pot core

The original design was bifalar-wound 7x36 Litz, for no rational reasons that I can see.

The wire was purchased Litz.

It simply has the added benefit of providing a bit

The leakage inductance usually has a significant effect on the switching lo sses - the flux that isn't linked to the other coil produces a decaying cur rent in the coil being turned off which manifests itself as a current throu gh a transistor, which has a significant voltage drop across it, and ends u p heating the junction.

In a Baxadall class-D oscillator this is less of a problem because the volt age is low at the switching point.

Why not try it and see? 300kHz is well into the region where the skin effec t in 30 AWG wire is significant, and Litz wire is recommended.

says that the skin depth at 300kHz is 139 micron, about half the width of t he 0.28mm 28AWG wire that you ought to be using.

says that you ought to be using 26 strands of 42 AWG wire in your nominally 28 AWG bundle.

The effect won't be all that big, but it won't be "tiny" or negligible, muc h as you'd like to save design time and intellectual effort by neglecting i t.

More surface area for the skin depth to spread over, no?

The input DC current will be under 1 amp worst-case. It might be prudent to use #28, if it fits.

Magnet wire is expensive these days! So is plain tinned bus wire.

On Sat, 07 May 2016 23:19:05 -0400, Martin Riddle Gave us:

Did you even look at the one he was buying?

Looking at the one he was buying, it must be several amps.

On Sat, 07 May 2016 22:20:05 -0500, Jon Elson Gave us:

No. he is right. Bolting up a gapped center pole core is a sure fire way to fracture the core.

pre-gapped core configurations are best epoxied together or epoxied together with a fiber or nylon washer filling the gap. Otherwise you are asking for breakage in high numbers.

On Sat, 07 May 2016 20:21:51 -0700, John Larkin Gave us:

Litz wire is more like 20 54. At the seven strand level it is very easy to configure your own, especially when the turns count is so low.

On Sat, 07 May 2016 20:27:38 -0700, John Larkin Gave us:

Hell depending on the turns count you could fit three windings of standard SPC PTFE in there at even 26 Ga.!

People tend to call multi-strand magnet wire "Litz" even of it's just twisted and not woven. It does have eddy-current advantages over solid wire.

MWS calls 3-strand wire "Litz."

The 7x36 that was originally used is equivalent to AWG 27.5. I'll get some Beldsol 28 solid and try that too.

I don't think I need stranded, and multistrand is harder to solder reliably.

mp

R_5.JPG

The reason for using bifilar winding was to maximise the coupling between t he two windings. The reason for using Litz wire, was to maximise the surfac e area of the wire - the skin effect is significant at 300kHz, and the core of a single 28 SWG wire wouldn't be carrying anything like its fair share of the current, and the effective resistance of the winding would be higher than it could be, leading to greater heat dissipation in the winding and l ower efficiency.

I've posted a link to a Litz wire manufacturer's web-site, which recommends 26 strands of 42 AWG wire for the 300kHz range.

Proper Litz wire isn't just twisted, but rather structured so that each wir e sees the same magnetic field, averaged over it's whole length. Bifilar wi nding of twisted pair does that for two strands, but you need something mor e complicated for more strands - braiding works, but you end up with a flat structure, which has other problems.

He might have done, but that tends not to create quite the desired effect.

You might halve the resistive losses, which won't make all that much differ ence to the efficiency, but would make the transformer run quite a bit cool er, though at 300kHz core losses can be significant too.

Scrambling the wire distribution in Litz wire is a tolerably sophisticated refinement, but it does seem to be a desirable one.

There's nothing wrong with that.

That will reduce your resistive losses - though not in proportion to the increase in the wire's cross-sectional area. Going back to Litz would make a bigger difference.

You don't need stranded wire, but it will deliver better efficiency and a cooler transformer. If you are having trouble soldering it, buy a solder pot - the Cambridge Instruments winding shop had one for each winder.

It's a very small pot of molten solder, with a tight-fitting lid with a wire-sized hole in the middle through which you can poke a single wire (or bundle of wires). Much easier to use than a soldering iron for tinning wire ends, and much faster.

The leakage inductance - the flux that threads one coil and not the other - means that you've got some current flowing through the coil being turned off after the voltage across the switching transistor starts rising.

If the parallel capacitance of your winding is high enough, it will flow into that rather than through the transistor being switched off.

Simultaneous conduction is a good way to fry transistors, but getting rid of it doesn't get rid of all the mechanisms that can generate heat in the switching transistor.

It isn't so much surface area (which after all is the same for straight multifilar wire as for Litz) as it is the braiding. This forces the current to be nearly uniformly distributed across the cross-section of the bundle, despite the magnetic effects. Every strand alternates between being inside the bundle and outside, so the skin effect affects them all equally.

Cheers

Phil Hobbs

On Sat, 07 May 2016 20:48:02 -0700, John Larkin Gave us:

I could point to a thread years ago where I argued exactly that fact and you argued that "Litz wire had to be specially woven".

to

It's not so much concentrating the field as providing a short/easy path for the magnetic field so you get a lot more volts per turn out of your coil f or a given rate of change of current. The concentration of the path helps t he coupling between the turns, but it's essentially an incidental advantage .

On Sat, 07 May 2016 20:48:02 -0700, John Larkin Gave us:

Not if you get the solderable variety.

It looks like simple Teflon wire at 26Ga and SPC would work.

Well, with a toroid held flat against the PCB/enclosure, yes.

Neither the central metallic screw, nor any kind of external clamping, is a concern for the pot core. Very little external flux, and that which remains is at a high impedance (low current).

A pot core is just an E core with a nearly full volume-of-revolution. Copper tape is regularly used around the outside of E and RM style transformers, to short out what leakage remains. Those don't have any problems, and pot cores are only better still.

Tim

On Sat, 7 May 2016 23:49:52 -0500, "Tim Williams" Gave us:

Exactly. No need to worry about a "shorted turn" with a pot core.

clamped down) in.

But not for three strands - the constraint is merely that each wire has to experience the same magnetic flux change over it's total length, which happens to be trivial for small numbers of strands, where you merely need to twist it.

Why not find the thread and quote what I actually posted, rather than relying on your less-than prefect memory?

Is there a question here?

Supposedly you just need an isolated, unregulated DC-DC converter with the appropriate outputs, a replacement transformer for the current layout, or a supplier for the missing smd platform.

Where does fiddling with a pot core rewind enter the equation?

RL