transmission line transformers on a board

? Do

Huh, I've always wanted low capacitance coax. (I guess that's normal.)

George H.

You came up with a neat, practical construction for a handy circuit element. Cool (and thanks for sharing).

Cheers, James

But expensive.

Cool, but scarcely novel. If he'd put a price on the coax connectors, it might have been more interesting, but probably less attractive.

Are you suggesting that I wind my 50-ohm transmission-line transformer from that?

U.FL connectors are like $0.10, even buying in small numbers. You can get them pre-fitted to jumper leads in various lengths for a buck.

JL, Please *do* try to keep up. He's answering my question about whether such stuff is commercially manufactured, as is amply clear in the quoted context.

The UMCCs that I used are about 35 cents by the short reel, and the cable asssembly is about $2.50. I never expect to make many of these, maybe 100 channels a year.

The pot core pair is something we had in stock, ungapped for a forward converter supply. The pair costs $1.80. It's probably not optimum magnetically, but the breadboard worked fine so I didn't need to think about that any more.

All that, building and testing and documenting, took an afternoon. Our resident magnetics expert Sloman spent a decade or so on his custom-magnetics 2-transistor oscillator and, at last report, hasn't yet got it to oscillate.

I sure wish you guys would stay on topic.

The coaxial transmission line alone, with no core, acts like a transformer for fast signals. Fast here means signals faster than the prop delay of the cable. You can Spice that, which is fun. For the fast signals, the inner and outer currents inside the coax cancel, so there's no field into the core.

For slow signals, the system acts just like a conventional transformer with a magnetic core. With luck, the fast and slow mechanisms overlap nicely and you get a clean pulse response.

I'm making 50 volt pulses into a 50 ohm load, 1 amp, with about 2 ns edges, and transformer saturation limits my pulse widths, even with that big hunk of ferrite. A gapped core might be better, but I used what we have and it's good enough.

A gapped core would perform strictly worse.

Tim

Probably so. To get edge speed, I need to push small fets hard. So I have a lot of Rds-on drop, which I can overcome with more supply voltage. But magnetizing current then makes my pulses look droopey. So I want all the L that I can get, right up until the pulse is zapped by core saturation.

Complex stuff.

What's with the gold pastie over the works?

Cheers

Phil Hobbs

Hiding the private bits, no fig leaves were available.

GH

Trade secrets.

Only gold leaves?

It oscillates fine in LTSpice, and it's not different enough from stuff that does work in the real world to make building a real example an urgent issue.

And the current version doesn't have any custom magnetics. There 's a ferrite bead or two on the circuit diagram, but they are purely precautionary.

If there was any prospect of somebody buying an example of the circuit, I'd probably get more enthusiastic.

On a sunny day (Fri, 06 Oct 2017 08:55:36 -0700) it happened John Larkin wrote in :

I like the breadboard, but what is under that almost square piece of copper? Is it forming a transmission line?

I blanked out the new pulse driver circuit in that picture, paintbrush in Irfanview. I can't give away some of our tricks.

On a sunny day (Sat, 07 Oct 2017 07:32:34 -0700) it happened John Larkin wrote in :

OK, yes in deep reflection I realized it must be a pulse driver, and what came to me was that short transmission line one transistor circuit we discussed here years ago...