Twisted Pair or Coax for 1:1 Transformer?

AFAIK, the leakage inductance would be more dependent on the core itself. What coax would do for you is (possibly) have less interwinding capacitance.

Why not try one of each, and compare?

You can get really cheap coax (like free) by scrounging a scrap VGA cable. ;-)

Cheers! Rich

Yeah..I thought the same.. Some dude on E*bay has some nice teflon #20AWG core coax. Silvered too..

D from BC myrealaddress(at)comic(dot)com BC, Canada

Coax, of course. The primary and secondary are concentric. You can't do better than that.

It sounds like you're making transmission line transformers. What are you trying to do?

Jeroen Belleman

Isolation transformer for a power converter.

I suspect if I buy a poorly qualified coax, the leakage inductance maybe worse than twisted pair. The leakage inductance for coax depends on the inner conductor insulation thickness.

I did a bit of Googling for a twisted pair leakage inductance formula but no luck yet..

D from BC myrealaddress(at)comic(dot)com BC, Canada

Not even close. Leakage inductance in transformers with coax windings depends mostly on what you do with the ends of your winding. I've made transformers with leakage inductances in the sub-nanohenry ballpark, using bits of semi-rigid coax running through stacks of ferrite toroids. The coupling factor of these transformers was of the order of 0.999997. I've done even better, I think, but my network analyzer doesn't go beyond 6 GHz.

Anyway, that's not what you should worry about, I believe. What you are asking for is not at all in the same league. Your real worry is more likely to be capacitive coupling between the windings.

It would be helpful if you stated what primary and secondary inductance you need, how much leakage inductance you can tolerate, how many volt-seconds this transformer is going to see, how much coupling capacitance you can tolerate and what the voltage between the two windings is going to be.

And don't forget that whatever circuit this transformer is going to connect to likely also has inductance, which directly adds to the leakage inductance.

Jeroen Belleman

How about twisted quad? Best of both worlds and cheap to make.

--
~ Adrian Tuddenham ~
(Remove the ".invalid"s and add ".co.uk" to reply)
www.poppyrecords.co.uk

Why not use one of the TI or ADI isolator chips? Some of the ADI parts even have a power coupling channel. DC to daylight bandwidth, no hassles.

John

oops... forgot to mention this is a power application, not small signal.. The transformer is the main transformer in a smps passing 140Watts.

D from BC myrealaddress(at)comic(dot)com BC, Canada

I think I get it.

4 wires.. 2 parallel for primary. 2 parallel for secondary.. Results in 1/2 the leakage inductance if I don't mind 1/2 the magnetization inductance.

D from BC myrealaddress(at)comic(dot)com BC, Canada

True.. At some point, the inductance of interconnections, leads and part internal inductances become significant. So there's no point in making the super transformer if it's going to get washed out by the surrounding parasitics. I'm just trying to be aware of the choices there are for a winding structure.

I'm a newbie with coaxial wound transformers... I really don't know what to expect. Wow! 0.999997 coupling. That's encouraging... :)

I'm also new at smps development.. At the moment, I believe the capacitive coupling is only a conducted EMI factor. Dunno yet..

Lp = Ls = 50uH Lleakage = maybe

The critical point is that the "North" and "South" wires of the bundle are paralleled to form the primary, the "East" and "West" are paralleled to form the secondary.

Assuming that each of the two paralleled wires carries half of the total current, the effective current will appear as though it is in the exact centre of the bundle for both the primary and the secondary. Twisting the bundle before winding it will ensure that unbalance due to skin effects and the variation in the length of each conductor due to winding around a radius is effectively cancelled.

Alternatively, the two conductors in each circuit could be used in series to ensure accurate current balance, but this constrains the winding to an even number of turns and makes non-inductive terminations more difficult. With parallel 'star quad', the balance can be continued along the PCB tracks if necessary.

--
~ Adrian Tuddenham ~
(Remove the ".invalid"s and add ".co.uk" to reply)
www.poppyrecords.co.uk

ONE TURN?

Why?

John

It might be up to 3 turns..I'm still playing with the math.

I going for minimal turns for least leakage inductance.

D from BC myrealaddress(at)comic(dot)com BC, Canada

The compass description is not clicking in fast for me... I'll give it some thought.

D from BC myrealaddress(at)comic(dot)com BC, Canada

--
~ Adrian Tuddenham ~
(Remove the ".invalid"s and add ".co.uk" to reply)
www.poppyrecords.co.uk

Neato! Thanks.. :)

I think I get it..

To help understand, I considered it air core and pictured the primary field pattern.. I think it's a hour glass pattern about the primary pair. The secondary wires intrude on the primary field pattern. All the wires nearly have the same field in common.

While googling for star quad wire ...I found

showing field pattern.

For microphones

Yikes 500ft min at $491.94Can at Digikey for Belden 1804. But the insulation is probably excessive to make a power transformer.

Over to

huh..no star quad magnet wire standing out but toss'm enough money and I think they'll make the wire and cook breakfast.

By any chance you know a nice source for a say 10ft of star quad magnet wire. Say #23/conductor, single insulated.

D from BC myrealaddress(at)comic(dot)com BC, Canada

You can imagine a 'virtual' primary conductor occupying the space between the two 'real' primary conductors. Then think where the 'virtual' secondary conductor is situated - in exactly the same place.

I only represented it as a sheathed bundle to make it easy to draw, you can make it yourself from four lengths of single wire without a sheath and it will work just as well. For the length you need, that would be a lot cheaper - and you can choose the wire gauge and insulation thickness which suits your particular application.

To keep an even twist , I would cut four identical lengths of wire, make them up into a parallel bundle and wrap a turn of adhesive tape around each end. Then grip one end of the bundle in a vice and the other end in the chuck of a hand drill. Pull firmly and gently twist the wire up good and tight, so that the bundle shortens by about 20%. Release the drill crank but maintain the tension on the bundle, this will allow it to unwind a bit.

If you are worried that it might unwind further, work some flexible glue or silicone mastic into the bundle with your fingers.

There is also a way of minimising the stray inductance of the lead-out wires (if they are long enough to cause problems) and the PCB tracks by using the same principle - and this bundle would definitely have to be hand-made.:

Once again, it is easy to visualise what is happening if you imagine a 'virtual' conductor in the space between two identical 'real' conductors.

--
~ Adrian Tuddenham ~
(Remove the ".invalid"s and add ".co.uk" to reply)
www.poppyrecords.co.uk

Cool.. I suppose the F field gets countered by the S field on both sides.. The more field cancellation the less trace inductance..

Electronics certainly gets artsy when respecting the magnetics.

Thanks.. :)

D from BC myrealaddress(at)comic(dot)com BC, Canada

For 10 feet i would consider two things: Asking around for a sample. Making it myself.