Transistor history question

[Enzo]

Yes, of course there are situations where it matters more than others. But using your example, if you need to replace an ABC123XYZ and you have to select individual transistors for hfe, they still are the same part. Like selecting 12AX7s for low noise or low microphonicity. And even now, we find for example SOvtek 7199s won;t work well as phase inverters in Sunn amps, or some other brand won;t hold up in cathode follower applications.

I don't dismiss these old circuits as junk. What I do think about them is that they were not very sophisticated designs, and that is why they were so touchy as to which ABC123XYZ transistor was installed. Note that modern circuits, though having their own issues, do not usually have this problem. I am not picking on the designers, state of the art was what it was back in that era.

[Rhodesplyr]

Yes, of course there are situations where it matters more than others. But using your example, if you need to replace an ABC123XYZ and you have to select individual transistors for hfe, they still are the same part. Like selecting 12AX7s for low noise or low microphonicity.

I'm going to have to disagree there because it brings me back to my original point. If someone designs a circuit for a 12AX7, then I know that I can at least look up the general specs for the 12AX7. Individual examples will vary, but at least I have some concept of what the ideal 12AX7 is supposed to be. On the other hand, if I need to replace a Wurlitzer 137159 that's apparently supposed to be different from its neighbor, the 203719, I can't look up either of those numbers anywhere to find out what I should look for. They don't cross to any industry standard. It's a lot simpler to plug in several samples of 12AX7.

And even now, we find for example Sovtek 7199s won;t work well as phase inverters in Sunn amps, or some other brand won't hold up in cathode follower applications.

True, but that's partly David Hafler's fault Even the old stock 7199s and their 6AN8 predecssors didn't always work well or hold up very well in that circuit. Want a 7199 that's suffered terminal heater-cathode leakage in the triode section? I've got several. (I played the "tweak the ST-70" game for years before I finally admitted defeat and switched to a better driver circuit.)

I don't dismiss these old circuits as junk. What I do think about them is that they were not very sophisticated designs, and that is why they were so touchy as to which ABC123XYZ transistor was installed. Note that modern circuits, though having their own issues, do not usually have this problem. I am not picking on the designers, state of the art was what it was back in that era.

All I wish is that I had some access to those designers' intentions, but the practice of in-house part numbers prevents that. I'm lucky enough to be acquainted with someone who built Vox organs in England in the 1960s, and his tips have been invaluable in terms of figuring out how to select transistors to get the dividers to work. As for Wurlitzer, I think that they lagged quite a distance behind the state of the art, especially as they got into the 1970s. Just compare a 1970s Peavey amp with one of Wurlitzer's.

[Enzo]

A friend of mine while growing up, was into electronics, as was I. We'd get together at one of our homes and "electronics" for hours. I could see early on he was more motivated than I, and our directions differed. He at one point built a television. It was big and ugly in a plywood box, but his family watched it in their living room. We are talking maybe 1962 give or take a couple years. He built a little tape recorder from scratch. Even tried to make his own head, but eventually went ahead and got a commercial head. But he built the circuits and even designed his own crude transport. The last thing I remember he did before college was he built a curve tracer. He embraced transistors before I did.

The guy went on to become an electrical engineer and wound up in high tech designing circuits for satellites and other space program gear.

His first summer job was for some electronics company. For some circuit they needed matched paairs of small signal transistors. They bought the parts by the thousands, and part of my friend's job was to match them. Their process was to stick one in the left side of a fixture, and then plugging other ones, one by one, into the other side until one matched the other. Thus a pair. Then move on the the next pair. All those parts that didn't match were discarded. They didn't go through the same pile of rejects for each new match. He used to bring home bag after bag of these perfectly good but rejected transistors. I got a few of them, but he went on to make all manner of transistor things with these rejects.

SO now that I have rambled on, one wonders what went on at the matching bench at places like Wurlitzer.

I probably don't disagree with you as much as it may seem. Your 137159 probably does cross to an industry type, it is not likely Wurly had custom made transistors. But knowing that it was a 2N3906 or something doesn't help if they are selected for gain or for higher breakdown voltage or whatever.

DO you have the old HAB guide? Or whatever they call it now? My copy is abbreviated. Certain sections I never copied - our organ guy had the real copy. There was a section of manufacturer part numbers to generic. What I kept was generic to manufacturer, so if I really needed some type, I could see if a house number would fit.

SO a quick scan of my HAB shows me for example that a Wurly 130536 656204 were both 2N3414. SO if they were selected for something, I don't know, but I at least know one could possibly sub for the other. The HAB also suggests alternative generics as well as other OEMs using that same part and their house numbers. But to muddy the waters, both those type numbers also cross to 2N3642. And that makes me wonder if maybe the selection criteria were more important than the actual generic xstr type within.

By the way, the HAB also covers Vox numbers, at least those from the Thomas era.


I spotted the 203719. It shows me it is a 2N4238, and offers a few generic equivalents. Maybe you already knew that and was just using the number as an example, but perhaps that is interesting to you.

[Enzo]

I found what I always called the HAB guide.

They apparently still sell it, but now on CDROM. COnsidering the parts of interest are alal ancient, it is a timely now as it was a couple years ago or a few years into the future.

Visit Organ Service Corporation - Home Page and there are some parts lists on the left menu. You can select for example Thomas and look up a part number, and transistors often describe the generic.

CLick on the Keyboard Cross Reference Master. There are some sample pages that show what I mean by listing all the various OEM part numbers for generics and vice versa.

May or may not be worth $80, but the parts lists are valuable, and free.

For some reason I can't copy and paste URLs at the moment, oh well, enter by hand, I guess.

*Keyboard Systems Component Cross Reference Master*

[R.G.]

...If someone designs a circuit for a 12AX7, then I know that I can at least look up the general specs for the 12AX7. Individual examples will vary, but at least I have some concept of what the ideal 12AX7 is supposed to be. On the other hand, if I need to replace a Wurlitzer 137159 that's apparently supposed to be different from its neighbor, the 203719, I can't look up either of those numbers anywhere to find out what I should look for. They don't cross to any industry standard. It's a lot simpler to plug in several samples of 12AX7.
...
All I wish is that I had some access to those designers' intentions, but the practice of in-house part numbers prevents that. I'm lucky enough to be acquainted with someone who built Vox organs in England in the 1960s, and his tips have been invaluable in terms of figuring out how to select transistors to get the dividers to work.

It may help to put it in some context.

The guys who were journeyman designers then had all their experience with tubes. Tubes are a marvel of consistency compared to transistors, and especially early transistors. Worse yet, the electronic design disciplines and approaches which would let one make a transistor circuit tolerant of parameter variations was not yet invented at the time of the designs you're talking about. I had two full college courses in design of part-tolerant transistor circuits, including figuring out what the sensitivity of the circuit was to various parts having various variations. For well done transistor designs after about 1973, pretty much any transistor which has the right polarity, enough voltage, current, and power dissipation, and enough gain should work. In fact, for most modern signal circuits, I use only one - the 2N5088. There are remarkably few small signal circuits it won't work in, with even a little thought. It took the best of the EE designers thinking for about 25 years to develop circuits with low sensitivities to transistor variation.

The reason the original designers did all that rigamarole with selecting devices for circuits and making the circuits depend on critical parameters is that they didn't know how to do it any better. And in this case, the word "better" does have a definition: more immune to transistor variations. I know *why* they did it: they did what they had to to get the products out the door. But exactly what they did depended on what they had to work with and what their bosses would write checks for.

I find that some analysis and possibly some simulation of circuit fragments lets me re-fit most of the old circuits like this.

I ran into an interesting situation recently. Had an old transformer-driver power amp that I'd opened my big mouth and said I could get running. Replaced the outputs and the driver, as well as the pre-driver transistors, fired it up, and of course, it oscillated. I dinked with and simulated and hacked on it for quite a while before I decided to make it oscillate as badly as possible to see what affected what. So I pulled out all the compensation hacks and pf-range caps sprinked into it. It promptly became stable.

The older transistors simply didn't have the frequency range of the modern marvels I put in there, so it was oscillating *because* of the extra poles the compensation hacks added. Removing them all left the driver transformer itself as a dominant pole, and it got stable.

You'll run into similar things a lot. The old stuff can be updated. The trick is to find out what uglinesses the designer had to live with, and what you can do to relieve that.

[J M Fahey]

after about 1973, pretty much any transistor which has the right polarity, enough voltage, current, and power dissipation, and enough gain should work. In fact, for most modern signal circuits, I use only one - the 2N5088.

Amen to that.
To place things in due context, in the 60's tubes were 65 years old; transistors less than 10 , coupled to the fact that modern *production* techniques were barely being developed, plus the very important fact that leaky lossy "no highs" weak Germanium was a false start.
But after the early 70's (coincident with Germanium demise) .... things changed.
I would not use an 6SN7 instead of a 12AX7 nor this one instead of a 12AU7 and so on, while , as in the US example mentioned, a 2N5088 can replace *any* small signal transistor in 99% of cases.
Am I limiting it too much to make it easier for transistors?
No, the tubes I mentioned are all "small signal tubes" for that matter, and the functional equivalents.
The respected Elektor magazine does not even spec which transistor to use in most of its projects, in every issue they recommend using:
TUN = universal NPN transistor: 25Vce , beta 100, 500 mW dissipation.... or better. They suggest a list of likely candidates .
TUP= same in PNP
DUS= universal Silicon diode (1N4002/1N914/1N4148, etc.)
DUG: universal Germanium diode : any of the usual suspects, whichever you can get

Being in an European type country, I use BC546/BC556 for *everything*.

Well, as usual I quoted from memory, but for the real thing:
TUP- TUN- DUS- DUG

[olddawg]

I remember the paint on transistors meaning they were beta matched for difamp input pairs and such.