s-params help

There are other less inaccurate more expensive VNAs around these days. And they can be had for a fraction of what you had to pay for a HP or R&S VNA 30 years ago. A relatively small expense could be justified on this occasion, I'd have thought.

Possibly you might Fourier-transform a time domain measurement. You'd still have to separate the forward from the reverse signal, since TDRs do not usually do that. Also, TDRs take what's basically a step response, while S-paramaters are the Fourier transforms of impulse responses.

All that can be dealt with, but you'll be shocked about the poor S/N ratio of time domain data as compared to a frequency-domain measurement. VNAs can do so much better,

Jeroen Belleman

I just want someone to do this for me.

What frequency range? I might be able to assist here.

$880 wow! I can see why you would want to get a piece of that action. I'm afraid my 2 VNAs top out at 1.3 and 3Ghz respectively. They're both full-blown lab jobs made by HP but the frequency range is virtually DC by today's standards. If you want to look at how switches perform above 3Ghz then there's not much I can do, I'm afraid. I was thinking about buying a third VNA to go up to 6Ghz, but I think that would be a waste of money as I'm really not interested in anything above 1. I don't see how $1 telecom relays can *possibly* provide that kind of bandwidth. Are you sure of your testing methodology?

Yes, it'll do that, but not to commercial standard accuracy, of course. I bought one a while ago just out of curiosity to compare it with the lab ones and was quite frankly *amazed* that they've done all that for under a hundred bucks. It's not a serious instrument, but a huge shot across the bows of the traditional manufacturers nevertheless. They need to up their game or they won't survive.

I know you know what you're doing so I won't argue on that point. However, I personally wouldn't dream of using a TDR for this purpose. I would sweep the switch across say 100khz through 3Ghz with a VNA and if any impedence discontinuity showed up, *then* use the TDR to find where it is.

Looks like a lot of people who make RF switches just specify VSWR and isolation. That's probably all I'd need to measure.

I was doing some measurements late last year of microstrip, experimenting with using the NanoVNA-H4 port extension feature to flatten the S21 phase response and using that to estimate the relative permittivity of the microstrip substrate, and comparing against a university lab Keysight 5080.

In that role the NanoVNA looks pretty good up to about 800 MHz but at 1 GHz and above it starts to be pretty suspect and everything above 1.5 is complete nonsense. Haven't tried the NanoVNA 2, the -H4 has about 40dB dynamic range on a good day at 1 GHz and the version 2 claims 90.

The Keysight is IIRC specced at about 130 or something there but a 6 GHz class 5080 cost what, $100,000 new?

My main HP VNA (can't recall the model number off hand) did cost around 100k - and that was over 30 years ago! I've no idea what the equivalent would be today; no idea whatsoever. It's a fascinating hobby. You can characterize all sorts of things with a VNA: cables, adaptors, filters, dummy loads, microstrip, transformers, switches. Basically anything you might find in a signal path of a transmission line. It became something of an obsession in my case. Fortunately there are worse things one can get obsessed over and this is a totally harmless pursuit.

Used examples of the 6 GHz machine (they go up to 50!) are going for about 30k USD so probably not far off, they're a few year old now.

They seem to run Windows 7. Crash to Windows 7, too!

I can use the NanoVNA 1.0 to tell you from a ~30 cm piece of microstrip on FR4, that FR4 has a relative permittivity somewhere between 3 and 5!

Not exactly earth-shaking information but to do better in theory with that box you'd need a longer strip, but then errors from other sources start coming in..

One can buy cheap multilayer PCBs these days, and could pack many experiments onto one board.

I spent a bit of time playing with pcb layouts for edge-launch SMA connectors. I did that with ATLC and TDR. There are some pretty good $2 connectors.

This is the main one I have:

It is extendable up to 6Ghz but there's no point since both the S parameter test sets I have are limited to 3. The fixture on the front is for testing RF transistors, which was the last thing I used it for. If you want me to check out your switch, I'm happy to help. It's just that the postage to Yurp from SF will be a bit of a killer I'd imagine. If you plan on doing much of this kind of thing you're going to have to bite the bullet and splash out on one to keep, because doing the necessary tweaks to get your DUT in spec will necessitate repeated testing which is impractical if you don't have a VNA to hand and have to send the thing to someone else every single time!