Help me diagnose a terrible new noise from my tube amp!!!


https://photos.app.goo.gl/QCjYHvuMhkArQiH88  

This is the left monoblock of my 100wpc Wavestream Kinetics Push-Pull amplifier, made around 2010, but diagnostically checked by amp's designer when I purchased it earlier this year.  It uses KT88s driven by EL34/KT77s and 6922s for input tubes.  Each EL34 drives two KT88s.  Bias is manually adjusted with dedicated knobs and a central meter.  

This noise is transmitted through the left speaker and is accompanied by flashing power tubes to the rhythm of this noise and the bias meter for that tube bouncing with that same rhythm.  No blows fuses.  Noise happens even with amp disconnected from preamp.  

This sound started on the left monoblock during loud passages in music last night like a temporary clipping or distortion sound, then overtook the music when played at higher volumes.  I traced the problem (or so I thought) to a bad EL34 tube, so I replaced it (and its counterpart as well) with KT77.  This made the noise go away for a while.  However, when I check the bias of this new KT77, it was near zero.  When I attempted to bring it back to its spec of 40mA via dedicated knob, the horrible sound came back.  If I leave the bias at 20mA for that tube, noise goes away. 

What the hell?

Now, for context, I replaced the coupling caps on both of these monoblocks a week ago (with caps to designers spec) and I may have screwed up the soldering somehow.  But I played these amps for several days at normal bias without any issue until last night.
redwoodaudio
I'll be totally honest I've never heard a noise like that from a tube amp before I say get it to a good tech to check out ASAP.
Before you do anything, redo the soldering on the capacitors.

I’ve had shoddy soldering go bad over a period of time. I’ve never had a tube flame because of that, but the amp made a loud scratchy noise. A dry solder joint can look 100% good - in your shoes I’d resolder just in case.
@fiesta75 - I'll be bringing it to a local tech tomorrow, then designer in San Jose next week if we can't solve it here.
"...I replaced the coupling caps on both of these monoblocks a week ago..."

That might be a good starting point. Are you really good at soldering? You might have smoked the caps. 
Now, for context, I replaced the coupling caps on both of these monoblocks a week ago (with caps to designers spec) and I may have screwed up the soldering somehow. But I played these amps for several days at normal bias without any issue until last night.
That sounds like an oscillation, as if the coupling caps had been installed in such a way as to invert the phase of the signal applied to the power tubes.


But whatever it is, the simple fact that this is the most recent change is significant so I agree with @russ69  this is the first place to look.
So I think this is “motorboating”, but damned if I know what that means technically.  Fortunately, there is the internet…

from a 2008 post on ampgarage.com:

This is from the Tube Amplifier Debugging Page: http://www.geofex.com/ampdbug/ampdebug.htm

"In an amplifier that has ever worked correctly once, motorboating is almost always a signal that the decoupling capacitors in the B+ lines of the preamp section are going high impedance, not decoupling properly. Replace the B+ decoupling capacitors at least for the preamp. Since the other capacitors are old, also, consider replacing ALL of the electrolytic capacitors in the amp (doing a cap job - see the Tube Amp FAQ at http://www.eden.com/~keen for info on the what and why of cap jobs.)"

Or, from same thread:

Motorboating can be caused by having resistors of too low a value in the B+ supply. I.e. if you have a 1K5 between two preamp stages’ filter caps, you may get motorboating that could go away if that R was increased to 10K.

It can also be caused by a power supply that is not decoupled enough (too many stages being fed by the same PSU node).
And 

One caveat, motorboating happens too fast to see in a tube as a rule. If the usual techniques don't help, I'd also look for a wrong connection in a gain stage or going into the inverter. That can also produce a popping but much slower. A quick check for that is adjusting a control will reduce or intensify it.

Or alternately, from geofex.com:

Sound cuts out or squawks on loud notes

This is why there are professionals…
Coupling capacitor vs. decoupling capacitor. Different things.

BTW, This is the first time I’ve heard of motor boating. Cool beans - I’ve just learned something new. 🙂
One other way motorboating can occur is if the coupling caps are a larger value than they should be. If they exceed the frequency poles in the power supply, the amp will be able to modulate the supply and the result can be a low frequency oscillation- motorboating.


Whenever I replace coupling caps in something I'm servicing, I replace them with the same value.
Interesting. Amp's designer told me to replace 0.22 caps with 0.1, so they're actually smaller value, unless there's a decimal point mixup somewhere. 
I agree. Go back to your original configuration and start from there.  If the problem didn't exist before you changed the coupling caps, then that is the problem. Take them completely out and start again.

Enjoy
@minorl I plan to have the designer take it from here. He's a few hours away...

I did same thing on both monoblocks, and only one has a problem. I carefully compared the work just now and couldn't find any meaningful differences. But I'm a complete amateur, so... 
Its only in one channel.

So the parts are off the hook.


I suspect a problem with the installation still.
I will post the final results of the diagnostic and repair work here when I get them. Amp is in professional hands as of yesterday.