The differential amplifier

[Euthymia]

Got a couple of solid state amps in the house right now and in both cases, as I chase a sine wave through the amp, it looks great up to a differential pair, then on the output side of the differential pair, the waveform is clipped or nonexistent.

My first question is: what are the signals supposed to look like on all the terminals of a healthy differential pair?

I get that the signal goes in on the base of one transistor (always depicted on the left side in amp schematics).

For instance, in the unit that's on my bench right now, a 1998 SWR Workingman's 15, the left unit of the pair is Q6, which has a sine wave on the base (.2VDC). The emitter (.8VDC) has a sine with the top chopped flat, and the collector (-52VDC) has a square wave. I'm sure something ain't right there, but what?

http://www.rawbw.com/~emiller/Transf...n%27s%2015.PDF

It occurs to me that some of the toughest SS troubleshoots I've had in the past, like the (in)famous Coli(form) Lead, came down to something screwed in the diffy pair. So I'm thinking that it's troubleshooting Kryptonite for me and I need to get a handle on it.

I've looked in my electronics textbooks, done searches online, but I still haven't found anything that explains it in a way that lets me predict what I should see in the way of voltages in and out.

It's funny how people can put so much effort into a really nice theoretical explanation and still not cover "trivia" like what to expect from observing a functional example of the circuit element being described.

It's like teaching someone to drive starting with a detailed description of how a carburetor mixes air and fuel, the linkage, etc, etc, but omitting the part where the car goes faster when the driver presses down on the accelerator.

[Don Moose]

The short version is that, given a sine wave input to a diff pair, you should have 2 outputs, out of phase with each other - mirrored about the X axis. In preamp-level diff amps, both sides will usually be full sine waves (think push-pull, class A).

A power amp stage may well clip the side it's not interested in (think push-pull, class AB2 or B), but both side will appear to clip the 'top' or 'bottom' of their respective waveforms, delayed by 1/2 cycle from each other.

The key is that a difference between the 2 waveforms is most always a sign of badness. The trick is that both sides being the same doesn't mean it's all ducky.

Hope this helps, but I'd be surprised.

[Enzo]

In a normal healthy amp, the waveforms all through it look like the signal.

SS amps are all fed back - massively - and with any break in the chain, the diffy pair will try to "correct" for the missing parts. In other words if the output can;t make say the positive half of the waveform, the diffy pair will increase the gain on that part of the waveform to as far as it will go.

WHile i do find the occasional diffy xstr bad, the pair rarely is the source of the troubles. In my experience anyway. For the noobie, I abbreviate transistor as xstr. Easier for my tired fingers.

And when I mention bias, forget anything you assume from tube amps, the only thing they have in common is the name.

The solid state amp:
there are two power rails positive and negative - ready to supply current to the speaker load. There are output transistors from the output bus to each rail. Then driver transistors control the output xstrs. Turn them on and current flows from the rail to the load. Of course if both sides turn on, then current just flows from rail to rail and fuses blow. So we need to pull their bases down to keep them mostly turned off. This is the job of the bias xstr, Q5 TIP29A.

The bias xstr pulls the + bases lower and the - bases more positive. Basically pulling them towards each other. We turn this xstr on to the point it holds the drivers and output just at the brink of conducting. The slightest little signal will turn them on, but at idle they are just barely off.

This whole totem pole of xstrs will sit there, output bus at zero. Now if anything can grab this bias point and move it, it will control what comes out to the load. The bias xstr merely establishes and holds a differntial voltage between the opposing bases. Move one base more positive and the opposing one will go that way too. I can grab the whole bias xstr circuit anywhere I want and yank it up and down with the signal, and the output will reflect that in current through the load. Q8 does this yanking. We call it the voltage amp.

I can't see the letters clearly, it LOOKS like Q8 2SC4382? Q8 is referenced to the neg rail, but its job is to yank the bias circuit up and down to the tunes. So the music has to come in through its base.

So far, if the output xstrs are not shorted, if the output is stable and not blowing fuses, we can usually say everything to the right of Q8 is OK. If the voltages at the bases - ie the bias voltage - is slammed over to one end, so will the output be. If things are OK, then the output will be centered at zero. Q8 determines where that rest point will be. If it opens, then R23,24 will pull everything positive. If it is shorted or stuck on, then the output will be pulled negative. WHen its circuit is working as planned, the output centers.

DO you have pretty much +rail at the bottom end of D9? (across C12)

The signal inputs to the base of Q6, it inverts and drives the voltage amp Q8. Meanwhile the output bus is sampled through R35 and fed into Q7. Q7 acts like an emitter follower trying to control the current through R30 which is shared by Q6. SO Q7 tries to adjust what flows through Q6 to keep the output matching the input.

At this point you have a large signal at collector of Q6. Is it at the base of Q8 then? They should be wired together after all. And if that is the case, why is it not at the collector of Q8 and thus on the output bus? Q8 open?

[Euthymia]

In a normal healthy amp, the waveforms all through it look like the signal.

SS amps are all fed back - massively - and with any break in the chain, the diffy pair will try to "correct" for the missing parts. In other words if the output can;t make say the positive half of the waveform, the diffy pair will increase the gain on that part of the waveform to as far as it will go.

I've printed out the rest of your reply (highly useful as always) and will be hurling it at the circuit in question, but I have one question about the above statement.

If I'm getting the drift here, you can only chase a sine wave through an amp like this as far as the differential pair, because whatever is messed up after the differential pair will cause the differential pair to go bonkers trying to correct for it.

If that is so, no wonder I've spent so much time going around in circles trying to troubleshoot power sections of solid state amplifiers. I've always tried to figure out what's going on by jacking a sine wave into the front end (or into the power amp in socket) and examining each stage with the oscilloscope and figuring that the first time I see the sine wave not a sine wave that the problem was in that stage.

Uuurrrgggghhhhhhhh....

[Robert M. Martinelli]

Hi,
just a footnote - the rectifier bridge polarity in the upper left part of the schematic is wrong, the cathodes are shown on the negative rail, the anodes on the positive ( backwards ).
Regards
Bob

[Enzo]

Right you are, I completely missed that.

You cannot tell from chasing a sine wave through there where the amp fails because it is so massively fed back on itself.

In the past I might have lifted the right-hand end of R35 from the output bus and clipped it to ground. That would provide a DC ground refernce for Q7 while breaking the feedback loop. That could be useful when looking for DC problems. But I almost never do it anymore since it rarely tells me what I can't find other ways.

Now that I stick my neck out, you will come back and tell me it isn;t the volt amp after all.

[teemuk]

Hi,
just a footnote - the rectifier bridge polarity in the upper left part of the schematic is wrong...

And another footnote....

The rectifier polarity is not all that's drawn incorrectly. Follow down the negative rail further and you'll see more mistakes: If circuit would be constructed as is shown in the schematic, the negative rail bootstrap for the voltage amplifier stage could never work and the 270-ohm resistor would seriously limit the current the "PNP side" power transistors could feed to the load.

[bnwitt]

What a great thread. Thank you to the questioner and to the gurus of SS.

[Steve Conner]

I don't know if Enzo said, but a flip side of the massive amount of feedback is that, when the amp is working properly, the signal at the collectors of the differential pair is tiny and distorted. It's the error signal: the difference between the amp's output and its input.

You could easily miss it altogether.

The key to troubleshooting SS, once you've got it to power up without smoking, is to work your way round the feedback loop looking for something that doesn't make sense. For instance, a transistor that's conducting hard even though it has no base-emitter voltage: it's probably shorted and the NFB loop is doing its best to compensate for that. You can do this with a multimeter while the amp is at idle, and only apply a signal once the DC conditions are right.

Getting back to the original post, the square wave implies that the diff amp is seeing a big error signal and yelling down to the engine room for drive, but nothing is happening. So probably the good ship Workingman took a hit down there. In a classical SS power amp, the next stage should be a common-collector amp, that Douglas Self calls the "VAS", for Voltage Amplifier Stage. What's going on around that?

[Enzo]

Which agrees with my thoughts:

Q8 does this yanking. We call it the voltage amp.

Q8 open?

No, I didn't mention the collectors as the error signal. For some reason I never thought of it that way. I learn something new every day.

[Marina]

This is great! This really helps me to understand this type of circuitry. Thanks for sharing your knowledge & insight. I love this forum!

[Enzo]

I am a big believer in: if you know how and why something works, it is a hell of a lot easier to troubleshoot it.

Once you have been around a while, we will teach you the magic stuff.

I once did a service call for a jukebox company - I was the guy they'd call when something was beyond them technically. The box was rejecting records in the middle of a song, randomly. I met their guy at the loaction and I went all through the thing. There are about four or five things that will reject a record - a button behind the bar, one on the rear of the machine, one when the record runs to its end, missing record detector, and some others. This was an intermittent problem, but I serviced everything that could make this happen.

I then took out my business card, and waved it over the machine while saying, "Woooooo." I handed the card to the guy and said, "Bob, you will never have trouble with this machine again."

And they never did... see? magic.

SOme months later they called me again on some other jukebox, different problem, but still over their heads. Bob calls me, "Enzo, I've tried everything. I checked the voltages, I tightened the connectors, etc. etc....and I even waved your business card over it."

At least Bob believed in the magic.

[jeeptechfred]

Hello,
I know this is an old thread but I found it yesterday and it was a huge help to helping me understand the operation of the power amp section of solid state amps. Enzo, your simple description of the circuits basic operation is almost impossible to find anywhere and I thank you hugely for it. It gives me the "jumping off point" I've needed.
To expand on what was discussed here I am looking at the schematic... Is the purpose of Q1 and Q2 to be a "power supply for Q7 and 8?
Thanks, Fred

[Jazz P Bass]

Good question.
What exactly is the purpose of Q! & Q2?

Workingman 15.zip

[Enzo]

Well, if my seat of the pants explanations helped, great. Now go over to www.ssguitar.com, and in the amp discussion, up among the stickies, is Teemu's book about solid state amps. I highly recommend downloading it and reading it.


Q1 and Q2 in the SWR are a mute circuit.

AC at W12 charges C7, which once charged powers the base of Q1, turning it on. That pulls current through R22 and R20, pulling current through the base of Q2, turning it on. Q2 conducting allows current through R25 from +HV, thus powering the differential pair and enabling the amp. Power amp won't be on at power up right away, so any preamp turn-on noises won't hit the speakers.