Klopfenstein taper

Bit confusing. At first they say they're aiming for 20dB return loss or better to 40Ghz but then later they say they didn't quite achieve the 40dB RL they were designing for. :-/

I had been Spicing a tapered line as a string of series TLINE segments, but this at least gives me a hint that there is a better way than a linear taper of equal-length chunks.

It's still a multidimensional problem.

That's a valiant effort and certainly better than I could achieve, but Dremeling's really not suitable for the kind of frequencies you have been typically talking about these recently. I know you're primarily interested in very sharp dv/dt, but have you VNA'd it? I'll bet it looks horrid at anything over a couple of hundred Meg.

You could do the math. ;)

The optimal taper is a straightforward calculus of variations problem.

The optimal taper has to be self-similar, at least in the adiabatic () limit, so as long as it’s several times longer than wide, an exponential will be pretty close.

Cheers

Phil Hobbs

I'm not interested in s-parameters. My deliverable will be DC-coupled pulses.

Could a VNA tell me what the impedance is as a function of position? I guess it could if I transformed the s-params into a TDR, but I have the TDR directly now.

The bummer with that one is that it had to be thin dielectric to keep the dimensions reasonable. I had some 0.032" thick FR4 and sheared up a nice little piece and started dremeling, but it turned out to be 2oz copper, which needs a jackhammer rather than a Dremel to cut.

I ordered a bunch of 32 mil 1 oz copperclad from Amazon. Amazon has everything.

Half-ounce copper would be even better. Skin depth dominates at my frequencies.

I have prototyped various patch antennas for frequencies up to 3GHz by cutting copper away from FR4 with a scalpel and then if some needs to be put back using adhesive copper tape. Reflowing solder over the join makes it almost as good as new. I monitor with a VNA or tracking generator, directional coupler and spectrum analyser or power meter. Sometimes I implemented design equations from Balanis in Matlab, but it is usually possible to make a reasonable guess and take it from there by measurement. For real-time estimation of efficiency the easiest way is to move my hand over the antenna perhaps 0.5m away and look for variations in return loss from the reflected signal. This can be done more formally with a pair of identical antennas pointing at each other. I used a linear taper for impedance matching in one air-dielectric patch antenna design with good results. A more complex curved shape might have given wider bandwidth, but this was good enough for the requirement. John

I certainly did, at least for the trace impedances.

I don't want the overall product to be too big, so a really gradual taper isn't an option. And the taper will not be a featureless wedge; it will do stuff along the way.

I wish I had a really good 3d e/m simulator, one of those multiphysics things. But it's good to get away from mice and screens and go shear and drill and dremel and solder once in a while.

I might guess and make the taper nonlinear, with a few line segments. It would be harder to dremel, or lay out a PCB, with fancy curves.

This half a hobby project for me. I'm overflowing with interns for a while longer, until they go back to school, and I've delegated the most interesting stuff.

Related to this project, we did a bunch of rigid and kapton flex boards from JLC, based on your reviews. We were impressed with the quality and speed and price.

https://a.co/d/d6AOfIA

The TDR is probably the best investigative instrument out of the lot for what you're trying to do with this transformer. What sort of reflections are you seeing? Any chance you could post a screen shot of that board you made under test from your scope also showing its settings?

Good point. I need to get some of that stuff myself. When you Dremel, do you use the tool as-is or have you got one of those pen-like extensions on a flexible drive for finer control (I need to get one of those, too!)

Have you checked Tom Lee's book 'Planar Microwave Engineering' (IIRC). There should be a *lot* of info on this very subject in it.

I just hold mine, under my Mantis magnifier.

Use a carbide dental burr at high speed.

I have that somewhere. It's not useful to me... just pages of equations that would take hours or days to actually use.

Am 07.09.25 um 22:37 schrieb john larkin:

Should we talk about that in public?

ROTFL, Gerhard

Here's the TDR. The giant negative spike is at the end of the taper.

It looks a bit nicer if I add vias between the top and bottom grounds. A real PCB would have a hundred stitching vias.

You should be able to improve on that quite markedly I'd have thought. I'm just wondering how much those rough edges are contributing to the general untidiness. I mess around with this kind of thing very regularly as a hobby. It's fun and challenging (but time consuming!) to re-do boards over and over again eking-out improvements (hopefully but not always) with every fresh attempt. Any way you could somehow 'polish' away the rough edges?

Yeah, there *is* a lot of that in there for sure. However, there is

Am 07.09.25 um 23:23 schrieb Gerhard Hoffmann: ....

BTW is there an English equivalent to the German word "Kopfkino" i.e "cinema that is triggered inside one's head"?

Not really. 'Daydream' would be the about the closest I can come up with right now.

I am reminded of the motto of The National Lampoon:

"Good Taste Isn't Everything"