• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Cathode feedback

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I'd like to use balanced cathode feedback in a push pull output stage like Audio Reaserch did in the d40:


My question is: I've got an Hammond 1608, 8ohm speakers and el84 power stage. Can I use the secondary to this purpose? There are two windings in the secondary, 0-4ohm and 0-4-8 ohm. Look here:


What's your thought guys?
Thankyou for your support ;)

Hi Marco,
I've read now Your email (my server is down).

To use this topology the winding of the transformer must be balanced and this strongly depends on the manufactoring.

Electrically, simply tie the 4 ohm (Gn/Yell) wire to ground and connect the 8 ohm speaker to 0 and 8 ohm wires is the only possibility I see.
The other 4 ohm winding can be paralleled experimentally to one of the previus or left floating or connected to some compensation reactance, to find out the best balance between the two halves of the amplifier.

Joined 2003
Yes, you can do it, but you need to connect the transformer in a slightly non-optimum way because of the way that Hammond implements 8 Ohms.

Connect the two windings in series as if for 16 Ohms as per Hammond's diagram. so that BLK/YEL goes to GRN. This is your centre tap. Connect BLK to one cathode and GRN/YEL to the other. If you have a shared cathode bias resistor, it goes between centre tap and ground. If you have individual cathode bias resistors, the centre tap goes to ground and the resistors are between the cathodes and transformer. Connect your 8 Ohm loudspeaker between the centre tap and YEL.

Check for instability with a 10kHz square wave on an oscilloscope.
Thankyou guys.

Ec8010, that's the way I've thought to connect the secondary. The question is, which are the problems or performance losses that I'd probably meet in that way? And the differences in power handling or frequency responce? I'm thinking to connect the speaker to the 16ohm taps, I'm only afraid because the reflected impedance should be too low for the 84s.
Or, due to the fact that 8ohm is only a nominal impedance, to connect the speaker to the 4ohm tamp... I think that only a test with my equipment could answer to these questions.

Piergiorgio...what's the optimum load for the 6005?

Joined 2003
If you connect an 8 Ohm load across the 16 Ohm winding you will halve the load the output valves see. The cathode feedback will make them a little more tolerant of that load but you will still lose power and gain distortion. Unless I know exactly what the impedance of a loudspeaker is (having measured and plotted a graph of Z against f) I normally consider all loudspeakers to be 4 Ohm. The loss in power is insignificant. Using the transformer as I have suggested is likely to increase leakage inductance (affects HF response) and increase DC resistance (slight loss of power and reduction of loudspeaker damping). It doesn't cost to experiment, so try it and see what you think.


Joined 2003
Paid Member
> the drive voltage requirement for the output stage will roughly double.

In this case: Not even. The secondary voltage is about 15V peak, the 6L6 are biased about -30V. That's not much increase in necessary drive voltage, nor is it much feedback.

To get a big increase in drive need (and lots of local feedback) you need hi-Z cathode winding or tube(s) with much higher power sensitivity than a couple 6L6. Massively parallel works great with 4/16Ω winding.
6005W is like 6V6, apart slighty less ratings on voltage and power. Any load suitable for 6V6 will fit also 6005.

The reccommended Raa is in the range 6600-10000 ohm, most application user near 8000ohm anode-anode.

Optimum for power or for distortion? I think that with some cathode feedback You can use profitably a lower load than the usual tabulated one, getting more power at the same distortion.

Pay attention not to exceed the max rating, especially on the screen current.
Thankyou guys.

Sy maybe the drive voltage requirement will be double but, if I don't use the local feedback, then I will need some global feedback...and the driver stage should be in any way with higher gain. So I'm planning to use a driver stage that could provide enough gain to allow the use of some kind of feedback. I'll split the signal with an input transformer ( I'll use the Lundhal 1544A), a differential driver stage dc coupled to the power stage. I think that an ecc88 should be enough to drive the 84s with a little amount of cathode feedback.
I'll do the calculation using these formulas:

But...how to use them? I'll use both ultralinear connection and cathode feedback. So, to find the gain, have I to use first the formula for the ultraliner then the one for cathode feedback? Have you other formulas? I need to learn to use microcap...

Yes, the feedback is suboptimal but maybe for an ultralinear power stage it could be enough to improve the damping factor, that is my first care with UL output.
However with the symmetrical connection I should be able to add some parallel feedback in the differential driver stage, if the distortion will be too high. I'm only worried about the frequency response of the OT with that unconventional secondary connection.

EC8010's warning about oscillation should be regarded. But if everything looks stable, you'll probably find only nominal source Z and distortion reduction from this connection. Balancing for parallel feedback strikes me as a much better motivation in this case.

Where the cathode feedback really shines is with circuits like AR's later amps using 6550. You can actually get 5dB or so of feedback with a single pair of tubes, which results in a noticeable improvement in source Z and distortion.
SY said:
Balancing for parallel feedback strikes me as a much better motivation in this case.


SY said:

Where the cathode feedback really shines is with circuits like AR's later amps using 6550. You can actually get 5dB or so of feedback with a single pair of tubes, which results in a noticeable improvement in source Z and distortion.

Mhmm...and this is because the load impedance for the 6550 is lower than in this case. The lower the primary impedance the bigger is the feedback, of course. So this seems a good solution when the load is low. I think could be really good for a 6c33se, no?

It's really more of a power sensitivity issue. If the output is swinging 20V each way and the output tubes need 20V of drive, you can get 6dB of feedback. Where this really worked well was in the massively-paralleled 6550 designs.

I've not actually used a 6C33. I'll have to look at the spec sheets.
Ahhh...Sy thankyou for your teaching.
First I'm thinking to the coupling ratio.

I'd like to try the plitron Ot but they cost too much. The only solution seems to find a good winder and ask for an OT with 10/20% tap for cathode feedback.
I see that Lundhal has already made a new verion of the 1620 with cathode feedback.

PS: the idea above is not for the 84s...

Comments/Opinion Please

I wanted to do this - cathode feedback from the secondary plus Ultralinear (What Menno VdV calls "Super Triode") using a Hammond 1650T and 2 parallel pairs of KT88 and connecting 4 Ohm speakers.

When I asked Hammond about this they said it can't be done because the 2 off 4 Ohm secondary connections MUST be wired in parallel to get full power bandwidth - a wiring interleave issue.

I had not considered this - my query to Hammond was regarding current/power rating of the windings - that is getting the full 120W from just one of the secondary windings. Not that I'm ever likely to be pushing 120W anyway, I don't think insurance policy window glass replacement would cover this. I can just see the insurance assessors expression when I say "well I was just listening to the HiFi and the windows blew out ...".

In the mean time I'm running the amps in Ultralinear ONLY with the secondaries in parallel and 6dB of global feedback.

Anyone tried using Hammonds with the 4 Ohm secondaries in series for cathode feedback? - or would like to venture an educated opinion/ guess?



Joined 2003
Paid Member
> more of a power sensitivity issue.

Exactly so, assuming that load tappings are fixed (such as 4-8-16 ohms).

> to find the gain

For a good approximation: find the grounded-cathode gain (whatever you do with the Plates and Screens), find the peak voltage on the cathode winding, and add.

What do we have? I think I misunderstood the first post and was thinking 6L6. Seems to really be EL84 here. In UL mode, book says 11 watts out, 11V peak per grid. 11 watts in 4Ω is 9.3V peak. So the peak grid swing with cathode on a 4Ω winding is roughly 11V+9.3V= 20.3V peak. So it does almost double and you do get ~5dB feedback just from the cathode tap. Damping factor may be slipping above 2 (UL is about DF=1, and the cathode feedback almost doubles that).

For "real" effect, you custom-wind a higher-ratio winding and increase the drive accordingly. Taken to Mcintosh extremes, you have to account for the power being delivered into the cathode winding. That's half a Mac's power but only like 9/250 or 4% in this case, negligible.

You know, you get tricky with UL-taps and cathode feedback, you are just going in the direction of a Triode. But not going All The Way. Why not? The 6BQ5 in P-P Triode will do 5 watts, not a lot less (to the ear) than 11W-15W in UL mode. And it is all done honestly with electrostatic fields, not by looping around a soft iron core and coming back again.

> Hammond ...said it can't be done... secondary connections MUST be wired in parallel to get full power bandwidth - a wiring interleave issue.

Uh, now we are on quad KT88?

Cathode feedback is tricky enough already. The cathode is a VERY broad-band output. 10MHz oscillation is possible. And push-pull amp windings must be well-coupled side to side or they spike when one tube goes to cutoff. If Hammond is saying it won't work, I'd wade into it very cautiously, with a very wideband 'scope at my side, always asking if it is worth the trouble. Transformers are tricky enough even when the windings are well-coupled.

And from an old-timer's point of view: Hammonds were always workhorses, and I respect them for doing what they say and for keeping the old tooling working. But Hammond and Stancor were not generally considered "high fidelity" even at the Eico/Fisher level. Yes, many "good brand" hi-fi amps used Hammond-made transformers but made to custom specifications. The Hammond brand was mostly radio and replacement iron; UTC, Peerless and Dyna sold the "better" iron.

> current/power rating of the windings

You are not going to melt-down, certainly not on unclipped audio, probably not on all-day sine-tone.

Your losses may increase from ~10% to ~15% or more. You lose ~5Watts, a non-issue on quad KT88.

If their story is correct, you lose top-end too. It could be as poor as 5KHz, a typical value for a non-interleaved winding. What may really happen is you get 30KHz on the push side, 5KHz on the pull side, leading to large 2nd harmonic at the top of the audio band. And possibly very nasty action at the cathodes because they don't work true push-pull, but tend to flop separately.
Ok, I'll looks for other solutions. I think that cathode feedback is really interesting issue but to make it works I suspect that I'd need a custom OT.

Ok, I'll look for other solutions. I think that cathode feedback is really interesting issue but, to make it works, I think that a custom OT is needed .

PRR thankyou for your considerations. Yes, what I'd like to obtain is a triode like output stage but with more power. However if I'll be able to obtain a better damping factor I'll be even happier ;)

I said that I'll use an input transformer as phase splitter. In this way to apply global feedback is not so easy. And I'd like to use only local and short loops. What do you think about plate to grid feedback, in the output stage? The internal resistance of the tubes will come down. The drawback is the input impedance that will go down too and the gain of the power stage. I'm thinking to use a differential mufollower driver stage, with a ccs on the cathode. Maybe the mu follower could be an hybrid one using the Gary Pimm's ccs. In this way I'll be able to obtain a full mu from the driver tubes and a really low output impedance.

I'm crazy :joker:?
I've used the Plitron specialist range toroidal output transformers (4070CFB) with success. I've been very satisfied with my push-pull KT88s in triode mode with the 4070CFB. (Note: I use a 6c45pi in SE to drive a Lundahl interstage transformer to phase split and drive the KT88s in push-pull.)


If you want to run SE with CFB, there is a great article at:

Hi Kashmire,

I've read Byrith's article.

Regarding Plitron transformers I have to say that I've already the trafo, hammond 1608, and I want to use them.
Maybe the only way to apply negative cathode feedback is using split load, as in QuadII.

However I'm reading about partial feedback. I'd like to have only two stages in the amp: input trafo to split the signal, voltage amplifier and output tubes.
The way to apply feedback are:
1) plate to grid in the output stage
2) plate to plate over two stages ( this is a different version of plate to grid I think)
3) plate to cathode of previous stage. I've found a full explanation in Crowhurst's "Designing Your Own Amplifier" articles.
4) Plate to grid of the previous stage ( crossing the loops)

Now, which is the best solution? Pros and cons?
Resources for the involved math?

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.