Fender 212R power amp blown

[lowell]

Right, ok. I cut the traces, then had to reconnect the collectors of Q12/13 with a wire. Still DC offset at -2.15v

[J M Fahey]

Ok, thanks.
Now an extra step that I forgot, to achieve *full* independence between "driver" (left) and "driven" (right) halves, please lift one end, (any will do) from D28 and from D29, so they stay "in the air" by one leg only, and re-measure if you still have offset.
I don't know what those diodes do, but I want them out of the picture.
If you still have offset, please re-measure (and post) voltages across R91/92/94/95/96/97/99/100.
(Not relative to ground).
I'd *love* to see perfect symmetry along that resistor (voltage divider) chain.
Good luck.
PS: let's finish this test first, but we still have two other nets which affect symmetry. We'll check them later if we find nothing here.
They are R101, D34/35/36/37, R105. Since D35/37 are reverse biased, *in theory* that current gets diverted by R107 to 112, but letīs check that later.
We'll also have to check that D30+Q14 drop the same voltage as Q15 + D31, but one thing at a time.
How I love simple amps !!

[lowell]

"Ok, thanks.
Now an extra step that I forgot, to achieve *full* independence between "driver" (left) and "driven" (right) halves, please lift one end, (any will do) from D28 and from D29, so they stay "in the air" by one leg only, and re-measure if you still have offset.
I don't know what those diodes do, but I want them out of the picture.
If you still have offset, please re-measure (and post) voltages across R91/92/94/95/96/97/99/100.
(Not relative to ground).
I'd *love* to see perfect symmetry along that resistor (voltage divider) chain.
Good luck."

Disconnected D28/29... still offset.

R91:31.5v
R92:10.4v
R94:670mv
R95:600mv
R96:600mv
R97:670mv
R99:9.18v
R100:28.5v

Pretty much as it was before in my previous post with these voltages. Seems that everything from the top of R94 to bottom of R97 is symmetrical. I still have C55/56 removed as I thought maybe they were playing a part in the offset.

[J M Fahey]

Where do you see symmetry?
The upper half adds up to 43.17V ; the lower one 38.95V ; the difference is -4.22V, the offset you have been measuring from the beginning.
Don't you feel any curiosity about that?.
Please:
1) Measure if there is any voltage across R103, there should be nothing.
If there is some, lift one leg and recheck offset, it might be injecting current into the centerpoint for some still unknown reason.
2) If there is not, solder a 10K resistor in parallel with R91. This should rise the speaker rail voltage a few volts.
We would be *very* lucky to balance the offset on this first try, but I want to check that the output transistors "follows orders".
If so, I'm too lazy to calculate the exact value to put in parallel with R91; if somebody wants to do so, he's welcome.

[lowell]

R94:670mv
R95:600mv
R96:600mv
R97:670mv

This part is symmetrical is what I was saying.

R103: 0v

A 33k gets the dc offset to a far better -.19v

[lowell]

Is this where you are going with this JM? Are we "bandaid" fixing this thing? If so, is the -.19v offset acceptable?

[J M Fahey]

If you want to use that word, use it, no problem with me.
I would put the emphasis on the second word: "fixing".
To see it from another point of view, it is as much a "bandaid" as a bias pot, which requires human measurement and tweaking to achieve an acceptable result, or an offset centering pot, same considerations.
In fact, I referred to this design as a poor design, which might have been tweaked, as everybody does (that's why major brand new designs take from 3 to 12 months, between original conceptual idea/sketches-on-a-napkin and full production) but instead lots of "bandaids" were added (servo amp, delayed turn-on, symmetrical bias, etc.).
Somebody mentioned earlier that he saw this as a "hurried" project, to meet some deadline. Maybe so, it looks like that.
I think the designer fell in love with the turn-front-end-off muting, got the idea accepted, then didn't have time to properly correct it.
There is an "Amp sickness" called "Frankfurterosis" (there should also exist NAMMitis) which refers to prototypes made in haste to be presented there. You could also call them "late Friday jobs", although in this case we are talking longer times.
What I asked you to do, trimming the value of an offset controlling resistor, is conceptually not different to adding , say, a 1K preset, with 22K from one end to +40V; 22K from the other end to -40V, and, say, 4K7 or 10K from the cursor to the "input point", the one I asked you to separate from the driver collectors.
I find that offset *very* acceptable, I would now work backwards, reversing everything we made (lifting diodes, separating both halves, etc.) until we have the amp "as original". It should work and have no offset.
Of course, leave those 33K there, that's the "trimmer".

[lowell]

Ok right on. I wasn't sure if this was an acceptable finished fix. I trust your expertise and it's good to know that it is indeed a good fix for a bad design. I'm gonna reconnect everything and check the offset again.

I wish we could figure out why the pa is not symmetrical. Must be that the negative side is "on" more than the positive side right? It's just difficult to narrow it down I guess. It seems to me that the diode string D34-D37 might be the cause. I don't quite understand how it functions though. I can't see how D37 could ever be turned on, or D35 for that matter. D34 drops 5.1v between R101 and D35. I'm confused down stream from there.

Anyway THANK YOU everyone for all the input and help with this... and patience with my posting etc...

[J M Fahey]

You're welcome.
Let me put it this way, don't want to sound harsh at all, maybe there is a better way to say it:
I do not understand a lot of what I see in this design, and do not want to.
It's not that it's *complex* but that it's uselessly so.
Maybe, as was said, those diode strings are there to make it clip in a softer way, more "tubelike".
As you correctly noticed, D34 is forward biased (so it does not "Zener", and D35 is reverse biased, blocking any current through that string anyway. The other half is symmetrical, of course.
The designers are not nuts (I never even implied that) , so probably at some point during the output waveform swing , at a certain power level, say -6V into the speaker, that pair of diodes gets reversed polarity and from there on it gets forward biased, maybe triggering Q16 making it clamp somewhat the waveform ... or maybe it's an add-on to make the short circuit protection ( that's what Q16 is "officially" there for) follow the second breakdown curve of the output transistors somewhat closer, don't know.
If it's a waveform bending trick, it's useless because it's within the negative feedback loop, which will flatten whatever it does .
Although a minimalist at heart, I love a complex circuit if it delivers what it promises.
I find excellent the Peavey Transtube power amps.
There you'll see that all the mojo is created in a block placed after the preamp and before the *real* power amp, and that signal, which reproduces what a couple overdriven push pull pentodes do, including output iron non-linearity, is then faithfully reproduced through a basically conventional power amp (guitar style, of course).
I would spend months analyzing Mr Pritchard's circuits, as well as the very complex Bluetone amplifier,or that famed Sans Amp rack preamp ... because they deliver.

[R.G.]

I have to insert a note about artistry and elegance in design.

Engineering is a potentially dirty business, just like fixing and keeping the results of engineering design running can be dirty and annoying. Ask any car mechanic what he thinks of the design of the late 60's small block V-8s as opposed to maintaining the rust buckets (for the most part) that now contain them.

But there is an elegance and an art to it, when that's allowed to show through the dirty and gritty business of getting the design done and on the market in time to sell and make enough money to feed all the people in the company.

OK, you may all prepare to laugh now. Back in my early 20s, my Dearly Espoused convinced me that we should together take an adult beginner ballet class. This was funny to be in as well as to watch, and not only because of my antics. I could salve my wounded pride with the idea that at one point making football players take ballet classes was thought to be good. 8-)

Anyway, something one of the incredibly graceful instructors told me in a class stuck with me. I could not raise my leg high enough in a kick step while keeping my hips properly and gracefully level. She told me that in class, everything was to be done properly, even if the kicks could not go high enough. Class was for learning to do it right. In a performance, the kicks went to the right height even if you could not do them perfectly. After all, that's what costumes are for.

You practice your art to do it perfectly. When the crunch comes, you get it done. Full Stop. Sadly, there are those who never learn that trying to do it perfectly is what you do so you can perform, and they let themselves wear a costume all the time.

[J M Fahey]

Congratulations on your good taste.
As to what the instructor said, I fully agree, i guess we must try to do things as good as possible, within our limitations, and progress step by step.
One of my favorite tortures is hearing a guitar player trying to touch 20% faster than what his maximum speed is or a singer trying to go beyond their range.

[Steve Conner]

I checked out the power amp circuit. Here's my best guess as to what all the unconventional bits do.

D21-29 are basically Baker clamps to let the amp recover from clipping quicker. They stop any of the transistors from saturating. In order for this to work on both positive and negative clipping, they had to use two VAS transistors each with its own compensation cap (Q12, Q13, C51, C52) and a 1:1 current mirror (Q11, D24 etc.) to send the drive down to the bottom VAS.

D34-37 are for dual-slope SOA protection as JM suggested.

I don't know why they used two bias generator transistors: it seems to me like the circuit would work fine with only one. Unless one is in thermal contact with the driver transistors and the other in contact with the outputs, but that's an advanced trick that belongs in some sort of audiophile hi-fi amp.

The power amp has both voltage and current feedback, but the current feedback is AC coupled, so it can't be related to the DC offset problem. I'm not sure quite what the two op-amps do, except that I think U6A might be the DC servo, hooked up as a "non-inverting integrator".

Since both op-amps are DC coupled straight to the front end transistors, and we have a DC offset problem, I'd be tempted to blame the op-amp. Is it receiving its proper supply voltages?

Did we check that the amp was making its full output? Another explanation could be a resistor burnt open in the power stage, causing a loss of power to one half of it.

[lowell]

One of my previous posts stated U6 supplies were good. The amps sounds good to me. Wouldn't it be clipping and low volume if one side of the power amp wasn't operating?

I guess I have 1 last question here. Why would a designer go through such "inelegant" complexity to make a solid state power amp? I guess maybe they wanted it to be more "tubey" (doesn't sound tubey at all!!!) or something. It seems a simple power amp that is a tried and true tested design would be the way to go, and then create "tubey" like effects in the preamp, or between the preamp and pa.

[Enzo]

I mentioned this before, and have not had my mind changed so far. I think what Steve said even backs it up, at least to an extent. When we talk about preventing transistors from saturating and recovering more quickly, I still have to think the whole point is to wring more performance out of less amp. They throw all these tricks and oddball accessory circuits into it so they can have this thing acting as though it were a more powerful amp than it really is.

I have a Fender Bassman 250 chassis in here now. SUffered a drink spill, and has.... DC on the output. The circuit? Same damned power amp circuit as this.

[R.G.]

I have a Fender Bassman 250 chassis in here now. SUffered a drink spill, and has.... DC on the output. The circuit? Same damned power amp circuit as this.

You can run, but you can't hide. 8-)