Frontman 212R Power Amp Problem

[Zephyr]

Hi All,

Am trying to troubleshoot a friend's 212R.
Hope this isn't too long-winded...

I've built several pedals and fixed a few solid state amps for friends,
but this is more complex than what I've done before.

The problem is in the power amp. Pretty sure.
Hooked the pre-out to an old receiver and used it to drive the speakers.
No problems after several checks this way over a week.
But an intermittent problem in the power amp. It might be fine when first powered
up or it might not. Sometimes cranking it up will make it act up.


This is where I'm at...

I set the amp up with a 3mV 1kHz sine wave into J1 with a 4ohm dummy load.
per schematic instructions.
The signal is good out of the TL072 op amp into the differential amp.
When it's acting up the output is a half wave rectified waveform at the
dummy load.
At the emitters of Q9 & Q10, or below at the junction of R85 & R86
it's always sine wave. (The scope is the Conar 255 I built 30 years ago in tech school !)

Checked DC voltages and found that when it's not acting up bias voltages seem ok.
When it's acting up there is approx +0.5vdc shift.
Checked DC voltages and put them on the schematic. Seems the bias point is shifting

What's got me stalled is that I've found that any attempt to check D24 voltages will
snap the bias voltages back - cathode or anode, touching the probe to either and
the wave form at the output goes from half wave rectifier to sine wave but not
vice-versa. And the amp will work well... until the next time.

R88 and R90 and the base collector caps seem pretty warm when it's operating
correctly and cool when it's acting up. Not hot enough to blister, but definitely uncomfortable.
The data sheet gives 150C max temp so this might be ok...don't have that much
experience with this stuff.
Re flowed the solder joints... if this is non lead solder - it sucks bigtime!
Am thinking about getting out the solder sucker and replacing it with the good stuff.

So I'm thinking it could be D24 or one the others, any one of 14 transistors. cap
or an offset in the differential amp.

The power supply rails don't seem to have ripple etc and the voltages are pretty
much constant

Can't help thinking I'm missing something obvious.
For now I've still got all my hair.

[Enzo]

OK so you are losing half the waveform at the load, yes? A push pull amp has a positive side circuit and a negative side one, so the amp is no longer working on one side, yes? By the way, which half is missing? positive or negative?

In the solid state amp, the bias is the difference between opposing bases, not just the voltage to ground. Your bias isn't changing, just the voltage on the bases. Look at the outputs, half a volt on each, so 1v between them, after your shift, there is still 1v between them, it has just moved off center. Look further back at Q18,19. Now we are two drops away from the output, so 1.1v plus and minus, for 2.2v, and your report has 1.6v and .6v, so 2.2v.

SS amps are massively fed back on themselves, the output is sampled into the invert of the TL072, and whatever the input wants, the system watches the output, and if it doesn't see what it wants, it tries to correct it. If half the waveform is missing, then the thing TRIES to restore that missing part, and the only way it can do that is to feed it more signal. I could be wrong, but I suspect that is why your offset is happening.

If half a waveform is missing then clearly this is happening with signal present. But while that is happening, do the same DC offsets remain with signal removed? Oh, and you have no offset on your output, but that is with a load. What is there with no load, and does the symptom ever occur with no load?

You have half a waveform at the load/output, but what does it look like at the output bases? At the driver bases? etc.

[The Dude]

You might try freeze spray. If it's an intermittent component, sometimes that will suss it out.

[Enzo]

yep.

As to D24, touching it triggers things, but while it might be the diode - 2 cents and you can replace it anyway - I suspect the impedance of your meter is enough to overcome some floating circuit element. Instead of touching the diode leads them selves, to take the reading, touch some place the diode connects to. For instance use the base of Q11 or the end of R84. Does touching those points also do the trick? If so, it may not be the diode, just the point in th circuit. I am in no way ruling out the diode, just warning.

[Zephyr]

Diode

yep.

As to D24, touching it triggers things, but while it might be the diode - 2 cents and you can replace it anyway - I suspect the impedance of your meter is enough to overcome some floating circuit element. Instead of touching the diode leads them selves, to take the reading, touch some place the diode connects to. For instance use the base of Q11 or the end of R84. Does touching those points also do the trick? If so, it may not be the diode, just the point in th circuit. I am in no way ruling out the diode, just warning.

Meter Impedence... had thought of that.
The diode is a 1n4444 with a forward drop of .555v ( checked w/DVM) Think I could get away with a 1n914 or 1n4148 with a higher forward drop?
Or would this unbalance things?

[Zephyr]

Heat and Cold

It has acted up with the amp cold. Off all night and was fubar when first powered
up. Also does it warm. Would this rule out temp?

[The Dude]

Mostly, yes, but I would say not completely. You never know. Whether heat related or not, freeze spray can still help to detect intermittent components because of its compression/expansion effects. In other words, even if the failure is not heat related, freeze spray can still often help find such problems.

[Enzo]

Fender is pretty clear they want 1N4448 instead of 1N4148, so I stock them. They are cheap like any diode.

Did you try the test I described taking voltage readings at other points rather than right on the diode leads?

[Zephyr]

Load No Load

If you mean with no dummy load, I don't think I have checked voltages
with or without signal.
I can't remember if I checked with no signal input with the load connected.
But I will check this next. I can see now how that might show up
something masked with a signal applied.
I was looking to see what was happening under operating conditions.

The junction of Q12 Q13 collectors connected to R103 seems to be the reference
point for Q14 Q15 emitters. Maybe a sort of virtual ground.
It looks like it sets the operating points of the bases of Q18 Q19.
Maybe not. This is way more complex than what I have done before.
This is what I meant When said bias.
Maybe not the best use of the term bias.

[J M Fahey]

Don't worry about yourself, this *is* a complex amp.

FWIW I always analyze every new amp schematic I see until I understand it, consider it a bad case of OCD if you wish , one of the reasons I hang out here,because besides the amps I actually lay my hands on, I also study those posted here .... yet, ... over a year and I still do not understand it, go figure.

That said, let's give it a try.

1) still unanswered, which half sine remains, which dies?

2) what about no load/full load?
Sometimes the voltage is present but with no muscle (current capability) to drive a few amperes into a resistor.

3) DC voltages look "shifted"+1V or thereabouts, that's weird bacause the circuit is symmetrical and to boot has some kind of DC servo ... or something similar

So far I suspect "something" opens in the signal path from input to speaker, in the missing half side , but nothing more precise than that.

[Enzo]

I knew what you meant, just nit picking terms.

The way it works is the bias holds the opposing bases apart by a constant amount of voltage - eliminates crossover distortiin mainly. Then the voltage amplifier stage - VAS - moves that whole thing up and down to the music, and the output follows. All those outputs and their drivers are amplifying the current, not the voltage.

if you operate the amp without a load - speaker or dummy doesn't matter - then the output is not asked to deliver any current. The load of course DOES demand current. This difference is a major one, and how the amp reacts tells us a lot.

[Zephyr]

More Load-No Load

I'll try to answer the last two posts with this.
And many thanks for taking the time.

FWIW, I was looking at this amp section overall like a simple amp with
a split supply and an input biased halfway between the power supply rails.
Maybe I should have been thinking more like it is two separate amps driven
by two separate inputs (the differential amp) from the op amp.

JM,
1) Must have had a senior moment. The negative half dies. looking at the bases on the transistors
that are negative to ground, with bases shifted +.5vdc I was thinking they are being cutoff.

2) Did the no signal and no load checks. Nothing changed with no signal with load. But then it is intermittent.
The no load and no signal. Got it to shift pretty much to the voltages I recorded in red on the schematic.
Turned on the signal generator to look at what was on the bases of the differential amp.
There were sine waves present but one was much larger (IIRC) ampltide and had a dc offset negative.
The other was pretty much at the center of the crt. The scope jumps around a bit when switching
between v/div so it's only a ballpark as to dc offset. But i think it was.

3) What I see when looking the voltage shift when it's acting up is the +.5 volt shift at the junction
of R103, Q12, & Q13 and the voltages at the bases of Q14, Q15, Q18, Q19, Q20, & Q21.
They are all shifted +.5vdc relative to ground.

Enzo,

As I said, maybe I'm not approaching this right. Looks like the two halves of the op amp are
each controlling either the positive or negative half of the output. I THINK I saw a dc offset
at the base of one side of the differential amp. Problem is, that when I tried to check again
it was just like touching D24, the problem cleared right up.

I'm going to do the no-load check tomorrow and look at the op amp - diff amp also,
any ideas on how to get useful readings without clearing the problem up... it's bit of a Catch 22.

[Enzo]

I am not as concerned with your half a volt offset as you are. You are focused on it laser-like. I might be wrong, but I think it is an artifact of the underlying problem.

remember, the whole power amp is one big servo, the thing is a huge loop. The output is fed back to the differential, and the thing tries to correct any problem it sees. if the output cannot swing negative, the diffy tries to increase the negative drive to compensate, until it runs out of room to do so. the output stages are all cconnected. If the base of Q18 is high, then so will be the base of Q19. And what they do, the outputs follow, so the bases of Q20 and Q21 will also all be the same higher. The output follows the drive.

You've lost the negative side output, so apparently Q21 isn't conducting, or R112 is open or a connection through them. That is why I wondered what the signal looked like at the bases. You know what the output looks like, so does the base of Q21 look like the output? Or looks like the signal? And then the same questions for the base of driver Q19.

Wait, Q11 has 39v on the collector. D25 has -39v on the anode. is that a typo? Those points are connected together and should have the same voltage, unless the path is open.

[Zephyr]

Enzo,

Gigantic servo... thanks for this way of looking at it.
This is the nudge in the right direction I needed.
I was looking at it based on the limited experience and knowledge I have.
Will check those wavforms on Q19 and Q21 and see what that might tells me.
Might be a way around that catch 22.

39.5v -39.5v ? Typo sort of
I put the 39v under the line next to the diode. Wasn't for the collector. Not the best choice.
Like you said I was focused on the base voltage shift and the voltage at R103.
Just double checked. -39.5 on the collector.

[Zephyr]

Enzo,

Put the scope on the output transisors and the waveforms were the 1kHz sine waves.
But the amp was behaving itself so I wasn't surprised. I'll have to catch it acting up
to make those checks.

I did check the waveform on the bases of Q9 & Q10. Something didn't look right
comparing them so I slowed the time base down to 5ms/div .

Took a couple of snapshots. Q10 doesn't look right at all...