Need good link to output transformer design.

[Alan0354]

Hi
I am studying RDH4 chapter 5 in transformers. But it really does not explain where the equations come from and I have questions in how it came up with the formulas. Can someone give me good links on the web in how to design output transformers?

Thanks

Alan

[Jazz P Bass]

output-trans-theory
Our very own R G has an excellant site. http://www.geofex.com/article_folder...es/xformer.htm

[sgelectric]

This is from Group DIY (formerly Prodigy DIY):

META-Transformers

SG

[Alan0354]

Thanks for the links, I had scanned through quite a few when I was searching, most are not as detail as RDH4. Problem is if I don't agree with RDH4, I have no other resource to verify with. I guess this is really not a popular topics and very few people interested in this. There's just no detail analysis in output transformer that I can find so far.

[jazbo8]

Thanks for the links, I had scanned through quite a few when I was searching, most are not as detail as RDH4. Problem is if I don't agree with RDH4, I have no other resource to verify with. I guess this is really not a popular topics and very few people interested in this. There's just no detail analysis in output transformer that I can find so far.

On the contrary, there is a lot interest (ability is something else ) on the subject, a lot discussion at diyaudio and yahoogroups... Anyway, output-trans-theory has a good tutorial, I have to admit, even reading it a few times, I am not sure I got everything, and I am certainly not going to wind my own...

Jaz

[Alan0354]

On the contrary, there is a lot interest (ability is something else ) on the subject, a lot discussion at diyaudio and yahoogroups... Anyway, output-trans-theory has a good tutorial, I have to admit, even reading it a few times, I am not sure I got everything, and I am certainly not going to wind my own...

Jaz

I actually read through some of the pages in this article. It really does not give the theory and proof. My most frustrating questions are in the last two post in my RDH4 thread. I think those presented in RDH4 is can be very useful, but so far, without anything to back it up, I don't agree with the derivation. I have not been able to find any article that is more involve the double check RDH4, not even close.

If people are so interest in OT, why nobody stop and look at the theory and resort to just using feeling and common sense. Even in your article said the author refuse to put in all the math. There are parts that you can wing it, BUT there are parts that you really want to know the exact response. And I totally disagree with the idea that deriving formulas and calculating with math is useless like the article said. That's how you move a big step beyond tinkering and into the design world. I've been on that side for 20 years and I can assure that you can go a giant step beyond by digging into the theory and do the math. Frequency response and distortion are such an important topics that you don't wing it!!!

Case in point, in your article, right above Fig.1, it said the winding loss of the secondary referred back to the primary is 390ohm. This is not negligible, and it is very import to know which side of the L0 it is on as I described in the RDH4 thread!!! You don't wing it and lump it into RW. It will cause like 10% error!!! Then the RDH4 ignore the capacitance and do strange combination for the upper limit of the OT.

I can assure you if you are dealing with BJT or MOSFET, you can find all the theory and math to back up the circuit.

Could it be because everyone "wing" it and missing 5% here and 10% there that create the "black magic"? We gone too far in the last 30 years to just chuck everything to "black magic" and "touch". We all know a few % change of value changes the sound completely. You don't just wing the 10%!!!


Sorry, I am just very frustrated with this that I can't find anything to resolve this and I strongly disagree with the RDH4. I even posted one of the two question in Physics Forum and so far no catch yet.

[Steve Conner]

10% error is nothing compared to the manufacturing variation in core steel permeability, and the variation of permeability with flux density. The models in RDH4 are very simple, and your concerns about them are quite valid from a theoretical point of view, but in practice you're wasting your time because the actual equivalent circuit is different even to the more complex model: it includes multiple high frequency poles from distributed capacitance. If you care that much, you should be building transformers and measuring them.

[jpfamps]

10% error is nothing compared to the manufacturing variation in core steel permeability, and the variation of permeability with flux density. The models in RDH4 are very simple, and your concerns about them are quite valid from a theoretical point of view, but in practice you're wasting your time because the actual equivalent circuit is different even to the more complex model: it includes multiple high frequency poles from distributed capacitance. If you care that much, you should be building transformers and measuring them.

I agree with this 100%.

Whilst leakage inductance and interwinding capacitance are conceptually important in understanding output transformer function, they are impossible to calculate with any practical degree of accuracy.

Not only as Steve points out is the core variable, also there will enough variation in the windings to effect capacitances and inductances. Even the tightness of the stacking effects the transformer parameters, as does varnish impregnation.

You can't even wind an accurate iron core inductor from theoretical calculations, so your chances with an output transformer are zero.

Regardless, the RDH4 provides several references for the formulae they provide if you really need to know how they are derived.

The best "cook book" for transformer design is Robert Wolpert's Audio Transformer Design Manual.

His approach is to make a "guestimate", wind the transformer, then measure it.........

BB Babani's Coil Design and Construction Manual is useful too for practical construction details.

[Enzo]

If people are so interest in OT, why nobody stop and look at the theory and resort to just using feeling and common sense.

But that is exactly what we do, and you object to the lack of mathematical rigor.

And I totally disagree with the idea that deriving formulas and calculating with math is useless like the article said.

SO which do you want, people using feel and common sense, or people deriving formulae and calculating with math?

[jazbo8]

Case in point, in your article, right above Fig.1, it said the winding loss of the secondary referred back to the primary is 390ohm. This is not negligible, and it is very import to know which side of the L0 it is on as I described in the RDH4 thread!!! You don't wing it and lump it into RW. It will cause like 10% error!!! Then the RDH4 ignore the capacitance and do strange combination for the upper limit of the OT.

I did not read it that way, I think what the author said was to add the winding resistance to the Zpri, in fact, it is a commonly given spec for OPT.

Could it be because everyone "wing" it and missing 5% here and 10% there that create the "black magic"? We gone too far in the last 30 years to just chuck everything to "black magic" and "touch". We all know a few % change of value changes the sound completely. You don't just wing the 10%!!!

I seriously doubt the transformer designers like Mr. Turner would "wing" it, it is likely he does not want to disclose what he knows and/or does not want to bore the readers with the math (as he said in the tutorial), since most are more interested in practical as oppose to strictly theoretical knowledge. I admire your desire to get to the bottom of things, but that's just not where most builders spend their energy. Look at it this way - whether you agree with it or not, trial-and-error or "winging it" as you put it, is a valid empirical approach to the research and development process.

Jaz

[R.G.]

I designed transformers as part of my job for some years, so let me give a bit of my insight.

Alan, you're doomed to frustration, because Steve is correct. The real math behind it is too complex even to derive. In actual fact, the location of every wire in relationship to every other wire, the core, and the surface the core is eventually mounted upon matters, at least a little. Maybe by now something along the lines of intense computer modelling would allow some clear answers, but I have never seen a straightforward set of equations, no matter how complex, that did a good job of design. About the best I've seen so far would be to cut the number of iterations of build-and-test.

As an example, the air gaps in a transformer have a big effect, yes? The air gaps are determined after the windings are done and the laminations are stacked by how much care is taken in jogging and tightening the laminations as the clamping bolts are tightened. Which equation takes that into account?

What the professional transformer designers do is to keep a detailed history of the measured performance of every transformer they've ever worked on. This lets them make decent guesses to start with, and cuts the number of build-and-try cycles.

A real problem with DIY on transformers is that a DIYer does not have the experience, equipment or materials to make even one transformer, so making several iterations is a monumental task for them. The pros have the equipment, materials, experience, and probably skilled laborers to do the two, three, or four iterations it takes to make a good enough design for production.

[Steve Conner]

Exactly!

If you are interested in computer modelling of coils and transformers, the most interesting work I know of is this work by Paul Nicholson:
Tesla Secondary Simulation Project

I've used his computer simulation code to design Tesla coils (as science park exhibits and the like) and the real coils come out within a few percent of the simulation.

What he's doing amounts to calculating effective inductance, self-capacitance and self-resonant frequencies (there are several) for a single-layer air-cored coil, as well as coupling coefficient to another similar coil.

Simulating the exact frequency response of a guitar amp OT would use the same math, only worse, because there are multiple layers, the coils aren't circular, and there is a non-linear magnetic core shoved up the middle of it all. (which as RG suggests, may lack magnetising inductance unless whacked just right with a NOS leather mallet during stacking.)

Hopefully when you see how complicated it all is, you can appreciate the value of the oversimplified models shown in RDH4. Williamson probably cursed these same models when trying to stabilise the feedback loop in his eponymous amplifier.
Welcome to the most complete do it yourself guide on the D. T. N. Williamson tube amplifier.

Someone once said that "all models are wrong but some models are useful".

[J M Fahey]

R.G: : clap clap clap ! Bravo! (standing ovation).
You couldn't have said it any better.
*Clearly* you have years experience, thanks for sharing.
Fully agree.
I'm not dissing theory, FAR from that.
In fact I often include (somewhat simplified) Math in my answers here, to avoid "opinion" as much as possible, but let's put things in due perspective.
Math (plus proper use of datasheets) puts you in the ballpark; say, within 10% to 50% of what you want to accomplish.
Which is very good !!!
Then (and this is very useful) , when you start tweaking (say, rewind using 10% more or less turns than calculated), *measuring* and *recalculating* lets you know for certain whether you are approaching your goal or going away from it.
SAVES A LOT OF FRUSTRATION !!!
Compare it to a missile: first you need to reach a point *near* the enemy plane, then "terminal homing" lets you hit it (or pass within a few yards).
So in practice you can't say "Math is useless" nor "Math should give me the perfect answer, on the first try" ; both extremes can be frustrating; use both tools which, by the way, are synergistic: one of them integrates with and helps the other far more than expected.

Menu 1) If your goal is have a perfect understanding of Theory, fine with me, but besides RDH4 you'll need to study *a lot* of other Tech books.
Maybe I'm killing your Social and Family life for the next 10 years, but here it is:
Technical books online
Enjoy .

Menu 2) If you want to make a very good transformer for some project (personally, that's my goal, I make a living out of this) , design (using simplified equations, and, more important, properly estimating a lot of parameters your supplier won't give you .... because *he* ignores them), build *one* , measure it, and start tweaking, but not at random, use measurements to correct estimated parameters (such as :"I had thought iron saturates at 15000 Gauss, but this crappy EI lamination I bought must be saturating at 12000 , only way to explain what I'm measuring") and so on.
As RG said (and Steve hinted), you'll probably have a so so working transformer on first try, and a very good one by the 3rd or 6th.
Good luck.

[Alan0354]

Thanks guys, I don't know transformers, as I mentioned before, I managed successfully to totally avoid touching transformer in my whole career. This is the first time that this become essential for me. I have no intention making an output transformer, I just want to understand it, particular I want to design the off line switching supply. I did not realize there is so many variables, I guess I have to forgo the ways I usually study!!!

[R.G.]

If you want to design off-line switching power supply transformers, the stuff in RDH4 is essentially useless. There are much better references for doing switching power supply magnetics, whole textbooks devoted to the topic. These will include math, modelling and in-practice hints as well. Skin effect, proximity effect, distributed voltage gradient, and core losses DO matter in switching power supply design. Worse yet, you're usually prevented from minimizing leakage inductance much by the need to provide 4000Vac isolation and 8mm creepage and clearance distances between primary side and secondary side inside the same transformer. It's an art - but a different art from output transformers.