transmission line transformers on a board

I don't need HV isolation in most applications. The difference between coax and twisted pair is that coax is a 50 ohm line, and the impedance of a twisted pair is generally over 100. A transformer wound with twisted pair is a low-leakage-inductance transformer, but it's not a

Twisted pair windings will also have all sorts of interesting capacitive couplings that coax doesn't.

Enamelled copper is good for 500V. You can get double enamelled wire if you want to go higher. Heavy coax can cope with higher voltages.

Not exactly.

Ghiggino, K.P., Phillips, D., and Sloman, A.W. "Nanosecond pulse stretcher",Journal of Physics E: Scientific Instruments, 12, 686-687 (1979).

uses uses two turns of Filotx VMTX 75R sub-minature coax. The voltage being stood off was only a volt or so. Since it was a lab one-off I wasn't motivated to make a twisted pair equivalent (as I was in 1988).

John Larkin is congratulating himself for having re-invented the wheel (which he does rather too often).

My paper cited Matick R.E. "Transmission -line pulse transformers - theory and applications" from the Proceeding of the IEEE voume 56, pages 47-62 (1968) - so the idea goes back a bit.

Of self-advertisement?

So get low impedance twist, duh. :-)

Interleaved (quadrupole) twist is even "better", typically 25 ohms. Do that with 2kV wire and you're set for anything.

Ironically, such wire is available because audiophiles think it's better than twisted pair. Well, they're not wrong, for once.

The Navy uses the concept, extended to absurdity, in their railgun. Ever wonder why those huge cables don't fly apart when the thing fires? (Sure they move, because the gun moves, but not because of their own current flow.) _Kickless cables_ aren't cheap, but they're darn useful when you need extremely low impedance transmission line.

What's that supposed to mean? Transmission line is transmission line.

Tim

You can even get the "triple insulated" variety, which is mains rated. Darn useful stuff, don't need tape between layers. You can do very useful, and otherwise naughty, things like bifilar windings with mains isolation.

Reminds me of an audiophile application: McIntosh's most famous tube amplifier used a split-phase configuration, where half the output transformer primary is in the plate circuit, and half in the cathode. The poor thing will oscillate unless bifilar winding is used. But there goes your 500V rating in one fell swoop. Unsurprisingly, they're notorious for failing with shorted transformers. Double enamelled would've been fantastic there. (Restorers may well use it, I don't know.)

Tim

You can also buy 75R coax and 95R coax.

Twaddle.

Most designs I've seen (and I've searched quite a bit) seem to be based on Joel Dunsmore's design. The scanned images are available if you search, but I made this handy PDF:

The key thing seems to be getting good coupling in the cable at the highest frequency of operation, then using ferrites of increasing mu to extend the low-frequency response. It's not very hard to get 30KHz-3GHz, as long as you can calibrate out any discrepancies in your instrument.

Clifford Heath.

In other words, twist and strip and wind and solder multiple pairs in parallel. The cost and messiness just jumped up, and the capacitive couplings just got way more interesting.

Audiophiles are mostly idiots, and don't need clean picosecond edges.

That's millisecond stuff. I do picosecond stuff.

Good grief.

Enameled wire is a pain to strip and solder. Heavy formvar is seriously nasty to work with. Triple insulated would be a nightmare. With a twisted pair (or even more than two wires!) you have to keep track of the polarities at the ends somehow.

We buy cheap surface-mount tape-feed UMCC connectors from Digikey. We pick-and-place them with all the other parts. Then we buy a cheap connectorized cable assembly, stock from Digikey, plug in one end, twirl around the bobbin, and plug in the other end.

I see a pattern here!

Thanks. Very nice. It's quite lossy (resistor bridge), but if you're doing small signal measurements, that's fine!

Resistor bridges are a standard among hams as well, though that gets tricky when you're trying to tune up a big PA. :)

Tim

Why are you /making/ transformers, anyway? Buy them from a manufacturer. Even custom samples are pretty cheap. Much cheaper than your production people at the same task, I would bet.

Audiophiles like it so it must be bad, huh?

Guess I better go delete all that favorable test data I measured when I was doing induction heating.

Well, microsecond.

So cables the size of shipping containers can't be scaled down proportionally in size and time. Hmm. Whatever you say. You must be right, people buy things from you.

I'm just guessing you haven't had your coffee yet.

Tim

Nobody makes what I need here. It's way easier, and cheaper, and faster, to do what I'm doing here, putting about $4 worth of parts on a board, pick-and-place with a little screwdriver assembly.

Bad and expensive and imaginary.

What frequency was that? What amplitude precision? Did you separate and strip and solder many strands of wire in your coils? Did you have to keep track of each strand, to keep them in pairs?

Are you still doing it?

And I want to make money if I can. The Navy doesn't have to do that.

I had to get up at 6AM to compose a nastygram to a customer before that post. But I did absorb a big cup of Peets first.

I suggested an easy way to make and terminate cheap picosecond-grade surface-mount transmission-line transformers. And the usual gang of idiots instantly decide to whine rather than think.

Coax pulse inverter:

Dropping some ferrites over the cables extends the lf response.

Hey, if ever I needed a 25 ohm TLT I think I'd now build it like JL but with two lengths of 50R coax paralleled.

piglet

Heh, I remember "inventing" that concept for routing audio power amp wiring, and made my own. I wasn't aware that it was commercially available! I hadn't thought about using it instead of pair or co-ax for a TLT.

Where? Also in small sizes? Where can I see impedance data, or is that common to this geometry?

Clifford Heath.

Since a pair is usually around 100 ohms, a 25 ohm bundle would probably need to be four pairs in parallel, which is 8 wires to terminate on each end.

Most coax has a fair amount of inductance, which could barely matter in driving a speaker. I've made super-low-L coax to drive NMR gradient coils, a fat conductor with very thin insulation, then a tight outer braid, then shrink tubing.

Yup. If you keep the fields of each pair separate, you need a lot.

You can cut that about in half by overlapping the fields in a quadrupole configuration.

The trade name is "star quad". Canare and Belden make it, apparently. Or maybe Canare is a rebrand, I don't know.

Similarly, ribbon cable gets a lower impedance, more quickly, if you alternate plus and minus, rather than using pairs with undriven wires separating them.

I've made super-low-LF-loss coax by wrapping Litz braid over a Litz cable.

Tim

Kapton flex PCB constructions are sometimes used to drive diode lasers. Zo of an ohm or two are feasible.

Example:

Probably, most 4-conductor cable is of this configuration anyway, it's just not sold as such (I suppose the order of colors might differ), and not tested in the quadrupole configuration.

The relatively low impedance is typical, yes. I think I measured 25 ohms for enameled wire in that configuration, of course you'll have an even harder time finding that on a spool. Still, 35-40 ohms for the cable is not bad.

Tim

Transmission lines don't have external fields. My twisted pair transmission line transformer used green and red enamelled wire in the twisted pair. Th e different colours in the enamel weren't just decorative, but the differen t kinds of enamel both worked fine in my application, and made life easier for production.

that

But they might be able to get them anyway.

Millisecond stuff has been known to deliver picosecond performance, when us ed intelligently. You may be right to avoid it. Others might do better.