tripled coupled inductors

They also have a hexa-winding equivalent, which is very handy for obscure supply ratios.

Tim

-- Seven Transistor Labs, LLC Electrical Engineering Consultation and Contract Design Website:

wound (OK, trifilar wound), very useful.

good.

Those would be essentially transformers but with lower inductance due to gap or material.

I wonder if you could run DC through one winding to counter the saturation from DC in another winding?

Another possibility might be a saturable core reactor.

Paul

What's special about multiple windings on transformers?

They go back to the origins of electronics. Bifilar and trifilar windings give you particularly closely coupled coils (more so on ungapped high-permeability cores), but twisting four wires together would also work.

With more wires, simple twisting doesn't give you exactly identical paths around the core, but that's a known problem, with known solutions.

I recently used a little custom wound toroidal quad-filar (?) common mode choke to deal with some issues on a device with +/-6V with analog+digital grounds (4 power lines).

--sp

I wonder if you could run DC through one winding to counter the saturation from DC in another winding?"

Yes. Absolutely. Think of a tube output push/pull audio output stage, it do es EXACTLY that.

Those were used in the early color TVs in in the 1960s when they came out w ith a (almost) square CRTs. In the beginning of color TV the CRTs were roun d. Monochrome CRTs used simple magnets for pincushion correction, which is needed on any square CRT. This is a function of geometry.

Back then (see the print of a CTC25) to use electronic correction was to cu mbersome. So they modified the waveform of the vertical sweep output to go to this saturable reactor to straighten out the sides of the raster. They a lso used some waveforms derived to make the top and bottom straight, rather than pincushioned. They imposed a waveform on the vertical yoke windings t o accomplish this. I consider this some pretty damngood engineering, try it .

Another technology, back in the days of iron power supplies was the ferrore sonant transformer. To this day I do not completely understand how those th ings worked, did what they did and were relatively efficient. Especially th e efficiency part.

The transformer ran right off of line voltage and that line voltage put out enough to totally saturate the core. The secondaries put out square waves. it was already regulated well enough for most supplies in Zenith TVs. I do not understand how they were efficient, but they were. There was a step up winding with a capacitor across is somewhat like a motor run capacitor. If that went open all the voltages dropped and if you got anything at all it was a small raster and, well it was like having low batteries. All the volt age were too low.

But those things were durable and I wish I had a couple of them now. they a re also inherently current limiting. The perfect source for a bench power s upply, regulated after of course. But try to find one now. You would have t o look for a power transformer for a 1977 Zenith 25" console TV with like a 25GC50Z chassis or some shit. The odds are none because slim left town.

But really with a couple of these you could build a damn nice bench power s upply and not even worry about current limiting because they do it. This is the part I do not understand. That gets into physics which really is not m y field. I thought when the core saturates the inductance goes down and the refore impedance which increases power drain. These things seem to be unaff ected and will work into a dead short for hours without batting an eyelash.

Anyway, designing a circuit using saturable reactors requires a knowledge o f physics on tip of electronics. No school teaches anything of the sort the se days. But then it does not pay.

But then there is a thread around about saturation of the core of an SMPS t ransformer during startup. Maybe one of the 120 year old engineers can answ er it.

give you particularly closely coupled coils (more so on >ungapped high-per meability cores), but twisting four wires together >would also work.

around the core, but that's a known problem, with known >solutions."

You might be opening a can of worms here, and actually I am pleased to be i n an electronics discussion here for a change...

I sometimes beef with audiophiles. Some of them claim that woven finely str anded but heavy gauge speaker wire sounds better. Now I agree as it has bee n demonstrated to me that long runs of shitty speaker wire do actually degr ade the signal, I heard it myself.

Seems to me that a single strand of wire would be the best. but now you bri ng this up, that having those on-identical magnetic paths might be better. At least that is hoe I am reading it and I DO see many multifilar transform ers these days.

I have advocated on audiophile groups jut using Romex. Solid copper wire. T he minimum inductance and the minimum bullshit for your precious audio sign al.

I want to know exactly how multifilar widongs on a transformer enhance its performance, ESPECIALLY in light of its obvious extra cost.

Really, if it has a demonstrable advantage there it might in audio. not tha t anyone but three in ten thousand people can hear it, but whe you shoot fo r it all you shoot for it all.

It really does seem to me that a single strand would be better. This Litz w ire and all that, just what does it gain you ? You got more insulation and therefore more DC resistance. How does the gain overcome that ? this is one of the things I have a hard time wrapping my head around. How can a not so lid be better than solid ?

I am sure it has something to do with magnetics and materials, not really e lectronics.

Does mixing it up a bit reduce core saturation ? And really, does that tota lly offset the deleterious effects of the higher DC resistance ? Has this c ome to pass due to the higher frequencies in the SMPS rather than using the old "iron" power supplies ?

Anyway, maybe it is outdated, but I learned that stranded wire was made so that it is flexible and won't break.

Stranded wire is explainable, but in a transformer ? It is not supposed to move.

What is the advantage ?

Litz wire reduces the ac resistance due to less skin depth loss

Look up skin depth

Cheers

Klaus

Conductance increases proportional to conductor diameter if it's skin-depth limited, but cross-sectional area of the conductor is proportional to diameter squared so it's usually better to use Litz wire to fill a given window area if the frequency is high.

--sp

with a (almost) square CRTs. In the beginning of color TV the CRTs were ro und. Monochrome CRTs used simple magnets for pincushion correction, which i s needed on any square CRT. This is a function of geometry.

cumbersome. So they modified the waveform of the vertical sweep output to g o to this saturable reactor to straighten out the sides of the raster. They also used some waveforms derived to make the top and bottom straight, rath er than pincushioned. They imposed a waveform on the vertical yoke windings to accomplish this. I consider this some pretty damngood engineering, try it.

IIRC convergence was even more painful with delta gun tubes in earlyish set s. ISTR an entire panel of controls.

resonant transformer. To this day I do not completely understand how those things worked, did what they did and were relatively efficient. Especially the efficiency part.

aka constant voltage transformer

NT

gs give you particularly closely coupled coils (more so on >ungapped high-p ermeability cores), but twisting four wires together >would also work.

hs around the core, but that's a known problem, with known >solutions."

in an electronics discussion here for a change...

tranded but heavy gauge speaker wire sounds better. Now I agree as it has b een demonstrated to me that long runs of shitty speaker wire do actually de grade the signal, I heard it myself.

a lot of people have heard things that turn out not to exist. It's one reas on why audiophoolery is so widespread.

ring this up, that having those on-identical magnetic paths might be better . At least that is hoe I am reading it and I DO see many multifilar transfo rmers these days.

The minimum inductance and the minimum bullshit for your precious audio si gnal.

s performance, ESPECIALLY in light of its obvious extra cost.

wire and all that, just what does it gain you ? You got more insulation an d therefore more DC resistance. How does the gain overcome that ? this is o ne of the things I have a hard time wrapping my head around. How can a not solid be better than solid ?

o move.

Lower R at rf due to skin effect

NT

gs give you particularly closely coupled coils (more so on ungapped high-pe rmeability cores), but twisting four wires together >would also work.

hs around the core, but that's a known problem, with known solutions."

in an electronics discussion here for a change...

tranded but heavy gauge speaker wire sounds better. Now I agree as it has b een demonstrated to me that long runs of shitty speaker wire do actually de grade the signal, I heard it myself.

Might be skin effect, which Litz wire helps.

As Wikipedia says, it's frequency dependent, and the 20Hz to 20kHz audio ra nge is a bit low for it to be all that important, but audiophiles can hear quite small differences.

ring this up, that having those non-identical magnetic paths might be bette r. At least that is how I am reading it and I do see many multifilar transf ormers these days.

. The minimum inductance and the minimum bullshit for your precious audio s ignal.

ts performance, ESPECIALLY in light of its obvious extra cost.

The poster children for multifilar transformer windings are the ratio trans formers, more formally known as inductive voltage dividers, used in standar ds laboratories.

Coaxial AC Bridges, by Kibble, B. P., Rayner, G. H. (1984) points out that a properly wound ratio transformers divides accurately to about one part in 10^7.

hat anyone but three in ten thousand people can hear it, but when you shoot for it all you shoot for it all.

wire and all that, just what does it gain you ? You got more insulation an d therefore more DC resistance. How does the gain overcome that ? this is o ne of the things I have a hard time wrapping my head around. How can a not solid be better than solid ?

When the high frequency current sticks to the surface of the wire and ignor es the core, as it does.

electronics.

Electromagnetics.

Not at all.

r DC resistance?

Yes.

n using the old "iron" power supplies?

Skin depth is frequency dependent, so this is what's going on.

o that it is flexible and won't break.

Another advantage.

o move.

Skin depth.

Saturating resonant circuits exhibit clamped hysteresis.

That is, if you have an inductor (in an LC resonant circuit) which saturates around the signal level you test with, then when it hits resonance, its resonant frequency rises -- as if to shy away from your input. You can never quite hit resonance, because it's always pushing away.

If you subject it to a rising sweep, you'll find the amplitude remains fairly constant over that range, until the frequency is high enough, and the amplitude isn't high enough, to maintain saturation. At that point, it snaps out of saturation, and resonance disappears.

If you use a falling sweep, then nothing happens, until you approach the unsaturated resonant frequency. As soon as you touch it, energy accumulates and the inductor saturates away from your stimulus. Only a blip is transmitted.

If you sweep in both directions, at any speed slow enough to 'hook' onto resonance, you observe a hysteresis loop, where there's a wide, flat, sharp-high-frequency-edged 'peak' in the rising direction, and a tiny blip in the falling direction.

Suppose you supply such a network with constant frequency instead. As the amplitude varies, the saturation and therefore the resonant frequency also varies with it. The effect is to regulate amplitude as a function of frequency -- constant flux.

The full circuit used is a primary winding, followed by a series leakage inductance, followed by a secondary winding with a resonant capacitor. The capacitor enforces a sinusoidal waveform, rather than the clamped sine that hard saturation would give (i.e., a saturating sine wave looks like the opposite of a TRIAC phase control waveform: the early part is transmitted and the later part is shorted out by saturation).

The series inductance (constructed by using magnetic shunts in the transformer core) prevents saturation from shorting out the mains supply, and yields an L-LC equivalent circuit, allowing the resonant tank's voltage to vary independent of the supply.

The resulting supply is pretty stable, but the load regulation range is dependent on the input voltage. The output voltage is proportional to mains frequency, though that's usually not a problem as mains is very stable in most areas (this won't work as well on a generator!). There's also a tempco due to the variation of Bsat with proximity to Curie temperature, but this is neatly solved by steel's high Tc (~800C). A ferriresonant transformer (??? I mean, one made from ferrite, that is), wouldn't be so nice, because ferrite has much lower Tc (~150C).

Because Bpk is necessarily being used at saturation, core losses are maximal for the material. Normally, thinner sheet steel is used, but the transformer still ends up running hot.

Tim

It may be important to differentiate between transformers and coupled inductors, as one cannot always serve as the other.

A simple transformer is not normally designed to store energy, while a basic inductor may have no other purpose.

The fact that flyback and Cuk magnetics rely on energy storage, and that these are popular at low power levels, may encourage the belief that any multiply wound structures can be used as coupled inductors.

Commodity flyback transformers, and commodity chokes with multiple windings may be used as simple coupling transformers, with some precautions, but signal transformers will not be satisfactory chokes or flyback transformers.

The multifilar winding of commodity chokes can be a disadvantage in their use as coupled inductors. Leakage terms are frequently used (are required) to absorb imperfections in channel matching or voltage ratios, load imbalance and to steer ripple current.

Multifilar windings may also be misplaced in situations where reliable isolation is required.

A more common application (in the past) was in multi-output forward or bridge converters, where the multiple output chokes could be replaced by a single multiple-winding inductor with the correct phasing and turns ratio. This allowed the higher voltage channels to absorb ripple current and reduced individual channels' minimum load requirements.

I'm not sure it's a profitable persuit - figuring out how to use commodity parts for other purposes, where parameters are not specced by the primary vendor, never mind the possible alternate sources.

Doesn't a commercially viable design justify purpose-built magnetics?

RL

John Larkin doesn't think so. There are enough variables in a transformer or inductor design that you'd expect most applications to need purpose-built magnetics, but the design process is complex enough to frighten off the faint-hearted.

The Versa-Pac parts have six windings.

Price is OK, better than custom parts in lowish quantities.

Dunno about that. It keeps George and Spehro and me in beer and skittles, and probably some others that hang round here.

Cheers

Phil Hobbs

but when you shoot for it all you shoot for it all.

You got more insulation and therefore more DC resistance. How does the gain overcome that ? this is one of the things

I have a hard time wrapping my head around. How can a not solid be better than solid ?

I'm sure you have heard of skin effect, if not look it up. Litz wire is all about keeping Rac as close to Rdc as possible for an equal cross section of copper. As frequency goes up the AC resistance of wire goes up, because the internal portion of the copper carries less current than the surface portion. The effect of this is very small at audio frequencies but is substantial at 1500kHz. The adults in crystal radio are getting Q's over 2000 on inductors in the 150uh to 220uh range (aircore), Q's of 1400 with ferrite. But somehow, I think you know this. Mikek

PS, for audio, I don't think it matters. But do use gold plated connectors, and keep the romex off the floor at least 4" with the most expensive pyramid shaped pedestals available. :-)

Sola still makes those.

Cheers

Phil Hobbs

Wurth has some nice parts with lots of windings. The sample kit is worth having on hand.

--sp