1976 100W Fender Twin Reverb SF: "feedback" after modding PI to AB763 specs

[Lambertini]

Hi!

I'm converting the Phase Inverter and its coupling cap from the Twin Reverb 100W Silverface specs to AB763 to tighten and tam the excessive bass, as well as get a bit high ends back.
I did the mods described in the attachments (changes are in orange in second pic).

The result is perfect... for 15min playing. After that time, once nicely warmed up, the amp enters into a sort of feedback/oscillation loop.
The attached sound is short as I don't want to let the amp run in this state, but if I leave it on few seconds more, the "feedback loop" volume increases progressively.

https://soundcloud.com/ekovah/20231223-trsf-feedback

I reverted the mod, played the amp for 1 hour, no feedback loop but overbearing bass and lacking treble again.

Might be a bad new (carbon comp) resistor? All values are ok but might badly react to heat?

Might be the fact I didn't retrofit the BIAS circuit to AB763? Don't see why but I'm not that comfortable with circuitry. I would like to avoid extra mods without fixing this first or at least understanding why it's happening.

Any idea what could be going on?

Any advice on how to proceed to identify the root cause? Should I replace some resistors step by step and do some tests between each? If so, in which order?
Thanks!

[sluckey]

So many colors. So many words. Hurts my eyes.

[Lambertini]

It's my learning path ;) Focus on the PI circuit around the 12AT7. I've changed the coupling cap + resistors value that are in orange around it, and the grid leak between bias pot and 6L6.

I've added green indicators here.

[Lambertini]

I've done many tests and came to one more problem: with the Orange Drop 0.01nF, 50 cycles hum is very low, whatever the resistor configuration is.
As soon as I switch to the 1000pF Mica cap, the 50 cycles get +20dB, anything else is the same.
Am I doing something stupid by using this cap for coupling the PI?

[sluckey]

Try a .001µF (1000pF) but not a silver mica type.

[Lambertini]

I wish I had one. New order again...

[Lambertini]

Did many tests again, changing carbon comp resistors to metal film, retrofiting the balanced bias circuit to the AB763 specs, tried different caps...
The only stable PI circuit in my amp is the original Silverface. I don't understand why but it's a fact :/
I don't like to do this but I'm giving up.

[Esquirefreak]

Silverface lead dress is usually not as pretty as blackface. That, in some cases, can lead to oscillation.

/Max

[sluckey]

Try a .001µF (1000pF) but not a silver mica type.

I used an orange drop in my amp. Fender used a ceramic cap.

[Lambertini]

Thanks sluckey, I've ordered a bunch of various 0.001 cap: mustard, orange drop (2 kinds), roderstein MKT, jupiter, mallory. I hope to find the right one ;)

I also discovered that in the 68 Custom Twin Reverb, the 82K and 100K plate resistors are 1W. I was using 1/2W. Would rather match with the "after 15min" oscillation, wouldn't it? Hopefully placed in the same order...

[sluckey]

Fender used 1/2 watt resistors in his AB763 amps. I doubt they are the cause of your oscillations.

[BrownIsound]

Silverface lead dress is usually not as pretty as blackface. That, in some cases, can lead to oscillation.

/Max

Yep to really blackface the amp, the lead dress needs to be changed to bf as well.


Id personally rather keep a spade a spade and go with what you got, unless it’s a total basket case.

[Lambertini]

@BrownIsound, @Esquirefreak, I don't really get your point, can you be more specific please? Like rewiring the whole circuit?! With no fact or proof that it will fix anything? I'm sure you don't mean that and my bad english doesn't help me to understand.

The goal is to tighten the bass and get better highs. I'm experiencing the difference you can here in this video @ 3:20.

FYI I'm using it with 2 different sets of speakers:
Set A - EVM12L + Cannabis Rex
Set B - 2 x Jensen P12N

Set B is rather ok, very vintagy tone, bass not too muddy and trebles are nice. I can play the amp as it is now with this set.

Set A could be really great, very modern, rich, smooth and clean, but the bass isn't tight enough and the treble is a bit lacking. The bright switch can do the job for highs for chord progression, but then it gets too bright for higher strings and single notes.

When I convert the PI circuit, it works great for both sets, for 15 minutes before entering into oscillation...... The change between the 2 configs are 7 resistors.

[BrownIsound]

@BrownIsound, @Esquirefreak, I don't really get your point, can you be more specific please? Like rewiring the whole circuit?! With no fact or proof that it will fix anything? I'm sure you don't mean that and my bad english doesn't help me to understand.

Well, with the evolution of the silverface circuit changes, the lead dress had changed as well, with a lot of added lead length and different routing of the circuit. Changing the values of a sf fender to a bf fender by changing the component values by the schematic does not magically turn that 1976sf into the tone of a 1966bf because of the lead dress differences. To do it right you need to rewire a lot of the leads to match a bf, and even then you’ll never be 100% there.

It’s your amp and you can do what you want to do with it, but my prior comment was that I prefer to just leave sf as a sf and not try to make it something else, one of the reasons being the lead dress issue. If I want to play with more gain or crunch, there’s always pedals, or just choose a different amp.

[Lambertini]

I'm not looking for any crunch at all. What made you think about that?

I'm an archtop player, looking for a great jazz tone out of this Twin. As I've already invested few pennies on this one and solved a tremendous amount of problems such an old and poorly cared amp can have, and as I can't afford an original 64 BF Twin Reverb, I'm tuning it to my taste. I'm almost there. I'd just like to tweak a bit the bass response to something tighter, and if possible, having a bit more chime in the treble (not a mater of EQ I think).
To give you an idea of the fine tuning I'm after, I'm still not sure if I prefer a 820/47 or a 820/100 NFB divider...

I've understood your point regarding lead dress. I've done a partial rework on this as some wires were old and messy, especially the heaters who were causing a crazy amount of hum. For the rest, wires look more or less correct.
I can of course rewire V6 if you truly think it might fix my issue but I'm skeptical the oscillation could come from that.

[sluckey]

I'd just like to tweak a bit the bass response to something tighter, and if possible, having a bit more chime in the treble
To give you an idea of the fine tuning I'm after, I'm still not sure if I prefer a 820/47 or a 820/100 NFB divider...

Replacing those two .1µF coupling caps between the PI plates and 6L6 grids is a popular mod to tighten up the bass (less boomy) and increase the treble. Most people use .047µF. Give this a try. Just might be what you are looking for.

As for NFB resistors, Fender used 100Ω for a 4Ω speaker load and 47Ω for an 8Ω speaker load. But you can experiment in this area too. Another popular mod is to include a switch that can select between 820Ω and some higher value (like 1500Ω). Still others will use a 10K pot.

Most people don't have a problem with not enough treble. My TR is plenty bright. I have to cut back on the treble quite a bit. Speakers make a world of difference with treble and bass on a TR.

[Lambertini]

Thank you @sluckey, I'm going to try these .047 caps.

On the NFB, I'm actually looking for doing this with a 3-way switch (mod from Rob Robinette), resulting in 3 selectable NFB at 530, 820 and 1K5.
https://robrobinette.com/Silverface_Amp_Mods.htm#3-Way_Negative_Feedback
It could be replacing the disconnected master volume if the wires from the NFB resistor eyelets to this place don't catch too much noise on the way. Will put some shielded ones.

For the treble, you're right regarding the speakers. But I'm not exactly lacking treble. Actually I have plenty of treble. I'm missing "air" or "chime".
If I try to guess frequencies, I'd say I'd like more 4.5K ~ 5k+ (maybe even higher), without adding more 3, 4K. I could get a Source Audio EQ2 for this, but I'd like to achieve that the analog way, permanently in the amp.
I get the air if I increase the treble knob, even better air when I switch the bright on, but then I also have much more 2, 3 and 4K, which become too much...

[BrownIsound]

Since you are playing with the NFB values and are looking for a treble boost, another thing to consider is adding a cap from the FB insertion point at the pi to ground (ala tweed bassman/marshall). As to what value to use, the smaller the cap, the higher the treble frequency will be focused.

[Lambertini]

You mean here? (see pic)
If so there's already a cap there. I'm not clear with the cap value: as it's out of phase, a smaller cap will remove more treble than bass no?

[BrownIsound]

No, it would bypass the 100ohm resistor.
A 59bassman/marahall utilizes a pot to make it a variable presence control at this spot.

[Lambertini]

Ok, I think get it. The cap let bleed NFB high treble to ground, so they don't cancel those of the main signal in the PI. Right? Easy to do, will give a try. Thank you.

[Lambertini]

Is that site reliable mathwise?
https://www.omnicalculator.com/physics/cutoff-frequency

Is that sim of NFB tail bypass cap relevant for a light high treble / air boost? (see pic)

[BrownIsound]

Someone smarter than and not suffering from Covid at the moment as I am can probably answer that.

One warning with doing this is that removing the HF from the feedback could potentially make it more susceptible to oscillation.

[Lambertini]

Sorry to hear you're sick. I wish you a quick recovery.
I think you're right regarding oscillation. This should probably be coupled with another filter letting pass anything beyond 12k to the PI.
How is still a mystery but I'll do some research.

[Lambertini]

Meanwhile, I found a workaround while waiting for ordered parts...