transmission line transformers on a board

Does that stainless hardware skew the inductance? I usually see nylon used in these instances.

Cheer

Apparently not. It's an ungapped pot core, so there will be little field in the hole. A screw had no significant effect on the inductance. It's just a pulse transformer, so a little inductance change or loss wouldn't matter.

What's weird about it? We did the same thing back in 1988, and our transmis sion lines were twisted pairs, so we could wind them on a standard pinned f ormer, and mount the whole transformer as through-the board component. I th ink you can now get surface mount formers, but the mechanical strength isn' t as good.

We used the transformers to galvanically isolate Taxichip serial links (whi ch are obsolete now, but were remarkably quick at the time).

On a sunny day (Mon, 02 Oct 2017 16:21:24 -0700) it happened John Larkin wrote in :

Nice, looks like pot-cores? I have of those small coax connectors in my drone and its remote control, always have problems pushing those on... but seems to work OK so far (2.4 and 5.8 GHz).

Mini-Circuits says they make 260 types of inexpensive surface-mount transmission-line transformers and baluns.

For example, the T1-1T+ a 1:1 type is $3.80 (100) and has an 80kHz to 200MHz bandwidth, the T1-6T+ is more expensive with a 15kHz to 300MHz bandwidth. Both types feature a center-tap, I'm not quite sure how they do that.

However, consider placing a wave spring washer (or a similar flexible element) under the screw head. The pot cores are very brittle and the temperature expansion coefficients (screw vs. core) are not matched.

A more flexible mounting would provide both higher reliability (the core will not crack) and better temperature stability of the trans- former (no gap will accidentally open) compared to a rigid mounting.

Regards, Dimitrij

We do use a nylon nut, which has a tight fit on the screw but can't be tightened enough to damage the core. Production may elect to use superglue or hot-melt to tack the bottom half of the core to the PCB, and the bobbin inside that, to hold things steady as the turns are wound. It can get a little messy, winding the stiff coax on the bobbin with everything flopping around.

Yes.

Those little UMCC things are great. Several people make compatible surface-mount pick-and-place connectors, and stock cable assemblies of various lengths, an enormous labor saving. They are a bit hard to mate and un-mate by hand, but seem reliable once mated. I think there may be tools to help.

The mini-circuits things didn't work well in my high-power pulse application. My pulses are still limited by saturation of that giant pot core.

I didn't invent the txline transformer, but I did have the idea to use the UMCC connectors and stock cable assemblies for the windings. We used to do things like this:

which didn't make me friends down in production. The 2:1 version was even less well received.

Why don't you rotate the top halves of the cores, so that they align with the bottom halves? Would that allow the coax to move around too much?

Things are tidier when the slots don't align.

Makes sense. You mentioned earlier the difficulty of winding the coax on the core. How about winding it on a rod of suitable diameter then sliding it on the core?

It's wound on a plastic bobbin. Maybe the bobbin could be wound outside and then dropped into the lower core half; I haven't tried that. I did try winding one without the bobin; that just didn't work. The coax is stiff and springy, which makes it tricky to handle.

But my production people are awfully good at assembling things. I bet that a couple of them, with a little practise, will find it quick and easy. They can figure out the best way to do this. The old way, stripping and soldering coax, was nasty.

I can imagine other uses for the UMCC connectors and stock cables, like a pulse inverter maybe.

Ah, yes. I forgot. I'll have to go check them out again.

Thanks.

These appear to be standard transmission line transformers on a ferrite core, with bandwidths of 5Hz to 18GHz. So my idea was correct. It does work. But obviously, Minicircuits is the way to go.

Incidentally, one big difference between your pulse transformer and most of the references is they use twisted pair, which would have very poor voltage standoff capability. Only one example used coax, and that was more for phase matching.

Your use of coax for coupling to high voltage loads is unique as far as I can tell.

Good job.

Core is mostly irrelevant at very high frequency. Easy enough to understand. Start with the ideal (SPICE style) model of the transmission line: infinite common mode impedance. Connect a magnetizing inductance between the ends, and there you have it. The impedance is simply the impedance of the winding on the core.

You can make a perfectly serviceable TLT with no core at all; the impedance is simply that of the air core inductor you've made. :)

(You might ask: connect HOW between ends? Well, depends on the TL cross section -- if twisted pair, say, then split it equally between both lines. If coax, put it on just the shield.)

Works fine, but you need to get the coupling balanced properly. Just making a TLT is no magic bullet!

Tim

Minicircuits does a good job. What JL has done is completely different.

Sort of. The characteristic that the wider bandwidth, low frequency models have, that coupling gets terrible at high frequency, is characteristic of nothing-special transformers.

Here's mine (which I'd already been inside, because naturally I got the wrong one, 75 instead of 50 ohm, but it's only different by the termination resistor and connectors; that's two 100 ohms stacked in parallel). You can see the transformers are basically small cores with a handful of turns on them, not arranged in any particularly fancy way.

The best method I think, will be TLTs with shields. Suppose you have a ground plane wrapped around a core, so it makes a cylinder with a gap. Place windings on the inside and outside of this cylinder. This gets the CMRR up enough to be useful.

Voltage? Hookup wires twisted together are fine to 600 or 1200V. :)

I'm fond of using CAT5 pair, or strips of ribbon cable wired up appropriately, for various applications. Such things are often rated for only 30V, or 100V, but that's fine as long as you aren't plugging it into the wall.

Tim

Mine does a good job too!