EMRFD Message Archive 4487

Message Date From Subject
4487 2010-03-29 04:40:35 Tim Multifiliar to the max: 11-filiar!
Last week I was playing around with step-up voltage converters (I know, not radio!) for Nixie tubes. Goal was to go from 5V to circa 200V, using just stuff I had lying around on the bench. I had a big bucket of FT37-43 ferrite toroids handy and tried various boost and transformer (non-multifiliar, just random wound) configurations. Typical turn ratios were things like 5:70 etc. They all kinda worked, but efficiency was rather low, and I could never really break 20 or 30% efficiency. I didn't have any power MOSFET's that could be driven well with 5V, and my bipolar transistors were rather wimpy too. But I'm not sure that all the inefficiency is the fault of my crappy semiconductors: the randomly wound transformers did not really work like I'd expect them too, and had resonances in the range I was trying to drive them, and coupling coefficient seemed to be much less than the "perfect 1" I had hoped for.

I've been using bifiliar, trifiliar transformers in radio projects for years. I thought, "well, the transmission line transformer theory is pretty neat, but it must break down at some point and not be so pretty". I kinda suspected that maybe above trifiliar it wouldn't work so good.

But as long as I was at the bench, rather than just think that it can't work, I decided to build something and see if it could work. So I got a FT37-43 toroid, twisted a hank of 11 strands of #30 enamel wire, and wound a little 11-filiar transformer. I got 7 or 8 turns through the core. Hooked 10 of the strands in series, and declared the 11th to be the input.

Driving the input, I took an octal 74HC244 bus transceiver, wired 4 outputs to each of the input terminals of my transformer BTL-style, and drove it with two out of phase square waves. On the output, I put a voltage quadrupler (diagram out of a 62 ARRL handbook) made out of 1N914's and 0.1uF mylar caps.

The resulting waveforms were so "textbook" it was ridiculous. 5V square wave in, 50V square wave out. With no load at all there were some little jaggies, but with a load it looks exactly like a square wave over 3 or 4 octaves (circa 30kHz to 500kHz). Efficiency is very high, circa 90%, and I suspect that that the limit in efficiency is not the transformer but the 74HC244 driver.

It must be that what makes transmission line transformers so ubiquitous, is not just the textbook theory, but that they work so incredibly well in real life over a very broad frequency range exactly matching the textbook. I was sure that the textbook theory would break down visibly at the 11-filiar level. But it didn't, it held up. I am impressed.

Tim N3QE
4488 2010-03-29 04:58:57 w4zcb Re: Multifiliar to the max: 11-filiar!
Great experiment Tim, and great results. We all learned something.

W4ZCB


Last week I was playing around with step-up voltage converters (I know, not radio!) for Nixie tubes. Goal was to go from 5V to circa 200V, using just stuff I had lying around on the bench. I had a big bucket of FT37-43 ferrite toroids handy and tried various boost and transformer (non-multifiliar, just random wound) configurations. Typical turn ratios were things like 5:70 etc. They all kinda worked, but efficiency was rather low, and I could never really break 20 or 30% efficiency. I didn't have any power MOSFET's that could be driven well with 5V, and my bipolar transistors were rather wimpy too. But I'm not sure that all the inefficiency is the fault of my crappy semiconductors: the randomly wound transformers did not really work like I'd expect them too, and had resonances in the range I was trying to drive them, and coupling coefficient seemed to be much less than the "perfect 1" I had hoped for.

I've been using bifiliar, trifiliar transformers in radio projects for years. I thought, "well, the transmission line transformer theory is pretty neat, but it must break down at some point and not be so pretty". I kinda suspected that maybe above trifiliar it wouldn't work so good.

But as long as I was at the bench, rather than just think that it can't work, I decided to build something and see if it could work. So I got a FT37-43 toroid, twisted a hank of 11 strands of #30 enamel wire, and wound a little 11-filiar transformer. I got 7 or 8 turns through the core. Hooked 10 of the strands in series, and declared the 11th to be the input.

Driving the input, I took an octal 74HC244 bus transceiver, wired 4 outputs to each of the input terminals of my transformer BTL-style, and drove it with two out of phase square waves. On the output, I put a voltage quadrupler (diagram out of a 62 ARRL handbook) made out of 1N914's and 0.1uF mylar caps.

The resulting waveforms were so "textbook" it was ridiculous. 5V square wave in, 50V square wave out. With no load at all there were some little jaggies, but with a load it looks exactly like a square wave over 3 or 4 octaves (circa 30kHz to 500kHz). Efficiency is very high, circa 90%, and I suspect that that the limit in efficiency is not the transformer but the 74HC244 driver.

It must be that what makes transmission line transformers so ubiquitous, is not just the textbook theory, but that they work so incredibly well in real life over a very broad frequency range exactly matching the textbook. I was sure that the textbook theory would break down visibly at the 11-filiar level. But it didn't, it held up. I am impressed.

Tim N3QE



[Non-text portions of this message have been removed]
4490 2010-03-30 04:27:18 hanssummers2000 Re: Multifiliar to the max: 11-filiar!
4491 2010-03-30 07:18:45 Tim Re: Multifiliar to the max: 11-filiar!
I am not an expert in magnetics. But my experience with SMPS designers, is that they always talk about the air gap in the transformer or flyback, and getting this right is very fundamental to make a good SMPS rather than a crappy SMPS.

Certainly my attempts at using an FT37-43 core, without an air gap and without going multifiliar, produced crappy SMPS's. I do not blame that on the material but on my lack of understanding!

What's amazing is that the 11-filiar transformer worked so well, despite my lack of understanding :-). I mean, I know that multifiliar transmission line transformers have high coupling coefficients for broadband use. But I was honestly surprised that it worked so well. I'm not sure that a hank of 11 conductors twisted in my drill really qualifies as a "transmission line" in the sense that it has a known controlled impedance between any of the 55 pairings. But I think that the broadbandness of multifiliar transformers is what saves the day.

I should take some pictures of the scope traces. I am truly impressed that when I put a square wave in, a square wave came out, despite the 100:1 impedance transformation. I was sure that I was stretching transmissi
4492 2010-03-30 08:17:19 Leon Heller Re: Multifiliar to the max: 11-filiar!
4494 2010-03-30 16:11:31 Russell Shaw Re: Multifiliar to the max: 11-filiar!
Tim wrote:
> I am not an expert in magnetics. But my experience with SMPS designers, is
> that they always talk about the air gap in the transformer or flyback, and
> getting this right is very fundamental to make a good SMPS rather than a
> crappy SMPS.
>
> Certainly my attempts at using an FT37-43 core, without an air gap and
> without going multifiliar, produced crappy SMPS's. I do not blame that on the
> material but on my lack of understanding!
>
> What's amazing is that the 11-filiar transformer worked so well, despite my
> lack of understanding :-). I mean, I know that multifiliar transmission line
> transformers have high coupling coefficients for broadband use. But I was
> honestly surprised that it worked so well. I'm not sure that a hank of 11
> conductors twisted in my drill really qualifies as a "transmission line" in
> the sense that it has a known controlled impedance between any of the 55
> pairings. But I think that the broadbandness of multifiliar transformers is
> what saves the day.
>
> I should take some pictures of the scope traces. I am truly impressed that
> when I put a square wave in, a square wave came out, despite the 100:1
> impedance transformation. I was sure that I was stretching transmission line
> transformer theory beyond the breaking point. But it held up.
>
> I suspect that someone making class-E amplifiers and actually understanding
> magnetics and transformer design, could tell you how to use a computer power
> supply switching transformer at RF in a 160M amp with high efficiency. Or
> maybe he'd say why the air gap is wrong :-)

You were driving the transformer with a low-impedance source instead of
using it as an energy-storage flyback device. For this, no gap is needed.
The reason it works well is simply because the leakage inductance is low.