Post PI buffers........

[Steve Conner]

Reducing the NFB might stop the symptoms, but it's not necessarily the cure. A proper cure would be to get rid of the extra RC worth of phase shift that you introduced, then the amp will be stable with the original NFB network.

As Merlin said, you can get rid of either C3/C4 or C5/C6 in your original schematic by appropriate rejigging, and maybe adding an extra negative rail.

It's very important to use as few AC coupled stages as possible inside the NFB loop. Look at the Williamson amp for an example, it has about as many as can be tolerated.

[Bernardduur]

Reducing the NFB might stop the symptoms, but it's not necessarily the cure. A proper cure would be to get rid of the extra RC worth of phase shift that you introduced, then the amp will be stable with the original NFB network.

As Merlin said, you can get rid of either C3/C4 or C5/C6 in your original schematic by appropriate rejigging, and maybe adding an extra negative rail.

It's very important to use as few AC coupled stages as possible inside the NFB loop. Look at the Williamson amp for an example, it has about as many as can be tolerated.

Even now it works perfectly?

[Steve Conner]

Yep, unless you can prove to yourself that it's unconditionally stable and doesn't have a big gain peak at some subsonic frequency.

[Jazz P Bass]

NFB

I agree Steve. Not that you took the words out of my mouth. Far from it.
AC coupling in a NFB is tricky.
What I stated holds true , to agree, on a working amp.
The fact the "popping " got worse at a lower resistance tells me the amp was not functionong properly.

[Bernardduur]

I agree Steve. Not that you took the words out of my mouth. Far from it.
AC coupling in a NFB is tricky.
What I stated holds true , to agree, on a working amp.
The fact the "popping " got worse at a lower resistance tells me the amp was not functionong properly.

OK; great to know. So removing the whole buffer would be key here. Would this be negative on the amp as I will be driving 8 powertubes? (as I never work on amps with more then 4 powertubes )

[Steve Conner]

No, if you really want to have buffers, you can change to DC coupled CFs. Then there will be no more AC coupled stages in the loop than there were before.

I'm not sure the buffers are needed anyway. If the PI was already able to drive the power tube grids all the way up to 0V, then it's hard to improve on that. Unless you install some hefty buffers DC-coupled to the power tube grids, and try to drive them into AB2. But in AB1 the grids need next to no power to drive them, so I don't know why some amps used these CFs that are limited, by AC coupling or large stopper resistors ala SVT, to work in AB1. Maybe it was to overcome stray capacitance in the wiring to so many tubes, that would destabilize the amp at HF. Or maybe Merlin knows some other reason.

The RDH4 has a whole math-heavy chapter on how to stabilize a power amp with almost any number of RC-coupled sections inside the feedback loop, but my personal approach has always been to minimize the number of sections to start with, and check amps that I work on for any signs of LF instability or peaking.

[Bernardduur]

No, if you really want to have buffers, you can change to DC coupled CFs. Then there will be no more AC coupled stages in the loop than there were before.

I'm not sure the buffers are needed anyway. If the PI was already able to drive the power tube grids all the way up to 0V, then it's hard to improve on that. Unless you install some hefty buffers DC-coupled to the power tube grids, and try to drive them into AB2. But in AB1 the grids need next to no power to drive them, so I don't know why some amps used these CFs that are limited, by AC coupling or large stopper resistors ala SVT, to work in AB1. Maybe it was to overcome stray capacitance in the wiring to so many tubes, that would destabilize the amp at HF. Or maybe Merlin knows some other reason.

The RDH4 has a whole math-heavy chapter on how to stabilize a power amp with almost any number of RC-coupled sections inside the feedback loop, but my personal approach has always been to minimize the number of sections to start with, and check amps that I work on for any signs of LF instability or peaking.

Great! Thanks all for the insight, I've learned a lot! I'll try the amp without the CF. I already tried the DC coupled CF's but they gave me too much compression

Think I'll return to my ECC85 in the PI stage!

[jbrew73]

before you rebuild the driver section you might try putting the feedback network back to your original specs and reversing the primary wires and see what happens. you may just have positive feedback.

[loudthud]

Do you just have one 220K grid leak resistor for all the output tubes? I'm not sure what the effect is but most amps you see with 6 or 8 output tubes (like the Fender you copied) use much lower resistance like 33K. Steve, what does RDH4 say about that? That's why I like the direct coupled cathode followers with a 33K cathode resistor and you can use a much higher grid leak on the input to the cathode follower and a reasonable coupling cap.

[imaradiostar]

If you haven't the room for cathode followers try a mosfet source follower- I've been playing with them recently and I really like the results. the LND150 is cheap and you can use it for all kinds of helpful things in a tube amp without effecting the "tube" tone of the amp.

jamie

[Bernardduur]

Do you just have one 220K grid leak resistor for all the output tubes? I'm not sure what the effect is but most amps you see with 6 or 8 output tubes (like the Fender you copied) use much lower resistance like 33K. Steve, what does RDH4 say about that? That's why I like the direct coupled cathode followers with a 33K cathode resistor and you can use a much higher grid leak on the input to the cathode follower and a reasonable coupling cap.

OK, good point. Dunno...... I used 2 220k's (for both pairs).......

I checked the positive feedback thingie and no, I was not getting any positive feedback.

[Steve Conner]

Do you just have one 220K grid leak resistor for all the output tubes? I'm not sure what the effect is but most amps you see with 6 or 8 output tubes (like the Fender you copied) use much lower resistance like 33K. Steve, what does RDH4 say about that? That's why I like the direct coupled cathode followers with a 33K cathode resistor and you can use a much higher grid leak on the input to the cathode follower and a reasonable coupling cap.

Good point Loudthud. That's probably the reason. The spec for maximum grid leak resistance in the tube datasheet is per tube. So if you have a push-pull amp with 4 tubes per side, then you should use one-quarter of the resistance. The buffers are then needed to drive this low resistance without signal loss.

With just one 220k resistor per side, you're flirting with thermal runaway. I think EL34s were specced for a maximum resistance of 100k in the first place.

[Enzo]

Jazz, the last thing we need is an Enzo emoticon, especially for Steve and Merlin who know an awful lot more about this stuff than I do.

[Jazz P Bass]

Apology

My sincerest apologies to Enzo, Steve & Merlin.
As I posted it tongue in cheek, I did not mean to offend.

[Bernardduur]

Good point Loudthud. That's probably the reason. The spec for maximum grid leak resistance in the tube datasheet is per tube. So if you have a push-pull amp with 4 tubes per side, then you should use one-quarter of the resistance. The buffers are then needed to drive this low resistance without signal loss.

With just one 220k resistor per side, you're flirting with thermal runaway. I think EL34s were specced for a maximum resistance of 100k in the first place.

Wow, that's why I love it here! I'll try!