Blackface-15 Amp Feedback / Oscillation (Solved)

[waldner]

On the heels of PDF64 solving one of my most frustrating amp problems today    I have another amp that has an issue that I can't figure out.  This is basically the normal channel of an AB763 circuit built with the Universal PCB from the Wattkins amp forum.  I made a short video demonstrating the problem at the link below.  Basically, when the volume is turned up to about 8 1/2 or higher there is a squealing kind of feedback sound that isn't super loud and the 6V6 tubes start to glow blue near the bottom of each glass envelope. 

I have attached the schematic and layout for this amp below.  Any ideas?

[waldner]

Here is a gutshot of the amp in case it helps.

[pdf64]

It’s oscillating, probably due to the speaker output signal being routed back close to the inputs. That’s a big layout error.
Do the treble / mid tone control settings affect it.

[waldner]

It’s oscillating.
Do the treble / mid tone control settings affect it.

Just tested it and YES, the treble control does, not the midrange.  I can make it go away with the volume cranked up to 10 if I lower the treble down.  As I bring the treble up, it comes back.

[PRR]

Find a wooden "lead" pencil. And dry rubber sneakers. Bring the amp to the edge of squeal. Pass the pencil around and between the volume and tone pots and wires. What makes the squeal worse? Better? See if moving a wire a cm or so "fixes" it.

I had to do that on an AA-Champ with just this vol/tone scheme and it was easy to find a better routing. Almost looked the same, but was stable at all knob settings.

If you had done something "wrong" no little tweak would work. But it is a known-good plan and I don't see a major deviation, so it is probably just a wee bit off.

[pdf64]

Move the NFB network over to near the speaker sockets, away from the knout sockets.

[mresistor]

It doesn't look like he has the neg fdbk wired up in the actual amp.

[waldner]

Well, I tried moving wires around near the TMB pots and inputs and couldn't get the oscillation to go away.  Perhaps I need to move things away from that area as PDF64 said. 


With regard to the NFB not being connected, that is correct.  I built this amp awhile ago and then wanted to put a switch in with 3 NFB options: stock fender, less and none.  When I wired up the switch and tried it no change occurred.  I then realized that the way the layout is drawn doesn't work and isn't connecting any NFB.  So using the schematic I figured out how to wire it properly and it worked with my 3 way switch.  The problem was the amp didn't sound anywhere near as good as the incorrect layout wiring.  It was noisier, quieter and less dynamic.  So I scrapped the NFB 3 way switch and reconnected it in the most simplified way to how it was originally.  Frankly I'm still confused as to why the amp sounds best the way it is wired now but has this oscillation.  There is no NFB connection. 

In case one of you guys can better understand what's going on, I've attached a diagram of the board traces along with the method I used to make the NFB work properly according to the schematic.  Again it worked but didn't sound nearly as good.  Take a look at the area around J13 and the terminal strips.  Maybe someone can help me get it to sound as good as it does now and get rid of the oscillation.
 

[mresistor]

Have you tried swapping the brown and blue leads from that OT?  (the primary side leads)

[waldner]

Have you tried swapping the brown and blue leads from that OT?  (the primary side leads)

I believe I did when I first built the amp, but will try again now and see what happens.  Thx.

[waldner]

I swapped the OT wires- no change.  I also replaced some of the clutter of the resistors on the terminal strip that were connected to J13 with just a jumper at J13. 

[pdf64]

Is that the 220k to 220k potential divider mounted right by the input socket?
If so, try moving it.

[waldner]

Is that the 220k to 200k potential divider mounted right by the input socket?
If so, try moving it.

Ok, will try to move that.  Thx.

[mresistor]

I don't see why the feedback loop doesn't work as laid out.  Reading over at Watkins on the Blackface page(s) for the board there isn't anyone talking about troubles with the feedback topology. I think your layout above is wired ok   but the amp itself isn't. 

here's the set-up for the board for a LTP PI

[waldner]

I don't see why the feedback loop doesn't work as laid out.  Reading over at Watkins on the Blackface page(s) for the board there isn't anyone talking about troubles with the feedback topology. I think your layout above is wired ok   but the amp itself isn't. 

here's the set-up for the board for a LTP PI

With regard to the wiring, mine has a balance bias pot as well as a bias level pot, so there is some difference there.  Also I added a high frequency reducing cap on the V2 tube. 

Regarding the NFB as shown on the LAYOUT, the 22K feedback resistor is connected to one side of J13 that is then not connected to anything. Also the 47ohm tail resistor is not connected to ground as it should be.

So I redrew it to match the schematic as shown in your 2nd attachment above.  It then worked but didn't sound as good as no negative feedback.  It also introduced more noise.  So I've now simplified it as much as possible with no NFB.  J13 is just a jumper wire.  The amp sounds fantastic- except for the oscillation that occurs at near full volume.

I will double check the amp wiring that you pointed to.  Thanks mresistor.

[waldner]

I've managed to clear off the terminal strip completely that is mounted close to the input wiring.  J13 is just a wire jumper (eliminated the resistors that weren't doing anything given no more NFB.  I also moved the 220k resistors directly onto the board by soldering them with the ground wire coming off the top in a tepee shape.   See attached pic.


I even tried using some aluminum foil surrounded by plastic placed between the input wiring, pots and the other terminal strip and PCB, still have the very same oscillation.  Again, the amp sounds utterly fantastic up until about 8 or 9 on the volume control.   

[pdf64]

…
I even tried using some aluminum foil surrounded by plastic placed between the input wiring, pots and the other terminal strip and PCB, still have the very same oscillation…

So that was screening the front panel inputs and controls from the  board?

Did you try PRR’s ‘pencil and sneakers’ idea, in reply #5?

[waldner]

…
I even tried using some aluminum foil surrounded by plastic placed between the input wiring, pots and the other terminal strip and PCB, still have the very same oscillation…

So that was screening the front panel inputs and controls from the  board?

Did you try PRR’s ‘pencil and sneakers’ idea, in reply #5?

Yes, exactly.  I tried both and couldn't get rid of it.  I chopsticked the hell out of everything.  Moved every wire, etc...


Is there anything else I could do that might be a work around like increasing grid stoppers or something?  I notice the 6V6 tubes on this design already have 8.2K grid stoppers.  Might this be because the proximity of traces on the PCB require more than the normal 1.5K?  Could it be that I need to increase this a bit more?

[AlNewman]

Maybe you could try a very small value snubber cap off of the front of the phase inverter?  Like 100-250 pf from c17 to ground?

[waldner]

So I was finally able to make the oscillation stop by putting a 100pF cap on the V1 12AX7 from plate to plate (pin 1 to pin 6) .  I already had an 82pf cap on the V2 12AT7 between the plates, so I though why not try the same thing on V1.  Why does it work?  I'm not sure. Maybe one of you guys can explain it to me.  I also haven't had enough time playing the amp to see if there are any negative sonic consequences. 


I did start with a 70pF cap and it made the oscillation mostly stop, but there was a little squeal with the volume on 10 and the treble control above 7, so I went up to 100pF.  Not sure if this is a good solution or just a bandaid?


Thanks for all the help!

[pdf64]

It’s not even a medium gain amp, it shouldn’t be necessary to use snubbers.
Whatever works, works, is one way of doing things, and can be pragmatic / expedient.
But I’m not keen on fitting it between V1 anodes, as its effect will vary according to control settings. It should have most effect on rolling off treble with everything up to max.

In the first instance, if shielding isn’t feasible, then I fit them between anode and cathode of the input stage, pins 1&3 of V1 here. And 100pF would be a good starting point.

But shielding should take precedence over snubbing. Here’s a shielding plate I fitted to an AC30 that became unstable after a top boost unit was fitted. The input sockets and speaker output terminals are too close together without shielding, given the additional gain from the top boost circuit.

[waldner]

It’s not even a medium gain amp, it shouldn’t be necessary to use snubbers.
Whatever works, works, is one way of doing things, and can be pragmatic / expedient.
But I’m not keen on fitting it between V1 anodes, as its effect will vary according to control settings. It should have most effect on rolling off treble with everything up to max.

In the first instance, if shielding isn’t feasible, then I fit them between anode and cathode of the input stage, pins 1&3 of V1 here. And 100pF would be a good starting point.

But shielding should take precedence over snubbing. Here’s a shielding plate I fitted to an AC30 that became unstable after a top boost unit was fitted. The input sockets and speaker output terminals are too close together without shielding, given the additional gain from the top boost circuit.

Hmm interesting.... Let me experiment a little bit more.    Thanks PDF64!

[tubeswell]

Seeing as how its a PCB amp, if its been repaired/modded previously, someone who was doing the unsoldering could have been a little rough and torn parts off the board without proper desoldering, which could have damaged the little thin copper lugs/traces on the board with micro-tears/cracks. When that happens, it can wreak all sorts of havoc with ultrasonic problems.

When de-soldering bits from a PCB, you need to use a pencil-tip* iron that's not too cold and not too hot, and use a solder pump or desoldering braid and make sure you get the pad completely free of solder before taking the through-mount part off the board. Don't ham-fistedly wiggle the part out, if its still partly soldered in. Any residual solder contact is likely to tear the trace pad when you go to remove the part.

*Or the appropriate multi-pronged tip if replacing op-amps/voltage regulators/relays etc


Otherwise, it could be cracked jumper leads or multi-strand ribbon cable where the strands are torn/cracked

[HotBluePlates]

... the 6V6 tubes start to glow blue near the bottom of each glass envelope.  ...

Since no one addressed it:

   The blue glow is a non-issue.  That is simply the glass fluorescing when stray electrons hit the glass.