another bizarre audio circuit

I plugged this into LTspice and with a bit of twiddling appears to work well, at least under simulation. I used a supply V of 15V, transistor input resistor (R1) = 100K, input capacitor (C1) = 1uF, transistor emitter resistor (R2) = 3.3K, feedback resistor (R3) = 22K, cap in series with feedback resistor (C3) = 10uF, bypass capacitor (C2) = 2200uF (needs to be big or lose LF gain/damping) output capacitor (C4) = 470uF (let it charge before connecting) inductor size of 10 henries (for sim assuming perfect, 0 ohms) transistor collector resistor (R4) = 3.6K but can vary, mosfet source resistor (R5, parallel with C2) = 47 but can vary. Transistor = whatever (2N5550 in sim), mosfet = IRL530.

These values optimize for medium output power (570mW) into 50 ohms and reasonable power (around 300mW) into 32 ohms and 100 ohms but with off-center clipping. R4 and R5 can be varied to deliver the desired power into the desired load... some of the values I tried...

(PDQ is mosfet dissipation, PDR5 is R5 dissipation)

R5 sets the overall power level, then adjust R4 to achieve balanced clipping.

Output impedance is fairly low, not much variance as load changes. Gain is approximately R3/R2 plus a bit. Distortion increases as R4 (and open loop gain) decreases but it appears rather "tuby".

A 10 henry inductor is probably overkill, anything 1H or more will probably be fine, for a HP amp there's plently of overhead and the negative feedback will mostly correct for deficiencies, smaller inductors just have less output at 20hz. Could probably use the secondary of an output transformer with the primary insulated.. but watch out for core saturation. Should have fairly low resistance, preferably less than a few ohms (can tweak values to compensate). Someone in the thread said large inductors are "unobtainium" but that's BS, transformer windings ARE huge inductors, for this level of power a winding of a power transformer will probably work. For class A amps using an inductor or transformer output doubles efficiency and halves the supply voltage needed for a given output. It also presents a high impedance at audio frequencies so that only the load determines the impedance (thus the gain) of the output stage. It's possible to use a resistor load but won't perform as well.

But this one does make sense. There's a reason this basic circuit has been around about as long as transistors...

Terry

..

=A0Not

Not really. Most of the people who post here haven't used a 555 for years. Not because they don't know about it or don't like it but because the kind of problem that it was designed to solve started being solved in other wau=3Dys around 1980.

I don't impress me. I see no reason why I should impress you.

-- Bill Sloman, Nijmegen

:) Tempting to put that on my sig.

Ironically, it looks somewhat similar to the old style fuzz face guitar effect.

You do not know that, and you saying it does not make it true.

You are an idiot to think so.

Only if it is a collection of stupid remarks.

My agenda is, and always has been, to design electronics. My ego has been tuned to further that end. Electronics design requires a combination of arrogance (to believe you can do things other people can't) and humility (to avoid the thousands of possible mistakes) and compulsiveness (to get it all done, all right.) And, more than anything else, brutal honesty. Not many people an manage all that, and lots of other people don't like the people who can.

There's not many things more fun than doing this with other people who know how. Especially since the whiteboard was invented.

John

Except that it makes distortion, and a headphone amp shouldn't!

John

eamp

one amp...

e fun

not.

ong with

ng with

to pursue.

assign

tors?

clucking

how

ut.

le do

out your

er to

OK,

ch

ne.

es

Intersesting thanks. Is there a reason not to take the feedback from the output (transistor collectors) rather than the opamp itself?

George H.

amp...

with

with

pursue.

assign

clucking

your

Depends on what you want to do. In my gradient amps, I want the overall box to be a current source, so I don't use any voltage feedback. The upper and lower boost transistors are replaced by precision current mirrors, and I usually cascode the opamp supply currents up into the mirrors, bacause the rail voltages tend to be high.

As a boosted voltage amp, you'd generally want voltage feedback from the final output. The feedback can go into the inputs of the opamp, but I've seen cases where the feedback was applied to the *output* of the opamp.

John

I make mistakes all the time, and a lot of my ideas get paved over by somebody else's ideas. I work with some *very* smart people who, in their areas, know a lot more than I do. That's part of the fun of playing with ideas.

But if you want to argue over definitions, like whether something that's unboundedly large can be referred to as "infinite", that's just words, definitions, and doesn't matter. It certainly doesn't affect the electronics. A latching relay does what it does.

John

If you are working at the microvolt noise level, then yes, the same rules may apply, depending on circuit configuration. If you draw the equivalent circuit of the input stage, the base resistance is effectively in series with the source, so contributes to the equivalent input noise voltage.

At low frequencies, shot noise dominates. We had silect transistors selected by ti, part number BC315L, but we still hand selected from each batch in a jig for lowest noise figure for the input stages. In '74, op amps were nowhere near good enough and even now I would take any such claims with a pinch of salt until they had been evaluated in circuit, on the bench. We spent a lot of time working on input stages and found, for example, a close correlation between collector current and optimum input impedance. Of course, there's an optimum Ic for lowest noise figure as well. Older low noise devices like the 2n2484 or 2n930 were specced down to 10uA collector current, iirc. We used step up ratio transformers most of the time, as it was the only way to get the required match to optimise the noise figure against Ic from a 600 ohm source. Even now, I would suspect that it would be difficult to build a truly low noise input stage with 600 ohm source direct to input. Differential input stages degrade the noise figure as well, as you then have effectively two transistor's worth of rb contributing to the total.

I had a look last night for the file, but it's boxed up in store. Will try to dig it out in the next week or so. Had quite a bit of fun designing audio stuff, but then micros and programming arrived and I was hooked for good :-)...

Regards,

Chris

Of course you haven't seen this circuit before: I just invented it.

But inductors were widely used as plate loads in the tube days. Tubes were expensive and had low gains, so transformers and inductors were sensible. Early transistor amps used lots of transformers, for the same reasons.

John

It's a newsgroup, not life.

What is infinity/1 ? You seem to be arguing that nothing can ever be unboundedly large since any number can be divided by 1.

That's the way infinity tends to work. I was taught that infinity isn't a number, it's a limit.

( Lim (1/x) as x>>0 ) >> infinity

which works well enough in engineering.

John

A similar thread to this a few years ago prompted a bunch of responses saying exactly that, from people with a history in this group.

Obviously not. It does happen to be true, none-the-less.

You may think so.

-- Bill Sloman, Nijmegen