Characteristics of Guitar Amplifier Output Transformers, and criteria for design

[SoulFetish]

I'm trying to better understand magnetic circuits and have been doing some informal reading. I've found discussions and threads in this forum to be insightful; particularly in inspiring me to step back and get some better fundamental knowledge on the subject. Quite frankly, I had some uniformed assumptions that were just wrong. I'm sure I still have some (and there's obviously much more to learn) but I'm starting to piece it together and I'm getting those light bulb/"Oh Shit" moments when it makes total sense.
I've been mostly reading about Mains type power transformers, and have found the articles on Rod Elliot's site to be really great. They are a really good primer in laying some groundwork for understanding some of the other info I've found in some other sites.
But, I've been thinking about Output Transformers lately, and would like to have a no bullshit conversation about what we are really looking for in OTs for guitar amps.
Most of the "output transformer shootouts" that people have put out there aren't that informative at all, I find. Everyone I've seen pits one manufacturer against another and one is left to choose which manufacturer's iron sounds better.
I would rather understand the different material characteristics and construction techniques and set up the pepsi challenge under those conditions to see if I preferred one method/lamination grade/interleave over another.
Really, I'd like to try my hand at maybe designing one from the ground up. I think that will help me understand how different materials react under different conditions.
And I think you guys are just the sort of folks to talk to about it. We need, like, a bat symbol send out for R.G. with threads like this.




So, where do I begin? I imagine I start with deciding what the power and bandwidth requirements will be?

[Malcolm Irving]

This is a good source (but I expect you may already have found it):

http://jensen-transformers.com/wp-co...rs-Chapter.pdf

and this one:

https://ieee.li/pdf/introduction_to_...chapter_12.pdf

[SoulFetish]

This is a good source (but I expect you may already have found it):

http://jensen-transformers.com/wp-co...rs-Chapter.pdf

and this one:

https://ieee.li/pdf/introduction_to_...chapter_12.pdf

Thanks for those links, Irv. I also tracked down a pdf of Lee’s “ electronic transformers and circuits “.

[jmaf]

There's a text by RG somewhere where he analyzes old and new trafos and finds subtle differences in old iron and as usual for RG in the process he yields several knowledge pearls. I can't find it now, it's not the top results at Google. You should definitely give his other texts a look as well.

Sergio Hammernik of Mercy Magnetics talks about rust in transformers in a column for a mag, that PDF can also be found online, which is a second resource that comes to mind about old iron actually being good.

Just my 2 cents about one of the aspects that isn't discussed very much. Like some wines, good transformers age well, so that's a rich field for research. What's in old iron?

[Leo_Gnardo]

What's in old iron?

Impurities, those magic impurities. Like what's in old copper wire, for pickups and transformers. Our local pickup builder-repairman is constantly scanning for the oldest rolls of wire he can find "because that's necessary to get that old-fashioned sound." Modern metal is too pure, too good.

Good advice on the OT's, why reinvent the wheel. Besides being very time consuming, and wondering exactly what it is you're hearing (takes double blind, um, make that double deaf tests), bound to be expensive.

AND may I add it's good to see you back jmaf, where have you been? It's been a while.

[jmaf]

AND may I add it's good to see you back jmaf, where have you been? It's been a while.

Thank you, Leo. Great to chat with you again as well.

Been on and off several projects, times have been a bit rough here in Brazil. But as they say here, I'm still above the ground and below the heavens.

Certainly wanna be more in touch with the great discussions and folks on this forum.

Cheers!

[Gregg]

The short answer to your topic is: there's no specific criteria for designing a guitar OT. Few simple facts from the past decades illustrate that very well:
1/ Many different types of iron were used
2/ Many different types of winding configurations were used
3/ Most of the time the basic criteria for the OT (especially for the big manufacturers) was the price - the cheaper, the better.

That's why you can see and OT made from 0.5mm PT grade laminations, one primary/one secondary, without any interleaving. On the other side you can see (not very often) HiFi grade OTs made from M6 iron, many pri/sec, heavily interleaved.

My sincere advice to you is don't bother designing a guitar OT from zero. If you're into a specific type of sound better find out what specific OT was used in that amp. Also if you think that designers of a specific amp spent endless hours designing that specific OT to achieve that specific sound you'll be as far from the truth as you can get. However if you would like to learn how an OT is generally designed and learn new stuff that's entirely different story. There's room for many experiments in this field.

[R.G.]

Don't beat yourself up if you don't get transformers clearly in your mind quickly. It can take years for the information from the texts and tinkering to soak in. It did for me anyway.

Here are some quick and dirty maxims to think of.

> There is a big inductor on the primary side. This inductor is the model for the loading of the core that your incoming signal must pay to get through the transformer. It is the fundamental low frequency limit on the transformer. The low frequency rolloff point at half power happens at the frequency where the primary inductor's impedance is equal to the reflected load from the secondary.

Example: If you have an OT with a speaker load, and you've get the winding ratios right, then the speaker load impedance (e.g., 8 ohms, 4 ohms, etc.) is reflected into the primary side by the transformer. For a pair of 6L6's, the plate to plate loading is often set at 4400 ohms. So if you want a half-power bass point of 40Hz so that your guitar's 82Hz low string is a full octave above the rolloff, you instantly know that the primary inductance must be no smaller than
Zl = 4400 @ 40hz = 2*pi*40hz*L
or L = 4400/(2*pi*40) = 17.5 henries.
That's a minimum, and for hifi work they like the half-power point to be lower than a tenth of the actual signal response. In this case, for 1/10 of 82Hz, the inductance needed gets to be 85.4H.

Things get out of hand rapidly, as making ~100H inductors that can handle tens to hundreds of watts requires BIG cores and lots of copper windings.

That sets the low end. Doing this stuff well means picking big enough iron and stacking the Es and Is in a way that maximizes the magnetic properties of the stacked iron and minimizes the contribution of air gaps.

The above is for continuous AC transformers or push-pull. Single ended designs get yet more complex because the magnetic field in the iron has to be able to put out a half cycle of the maximum maximum output energy in a lowest-frequency half cycle. There's only one output device to pull power on its active half cycle, so the energy for the "relax" half cycle of the output device can't come from anywhere but the magnetic field in the core. This gets out of hand in the iron design, as a practical matter, because you must include an air gap, and designing the air gap is tricky and may well need multiple passes.

> For the high end, the fundamental limits are set by a combination of the leakage inductance and interwinding capacitance, as well as funny stuff like interwinding balance and cross-winding coupling. These are all properties of the winding method, mostly.

Leakage inductance is caused by the magneto-motive force ( or MMF; that's the current in the primary windings times the turns) forcing a magnetic field to form that energizes the core. Iron cores are not complete "conductors" of magnetic fields, and free space is not a perfect "insulator" of magnetic fields, so some of the MMF sent into the primary coils leaks out into space and does not get inside some of the secondary windings.

To combat this, special winding schemes are cooked up to intermix primary and secondary windings and layers interspersed with one another so that leakage MMF has a hard time getting out into free space without coupling with the intermixed secondary wires. Ideally, you'd wind the primary and secondary wires side by side, so that every primary wire was parallel with every secondary wire. That's not possible in a world where we want 20-30 times as many primary turns as we want secondary turns, unfortunately.

The next best thing is to wind a layer of primary, then a layer of secondary, then another primary, another secondary, etc. That is GREAT for reducing leakage, but is a real PITA to actually wind, and makes capacitive coupling worse. One of my first tasks as an engineerling was to derive the equations for leakage inductance from the physical geometry of coils of X width and Z winding height. Then I had to go wind the trans former in the lab and explain why I got it wrong. It was very instructive, and in only another year or so I found where I'd gone wrong with that.

As a practical matter (there's that word again) commercial ventures can't make layer per layer interleaves. They subdivide the primary and secondary into sections, and wind, for instance, three or four layers of primary, then a couple of secondary, then some more primary, then some more secondary until the right number of turns is on. This can be shown (I finally got the equations right) to reduce leakage by the square of the number of alternations between primary and secondary. Generally, splitting primary and secondary into two sections each or two/three will reduce leakage by a factor of four over just primary over secondary.

> Interwiring capacitance fights you at every step. If there is a voltage difference between two conductors and insulation between them, capacitive current flows through the space between the conductors. This is true for even two side by side turns of copper wire in a single turn of wire. The intra-and inter-winding capacitance is a distributed thing that depends on where each wire is actually put in the coil. Random winding makes this unpredictable (if smaller - no parallel wires) and different for each article. Layer winding maximizes it. Sigh.

> The general figure of merit in an OT was defined as the "Goodness Factor". No, I didn't make that up, that's what it was called in the OT biz back in the 50s and 60s. This was the ratio of the primary inductance to the leakage inductance. An OT with a primary inductance of 100H and leakage of 10mH (representative values, these) gives a GF of 100/0.01 = 10,000. This is the range you want to shoot for if you're designing hifi OTs.
Write up your last will and testament before trying to get to that with an SE design.

> Other complications happen if you're doing push-pull as anything other than Class A. When half a winding turns off, as it does once each half cycle for Class AB or B, that winding's coupling and magnetic effects (and inductance!) vanishes as it quits participating in the magnetic goings-ons. Not being very careful to make each half-primary have beautiful, complex interlayering/interleaving with the entire secondary causes little glitches to happen in the output right at the spots where current in the primary halves turn on and off. This imposes a fundamental limit to how low crossover distortion can go in a Class AB push-pull output stage. It is the fundamental reason for the "unity coupled" McIntosh OTs and the big hoopla in the hifi market it caused. Guitar amp designers mostly ignore this as an issue, as the silly guitarists LIKE distortion. Let'em live with it. And it's too expensive to fix, anyway, so the makers said.

Hope that helps rather than confusing issues. Yell with questions.

[SoulFetish]

My sincere advice to you is don't bother designing a guitar OT from zero. If you're into a specific type of sound better find out what specific OT was used in that amp. Also if you think that designers of a specific amp spent endless hours designing that specific OT to achieve that specific sound you'll be as far from the truth as you can get.

I appreciate the advice. Look, I've though about leaving the whole thing alone and walking away from the topic all together. There's a part of me that has a sneaking suspicion that when the dust settles, and we near the ass end of this conversation, I'll already be 3 orders deep and waiting for more supplies so I can wind my own OT.
...ugh. (Don't let me go down that rabbit hole, guys)

However if you would like to learn how an OT is generally designed and learn new stuff that's entirely different story. There's room for many experiments in this field.

Yeah, more this ^^ I want to be a little more purposeful in my approach to this area of the design. I generally think learning empowers creativity, and the creative part of this is one things I enjoy most about music electronics.
(even though I'm not breaking any new ground here)

[Chuck H]

Maybe a good place to start from an amateur perspective would be to reverse engineer the OT from an amp you like the sound of (as much as possible without destroying one I suppose) and then apply your research to HOW those physical properties affect the end result and WHY those results sound good. In doing this you're bound to discover some construction aspects you'd like to hear tweaked a little

[SoulFetish]

Don't beat yourself up if you don't get transformers clearly in your mind quickly. It can take years for the information from the texts and tinkering to soak in. It did for me anyway.

Yeah, I have a lot of questions that kind of stretch out over all the different factors at play. For instance, primary inductance is an important factor in audio transformers, where it is of no real design consideration in power transformers? The inductance needs to increase with loading impedance on the driving signal. This seems to indicate a relationship between inductance and magnetizing current? Also, making practical sense of BH curves in the context of voltage/current/frequency conditions. For instance, is there risk of saturation at low signal levels in high permeability cores? Anyway, I have many more, but at this point I'm trusting the learning process. I know it will come together. So let's start off where you did.

Here's what I'm designing for:
2XEL84s with a p-p load impedance of 10k2 and a peak sine wave output of around 17.5W. But square wave output at full drive, i'm guessing +25W maybe? Let's account for alternate tuning and the rare bass player who may want to plug in and set the -3dB at 20Hz.
That means I require a primary inductance of around 81.5H rated for min 25W. (Ideally)
Now What? (lets make some big iron)

[Chuck H]

Hammond 1609

The 1608 at 8k has a very good rep for 18W type builds and it's what I have in my personal (prototype) amp. I solicited Heyboer for the production model and the amps are apples to apples. The Hammond sounds better. Maybe try one and see what you think, then reverse engineer it if you want to play with that. The transformer is rated for 10W, but trust me when I tell you that it will handle anything 2Xel84's will throw at it under any operating conditions. I have that from a tech at Hammond and I've experienced it for myself.

[J M Fahey]

Also if you think that designers of a specific amp spent endless hours designing that specific OT to achieve that specific sound you'll be as far from the truth as you can get.

True.
Amp designers work on the *electronic* side of the problem , and just *order* transformers from an Industry supplier, leaving him to work the details.

[Malcolm Irving]

… . For instance, primary inductance is an important factor in audio transformers, where it is of no real design consideration in power transformers? ...

Primary inductance is in parallel with the load reflected from the secondary, so for an audio transformer the tubes supplying the OT have to provide current into that parallel inductance as well as into the proper load. At low audio frequencies, the reactance (2.pi.f.L) of the primary inductance is low and the extra current causes the primary voltage to drop (due to voltage drop across the output impedance of the tubes - if you like).

For a PT, the internal impedance of the mains supply is very low (say 1 ohm or so). The mains is almost a perfect voltage source. In this case some reactive current is 'wasted' into the primary inductance, but the primary voltage holds up and the secondary (load) side is happy. There still needs to be a reasonable primary inductance (and hence reactance at 50 or 60 Hz) otherwise the primary would draw very high current and overheat.

[Malcolm Irving]

... The inductance needs to increase with loading impedance on the driving signal. This seems to indicate a relationship between inductance and magnetizing current? …

Yes. We can think of whatever is driving the primary as a Thevenin Equivalent, i.e. voltage source in series with an internal impedance.

Then the magnetizing current is Vs / (Zth + Zmag)

where Vs = source voltage, Zth is source internal impedance, and Zmag = magnetizing reactance ( = 2.pi.f.L ).