Combining two single-ended output transformers to make one “push-pull” transform

[passaloutre]

I have acquired an old stereo console amplifier parts donor that has two single-ended 6BQ5 power tubes, each with its own output transformer.

In the spirit or experimentation, I am wondering if I can wire the output transformer primaries in series, apply my B+ at the center point, and use it as a push-pull output transformer. Then the secondaries could be operated in series with a single speaker? Would it work? What would be the (dis)advantages compared to parallel-single-ended or true push-pull?

The way I see it, it has all the pieces of the system, except reduced magnetic coupling from not having all the windings around a single core. I’m not read up enough on my fundamentals to know exactly what effect that would have though…

I’m not asking if this would be ideal engineering practice, obviously I could just buy a proper PP output transformer, but I like using up these old hifi parts, and these types of experiments help me learn.

I’ve read a lot about not using PP transformers for SE, but I haven’t found much about the opposite.

[HotBluePlates]

I haven't tried it.

I would expect to Not-Get:
   - The hum-reduction due to ripple-cancellation in a real push-pull OT.
   - The high peak plate currents when a real push-pull OT presents a "Class-B loadline" (that some authors describe as "Gm-doubling of the output tubes").

Both of those effects lean on how a half-primary is affected by signals/magnetizing-current on the other half-primary, when sharing a core.

[passaloutre]

Thanks HBP, that makes sense.

Would there be any advantage to doing this instead of just having parallel single-ended tubes?

My thought is: in class A, there’s no efficiency gained by running push-pull instead of parallel-single-ended, but the push pull arrangement lets me squeeze a little more power out of two tubes by giving the option of class AB?

The “downside” of the push pull circuit (traditionally) is the need for a phase inverter, though I do often enjoy the sounds produced by not-quite-symmetric phase inverters.

Thanks for letting me think out loud

[HotBluePlates]

Would there be any advantage to doing this instead of just having parallel single-ended tubes?

Not that I know of.

My thought is: in class A, there’s no efficiency gained by running push-pull instead of parallel-single-ended, but the push pull arrangement lets me squeeze a little more power out of two tubes by giving the option of class AB?

When a tube is run single-ended (and not as you're thinking, with 2 output transformers), then the tube distorts when driven to:
   1)  Plate-current cut-off, or
   2)  0v grid-to-cathode (which results in grid-current).

To get the biggest output, the tube is typically biased about halfway between these 2 points (if plate current cuts off at Grid at -35v, idle grid-to-cathode voltage might be -15v).


When tubes are run push-pull, plate current cut-off on one side may not result in distortion if the other side is still conducting.  It depends on whether the tubes are being run in "Class A," "limiting Class A," "a little into Class AB" or "deep Class AB."

Read Radiotron Designer's Handbook 4th Edition for an explanation of "limiting Class A," which basically means one side is cutting off just as the other side reaches grid-current.  Overall, Class A means peak plate current on one side is 2x that tube's idle plate current.

Whether the output tubes are running Class A, "a little into Class AB" or "deep Class AB" is decided by the overall balanced of supply voltage (including screen voltage), OT primary impedance, and idle bias.

Cathode-bias push-pull tubes can run "Class A" or "a little into Class AB" depending on the factors noted above.  "Class AB" mostly means peak plate current is higher than "2x idle current" (sometimes very much higher); obtaining the high peak plate current requires higher supply voltages, lower OT primary impedance, and fixed-bias.

The “downside” of the push pull circuit (traditionally) is the need for a phase inverter, though I do often enjoy the sounds produced by not-quite-symmetric phase inverters.

100 years ago, creating a phase inverter might have been a challenging task.  Today, it's pretty trivial.

[passaloutre]

Thanks again HBP. Sounds like Class AB probably isn’t likely with this arrangement, leaving the only remaining argument for doing it as “just because” which is still a perfectly reasonable justification in my book.

I will report back if I try

[tubeswell]

My thought is: in class A, there’s no efficiency gained by running push-pull instead of parallel-single-ended, but the push pull arrangement lets me squeeze a little more power out of two tubes by giving the option of class AB?

In a push pull output stage, Class A is what you get anyway - when both sides are conducting at the same time throughout the signal cycle (i.e., when your PI output is small enough that neither output tube is driven into cutoff. Its just that you've got 'Class A' on both sides, meaning that there is current through both sides of the primary (albeit that one tube is pushing positive current clockwise through the primary and the other tube is pulling negative current anticlockwise through the primary).


Class B occurs when the signal swing output from the PI is big enough that one side of the output stage enters the cutoff region and so there is now only one 'on' tube pushing current through half of the primary. In this condition (because the Pr:Sec impedance is effectively 1/4 of the normal plate-to-plate impedance, the 'on' tube is pushing far more current than its designed for, generating a much higher peak signal voltage during that part of the cycle. Conversely, when the other side becomes the 'on' side (and the first side is in cutoff), the another equally as high peak occurs on the opposite part of the cycle. The tubes can handle this* because their power handling is an average power handling, and for part of each cycle, each output tube is not conducting - thence keeping dissipation within acceptable limits. But this trick gives much more output power during Class B - far more than running both tubes at the same time in Class A.


*i.e., provided that the output tubes are biased appropriately to enable each tube to maintain average dissipation throughout the AB cycle.

[passaloutre]

Thanks. Yes I was at least nominally aware of the differences between Classes A, AB, and B, but not as well read on how to achieve the different classes.

It would seem anything beyond Class A is unlikely with the transformers that came with this amp.

Let me ask this then about parallel single-ended: if I want to run two power tubes and two output transformers, do I link both primaries at both ends (i.e. one end of each to B+ and the other ends linked together to both plates) or just feed one plate from each primary? Hopefully that makes sense, I’m on my phone and can’t draw a diagram

Say one end of the primary is brown, one is blue. Do I tie both browns together and both blues together? Or both browns to B+ and one blue to each plate, but not together? Then I can put the secondaries in series or parallel to get the impedance that I want? Say they are 4k:8ohm

[tubeswell]

if I want to run two power tubes and two output transformers, do I link both primaries at both ends (i.e. one end of each to B+ and the other ends linked together to both plates) or just feed one plate from each primary? Hopefully that makes sense, I’m on my phone and can’t draw a diagram

If you’re running 2 lots of SE output stages, each with a separate output tube, OT and speaker, that’ll probably sound pretty good - bearing in mind that if you have a single preamp tube driving both output tubes, the impedance bridge will be ‘lossier’ all other things being equal. So you might want to increase the plate load on the preamp driver to compensate.


If you’re running 2 parallel tubes Center-bias Class A into one SE OT and driving its own matched speaker load, you’ll want a gapped OT that can nominally handle 2x the power (as well as paying due attention to the preamp driver impedance bridging).


If you’re thinking of running two sets of output tubes each into their own SE OT, and then into one speaker that is suitably impedance matched to both OTs, you might or might not like the result.


As to what colours the primary leads are, you might not be able to tell if the colours will ensure the OTs are in phase or not. Sometimes even the same model OTs from the same factory will get the colours and the phases mixed up.

[passaloutre]

Don’t get hung up on the colors, that’s not what I was trying to illustrate (poorly, without a diagram). Assume I can get the phasing right ( this is all hypothetical for now). I was really just asking if I should tie the plate ends on the transformers together, or treat them as two separate units.

Single preamp into parallel output stages into one speaker.

[kagliostro]

This only as to share a curiosity


In 2014 a friend at DIYItalia got a 1958 Corland (an  old italian brand) guitar amp, restoring it we were astonished by the fact it was an el84 PP amp with two SE OT instead of a single PP OT


at the time my friend was able to contact Mr Camillo Orlando, the founder of the Corland Brand (that was 84 year old), when my friend asked for the reason of the two SE OT instead of the usual PP OT Mr Orlando specified that on 1958 was difficult to provide components for the construction and they used what they had under hands, for this reason later they begun to build theyr own transformers

[tubeswell]

I understand passaloutre is planning to run two SE output stages as a combined SE overall, not as a combined PP.


(@passaloutre, in either case, knowing the Pr:Sec phasing of both OTs is important - as you don't want opposing signal phases on the secondaries cancelling the signal out, but that may end up being a matter of trial and error depending on how the actual phasing turns out from the factory - you can't necessarily trust the colours of the leads)

[passaloutre]

Thank you Kagliostro, glad to see one in the wild!

[kagliostro]

Ciao Tubeswell

I understand passaloutre is planning to run two SE output stages as a combined SE overall, not as a combined PP

Yes but at the begin of the thread he was asking if it was possibile to use two SE OT in place of a single PP OT

Is It possibile: yes

Is It good practice and advantageous: no

Franco