Building a class A audio amplifier - no audio out

There's 101 ways to do it. Designing power amps isn't quite so simple especially when you want 1600W @ 0.02% THD.

Graham

complex at

and

Back then, few people knew C. A lot was done in PL/M which was Intel's own development language.

The job was offered on the basis of my programming skill on a 'BBC Micro' for computer animation. I recall one Xmas party at the film company I was working at then and I was explaining how it worked to some tech. He said "you can't do that". I responded "watch this then". He didn't quite fall over backwards. The company was part of a group, word got round that I knew how to program, hence the offer.

And to be honest, later I did a job in embedded Pascal ! How obscure is that ? I said I wasn't familiar with Pascal and one of the Directors said he'd loan me his book. Shortly after I was programming in Pascal. I ended up liking it because of its stong data types.

That's when I realised that if you can intuitively program, it doesn't really matter what the language is. I've even done some DSP assembler for audio effects.

I still hate C's syntax though. All those (void) s drives me nuts.

Possibly so !

together

That's the basic voltage gain stage. Then learn about emitter resistor degeneration ( a form of feedback ) and how bypassing it with a cap removes the feedback but still helps with DC stability.

used

instinctively

No, start with the basic one. Then move on to Darlingtons ( and possibly triples ) and only then the Sziklai pair and compare their differences.

I thought it might be but in practice I haven't seen any real advantage in an output stage. It went through a phase of popularity but I ended up going back a quite different design where the main output device(s) is actually OFF in quiescent mode and the output comes from some beefy drivers. It's all to do with trying to make the mutual conductance as linear as possible.

And don't let the load impedance reflect into the final gain stage or instability is likely to result, i.e. use LOTS of current gain.

Graham

I consider my PL/M to be nearly as readable as English to anyone with a programming mind especially with decent use of literals. I once 'converted' someone from assembler to PL/M. At the end of the project he said he understood why I'd insisted and he's quite a high flyer. Mind you he made loads of silly mistakes but his code was so easy to read because it was PL/M, it was easy to find the errors as they got reported. It ended up as V 2.4 though.

My later re-write for an enhanced version was issued as a 'beta' i.e V 0.9, thinking *someone* MUST find a flaw. Then we forgot about it and a year later the product was still shipping with V 0.9 software and it seemed pointless to change ! ;~)

And also with PL/M you can drop in 'assembler like' lines to to set registers directly. IMHO it combines the best of medium and low-level programming.

Yes, I totally agree. I think they were quite upset I didn't take the bait.

Typical HR nonsense. Ask for the impossible. Then wonder why either (a) you get no responses or (b) those who do respond haven't a clue.

Graham

VOID !

It drives me nuts. Also things like = and ==

Graham

look

C++,

C

here.

nt

mple

th 2

It's lowercase void, by the way. :D

Ah yes. I took a Fortran class (was required for all engineering majors at the time: 1998 or so) and I couldn't figure out why my code wouldn't work.

I had > instead of .gt. I had < instead of .lt.

IF a .lt. b, etc.

Oh joy.

In C, the open parentheses took some getting used to, also.

I guess C *could* have made void the default type, and then if you

*wanted* a function to return an int, declare it as int... oh well.

Regarding =3D and =3D=3D, Pascal forced :=3D to assign something, which I thought was weird too... but yes, I made many mistakes with =3D and =3D=3D too...

Michael

The type syntax can be a bit confusing if you are mixing prefix and postfix operators, particularly in conjunction with function pointers (but most people would just use a typedef), or if you want to specify the type without naming a variable (e.g. for a cast).

Huh? What's wrong with "void"? Would you have preferred "()" (a zero-element tuple) like in functional languages?

Are you advocating for ":="? Using "=" for both assignment and comparison isn't an option in languages where boolean expressions aren't relegated to the tests of if/while instructions.

or (c) those who respond are very experienced; very experience bullshit artists, that is.

But the most common form of "typical HR nonsense" is to specify very high levels of knowledge and experience but offer McJob salary levels.

"The position requires X, Y and Z, and will pay W" should be "The position requires that you are willing to work for W; having any of X, Y and Z would be an advantage".

s

Ah, I see now.

"Back in the days when audio kits were popular, there was a line of amplifiers called Tiger Amps sold by a company named Southwest Technical Products Corporation (SWTPC). Some of these amps used common- emitter output stages which had local feedback around the output stage to cause it to operate at a voltage gain of 4. This was achieved by adding resistors from the emitters of Q14 and Q15 to ground so that only a fraction of the output voltage is fed back to the emitters. These amplifiers were very unstable, causing them to oscillate, overheat, and blow tweeters. Walter Jung tested one of the SWTPC Tigersaurus Amps for the Audio Amateur. His concluding comment in his review was, "Dan Meyer, where are you?" Dan Meyer was the founder of SWTPC. His original company name was Demco, but the General Motors Delco Radio Division brought a lawsuit against him to force him to change the company name. The company went out of business after getting into the home computer market."

Michael

Even WORSE. A language that's case sensitive ! God preserve us.

Those characters didn't exist on keyboards of the time.

I would say so.

You don't make those mistakes with PL/M because they don't exist.

Graham

I don't need that shit for simple embedded programming.

I'd prefer PL/M's "return value".

You'd be amazed how unneccesary it is.

Graham

tm

as

The only units I built were the 2 Tiger amps running right at 60 volts power supply - I don't remember how far the supply drooped under load, it was 37 years ago. The finals would blow and take out the power supply fuse. The output cap (would never run a single supply amp again) kept any DC from the speakers. No speakers ever failed. I later worked in a stere repair shop and a customer brought in a Universal Tiger pair but I could never get it running properly for him as it had a nasty tendency to oscillate. I had some friends with Super Tiger amps and AFAIK they worked properly. My antique stuff now is a Hafler preamp, Technics parametric equalizer, Adcom power amp and an HK Citation 15 tuner

G=B2

obsolete

now by a

Bullshit. It is what it is: a jelly bean, general purpose transistor. The reason you don't like it, is that it is a JEDEC registered part, instead of your lame European numbering system. It was never designated as a low noise audiofool part.

Apparently, you do. Since it is available in multiple packages and mounting styles, the original number is the appropriate one. The people posting simple designs online have no idea which package will be used, and by whom.. If they want metal, they use the original. If they can use TO-92, they use the PN, and if they want surface mount, they use the MMBT. The online designs aren't expected to be exotic performers. They are simple circuits, to be built with easy to find parts. If you plane top build 10,000 of something, you can comb the world for whatever you want, but if you are building a single item and have no local source of parts, then the recommended parts are fine.

You can't get it through your thick skull that beginners need things to be simple enough to learn from. No matter how highly you think of yourself, you were not born knowing how to do anything other than breathe, and fill your diapers.

--
You can\'t have a sense of humor, if you have no sense!

"The Brute 70" I built this one in high school. I made my first PC boards for it. They took the circuit from the RCA Data book and drew up plans around them. I saw a lot of early commercial solid state PA amps built from the same basic circuit. Some had been ion daily use for 20 years before they needed repairs. The big change was the addition of a output transformer to match them to either a 25 or 70 volt line speaker system.

--
You can\'t have a sense of humor, if you have no sense!

can't do this in C

Procedure wrapper(x:integer) procedure recursive(y:integer) begin if something(y,x) then recursive(y+1) end; begin recursive(0) end;

ook

++,

ere.

t

ple

h 2

Probably not as written, but couldn't you move recursive() to outside of wrapper() ?

If it needs variables local to wrapper(), pass the addresses of the variables to recursive()...?

Michael

eed

PC

p

Interesting. Are Q6 and Q7 both supposed to be NPN?

And what is this RCA Data Book?

Thanks,

Michael

Yes. Those were times when high-power silicon PNP transistors were not easily available. The arrangement was sometimes called a quasi-complementary circuit because it's not a true symmetrical NPN-PNP complementary circuit. I don't know if RCA invented the circuit, but it was they who made it popular. The circuit was presented in the 1968 edition of RCA Transistor Manual, one of the first reference books I owned.

Mike is probably talking about the same book.

Yes, it's commonly called a 'quasi complementary pair' output. One side is a Darlington, the other is a Sz... pair ( never can remember how to spell it ) or 'compound transistor'.

This was done long ago because it was difficult at the time to make high power Si PNP power devices. It's not a problem now and no-one uses it any more because of the inherent gm mismatch.

The same as the one I've got by the looks of it ! '69 I think.

Graham

I have it too. It was a poor transitional phase to pure complmentary output stages. The method lasted some time though.

Graham