Characteristics of Guitar Amplifier Output Transformers, and criteria for design

[J M Fahey]

Yes, that´s the point.
That´s why I said winding a few will definitely improve your ... um .... "mechanical" skills.

As of designing, measuring, testing and comparing different designs, obviously it´s also possible, but it takes a lot more time and dedication.
And $$$$$.

[Tom Phillips]

I concur with Juan. You may wish to consider winding a transformer for practice and to find out first hand the skills needed and the hassles involved. I suggest the approach of re-winding an existing transformer so that you don’t have to procure all the parts. You could do a PT for the first project and you might even take the opportunity to make a special configuration that is not available off-the-shelf but which you need for a project. Donor transformers can be defective ones that you have removed during an amp repair or an old transformer that doesn’t currently have voltage ratings useful to you.

I wound a couple of PTs when I built my first amps and I just recently found the QST magazine article that I used as a reference. Attached is a copy of that article. It is very basic as it was written for the novice. The article along with a copy of the Radio Armature’s Handbook and the RCA receiving tube manual were my source of initial technical information. After a lot of scrounging on a high schooler’s budget and many interesting learning experience mistakes I ended up with a working 6G14A Showman head.

Tailor-Made Volts_96.1d.pdf

I also find that it is a good learning experience to take things apart. It's interesting to see how they were built even if you don't plan on repairing the thing.

[nosaj]

Tom you Rock. Thanks for pulling out these old articles for us.

nosaj

[R.G.]

That's the kind of stuff that hooked me on electronics (and transformers) back in the 1960s. Great article, Tom.

The bones of a rewind-your-driver-transformer article is up at geofex:
Thomas Vox Driver Transformer
It outlines some of the skills to rewind a small mains donor transformer into a replacement for the otherwise-unavailable driver transformers for the Thomas Vox amps. It includes some tips that make the process simpler, like dummy spacer winds at the ends of layers to preserve layer margins and edge stability, and some tips on impregnation. Of course, winding on a coil form with ends makes this much easier than winding self-supporting layers in a stack.

I had an article up on de/re-winding a semi-toroidal trannie to give many secondaries for isolated pedal supplies. I took that down because of the great lurking demon of everything electrical today: litigation. I became seriously worried about the possibility of being sued by the parent of some junior genius that could not be bothered to follow the safety warnings. That is one big difference between the USA today and the USA of fifty years ago.

[Mike Sulzer]

Of course it's not that simple. However, when you're trying to introduce complex, slippery subjects to someone new to the concepts, you start with the simplest views first, then add on modifications as they get the grosser concepts under control. I didn't get into the quantum mechanical view of how magnetism arises, either.

Yes, there is a truth there. Someone did the physics, and showed that the B-H curve, something that can be measured for a material without understanding the QM, is what you need to know to design a transformer. But no one said to would be easy, and it is not. Most people just pick out an output transformer for the job, and that is a good thing. If you figure out how much primary inductance you need and then pick out a transformer with that in mind you could be in big trouble. In a push pull design, at low signal, the inductance to support the magnetizing current is the low signal inductance, several times smaller than the high signal inductance, as Helmholtz most recently pointed out. The specs usually show high level inductance, but might not say that. Use the wrong inductance and you lose bass at low power.

A related issue is the myth that you have to have an air gap in a transformer for SE use, or it would be gigantic, yes GIGANTIC!. An air gap does two things: it postpones saturation to a higher current, but at the cost of lowering the effective permeability. These effects tend to cancel out, meaning you just have to wind more wire on to get the missing inductance back, BUT there is something very different about how an SE versus a PP transformer is used. The difference is where they operate on the B-H curve with no signal. Unlike the PP transformer, the SE transformer has a big dc current through it and thus is at a different location on the curve. Does that immediately tell you why you can lower the effective permeability with a gap? Not in any easy way, because this stuff is hard to understand. I doubt that most of the people who make transformers could explain exactly how an air gap in an SE transformer works, but they know that it does, and they know how to take advantage of that.

[Malcolm Irving]

If an air-cored output transformer was built it really would have to be huge, and would need to be kept a long way from the rest of the electronics (and anything else which is susceptible to magnetic fields). However, it would be perfectly linear – no hysteresis – no saturation – no funny business at low excitation levels!

I wonder if any hi-fi enthusiasts have done it?

An air-gapped core is a short air-core magnetically in series with the ferromagnetic core. A linear component in series with a non-linear component must improve the overall linearity.

EDIT: A push-pull air-cored output transformer would have the added advantage that you could check the DC balance of your output tubes using a compass.

[Chuck H]

A push-pull air-cored output transformer would have the added advantage that you could check the DC balance of your output tubes using a compass.

It would have to be a pretty fast compass! You could use it as a fan to improve cooling

[R.G.]

Interesting concepts.

One effect of air gapping that is fairly well buried in transformer design is energy storage. An SE transformer must store energy to release on the half-cycle where the output device is turning off. It has to store enough to complete a half cycle of the lowest frequency you want to come out undistorted. It being an output transformer, it has to store enough speaker energy to fill that half cycle, so this is a pretty stiff requirement. Looked at in this way, SE OTs are different in nature from PP in that they are energy store-and-release devices, not flow-through devices.

It turns out that energy storage in a transformer with an air gap is lopsided in favor of the energy being stored in the gap, not the iron. This is much like in a capacitor, where the energy is stored in the insulation, not in the plates. The iron is a "conductor" to conduct the m-field into the air gap. Much less energy is stored in the iron than the gap as you increase field strength by forcing current into the coils.

I have not done the math, but I suspect a non-air-gapped SE transformer would be much larger for the same output than an air gapped one because the air gap stores the permeability times more energy at the same flux density B. The iron helps you constrain the field intensity into a defined volume of space.

A purely air gapped SE transformer would be fighting another problem - leakage. One screaming advantage of combining iron and air (vacuum works, too ) is that the iron channels the field outside the air gap back to the air gap with mu times less leakage. A purely air core coil would have much worse leakage for the same cross sectional area times magnetic path length, so the losses to leakage at the upper end of the frequency range would be a lot worse. This is one of those things that it's hard to add more wire to fix. The bigger the stack of the coil, the worse leakage gets. Couple that with lower primary inductance and the need to put even more coils of wire to get inductance back up and it gets much, much harder to design to wide bandwidth audio.

[Malcolm Irving]

...
An SE transformer must store energy to release on the half-cycle where the output device is turning off. It has to store enough to complete a half cycle of the lowest frequency you want to come out undistorted. It being an output transformer, it has to store enough speaker energy to fill that half cycle, so this is a pretty stiff requirement. Looked at in this way, SE OTs are different in nature from PP in that they are energy store-and-release devices, not flow-through devices. ...

Yes, I've observed that effect in the waveforms of a SE guitar amp at low frequency. I think it gives another type of distortion, which people sometimes incorrectly attribute to saturation.

[J M Fahey]

Not in any easy way, because this stuff is hard to understand. I doubt that most of the people who make transformers could explain exactly how an air gap in an SE transformer works, but they know that it does, and they know how to take advantage of that.

Not really
The core of the answer is that yes, you lose inductance because of the gap, but core does not saturate and transformer is heavier than expected for the power handled because now you need more copper and iron.

If ypu do NOT use a gap, core saturates, period, so it BOTH loses inductance/Bass (tons of it) and to boot, saturated iron is terrible to handle Audio.

"Saturated" by definition implies "not change", while Audio is "all about change".

If you don´t use a gap (you still have one anyway, because of mechanical imperfections in punched lamination) , you need to use 5X or more than if you had used a proper gap to begin with.
I fail to see the savings/advantage.

[Malcolm Irving]

Iron will saturate when its flux density gets high enough. The increased overall reluctance due to the air-gap means that more ampere-turns are needed to get up to that flux density.

[Mike Sulzer]

Interesting concepts.

One effect of air gapping that is fairly well buried in transformer design is energy storage. An SE transformer must store energy to release on the half-cycle where the output device is turning off. It has to store enough to complete a half cycle of the lowest frequency you want to come out undistorted. It being an output transformer, it has to store enough speaker energy to fill that half cycle, so this is a pretty stiff requirement. Looked at in this way, SE OTs are different in nature from PP in that they are energy store-and-release devices, not flow-through devices.

It turns out that energy storage in a transformer with an air gap is lopsided in favor of the energy being stored in the gap, not the iron. This is much like in a capacitor, where the energy is stored in the insulation, not in the plates. The iron is a "conductor" to conduct the m-field into the air gap. Much less energy is stored in the iron than the gap as you increase field strength by forcing current into the coils.

I have not done the math, but I suspect a non-air-gapped SE transformer would be much larger for the same output than an air gapped one because the air gap stores the permeability times more energy at the same flux density B. The iron helps you constrain the field intensity into a defined volume of space.

A purely air gapped SE transformer would be fighting another problem - leakage. One screaming advantage of combining iron and air (vacuum works, too ) is that the iron channels the field outside the air gap back to the air gap with mu times less leakage. A purely air core coil would have much worse leakage for the same cross sectional area times magnetic path length, so the losses to leakage at the upper end of the frequency range would be a lot worse. This is one of those things that it's hard to add more wire to fix. The bigger the stack of the coil, the worse leakage gets. Couple that with lower primary inductance and the need to put even more coils of wire to get inductance back up and it gets much, much harder to design to wide bandwidth audio.

All output transformers store and release energy because of the magnetization current. At the lowest frequency of operation, the magnetization and signal currents in the primary are about the same. So this energy is of the same order as the energy needed for a half cycle. The SE is more demanding, of course. It has to store the energy from the dc current as well. But that dc current is nearly the same as the peak ac current, and so I think we have just a few times, not many times.

[Mike Sulzer]

Iron will saturate when its flux density gets high enough. The increased overall reluctance due to the air-gap means that more ampere-turns are needed to get up to that flux density.

Yes, and the increased reluctance means that the inductance is down, and the magnetization current is up, and so you get those ampere turns, if you can provide them. Transformer design must consider all the effects at once, and it is certainly true that an SE transformer is most efficiently done with an air gap. I think you have to go through the whole design to understand why.

[Roberto]

What the hell does "'teaching my grandmother to suck eggs" mean?

In northern Italy we say "teaching the cat to climb": basically try to teach something to someone who knows it way better than you. Until early decades of the last century a warm-fresh (sorry for the oxymoron) egg was the "redbull" of the era: just two holes on its symmetrical axis and suck the inner part.

[SoulFetish]

I've been bouncing between authors and whatever industry data on EI lamination standards im coming across. I'm piecing the design together as I'm figuring this out.
I'll say this though, "Audio Transformer Design Manual" by Robert G Wolpert (2004) is fantastic. Not only is it incredibly practical and informative, it's been serving as an outline, kind of guiding me through the learning process.
I've settled on M19/29ga lamination
*M19 0.014" (29 gauge); ASTM A677. 1.55 W/lb • @Magnetic Field 1.50 T • Frequency 60.0 Hz – 3.42 W/kg • @Magnetic Field 1.50 T • Frequency 60.0 Hz)
Nominal Silicon % 2.5-3.8
Approx. Permeability μ = 7,500
Max recommended operating flux density 12 to 13 kilogauss (but I've seen figures as high as 14 kilogauss)

One question I have is, when looking at the loadline and specifying the primary voltage and current, should I use RMS or peak numbers?