Many AC128 Transistor Questions

[Emanresu]

Hi guys,

I was browsing Ebay looking for a few AC128 transistors to use in a Fuzz Factory, after sifting through the fake chinese ones, there's a plethora of Tungsram AC128 Transistors being sold for dirt cheap (like 30p each). Amongst the Tungsrams I found this listing for a Mullard AC128 (Which I've never really come across), I'm not sure if it's genuine, but the seller seems legitimate and they look pretty real to me...

http://www.ebay.co.uk/itm/AC128-01-MULLARD-GERMANIUM-TRANSISTOR-NOS-/260904864515?hash=item3cbf244b03:m:mG3aOVGfxUpEFsTq11uILeg

Anyway, here come the questions. It's impossible to tell really without actually testing it, but would you hazard a guess as to whether they are real or not? Would it be worth it? £6 seems pretty steep to me for one, even if they are supposedly 'tested' considering the likely rate of usability, right? Although I guess they are getting harder to get hold of. Are the Tungsram ones really so bad? From what I've read, they often come in low gain ranges, and occasionally a large percentage are very leaky or noisy, but even if only 10% are good/usable it would make it better value than the other listing. Would it just be better to buy AC128s from somewhere like Banzai Music/SmallBear/DrTweak/Musikding? These are almost definitely going to be good and in usable ranges, but they don't appear to be NOS so I assume they're fairly new production ones, and I know it's almost more likely for there to be a greater difference between two identical part number/brand germaniums than there is between a transistor and an equivalent by a different brand, but it's also been said that the qualities that made these types of transistors so revered were the qualities that didn't appear on the specs sheets, so would you say these could be missing some of the characteristics/have different characteristics to the original germanium transistors that made them subjectively 'good'? Finally, I've read about a few, but would you guys recommend any different model transistors (Japanese models, military, new production, etc) in place of the AC128s that are more readily available, very similar or arguably better? Be interesting to hear what you guys have found.

Sorry for the wordy post! Peace!

[digi2t]

In my book, there's two ways to go...

A) Buy a cheap bagful of them from EBay, and sift through them with a test rig (see Geofex), or DCA. Out of 100, you might get 10 good ones. Been there, done that.

B) Go to Smallbear, and buy two good ones. They'll even match them up for you.

I've gone route A) plenty of times, but only when the price justifies the 10% rule. If you don't want the hassle, go with a known quantity.

P.S. I've never noticed anything particularly special about AC128's. I have several "good" ones, and have done tons of germanium fuzz projects. To date, they've only made it into one project. Nothing is set in stone. I've built a couple of identical circuits as well, Japanese transistors in one, Russian in the other. Both performed really well. At the end of the day, it's a crap shoot anyway you go. Your ears will give you the ultimate thumbs up. If ears had thumbs that is. That would look funny.

[chuckd666]

In my limited experience, playing with Germanium transistors can be either really fun or a total shitfight. If you want a painfree and probably (in the long run at least) more economical experience, I'd buy tested ones (from Mr Bear).

If I was to buy a large bunch to sift through, I'd go to that minifux eBay seller and get some weird Japanese things while cross checking against datasheets.

[PRR]

In Europe, at some time, if you needed a transistor with no particular specs (high voltage, high frequency, high gain), you stuck in an AC128 and got a no-particular-spec transistor.

The original parameters could vary all over the place. They were so popular that they were made in many factories, some better than others. Demand varied enough that one factory would make them with another company's brand for sales through that brand's dealers.

For some uses, if you needed a few special-spec transistors in a many-transistor device, you bought a crate of AC128 and picked-out the few which met your special spec, used the others where performance was not fussy. (And this practice was never noted in the schematics.)

I am not aware that any have been made in decades. Much 'better' transistors appeared in the 1970s, why make the old junk? There seem to have been huge warehouse stocks; I suspect that "AC128" was often the fallout type-number for transistors which failed to make the grade under another (higher price) part number.

Those old-stock cartons have probably been re-sifted many times to sort-out the "better" parts. So yes, you could end up with 50 pieces, none with enough gain to wipe their own nose, or breakdown voltage lower than a 9V battery.

[smallbearelec]

they don't appear to be NOS so I assume they're fairly new production ones

AFAIK, the only company that was still making germanium transistors was Germanium Power Devices. Their web site now emphasizes optoelectronics, so IDK if they are still making 2N404 and such for the military as they used to.

There are still some decent-size lots of '60s and '70s vintage germanium available (even some NPN types), but the brokerages that have them mostly don't want to sell small quantities. For me, when I can find a small "mother lode", it's an investment that pays off like an annuity.

If you want to save some $ by sorting your own, rather than frustrating yourself with the 10% (or worse) yield from eBay, check out my bulk "rough-sorted" devices:

http://smallbear-electronics.mybigcommerce.com/bulk-rough-sorted-lots-1/

You'll find parts that were made by Name factories and "house"-numbered for equipment makers. While they are not cheap, they have not been "cherry-picked" like so many lots of popular type numbers. No minimum order, so try a few of a type till you find one that curls your fur. Many lots cover a pretty wide range of gains, and you have to live with that. That said, because they are not audited, I will exchange any that are hissy as long as they are not soldered to and have full leads. USE SOCKETS TO TEST!

[Emanresu]

A) Buy a cheap bagful of them from EBay, and sift through them with a test rig (see Geofex), or DCA. Out of 100, you might get 10 good ones. Been there, done that.

B) Go to Smallbear, and buy two good ones. They'll even match them up for you.

Yeah I figured as much, as usually, I'd try and go the first route. Always seemed like fun to me rather than a hassle, especially finding a few weird ones in the pack, but then again I haven't had to sort through more than a hundred at a time. Smallbear is always the easiest way to go, but I just saw them on Ebay and figured hey, if they're tested and 'the real thing' (whatever that means) it would be interesting to see what they're like, even if just to appease my curiosity.

P.S. I've never noticed anything particularly special about AC128's. I have several "good" ones, and have done tons of germanium fuzz projects. To date, they've only made it into one project. Nothing is set in stone. I've built a couple of identical circuits as well, Japanese transistors in one, Russian in the other. Both performed really well. At the end of the day, it's a crap shoot anyway you go. Your ears will give you the ultimate thumbs up. If ears had thumbs that is. That would look funny.

Yeah of course, I mean the only real reason they're so popular is because they were used in a few fairly famous pedals around half a century ago, and the chances they chose the best sounding transistors out of all of the ones out there is very slim anyway. It sort of makes me think that in another 50 years, if/when they've created quantum computing and you can buy pedals with antimatter ICs for like, £20 (although in 50 years when we get taken over by the aliens, I suppose we'll be using credits, if they let us have any freedom at all), that perfectly emulate octaves or harmonies or whatever without the issues that come with it currently, whether there'll be a group of people paying hundreds or more for the old dsp chips they found in the pogs or the whammys because they need that latency or cheesy sampling to sound like Jack White/Tom Morello/whoever. Anyway I apologise for the fairly substantial digression, but have you tried any particular Japanese/Russian/whatever replacements that you really liked for whatever reason, be it they sounded just like the originals, or because they sounded totally different?

Wait, so your ears don't have thumbs? Surely that looks more funny, right? Mine wiggle when I'm excited. 

In Europe, at some time, if you needed a transistor with no particular specs (high voltage, high frequency, high gain), you stuck in an AC128 and got a no-particular-spec transistor.

The original parameters could vary all over the place. They were so popular that they were made in many factories, some better than others. Demand varied enough that one factory would make them with another company's brand for sales through that brand's dealers.

Ahh well that's interesting, and slightly disconcerting 

So, out of curiosity, is there (Or more likely 'was there') an AC128 made in a certain factory at a certain time that is known as the sort of, gold standard of AC128s? Or is the variation between even the same factory/brand/time period/type germaniums so big that it was totally by chance that you'd get a pair that sounded good for whatever reason?

I am not aware that any have been made in decades. Much 'better' transistors appeared in the 1970s, why make the old junk? There seem to have been huge warehouse stocks; I suspect that "AC128" was often the fallout type-number for transistors which failed to make the grade under another (higher price) part number.

Those old-stock cartons have probably been re-sifted many times to sort-out the "better" parts. So yes, you could end up with 50 pieces, none with enough gain to wipe their own nose, or breakdown voltage lower than a 9V battery.

Strange, I've heard it mentioned, seems strange though for so many different places to be selling fairly new looking AC128s for so little without advertising them as NOS, as I'm sure just those three letters would make them much easier to sell, but I suspect you're much more knowledgeable on the issue than I am 

Well that's sad, almost makes me want to buy them all, thinking of all those poor, unloved transistors that were told they weren't good enough to be used and were left behind

Good point, I suppose it is just much easier to just get a ton of alternatives and just audition each type to see which ones come closest/sound the best, I was just wondering what the whole deal is with the AC128s, with regards to whether there are still any 'good' ones available and whether even the 'good' ones are even really that good, if that makes sense... 

There are still some decent-size lots of '60s and '70s vintage germanium available (even some NPN types), but the brokerages that have them mostly don't want to sell small quantities. For me, when I can find a small "mother lode", it's an investment that pays off like an annuity.

If you want to save some $ by sorting your own, rather than frustrating yourself with the 10% (or worse) yield from eBay, check out my bulk "rough-sorted" devices:

I was actually checking these out last night, along side the AC128s, my thinking was that if I could sort of A/B them see the variations between each and see if there was anything that justified the higher pricing on the AC128s, which I somewhat doubt, but it'd still be interesting to try and see!

Cheers guys for the insights, they're really helpful!

[Electric Warrior]

they don't appear to be NOS so I assume they're fairly new production ones

AFAIK, the only company that was still making germanium transistors was Germanium Power Devices. Their web site now emphasizes optoelectronics, so IDK if they are still making 2N404 and such for the military as they used to.

Magnatec was making AC128s until recently. New Jersey Semiconductor still is. DSI and American Microsemiconductor are also making germaniums.

[R.G.]

Let me see if I can shorten your wondering list. I've typed something similar so often that I get thought fatigue doing it again.

1. There is no magic about a transistor with a given part number. Read that to yourself about twelve times until it starts to sink in.

2. Any given transistor is NOT a part number. A part number is a DATA SHEET. Any chunk of semiconductor that meets the mins and maxes on the datasheet may be printed with that part number. Some 2N5088s will meet 2N3904 specifications, so they ARE in fact 2N3904s. Some 2N3904s will meet minimum 2N5088 specifications, so they ARE in fact 2N5088s. It's like legally having dual citizenship. You are a citizen of both countries, but are a particular instance of each, while still being a single person. Some 2SB75s will meet the specs for AC128, so in a sense they ARE AC128s. And the reverse.

3. Germanium was a "training wheels" semiconductor that humans learned to make transistors with. It is easier to make transistors with Ge than Si, but the results are poorer-performing in the technical sense. These poor-nesses happened to be what was used with early fuzz boxes, and that got the the reputation.

4. Germanium processing in the 1960s was (by today's standards) sloppy, trashy, clumsy and inept. We humans had no clues about how to make tight, fast, high performance transistors. One early manufacturing technique for Ge transistors was to take little slabs of refined Ge, put indium slabs on each side and place many of these "sandwiches" on what amounted to a cookie sheet then to bake them until the indium diffused into the Ge "base" to make transistors.

5. As a consequence, Ge transistor results were all over the map, on top of the variations in the material itself. What actually happened was that the manufacturer would bake up a bunch of transistors, then test them and see which datasheet they fit under. The best, fastest, highest gain (and therefore most expensive to buy) devices were given one part number, the next sloppier ones the next part number, and so on down to the floor-sweepings.

6. Transistors don't have a sound. CIRCUITS have a sound, depending on the interactions of the parts in the circuit.

7. Correspondingly, you cannot make a circuit sound better by rubbing germanium transistors on it.

8. It's even worse than that. The characteristics that make any particular germanium device sound good (however you define that word) are not on the datasheets. Musical electronics is often about the side effects of parts that are not documented.

9. Special ebay consideration: the huge popularity of making your own pedals and the general name-reputation of the AC128 in particular and PNP germaniums in general has led to thousands of people scouring the planet for leftover stocks of cheap germanium transistors that can be sold for high prices.  If you're a money-making kind of guy and you buy a bucket of germanium PNPs to test for "sounds good in fuzz face"-ness, you're obviously going to keep the ones you like. What do you do with the rejects? Throw them in the trash? No - you SELL THEM ON EBAY to the next guy, the guy who was a little late to the party. Think about that. Professionals have been looking for, testing, and rejecting germanium transistors for use in effects pedals for nearly 20 years now.

So if you want to experiment with messing with tinkering with transistors that may or may not really be AC128s in either sense of that label, go for it. I tinker with lots of stuff just to see what happens. It's cheaper than a movie these days and provides more hours of entertainment. But don't count on either finding magic transistors. Count on finding a few that are OK for your uses. This may be more expensive than buying one or two from a professional that has already found hidden caches of germanium and tested them for you. It's often cheaper. Just saying.

[digi2t]

7. Correspondingly, you cannot make a circuit sound better by rubbing germanium transistors on it.

Transistor before rubbing...



Transistors after rubbing...



Oh darn.... I read that wrong. Sorry. 

[Emanresu]

Hey RG, I've often come across your posts particularly pertaining to this issue, and a lot of the info you've shared in the past has been very useful. I've also seen the frustrations you've gone through regarding it, so I'll do my best to avoid boring you or frustrating you with any naïvety.

I totally understand there are no magic transistors, and I understand that had the original fuzz face used 2sbxxx transistors or whatever else then there'd be people selling them for inordinate amounts of money instead and AC128s/NKTs/Whatever else would be almost unheard of. I also understand that a transistor is not a part number and that variations between the 'same' transistor can be so great it's basically pointless to look at them as part numbers because you're more likely to have more similarities between two transistors that are almost completely unrelated.

Anyway back to the point, I know how poor and unreliable the production of germanium transistors was (although, I didn't know they baked big transistor cookies, sounds delicious!). Now correct me if I'm wrong (which I'm fairly sure I am), but wouldn't the manufacturing process vary from place to place? For example I read (I believe it was something you said a long time ago) about different single malt scotch distilleries replacing barrels and creating holes in the barrels and filling them the same way they had with the original barrel they were replacing, because they weren't sure what the secret was that made their scotch the way it was. So in the same way you might hate one single malt, but love the next for whatever reason, wouldn't you say that while neither contributes to a bad sound objectively, that a Japanese transistor and a European transistor that are considered identical in all measurable ways, would sound different (in the same circuit) due to the different materials/machines/processes used in each country? Or would you say that the variation between even two germs from the same factory, or even from the same batch with very similar gains/leakages and whatever else would likely be so different anyway that it's totally pointless trying to buy NOS AC128s because germaniums are all so different you're just more likely to find two you like or 'sound more like AC128s' (if there is such a sound) amongst 200 japanese transistors than 20 actual AC128 part number transistors?

Also, surely transistors do have a sort of sonic character they impart on the circuit? I know adding some NOS £200 rare 50s germaniums to an inherently bad circuit isn't going to magically make the circuit good in any way, but seeing as a transistor doesn't amplify a signal totally cleanly/transparently wouldn't you say that has an effect on the sound produced at the end of the circuit?

Anyway, yeah I understand that most of the good ones are pretty much gone for good by now anyway, I just haven't really been able to experiment with different types, I was just curious whether if there were 'good' AC128s out there ie good gain ranges/low noise/low leakage/reliable out there, whether it would be worth having them as well as the equivalents not because they sound better per se, but because they might react differently in a circuit and therefore sound different, due to where and how they were made.

I'm not disputing any of your points by any means, at ALL, I'm just sort of sharing my understandings/misconceptions, and wondering how far off I am? 

I'd also like to say, I'm not some sort of elitist, trying to slavishly replicate the Hendrix or whoever tone by copying every component of every amp, guitar, and pedal he ever used or something, honestly, that always seemed a bit pointless to me, he already made those sounds do I really just want to copy them, like tracing a Picasso painting or something. Just curious about whether there is any relationship between where/how/when different germaniums were made and how that affects different aspects of the way they work, if that makes any sense! I apologise for bringing up this topic again as it's been discussed a lot, if anything I want to know more about alternatives that are totally different to the AC128s than the ones that are the same! 

[R.G.]

I totally understand there are no magic transistors, and I understand that had the original fuzz face used 2sbxxx transistors or whatever else then there'd be people selling them for inordinate amounts of money instead and AC128s/NKTs/Whatever else would be almost unheard of. I also understand that a transistor is not a part number and that variations between the 'same' transistor can be so great it's basically pointless to look at them as part numbers because you're more likely to have more similarities between two transistors that are almost completely unrelated.

OK. That's a good place to start. I never know whether people get that. So I have to repeat it, not only for the individual but also for the current and future readers of these forums. The majority of people messing with effects "pass through" effects building over a year or two, and there is a constant inflow of people who haven't participated in the history.

Now correct me if I'm wrong (which I'm fairly sure I am), but wouldn't the manufacturing process vary from place to place? For example I read (I believe it was something you said a long time ago) about different single malt scotch distilleries replacing barrels and creating holes in the barrels and filling them the same way they had with the original barrel they were replacing, because they weren't sure what the secret was that made their scotch the way it was.

It probably was me. Scotch is a great example of practically everything.   

So in the same way you might hate one single malt, but love the next for whatever reason, wouldn't you say that while neither contributes to a bad sound objectively, that a Japanese transistor and a European transistor that are considered identical in all measurable ways, would sound different (in the same circuit) due to the different materials/machines/processes used in each country?

That seems reasonable but really isn't. The problem is one of measurable complexity. Actual human tasting apparatus is several orders of magnitude more complicated than transistor performance. As one example, more of what you "taste" happens in your nose, not your mouth, and quite a lot of it happens in your brain, not nose or mouth.

Transistors are by contrast wonders of simplicity. They're still complicated enough, but not beyond description.

Or would you say that the variation between even two germs from the same factory, or even from the same batch with very similar gains/leakages and whatever else would likely be so different anyway that it's totally pointless trying to buy NOS AC128s because germaniums are all so different you're just more likely to find two you like or 'sound more like AC128s' (if there is such a sound) amongst 200 japanese transistors than 20 actual AC128 part number transistors?

There are several items in there to discuss.
1. Two germs from different batches, factories, and times that measure the very same gains, leakages, and also the whole slew of timings and capacitances as well as measured noise performance will likely sound almost identical when subbed into the same circuit.
2. The non-identical-ness will mostly be from how well packaged the actual transistor dies are and how contaminated the surfaces are. Early transistor noise was often dominated by surface contamination, and much of the leakages happened there as well. This is why "glass passivation" was a big deal in the late 60s and early 70s sales literature for transistors. Surface contamination was a big deal, and it exerted big influences on transistor operation and noise performance. Worse yet, it was not constant or predictable from batch to batch or time to time within the same factory. Contamination could make transistors fail by running leakage and noise up to intolerable levels. Contamination was largely a thing of the past once its effect was known and new processes evolved to fix it.
3. There is a time/history effect. Semiconductor processing is a game of eliminating all "contamination" except the very few intentional ones put there by doping the base material. Purification of the raw material, germanium ingots, got better and better over time. Crystal defects in the base material sometimes affects performance, but today those devices are simply tossed out. When a single transistor cost the equivalent of $10-$50 each, there was a high incentive to sell the tested-mediocre as lower performing, but still good devices, or foist off the rejects on unsuspecting "consumer" manufacturers since the military took most of the good ones. This changed over time too.

4. There isn't any "sound like an AC128".

Also, surely transistors do have a sort of sonic character they impart on the circuit?

No. Not in the literal sense of that statement.

I know adding some NOS £200 rare 50s germaniums to an inherently bad circuit isn't going to magically make the circuit good in any way, but seeing as a transistor doesn't amplify a signal totally cleanly/transparently wouldn't you say that has an effect on the sound produced at the end of the circuit?

Only in the context of the entire circuit. The parasitic and nonlinear capitances of the device will limit and change response, and it may or may not have leakage/noise/contamination effects. Worse yet, "sound produced at the end of the circuit" has to be judged. There is a long history of people hearing things in their heads that cannot be measured, in all senses of that statement. 

Anyway, yeah I understand that most of the good ones are pretty much gone for good by now anyway, I just haven't really been able to experiment with different types, I was just curious whether if there were 'good' AC128s out there ie good gain ranges/low noise/low leakage/reliable out there, whether it would be worth having them as well as the equivalents not because they sound better per se, but because they might react differently in a circuit and therefore sound different, due to where and how they were made.

Satisfying curiosity is always good, so you ought to go tinker and enjoy yourself. I completely and absolutely agree with that right up to "due to where and how they were made". The range of variation types and range within types is so large that you really can't say much about that.

There's a larger issue lurking there. Will the devices you can find and test be representative of what was produced in those locations and those times? "The blind men and the elephant" is a humorous statement for encapsulating this issue. You can't get your hands on a representative sample of devices from any given time/place/method for production of any transistor from this era, so whatever measurements you can to is limited to saying something about this one device here in my hand. Generalizing to what the production run it came from is not possible at best and wildly misleading at worst.

Say that your "AC128" is high gain and low leakage, but has high noise. Are all of the ones made in that factory (... um, which factory was that?) on that day, in that batch, from that one wafer, just like that? Was this one the best, worst, or just a middling one? Without more devices of known provenance, you can't generalize in any meaningful way. All you can logically say is "this one device right here is...".

I'm not disputing any of your points by any means, at ALL, I'm just sort of sharing my understandings/misconceptions, and wondering how far off I am?

And I'm not disputing your points, either. I'm just adding my own understandings and perhaps misconceptions too.

I'd also like to say, I'm not some sort of elitist, trying to slavishly replicate the Hendrix or whoever tone by copying every component of every amp, guitar, and pedal he ever used or something, honestly, that always seemed a bit pointless to me, he already made those sounds do I really just want to copy them, like tracing a Picasso painting or something. Just curious about whether there is any relationship between where/how/when different germaniums were made and how that affects different aspects of the way they work, if that makes any sense! I apologise for bringing up this topic again as it's been discussed a lot, if anything I want to know more about alternatives that are totally different to the AC128s than the ones that are the same!

I didn't get any sense of elitism or whatever out of your comments. I think you're exploring the possibilities. I went through a lot of that myself, but had the advantage of several decades of experience alongside the semiconductor industry and friends inside it.

We're almost in the business of semiconductor archeology here,  digging up the few bones that are left and trying to figure out what kind of animal that was. 

[Cozybuilder]

Any time you can draw parallels with Scotch makes for great reading 

[Resynthesis]

As alluded to by digi2t I'd definitely look at some of the Russian Ge equivalents. When I had more time to tinker I found that it was much easier (and cheaper) to get hold of Ge trannies with the right parameter ranges.