Single ended output through push pull transformer

[Stadler Guitars]

Can a push pull output transformer be used on one side in a straight class A single ended mode? I have a great old P.A. chassis that I want to play with by comparing a single ended circuit against a mirror image with both tied together in a differential pairing. I would like to minimize part acquisition and swapping. Are there any considerations I should make?

[mooreamps]

Can a push pull output transformer be used on one side in a straight class A single ended mode?

Yes.

Are there any considerations I should make?

You will still need two power tubes to balance the primary winding.

-g

[R.G.]

Can a push pull output transformer be used on one side in a straight class A single ended mode?

Maybe. The devil is in the details, and how strictly you define success.

Are there any considerations I should make?

Class AB push-pull is a special case transformer, as are Class A push-pull transformers. Class A single ended is yet different again. Knowing what they were designed for tells you some of what to look out for.

Class A single ended has DC current flowing through a non-push-pull primary all the time. The idle current through the primary limits the current swing in one direction (i.e. it can only decrease to zero, can't reverse) and usually the core and windings are designed so that 2x the idle is near saturation flux density at the lowest frequency of the design. This means that there is a unidirectional magnetic field in the core all the time, and so the peak to peak magnetic field in the core is limited to zero flux to saturation in one direction. Half of the core's magnetic capabilities can't be used, and is wasted. This makes single ended cores way bigger for the same power than push-pull.

Class A push-pull transformers sit at a net zero magnetic flux because each output tube pulls an idle current half of its maximum with no signal. The windings are arranged so that each tube pulls down on its half-primary, but the winding directions are opposite, so the magnetic flux from the two currents are equal and opposite - their magnetic effects cancel in the core. So the core sits at magnetic zero (mostly) and can swing to + saturation flux and - saturation flux; the core can handle twice the energy transfer, and so it's smaller than single ended for the same power output. The tubes limit power output here because each tube is biased to half its maximum current, and can swing up to maximum and down to zero while its counterpart does the opposite.

Class AB push-pull is the next step. The tube idle currents are dramatically smaller than Class A, so the idle currents still cancel in the transformer primary and magnetics, but each tube runs its half-primary while the other output tube is off for much or most of the signal swing. Current (conceptually) flows alternately in each half-primary. This lets each tube swing from nearly zero to max current, and the magnetics swing from + saturation to -saturation. So the tubes and transformer are both used to their fullest extent, and you get more power out of a pair of tubes and more out of the transformer.

Both Class A versions have a DC current flowing in their primaries all the time. The current is also large. Class AB has only the current of half the signal flowing at any one time, and the heating effect of the effectively half-wave rectified currents in each half primary heat them less than the heavy current Class A use. So manufacturers can and do make Class AB push pull transformers smaller for the same output power than either version of Class A.

Now we're getting down to the details. Using a Class AB transformer in Class A means (1) you have to limit the currents in the primary to avoid overheating the primary wires. It can still only do what heating it did in the Class AB it was designed for, so its output power when used in Class A is much, much smaller than it was in AB. You do this by cutting down the B+ and changing the secondary loading to keep the tube currents in bounds. There's math to tell you how much if you want to hear the math. Assume going to Class A push pull would mean a power output drop of maybe a factor of four to six to keep from burning the primary windings.

Going to single ended is another dramatic drop in power because the core will readily saturate at a low DC current. Commercial single ended Class A transformers are usually gapped to help prevent saturation, and this is in effect giving up bass response to keep from saturating. Going to Class A single ended can be done, but the power you can usefully get through it drops again, by a lot.

There is a way to cheat. You can run effectively Class A single ended if you use an output tube on half the primary and any means whatsoever to pull and equal idle current through the other half-primary. This idle side just pulls the transformer's flux back away from saturation and prevents the necessity of gapping. In effect, the idle side pulls the tranformer's magnetic flux back to zero to let you not have to gap the core.

This can be two tubes if you bias both output tubes to Class A idle currents but provide signal to only one grid. That's a horrible waste of a tube on the "idle" side. It makes more sense to use a power transistor or power MOSFET to pull a constant current through the "idle side". I think there was a commercial amp that did either the idle tube or idle current source. Can't remember the name now.

Or you can just whang it into single ended class A. You need to tinker with this a lot to get it to have a somewhat symmetrical output signal, and the output tube DC idle current will be very much limited by the saturation of the ungapped core. So clean power output will be very limited compared to the power ability of the same OT in Class AB push-pull.

[Stadler Guitars]

R.G.

Thank you so much. The mention of running a dummy current on the other side of the primary is what I was looking for, and will allow easy switching between the single ended mode and the diff mode. I plan on running the differential mode with the power section in class A push-pull to keep comparisons as equal as possible. I haven't tested the tranny yet, but it is an old Bendix running two 6L6s in cathode biased AB. I is very large for that application, and I bet it could easily handle 80 watts run that way. I'll find a tube type that will steer clear of too much current and work well with the ratio and loading. I'm keeping my fingers crossed that a 6550 will hang in there, but if it has to be 6V6s (or whatever), so be it.

And as anyone who knows me can tell you, i can whang it in there with the best of 'em.

[Ronsonic]

I am not going to disagree with RG, but will suggest alternate rather less sophisticated strategies.

While he is correct that the transformer is not designed for unbalanced operation and will be less than optimal, we are talking about using it in an application that severely derates it. Most transformers are not stacked that tightly and an oversized PP tranny will often work in a lighter loading SE application. We are talking about maybe 15W max from a 6550. So that's one option, just do it. Yeah, you'll probably get a crossover notch at any but very low power, but that isn't necessarily a bad thing. If the tranny is as overbuilt as you suggest this may work out well.

You also have the option of busting it apart and restacking the lams so that all the E's are on one side and all the I's on the other except for the ends. Alternate those to hold it together while you secure it. This introduces an air gap that reduces the saturation of the core from DC. This will allow you the full end to end winding of the core and you may want the higher primary impedance for a class A amp. We could do a lot of propeller-headed calculations here, or just go with it. Don't try to introduce any additional gap with spacers (though you can experiment with that if you feel the need) usually just restacking and reassembling as tightly and hand work permits will do the job well enough.

[R.G.]

I am not going to disagree with RG, but will suggest alternate rather less sophisticated strategies.

]
I have to admit that I'm very much more at home with electronic solutions than mechanical ones! I have all these thumbs...

While he is correct that the transformer is not designed for unbalanced operation and will be less than optimal, we are talking about using it in an application that severely derates it. Most transformers are not stacked that tightly and an oversized PP tranny will often work in a lighter loading SE application. We are talking about maybe 15W max from a 6550. So that's one option, just do it. Yeah, you'll probably get a crossover notch at any but very low power, but that isn't necessarily a bad thing. If the tranny is as overbuilt as you suggest this may work out well.

Yep. That's the "just whang it into SE class A" option. One note - Class A has no crossover distortion in any form. Class A never crosses over. SE class A never even "crosses over" from one side of the BH curve to the other, even giving that magnetic crossover does not generate crossover distortion either.

You also have the option of busting it apart and restacking the lams so that all the E's are on one side and all the I's on the other except for the ends. Alternate those to hold it together while you secure it. This introduces an air gap that reduces the saturation of the core from DC. This will allow you the full end to end winding of the core and you may want the higher primary impedance for a class A amp. We could do a lot of propeller-headed calculations here, or just go with it. Don't try to introduce any additional gap with spacers (though you can experiment with that if you feel the need) usually just restacking and reassembling as tightly and hand work permits will do the job well enough.

I'm proud of my propellors...

But the point is well taken. If the transformer you're working with is big enough, it won't saturate at some value of SE Class A.

However, that "big enough" and "some value" disparity can be dramatic. SE Class A transformers designed from the start to be that typically put out 1/8th to 1/32nd of the power that the same size iron can do in P-P A-B. But for tinkering, it may be fine. Restacking it to have a gap will help too. The gap doesn't have to be large, as you note. Transformer iron typically has ab out 8000 to 12000 times the permeability of air, so introducing an air gap or 1/10,000 the magnetic length of the core halves the primary permeability and primary inductance for the same windings. Using both sides of the primary in series quadruples the primary inductance (inductance scales by the square of the number of turns) but introducing gaps the size of a couple of thicknesses of paper cuts it in half to in fourth, too.

But that may be enough. This is, after all, a tinkering exercise. When tinkering, a good rule is to do the least work first. I refer to this principle as the Law of Universal Parsimony.

I can suggest some propeller-y things to do to figure out what you have and where you'll get with a transformer, but where's the fun in that?

[Stadler Guitars]

Tinkering. The operative word in this thread. I would stack the transformer with an air gap if I knew that single ended was the ticket, but before I do that I'm going to run one side of the primary solo, and when I hear saturation I'll run the dummy tube against it.

It'll be at least a couple of days before I can rip everything out and pencil a schematic. You guys are making me feel motivated.

The primary Z is 9,000k plate to plate. The secondary has 4, 8, 100, and 500 ohm taps. This is according to the stencil on the case.

[R.G.]

Tinkering. The operative word in this thread. I would stack the transformer with an air gap if I knew that single ended was the ticket, but before I do that I'm going to run one side of the primary solo, and when I hear saturation I'll run the dummy tube against it.

You go, guy. Remember, a monstrous mind is toy forever.

The primary Z is 9,000k plate to plate.

That's 2250 (1/4th) from plate to center. Watch the unused side - it wobbles around the CT in equal but opposite voltage to the driven side.

The secondary has 4, 8, 100, and 500 ohm taps. This is according to the stencil on the case.

Also remember that transformers don't have impedances - they have ratios. What that means is if you put a 4, 8, 100, or 500 ohm resistor on the appropriate tap, you get 9000 reflected to the whole primary, 2250 to the center-to-plate. If you put a 2x too big resistor on the output, the output tube also gets bumped up by 2:1.

Also remember that transformers are rated for X power at their lowest frequency, and that the power they can handle goes up linearly with frequency. What you don't have (probably) is the lowest frequency the amp/output trannie is specified for. That's one thing that sets the primary inductance, which does NOT change with loading. But if it was rated for, say 10W at 40Hz, if the lowest frequency you put into it is 80Hz, then the transformer alone can handle 20W because the frequency is 2X. This is why switching power supply transformers are so small - they handle their power at 100kHz and above.

But I'm getting ahead. Watch what you get out on the scope, and look for signs of distortion, then tinker the tube bias to get max undistorted voltage swing on the secondary at whatever load you put on it.

[Stadler Guitars]

I'm pulling this stuff out of the fog called my brain, but I was counting on tailoring the load to the tube/tranny. That's something I've been doing since hearing and understanding the reflected load (an old Traynor PA that would not hold down an EL34 or sound right). The primary is half and the reflected load is a square, hence 1/4th impedance. Thanks for another eye-opener, R.G., but I have to admit that transformers are still strange and mysterious to me.

[Chuck H]

Just for the sake of argument... What's wrong with Gary's idea? I assume by his post that his idea is something like:

Ground the grid on one tube.
Heat the bias.
Reduce drive voltage to the operating tube.

Would this do anything to correct the operation for class A? Balance currents, etc?

[Ronsonic]

Yep. That's the "just whang it into SE class A" option. One note - Class A has no crossover distortion in any form. Class A never crosses over. SE class A never even "crosses over" from one side of the BH curve to the other, even giving that magnetic crossover does not generate crossover distortion either.

True. What I was referring to is a step in the waveform in the crossover area. It will be there if you start pushing that tube and tranny combination. Just speaking from observation, I'll very happily leave the theory to you. In general, class of operation theory gets shot to hell when we start overdriving things as we do.

I can suggest some propeller-y things to do to figure out what you have and where you'll get with a transformer, but where's the fun in that?

ATMO, that's what these discussions are for. Approaching these things from different angles. Just because I recommend greasing it up and tapping it home doesn't mean I'm discounting the value of knowing what we're doing.

[R.G.]

Hmmm. Interesting. I'll have to go fire up one of the old single ended hulks I have and do some messing around.

My long-suffering spouse has had to endure me looking at *everything* through the lens of "hey, how does that work?" for so long that she knows what's going on in my head when I study the wiring conduits, lighting, structural arrangements, floors, everything, when we go shopping. I do have an affliction of needing to know how things work. I am an addict...

And you're right - we do things never intended for these amps.

Hmmm... I wonder what *that* does to them.

[mooreamps]

Can a push pull output transformer be used on one side in a straight class A single ended mode?

May I say just one thing.... Inside one of my guitar amplifier builds, it has a mode where only power tube is driven in Class A bias, and the other power tube is left to idle in Class A bias, for a total output power of only 6 watts.

-g

[Chuck H]

So, is it like I stated above or not? It seems like the obvious move in this circumstance but does it actually work as well as an OT specifically designed for single ended use, and why?

[mooreamps]

So, is it like I stated above or not? It seems like the obvious move in this circumstance but does it actually work as well as an OT specifically designed for single ended use, and why?

Ya, it works just as good. and the mode switch that toggles between 6 watts single ended to 36 watts push/pull is hot switchable using combination bias on the power tubes.

I'd like to buy a video camera, and put a video up on YouTube to show this. But the video camera I wanted costs 2000 dollars, and my wife just put a screeching halt to that idea.

-g