Where to add negative feedback? Schematic

Schematic attached.

Experimenting with negative feedback, but the stepped attenuator location causes issues.  Any thoughts on the best location to insert NF, global or local?

The cathode bypass cap option has been discussed. Any other thoughts?

WHY do you want "Global negative feedback"?

Is it a Marketing Requirement? Or is gain too high? THD too high?

The "obvious" path is to discard that interstage attenuator and bring adjustable NFB from output to V1a cathode. This will require a custom attenuator and some different biasing.

If THD is bad, or output overload is bad, fix R28!! Tack a 1K across it and see if it sounds better.

If you have enough gain.
You could move the atten switch to be in front of the wcf.
Tie a feedback network around the two first tubes.  Sec plate to first cathode.
And remove the bypass cap of the first tube.

Theoretically global feedback would be from the output through a cap and resistor to the cathode of the first tube but you would have to get rid of the bypass cap on the first tube. Note that any phase shift (such as because a bypass cap isn't large enough) could cause instability.

However, the biggest problem would be that the amp would be kinda boring. I think it is more common to divide the amp into two sections - the preamp that has lots of gain right on the input which is good for noise. Then the output stage would use feedback to tighten the response. For that, nfb would be from the output to the base of the second triode. But you would probably need a capacitor somewhere.

Then again, I'm not a tube guy.

I am not sure you are asking the right question. The topology as it stands is similar to classic radio consoles of the 1940s and earlier in using fixed gain stages with attenuators between them to set overall gain and none of the stages uses NFB. The advantage of this topology is simplicity and freedom from stability problems. Is major disadvantage is headroom and distortion depend primarily on input level. This is identical to the topology I used in my earliest mixer designs and that is now embodied in the 'Classic' versions of current designs.

There is no obvious way to add NFB to this topology and address the major disadvantages, especially if you want to be able to vary the overall gain over a wide range. simply becasue it has insufficient overall open loop gain.  I went though just this process quite some years ago and I concluded that the answer was to change the topology. The result was the EZTubeMixer design. It is by no means the only solution but I don't think just adding NFB is it.

Cheers

Ian

Thanks, all.  I suspect I’ve pulled about as much out of this design as possible.

Your design is almost identical to the amp on an LA-2A, so if you copy that it will solve your problem.  The pot will have to go before it as others have said.

DaveP

DaveP said:

Your design is almost identical to the amp on an LA-2A, so if you copy that it will solve your problem.  The pot will have to go before it as others have said.

DaveP

Click to expand...

Except the La2a is line-level.

BluegrassDan said:

Except the La2a is line-level.

Click to expand...

???????

The La2a potentiometer is before the first gain stage. I’d be concerned that attenuating a mic signal before the first gain stage would cause noise problems.

BluegrassDan said:

The La2a potentiometer is before the first gain stage. I’d be concerned that attenuating a mic signal before the first gain stage would cause noise problems.

Click to expand...

It is a trade off between noise and headroom. You current schematic with the transformer connected straight to the first stage tube and a potentiometer after it should have a better noise performance but this in paired with an equal worsening of headroom at lower gains. Because the gain is fixed and it has a fixed maximum output level, as the input increases the headroom decreases. Arguably a better solution is to have two pots, one before and one after the first stage. Every reduction in gain then both reduces output noise AND limits the loss of headroom. This is effectively what happens when you use the input pad.

Cheers

Ian