Minimum hfe

I thought you were buying batches of parts (perhaps floor sweepings or leftovers) of unknown ancestry. Yeah, my plan would require some sort of consistency.

Sure does, but listen numb brains, it was to show the output driver section. If you actually read the post you would of known that but that isn't you, open mouth and insert what ever it is you like to stick there.

The circuit only needs a feed back cap to correct the basic issue and if you really knew what you were doing you would of know that it's oscillating and it needs a feed back cap, you know things that are designed into OP-AMPS? Some where the term miller comes in there? Oh wait, too much for you to take all at once.

Btw, it also needs a current mirror if one was to be serious about using it. But then again, the poster stated it was to be a very basic system made from scrap parts? do you get it yet? You seem to be at the start of the line for degenerative defects, the type that droll and lick the windows on the bus. Don't worry sunshine, there is hope for you.

You, Slowman and Fields should shack up. You can talk trash to each other all night and none of you will have no idea what is going on. Sounds like a typical day in paradise, don't it?

Frigging window licker!

Jamie

using it. But then again, the poster stated it was to be a very basic system made from scrap parts? do you get it yet? "

You do not need a current mirror at all. yu mean the old usual circuit with the diff pair and hopefully a current source ? Even a resistor will do the re, but you don't even need that other transistor. The circuit will work fi ne by just putting the feedback to the emitter of the input transistor. Wha t, you want 0.00001 % THD out of this thing, ust using any transistor that comes along ? Ain't happening.

Fact is most people will easily tolerate 3 % THD, and many 5 %. And, even o rder distortion is more tolerable according to studiesa than odd order. But don't ocnstrue this to mean that a single ended stage is better. It is not , really.

Sorry about that, I made a quick circuit to show you the driver section. But I guess you can't see it. can't you run WINE in linux? You then could use LtSpice. Any way, I used a feed back to keep things hanging around the center point of the Vcc rail and btw, it looks like your output section is basically the same as I have, but that is to me very unstable with temperature changes?

Look for a differential transistor pair that you can add in for the feed back. some good reading for you.

I wish. Nothing I install from the software respository has yet run.

Is it? With temp:

a) Vbe shifts. -2 mv/C -> 200mV per 100C rise on Tr2 and 30C rise on tr1 would only give 60mV change Total delta V 0.26v on a resistor with 2.5-.65= 1.85v across it, a V change across R1 of 14%.

b) leakage i goes up, but I expect it to stay small compared to quiescent current And it doesn't matter if it shifts some from 1/2 Vcc. It only needs to work.

c) there is enough bias nfb in place to move the bias much more than those figures while still working ok.

d) In most cases the tr will be overrated, though I've yet to monitor the drink-can heatsink to ensure its adequate.

Why do you think that's needed? It would double to triple the component count.

Looks like beginner stuff to me.

I've totally omitted any frequency compensation or zobel network... so it'd be nice to spice it. Not sure I've got any hardware that could run win98 any more.

I'll see if I can do something about spice tomrrow.

NT

t will be built from random electronic scrap. There's a darlington pair cla ss A

to

wind.

powering up before its fully built.

transistors is probably different. In most cases people will have no data o n them at all. If it looks the right physical size, and isn't from a high V circuit, it'll get used. A bit different to what s.e.d folks are used to : )

Right, makes sense. Everything - absolutely everything - has to come from d umped garbage. Budget is truly zero. The people that make them keep the pro ceeds, they're not working for anyone but themselves.

NT

You don't need all that other stuff to show the output driver section.

At 100Hz? Miller was never interested in his capacitance at that frequency.

Yeah, it's difficult to understand your word salad.

So, add more parts to make the current mirror and let's see how it works.

Good. But I can't say the same for you.

Licking window Frigger!

You sure are wrapped up tight, aren't you Mr. Philpot Jr? Rather than spouting gibberish, why don't you show us how to fix it?

Of course it's not good, it's a crap idea. if you're not an audio Phool then I don't think it matters.

I remember the old solid state auto radios using a one power transistor output. Now that was doing it cheap.

As for the differential pairs, the current mirror gets added to help the fixed bias side from drifting out to far. The wilson style would work well here to help maintain it. I also had a variable Vcc in mind too, but as it stands the requirements aren't that critical. In any case it's obvious this isn't going to be for what you would call HI-FI!

Jamie

LTSpice will run on Linux under Wine, AIUI.

Looks is deceiving, you really should get a handle on it and then you'll see where it comes into good use. Two more transistors will make a whole world of a difference.

Your circuit can be altered to use it as an emitter follower into the load instead. That why you can place a voltage divider at the base of the first transistor to set the base bias. You will lose some output swing but I don't thing that will matter with this project, at least it'll stay on track better. Install WINE on your linux box.

If you would like to see an image of the schematic I have I could upload it to my website..

jamie

I told you how to fix it.

Jamie

A test article with sockets? If the part works, it goes in bin-A, else the circular bin?

If you can assume the parts have been tested and binned before being used, the whole point is cheap labour, right? then you can get any gain you want.

linear.com

If you're running linux you'll need "wine" too, exact details for that depend on which linux.

Parts are tested to find the pinout. There is no binning though, if it functions it'll be used.

The point is to enable some people in poverty to make money producing electronic equipment. The simpler the process is, and the more they can make from limited resources the better. Hence the very cut down design.

NT

It would if wine would run. It doesn't, and I don't know why.

NT

If I use an emitter follower, I've got nowhere to derive a stabilising bias feed or nfb from. Increasing bias gives higher V_out, not lower. So then I'd need a 3rd tr as you suggest.

Which brings me back to the question of what's wrong with the bias setup as it is. If the 1.2v Vb1 varies by 0.26v over 100C rise of tr2 and 30C rise of tr1: @1.2v Vb1, Vc2 = 2.5v @0.94v Vb1, Vc2 = 2.5*.94/1.2 = 1.96v. Its not perfect, but 2v output swing instead of 2.5v is fine with me. If it means one less tr, its great. In fact if I biased it to give Vc2=2.75v cold, I get 2.25v of swing over cold to cold+100C.

Is that wrong somewhere? (I know I've ignored Vcesat at this point.)

NT

If wine ran on this I'd have done it a while ago. Going to look at other options now.

My guess is 30 min for the NPN iffi a truly random small signal device; otherwise a minimum hfe of 100. For the power transistor, minimum hfe of 3 (sorry, they get pretty bad).

Cannout count on a minimum of 10 for a power NPN..