New law

[Jocko Homo]

I think is time for a new law, and therefore a new thread.

What is the law?

Don't use transistors that (used to) come in metal packages for audio.

That means no TO-5s, no TO-18s, and no (gasp!) TO-3s.

That means no 2N2222s, 2N2907s, 2N2218s, 2N2907s, 2N3904s, 2N3906s, 2N2102s, etc.

Why, you ask?

Because they are all lousy transistors, that is why.

These parts have been around longer that most any of us have been designing stuff. They are from an era when a breakdown voltage of 40 or 60 volts was hot stuff. They all have low beta, which means they have low gain. Which you need to eliminate the nonlinearities they generate.

In the 70's, power amp desingers discovered high voltage transistors with weird numbers like MPS-U10 and MPS-U60 had very low capacitance. This made them very desirable for the gain stage of a typical amp. Too bad thay had low betas, but they were fairly linear, and hey, back then we used tons of gain (and feedback), so no problem.

Today, they a lots of modern transistors with a small base region, which means they will have low noise and capacitance, and a high beta. These devices have breakdown voltages of 80-120 volts and are very linear. Their Ft is plenty high enough for audio, but not high enough to build 70 MHz IF amps with. (See earlier post on why not to use RF transistors. Come to think of it....most of them also came in metal cans.) These devices are made by companies like Toshiba, Rohm, Zetex, and others. They are out there, and not all that hard to find.

Now, as for TO-3s.........

I can't think of a worse package to try to work with. A senior engineer I knew used to swear everytime he was forced to use one. Blamed the damn military for coming up with them, and holding back the evolution of modern packages.

In the 70's all we had to work with was lousy, stinking TO-3's. And the parts themselves weren't much better electrically. Thanks to places like Toshiba, Sanken, Fujitsu, and others, we now have modern plastic packages that are much easier to work with. Plus, the parts are higher beta, more linear, lower capacitance, and more rugged. Don't believe me? Then why is Motorola ("On".....or whoever they call themsleves now) second-sourcing Toshiba 2SA1302s and 2SC3281s? (You didn't hear this from me, but the first few batches Motorola sold were rebadged Toshibas: they couldn't get theirs right yet. You heard this from my retired friend who worked there at the time.) So there should not be any reason to buy a MJ15003 or MJ15028 or whatever unless you have to fix some old boat anchor amplifier.

Speaking of lousy TO-3s and power amps.........

A dealer I knew was having troble back in the early 80's with a certain amp blowing up all the time. I was called in to fix the problem. The amp used 40 volt rails. The problem was that the output transistors were rated at 80 volts. At 25 degrees C., that is. Too bad the amp was much warmer that that.

I called the company that designed it and explained the problem. I suggested a better transistor with a higher breakdown voltage. Their response?

"Well, that transitor has a different slope in the SOA. I don't think it will work."

Appears as if they were smart enough to look at a data sheet, and even where to look. Maybe. Too bad they weren't smart enough to understand how to derate operating voltage as a function of temperature.

Guess what? They soon went out of business, and not a moment too soon. We made $$ fixing amps for dealers who couldn't get repair work done. They didn't come back after we finished working on them.

Let's see how much hate mail this generates.

Class dismissed.

 

[subwo1]

To start, I totally agree about the T03 cases. I always hated them too. But could it be that most transistors like 2N3906 etc are not really made and structured like they were in the past? Recently I bought some 1N4001 diodes which were able to stand 600 volts! I think that manufacturers often just relable similar transistors with the old numbers just to supply the market.

 

[rtirion]

Don't throw your to-3 out of the window just yet!
A lot depends on the type of heatsink you use with them!

Most of the time thermal specs and application is vastly superior to that of flat packs.

Yes, even the much touted toshiba 2sc3281/2sa1302 is a poor performer in the thermal department. Tried them in a class A amp (JLH 1969) version on massive heatsinks, very carefully mounted. They got up to 70 degrees within a few minutes. I even have them in a commercial amp, and they are not to my liking. Used the old to-3 2n3055E in TO-3 from motorola and RCA in this JLH and they are quit fine.

Mr. Nelson Pass uses IRF's in to-3 package and these work very nice. Ask around on this forum, and you will hear a lot of nice comments about his products and designs.

Yes, I am aware of better flatpacks, for instance the ones from Sanken, but did you compare the price per unit with for instance a nice MLJ 15XXX from motorola?

regards

 

[Damon Hill]

I think it's a non-issue.

I've been using MJ15003/4s for literally decades with no problems,
on 60 volt rails. Sometimes I had the bias turned up a bit high,
before I realized I didn't like the sound, and the silicon was
getting pretty hot. I'm currently using MJ21193/4s, which have
much higher breakdown voltages and even better SOA. Both
are reasonably rugged devices, more so than the plastic case
versions, and there are good choices there, too.

Sometime getting the heatsink drilled for TO-3 cases is a problem.
Plastic case devices may need at most one hole for the mounting
screw, and that can be a money-saver in a given design. Plastic
devices lend themselves to PCBs and designs with minimal or no wiring and no sockets, and that's certainly A Good Thing on several counts.

The above TO-3 transistors are packaged in steel (magnetic) cases, and audiophiles consider this A Bad Thing. I'm not convinced, and it's not easy finding every electronic component with non-magnetic construction. On my budget, I just live with it.

Given a choice of bipolar power transistors, I'd take a serious
look at the Toshiba devices as they're more linear and faster,
and reasonably rugged, too. But my present amplifiers have or
will use heat sinks drilled for TO-3, and I prefer to use them as
they can dissipate more heat and have a higher SOA.

Very good driver transistors are available in plastic cases, so there's no need to worry about metal cases any more, as are most signal transistors. Moot issue, and it's been that way for
a long time.


I don't see a need for any laws, new or old. Just a good understanding of the choices and how to design with them.

(Cheerfully tosses bucket of ice water on Jocko Homo)

 

[grataku]

Cheers to that Yoko!
I haven't been using motorolas since I found the way to put my hands on multi-emitter, extended beta sankens. Then I discovered that japanese drivers, ie 2sc4793pair, where also ALOT better (read beta within 10 units for both NPN and PNP, try that with the MJE15030-31!), full plastic case, and much easier to work with. Further down the line I found that there were low noise small signal trannies 2sa1015(I think) that were rated to 120-130V with the same excellent consistency in beta values at that point it was bye-bye to the BC550-560c.

Damon,
unless you got that "buy American" thing going, I think it's high time for you to move to the 21th century.

 

[dice45]

Originally posted by Damon Hill

....


The above TO-3 transistors are packaged in steel (magnetic) cases, and audiophiles consider this A Bad Thing. I'm not convinced, and it's not easy finding every electronic component with non-magnetic construction. On my budget, I just live with it. ...

Damon,



being a tube nuts (but with the mind open to use "sand" where it is useful), i cannot avoid using magnetic material, allthemore as i finally had to face that signal transformers (quite oposite to their reputation do not add to audible distortion and eat up detail resolution) sonically can behave like a piece of wire and/or add speed, juice, flavour to the sound.



But apart from excessive (and expensive) use of "iron" in my amps, i have learned to avoid any magnetic component, be it a capacitor, a resistor, a wire sort, a part of a plug, what ever. I even made listeningcomparisons between the same amp mounted into a magnetic, non-magnetic and non-metallic housing. Provided the amp needs no shielding, the non-metallic housing and the non-magnetic housing play in the same league, the non-metalic however wins. But both win hands-down agains a magnetic housing, Admitted, a non-magnetic case costs.



But we are in the DIY area here, right? Why not building the amp in a wooden box with the heatsink mounted to it? Why not using a small pocket magnet when choosing components, be they from the surplus store or slaughtered out from an old Tektronix? this cost a bit of care, not more. And there is no need to buy fancy NOS stuff llike Vitamin Q for $$$ and later complaining about the nastiness not trackable in the sound (just to give an example).



A year ago, i made a huge listening comparison of different coupling capacitors. I also measured them, and noted down physical properties, magnetism among them. Those caps i loved or atleast could live with all were non-magnetic. Those I loathed or could not stand all were magnetic to some extent. The cap I hated most (in fact my body insisted to escape) was an expensive hermitcally sealed Teflon Mil cap and its housing was very magnetic. The test was done in a scinetific manner, separating data acquisition from analyzing the data. However, before others second my experience, i stick with calling this magnetic thing a coincidence.



As said above, this anti-magnetic thing is not a cost issue, it is an issue of a ittle bit care and effort. And we all DIYers have heaps of that, right?



Damon, i dod not like how Jocko expressed or described his coincidental finding; for my taste he was a bit quick to call it a law and asking the community whether they can share the observation leads to a better open-mindedness to that topic for us all. At the moment it is nothing more than a coincidence although i can follow his logic. I have learned to track coincidences of that sort and to pay attention to them, often an explanation can be found later, turning the coincidence into a general rule.

 

[Jocko Homo]

"What is the law"

To subwo1:

They probably make them the same old way. Eventually they will stop making them period and suggest better parts. Somewhere I have an old National book from the 70's that shows the die geometry. No reason to change it, unless they have too. When they do, there is always someone who complains that the part no longer works in their design.

To rtirion:

I don't have to worry about price......availabilty is a more important issue with us. Any transistor will overheat, and yes 70 degrees is too high. But that doesn't mean the heat transfer was the problem. Ever check for oscillations? Nothing heats an output faster. My designs might be more conservative (not necessarily better, just different), so I don't encounter that problem.

To Damon Hill:

I hate cold!!!!!! Brrrrrrr. I think you may have missed the joke buried inside all the sarcastic diatribe. You are right though: DIY types would do themselves well to adhere to the basics, and learn to master them. I have seen many professional designs done in by overlooking the basics, namely good grounding. Perhaps you would like to share your expertise in those areas with the gang. No one is too old too learn.

You will note that the part I skewered used 40 volt rails and was only rated for 80 volts. I believe the part you are using is better than the one I was so riled over. Plus, you are probably smart enough not to drive it to hell and back. Dealers.........can't live with them, can't shove them in the trunk of your car......

To grataku:

Yoko......I like that. You were paying attention. Move to the front of the class.

To dice45:

I think you missed the joke too. I'm just tired of seeing postings with ancient parts that are lousy.

It has nothing to do with magnetic steel parts or any or anal-retentive audio neurophillia.

Let me put it this way:

L-I-N-E-A-R-I-T-Y.

Keep that hate mail coming!

 

[Chucko]

Jocko, might we ask: What are your favorite parts of the moment?

As an aside, I play with a lot of mass-market musical instrument gear, and it always amazes me what those manufacturers expect a humble 4558 op amp to do! Even a TL072 sounds better in many of these applications, and that's not saying much.

 

[Damon Hill]

Jocko:

In the eternal search for The Better Part, I am VERY curious
about the 2SA1302/2SC3281 and how they're constructed
differently that allow them to have much better bandwidth
and linearity than most audio power transistors. Toshiba's
web page hasn't offered any clues.

I'm using TO=3 cased devices because that's the type of
heat sink that I have to work with at present. At some point
I want to build a professional-looking implementation of Leach's
amplifier design, and the Toshiba transistors are candidates
for the output stage. I'd like to take best advantage of their
characteristics, if I can; that might include a redesign of the
PCB layout for that type of case, to eliminate the wiring and
sockets, if I can. In the short term, the most I can do is
obtain some samples and try them on a prototype to verify
stability, etc.

But I'm not dismissing Motorola/ONSemi devices out of hand,
either. I recently obtained 10 each of the MJ21193/4 devices
and found them pretty tightly grouped for gain; they're more
linear and have twice the Ft of the MJ15003/4. And they're
pretty rugged, partly because of the TO-3 case.

Does one really need an Ft of 30 Mhz in an audio output transistor? Sound like asking for trouble into a highly reactive load.

Sanken and HItachi, among others, make some interesting
bipolar output devices, but I know next to nothing about them.
Sanken likes to use a package for which I cannot find insulators.

Don't know what to tell you about dealers and repair shops; some
of the things I/we know in this forum are very specialized.

grakatu:

Noted on the transistors you like; those numbers will go into my
little black book for further investigation.

Dice45:

It's really difficult to find electronic components that are always
non-ferrous in construction. Some of them are expensive, too.
I used Black Gate electrolytics in a couple of places in my present
build of Leach's amplifier where I thought the application was
critical enough and might have used Mills wirewounds for the
emitter resistors; I found some good quality IRC wirewounds
locally that appeared to be a distinct improvement over sandcast
devices, but they do have ferrous end caps. I attempted to
select metal film resistors that were non-ferrous from my local
supplier, but was only partially successful and I can only afford
ten cent parts, not seven dollar tantalums or bulk foil! But it
would have been awful nice...

I made other tweaks too, but I'm under a tight budget and couldn't afford all the best parts; I had to select the best ones
I could afford and the rest came out of my collection of parts.
I just have to hope that I made good compromises.

Building with wood is out of the question. Don't go there. To get
a decent chassis I may have to cannabalize a Hafler; I don't have
access to shop tools for metal fabrication. That's always been a
big headache for me.

 

[rtirion]

Jocko,

Oscillation was not the problem at all. Checked with scope and spectrum analyzer. Dissipating 35W per device continuossly is what TO218, TO247, TOP-3 simply can't handle reliable.
Getting Sankens, Toshibas and the like is no problem what so ever in the Netherlands. Even mica insulators or silpads are no problem. 15$ for a single 2sc2922 and 1$ for an insulator is what I think a little expensive. When you need 4 or 6 per channel things add up fast.

I also found the production span of a lot of japanese transistors to be short. So if you think you found a good one, you must be carefull it is not out of production within a couple of years.
Famous hiraga's are all prone to this.

 

[Jocko Homo]

rtirion:

You are right: J. A. Pan does make brief runs, and when they are gone, that's it.

35 watts is a bit much, but at least the plastic parts use copper instead of steel to get the heat out.

You are unfortunately paying much more than I am. I use perhaps a dozen devices per channel. And I am not rich.

Damon:

Right now all my data on that stuff is packed away, so you will have to trust me on this for a while. If the data is available, you should notice how low the Cob is. Think phase margin......

The higher you can push the point at which the gain goes to unity, the more likely you are to have enough phase margin to keep the whole thing from blowing up. Especially if a reactive load shifts the phase even more.

I believe I have a stash of mica washers for the Sankens here somewhere. Email me, and perhaps we can work out something in the way of a swap. I don't think I have very many Sanken devices, though.

Motorola has learned much by cooperating with........well, can I say......the enemy. You might notice that a lot of those TO-3 types they used to make are now gone. They are a volume driven firm; if the demand dries up and they have something else in its place, then say goodbye to your favorite part.

chucko:

Anything in specific? Or just in general? BJT or JFET?

 

[rtirion]

Jocko,

Copper to get the heat out? Where?

Not on Sankens, not on Toshibas.

All have nice steel backs. Come to think of it, I never saw a power BJT with a copper back. If you can show me one (photo), I will believe you.

Medium power, Motorola BC362 (very, very old) has a flimsy little copper/brass tap. This is the only transistor I ever saw with copper.

So what's wrong with this story?

Regards

 

[dice45]

Re: "What is the law"

Originally posted by Jocko Homo

To dice45:





I think you missed the joke too. I'm just tired of seeing postings with ancient parts that are lousy.





It has nothing to do with magnetic steel parts or any or anal-retentive audio neurophillia.





Let me put it this way:





L-I-N-E-A-R-I-T-Y.





Keep that hate mail coming!

Jocko,



Your call for linearity tries to break-open doors already open, but linearity is not all-and everything and the magnetic issue touched here (serving only as an example) might be quite a challenge to measure and fit into any signal theory but it is audible.



DTopic,

now you sign your post a 2nd time with "Keep that hate mail coming!"

Do you enjoy that? Do you intend to start flaming around in this friedly forum?



You obvioulsy are very competent in the semiconductor electronic field but being a master in one field does not make you automatically a master in all other fields. One of them being human communication.



I read your post over and over again and , normally being very receptive for jokes, i could not find the joke, not even even traces of it. The fact you had a joke in mind is not enough for all to reckognize the joke. Email and forums are tricky as they do not carry emotions at all. Misunderstandings happen often from that. So may i recommend the use of smilies/emoticons as a beginner's technique for indicating if you do not mean it that serious or intend a put-on or whatever?



Besides that, better get your neck soft. smooth and supple again; it can be very nasty on the long run if the neck remains stiff in that position and it is permanently raining into the nostrils besides that fists with flight height of ~6' cannot be seen early enough

 

[dice45]

Hello Damon,

Originally posted by Damon Hill


It's really difficult to find electronic components that are always



non-ferrous in construction. Some of them are expensive, too.



I used Black Gate electrolytics in a couple of places in my present



build of Leach's amplifier where I thought the application was



critical enough and might have used Mills wirewounds for the



emitter resistors; I found some good quality IRC wirewounds



locally that appeared to be a distinct improvement over sandcast



devices, but they do have ferrous end caps. I attempted to



select metal film resistors that were non-ferrous from my local



supplier, but was only partially successful and I can only afford



ten cent parts, not seven dollar tantalums or bulk foil! But it



would have been awful nice...

you are exposing an exquisite taste, if biased me may say so


Since a buddy of mine came up with this magnetic thing, i have a strong samarium cobalt magnet in my rucksack when i go shoping components. I build with ordinary components until i know which values i need, then i take from my collected stock what i came across over the years and the rest of expensive parts i order. Have tried out all form your list, i like the Mills, Vishay bulk foils and tantalums very much. I can afford this rest as i invest some time in slaughteing old Tektronix equipment and my stock of non-magnetics is quite big meanwhile

I made other tweaks too, but I'm under a tight budget and couldn't afford all the best parts; I had to select the best ones



I could afford and the rest came out of my collection of parts.



I just have to hope that I made good compromises.

There are some expensivish components like Hovland and Holco and Teflon Mil Caps and Audionote copper and silver foil in oil which were quite a disappointment for me. But i often find stuff sounding 1st class from non-audiophile sources. My preferred coupling cap is primarily used in military, railroad, avoinics applications, not cheap but affordable: 1µF @ 500V~ costs about 10$, compare that to Hovland or the like.

Building with wood is out of the question. Don't go there. To get
a decent chassis I may have to cannabalize a Hafler; I don't have



access to shop tools for metal fabrication. That's always been a



big headache for me. [/B]

If you are on a tough budget (so am i), plywood boxes carrying components make your amp sound better and cost less, freeing the money for better components. Consider it


So did i and ended up using an old wine bottle box to build a prototype of my PP AD1 amplifier with Tango OPTs. Super expensive components (collected over the years, could afford none of them today), built in a wine box! (Bordeaux, of course )


The beauity with this box is, i can fold the p2p-wired monster open and have access to every solder joint at once, not directly German style engineering and if the box is closed, no lil'critter can crawl in where 400Vdc are alive and check out its own conductivity





Give your self a bit of time to get accustomed to the concept

 

[traderbam]

I just bought some IRF740As in TO-220 which have plated, solid copper heastsinks.

 

[Havoc]

You just need to use a torque wrench to set them on the heat sink. I prefer TO3 over TO220 any day.

 

[djdan]

The debate TO3 versus Plastic it is not so relevant for sound quality .

1. What do you want for a power amplifier ?
2. It would be some domestic HI-Fi or something like a tank for stage or dance club ?
( I am a DJ for 15 years and for me the nr.1 priority is " my power amps must work " )
3. Majority of people ( 99%) heare from 20Hz to 16.000Hz.
So why should I care if my amp will deliver full power at 100Khz or at 30Khz ?
4. If somebody heare 0.003% THD or 0.03% please send me a picture of his ear.
5. Why should I care if my amp deliver 0.003% THD if the most better high end speakers have a very very poor distorsion figure and linearity like an EKG from a man in the middle at heart atack.

"Buy american's" or "buy japanese" or " buy europeans " ?
Come on people ! We are in 21 th century !

 

[paulb]

If somebody heare 0.003% THD or 0.03% please send me a picture of his ear.

Not the ear itself, just a picture. Don't want any kidnapping scares here.

 

[HarryHaller]

Picture of ears

I got both ears in the shot....

H.H.

 

[Jocko Homo]

"What is wrong with what story?"

rtirion:

I have HUNDREDS of Japanese plastic "flat-pak" transistors. All with copper backs.

Sorry, but I don't have a digital camera to send you a picture of them. I hate digital.........(joke!)

I see that this thread is going astray. The original point was that why are there postings in 2002 that use ancient parts like 2N2102s? Parts like this should have gone away with the Apollo space program. But it gets posted, and no one notices. Instead, people reply with details of their favorite topolgies (which is fine), but no one points the guy in the right direction where he is probably in the need of the most help.

I realise that there a lots of people here who weren't building stuff in the 60's. But the point is that any transistor that ever came in those old metal cans is also old, and therefore should not be used in a new design/project. There are much better, and modern, parts at our use now.

If one were to ask "Hey, I have a schematic here, probably from the 70's. It has some thing called a 2N2102. Does anyone know what it is? Are they still made? Any better suggestions?", I hope that we could help out.

What if someone asked "Hey, which op-amp sounds better-741 or 1458? And which brands?", I bet we would howl with laughter. Then help the guy out. The same should go for old semiconductors of ALL types, not just obsolete op-amps.

Diceman:

I don't claim to be a master of human communication. Although I will point out my last boss (who thought I was a nut.....wait.....he KNEW I was a nut [joke]) always felt I was the best at written communication of all the engineers there. Bet you are glad you didn't work there [joke]. Wish I didn't either [also joke].) I have a warped sense of humor that comes from 30+ years of working with this stuff. And getting told I can't use a plastic transistor even though TI doesn't make it in a metal can anymore. (Very long story......not germane here......sorry.......[pseudo-joke]. Although the story itself is funny.)

Sorry you don't have the same warped sense of humor I have, and that my jokes evade you. I don't get yours either. So then the joke is now on me. Which is fine. And don't stop on my behalf, either.

I didn't intend to break-open already open doors. I do love lively debate. I don't like flame wars, and have no intention of starting one. If so, I could have a flame war with the last boss I had [joke]. Just trying to wake some people up and see who is paying attention.

Nor do I want to drive this debate in some obtuse direction. Like how to derate the operating voltage of tantalum caps. But I am curious, what factor(s) besides linearity do you consider important? Email me direct if you like. (Anyone else who has a specific question or need may also email. I do answer it all. Advice is free, just laced with snide remarks [joke].)

So let me now close with:

Let's stick to the original topic. Thank you.