Low frequency Feedback in AB763AB reverb circuit

[G._Hoffman]

I'm working on a Super Reverb kit, and I'm very happy with pretty much everything, EXCEPT, when I turn the reverb up past 8, I get a low frequency feedback through the mix circuit. 
I traced it down to the mix circuit.  I tried with the tank disconnected, and it still happened;  I removed the driver tube, and it still happened; I measured all the components, and they all measured right, and all the nodes rang out correctly. 
I finally tried pulling the 470K "attenuation" resistor, and it goes away.  So, it is clearly going through the mix circuit.  But where was the signal coming from, when the drive tube and tank were connected? 

Does it make sense that it might be coming from the HT node shared with V2?  And is there a way to fix this without adding another HT node to the power supply?  Or is there something I'm missing?

[pdf64]

Assuming no gross component value errors, it may be that you have a bad decoupling (cap / connection) at the D HT node, or the V4 shared cathode (don't attempt partial bypass here, unless also moving away from shared cathode).

[Merlin]

How low in frequency are we talking?

[SEL49]

Something to try...

V4A and V4B cathodes are tied together and share a single 820Ω || 25µF. Separate the cathodes and give each one their own 1.5K || 25µF. Using a bigger filter cap on that node may help too.

[G._Hoffman]

How low in frequency are we talking?

The main one is at 35Hz, and then a bunch of pretty high amplitude harmonics going up to around 1k.
(And just as an aside, my new digital oscilloscope sure does make figuring that out a lot easier.  I got my tax refund, and two days later, kinda like magic, a brand new Siglent SDS804 HD showed up!!!!)

[G._Hoffman]

Something to try...

V4A and V4B cathodes are tied together and share a single 820Ω || 25µF. Separate the cathodes and give each one their own 1.5K || 25µF. Using a bigger filter cap on that node may help too.

This does suit my personal prejudice against shared cathode circuits.  I'll try it. 

Oh, just did a bit of math, and the 35hz I'm seeing is (within a reasonable tolerance) also a harmonic of the filter frequency of the cathode RC.  That shows promise.

[G._Hoffman]

Something to try...

V4A and V4B cathodes are tied together and share a single 820Ω || 25µF. Separate the cathodes and give each one their own 1.5K || 25µF. Using a bigger filter cap on that node may help too.

That did it!!!!  Thank you so much.  That was the last thing on this amp, electronically, so now I just need to replace the baffle so the speaker doesn't hit one of the output tubes, and the amp is done. 

[Latole]

Interesting topic.
I have two questions why is the circuit called AB763AB when the original Fender circuit is simply AB763?

Perhaps the first question will answer my second?
Why does the original circuit work so well with both cathodes together?

Thank's

[pdf64]

...
Why does the original circuit work so well with both cathodes together?
 ...

It works fine provided the shared cathode (in a cascaded common cathode stages) is fully bypassed.
However when the cap eventually begins to fail, the bypass becomes progressively less effective and a positive feedback loop forms. Eventually this will cause oscillation at higher reverb settings.

I recall a thread here in which the builder used partial bypass on V4 cathodes in order to trim the low end, and got oscillation.

[Latole]

...
Why does the original circuit work so well with both cathodes together?
 ...

It works fine provided the shared cathode (in a cascaded common cathode stages) is fully bypassed.
However when the cap eventually begins to fail, the bypass becomes progressively less effective and a positive feedback loop forms. Eventually this will cause oscillation at higher reverb settings.

I recall a thread here in which the builder used partial bypass on V4 cathodes in order to trim the low end, and got oscillation.

Thank's ,

Why here OP need to separate cathodes a use another resistor and cap ?
If the issue is the cap ?

[G._Hoffman]

Why here OP need to separate cathodes a use another resistor and cap ?
If the issue is the cap ?

It was suggested first, so I tried it and it worked.  But it also suits my personal preferences.  It's about $2 extra in parts to separate the cathodes, and it removes an area of potential failure.  I had the stock value of cap in there, and I'm guessing it was a faulty cap (I don't want to fish it out of the trash to check), but it's an electrolytic cap - they fail.  Now, sure, the two caps may fail in the future, but it won't cause that kind of problem again.

[stratomaster]

Now that they're separate, I'd suggest using a very large cathode cap (100μF+) on the reverb recovery.  I've done this to great success on Princeton Reverb circuits to drive down the noise floor.  Fender did this on the standalone reverb units.

[G._Hoffman]

Now that they're separate, I'd suggest using a very large cathode cap (100μF+) on the reverb recovery.  I've done this to great success on Princeton Reverb circuits to drive down the noise floor.  Fender did this on the standalone reverb units.

It's pretty damn quiet, but I'll think about it if I have any issues. 

[Willabe]

Why here OP need to separate cathodes a use another resistor and cap ?
If the issue is the cap ?

The problem is de-coupling. Signal is sneaking back around through the cathodes and the power supply.

Separating the 2 cathodes fixes it, now decoupled from each other.