ΔIc/ΔVbe

I added some bold emphasis to one of the important "design principles" that is more common practice, and you'll note that the Fuzz Face does actually follow this, as the circuit "works" with any of the transistors that were in the originals, however, as lots of "mojo" claims led to people hand picking out of large quantities of FF, and as we have later learned what values of gain and voltage on the collector sound "like the best/original used by hendrix/people like the most", we're trying to pick transistors with the right gain, even the right material. eBay has a huge industry of people trying to sell germanium transistors because of the Fuzz Face alone. Anyway, I was kindly enough told about this design practice by R.G. so I've kept it in mind when trying to redesign the Fuzz Face for specific transistors. It's not something one would normally do, but with NOS and non-renewable transistor sources for "the mojo semiconductor", I think it might be something FF builders have to bite the bullet and design around. Joe Bonamassa did a Fuzz Face based on MP39B and some other transistor, low gain devices, so I feel a little more confident that I can push something out of some transistors that people don't usually think of when they consider what kind for Fuzz. Mostly, the backup stock for Ge transistors not associated with fuzz face are far, far cheaper.

[tjdracz]

True. It seems that you only need to mention fuzz face in the eBay auction title to automatically bump up the price to £5+ per transistor, even if they're Japanese 2SB's, not normally associated with this pedal. Obviously AC128 and such are the worst. I've actually seen TWO AC transistors on eBay with NOS/fuzz face claim going for £60! Madness!
I've actually talked about it with my dad the other day and he couldn't get his electronics engineer head around the idea that people go crazy about germanium mojo. In his mind, germanium is this less than ideal device, rightly replaced by silicons.
And Russians seem to be last cheap readily available types, but this cannot last. If people start redesigning effects around them, then they would get more popular, then more expensive. Supply and demand. That's why I'm researching the virgin land of polish transistors.
Anyway, I will experiment with this Russians in fuzz face, see how it goes and sounds after some trimmer based bias fun!

[zombiwoof]

True. It seems that you only need to mention fuzz face in the eBay auction title to automatically bump up the price to £5+ per transistor, even if they're Japanese 2SB's, not normally associated with this pedal. Obviously AC128 and such are the worst. I've actually seen TWO AC transistors on eBay with NOS/fuzz face claim going for £60! Madness!
I've actually talked about it with my dad the other day and he couldn't get his electronics engineer head around the idea that people go crazy about germanium mojo. In his mind, germanium is this less than ideal device, rightly replaced by silicons.
And Russians seem to be last cheap readily available types, but this cannot last. If people start redesigning effects around them, then they would get more popular, then more expensive. Supply and demand. That's why I'm researching the virgin land of polish transistors.
Anyway, I will experiment with this Russians in fuzz face, see how it goes and sounds after some trimmer based bias fun!

The vintage type larger-case AC125, AC128, and NKT275 trannies are the expensive ones, the ones in the skinny case with the bottom flange and tab indicator are the more modern ones, which are more abundant.  The vintage type large case type are in short supply these days, as guys like Fuller and Analog Man (and other boutique makers) have scoured Europe for supplies of those, and subsequently the ones that are left are expensive.  The Tungsram AC125/128 are the more modern ones, which are fine but supposedly the best ones are the older large-case versions.  Here's a good article on AC128's and similar trannies:

http://www.jacquesstompboxes.com/ac128.htm

The first version, which he calls "the real ones" are the now-expensive ones, and the type that were used in vintage Fuzz Faces and other vintage fuzzes.  I have a few of the AC128's in the "heat-sink" package, they are pretty good too.  I have one in a modded Dunlop Fuzz Face along with a Toshiba AC176 in the other position, and get a good sound with the pair.

Al

[Thecomedian]

Im waiting for one of those tube making factories to start producing old-style Ge transistors again. 

[skrunk]

The first version, which he calls "the real ones" are the now-expensive ones, and the type that were used in vintage Fuzz Faces and other vintage fuzzes.

I've never seen AC128s used in any vintage fuzzes.
I think they were used in the Crest Fuzz Face re-issues from the 90s, but that's about it.

[Thecomedian]

The first version, which he calls "the real ones" are the now-expensive ones, and the type that were used in vintage Fuzz Faces and other vintage fuzzes.

I've never seen AC128s used in any vintage fuzzes.
I think they were used in the Crest Fuzz Face re-issues from the 90s, but that's about it.

If only we had a time machine , but honestly what matters is how the waveform clips and the frequency response, if the requirement is "vintage sound".

I wonder how the simulation of leakage on the silicon fuzz face via BFH resistor between C and B could improve traditionally "not leaky enough" Ge transistors..

[zombiwoof]

Im waiting for one of those tube making factories to start producing old-style Ge transistors again. 

Actually, in the last few years, a couple of companies were making the old-style AC128's, NKT275's, OC81's, and other similar types in the large case versions, and I saw some in some catalogs for electronics companies (one of the manufacturers was Magnetec, the other I can't remember).  Small Bear tried out some samples, and found they were all in extremely low hfe ranges, and when he asked about gain ranges that would be more applicable for pedal builders, he was quoted some outrageous price for pre-selected gain ranges, so he gave up on them.  So there have been attempts to replicate those old transistors, but the results have not been too good.

As for the comment about AC128's never being in vintage fuzzes, the point about the article about AC128's that I posted was that the vintage transistors that were used in old Fuzz Faces (I think NKT275's were the most frequently used in the germanium versions), were not the modern skinny, bottom flange type that are around now, and the larger case versions (whether they are AC128, NKT275, or whatever) are the ones that are harder to find now, and thus more expensive, especially in preselected gains for Fuzz Faces or Tone Benders or whatever.  Analog Mike talks of scouring Europe for usable vintage NKT's for his pedals, and he would score large batches at a time, then have to sort through them for ones that would be usable in his pedals.  He had to pay a lot in the first case for the transistors, then there were many rejects, so the usable ones ended up being quite costly.  Mike Fuller of Fulltone used to use the vintage large case transistors in his '69 Pedal, I've seen NKT's and various Japanese large-can types in the older '69 pedals, but at some point he just stopped making that pedal, due the problem of finding those types of transistors in enough quantity to make the pedals.  I would assume (although I haven't seen inside them), that the newer '69 pedals he is making are using the modern type of germanium transistors, in order to be able to mass produce the pedal.  I would be interested in hearing how the change in transistors has impacted the sound of those pedals, if that is the case, I really haven't heard any side-by-side comparisons of his older '69 pedals to the newer versions.

Al

[jim68000]

I think Jacques' opinions on the AC128 are a bit eccentric. As I understand it, the AC128 is a European standard transistor type made by a wide range of manufacturers, as opposed to a branded tranny like the NKT series (although other people besides Newmarket have made the NKT275, especially since Newmarket's passing). Therefore what he's identifying as the original and best is probably just one manufacturer's output. Some manufactures are going to have better quality control than others. The hfe range for an AC128 that complies with the standard is 45 to 165, which gives you an idea how all over the place they are.

The case he identifies is a TO1 as opposed to a TO1H - the difference is a millimetre in height. Not sure what makes a tophat "modern". It might be cheaper to package things in something flanged that way which would explain the different reputations.

I have a bag of the Tungsram AC128s and another bag of Tungsram AC125s. From the general state of decay of the packaging etc I'm not sure if they count as modern. Possibly 80s, considering they were made in the Eastern bloc, but I'd plump for early 70s. Anyone know if there's a way to date these things?

The AC128s are pretty uniformly rubbish, by the way. The AC125s on the other hand have a sprinkling of mojo, especially if you pair them up with something genuinely magic.

One other historical note. The ACxxx sequence is part of the Pro Electron standard which was established in 1966. AC anything has to be '66 or later. NKT and OCs were available earlier - Mullard started making OC71s in 1954, just 7 years after the Bell invention. The fact that they were available and familiar to UK electrical engineers probably explains appearance of the UK branded types in the iconic effects of the period.

[Thecomedian]

Im waiting for one of those tube making factories to start producing old-style Ge transistors again.  

Small Bear tried out some samples, and found they were all in extremely low hfe ranges, and when he asked about gain ranges that would be more applicable for pedal builders, he was quoted some outrageous price for pre-selected gain ranges, so he gave up on them.  So there have been attempts to replicate those old transistors, but the results have not been too good.

lol isnt that life. When we try to replicate old technology, it often fails miserably because we don't have the same frame of reference. I wonder, though, what exactly is wrong with their fab process, why they cant go through expired patent documents to get the proper recipe? All the same, if they have some dirt cheap low hFEs, I'd support their agenda with my dollar-votes.

bah, their products are all discontinued.