Re: My TDR Evaluation of the SG503 012-0482-00 Cable has been REVISED


Reginald Beardsley
 

Thank you all for the compliments. This is slightly off topic, but I think the context makes it useful.

My major project for the next 2-3 years is FOSS DSO FW for COTS Zynq and Cyclone V based instruments. I bought my 11801 so I could measure bit skew across various FIR and IIR filter topologies, It's a long slog getting set up, but I'm getting there. Not helped though by new toys like the 8566B which came yesterday. Or the 7x14 Chinese mini-lathe I bought to make go/no-go gauges for checking RF connectors. Teaching that to hold 0.0001" tolerances will be a chore. Doable, but time consuming. And first I have to fix the damage UPS did.

I started to work on building a rack from some scrap angle which had been hastily dumped in the drive to my shop. After carrying the 2nd load to the concrete part of the drive to wash it off I reached down to find a ft long copperhead making threatening moves not far from my hand. Fortunately it was not close enough to strike, so I proceeded to kill it. I don't mind snakes, but poisonous ones close to houses are bad news.

I've bought an FT-991A and an FT-818ND and hope eventually to get around to getting licensed and an antenna put up. But my real passion is T&M.

My frustration 30 years ago trying to do RF work without basic instruments or any chance of access to them is still palpable. So my biggest interest is cheap alternatives to expensive gear. I'll turn 66 in 2 weeks. In the last 18 months or so I've spent about $15K on old HP and Tek gear. I now have a lab that would have cost well over $200K 30 years ago. I feel incredibly lucky to have it. My sole bad experience was an SD-32 which came with an SMA F-F adapter soldered to the 2.4 mm input. I got full credit on the return today.

I'd much rather be working in the oil industry, but the work I did is only economic if oil is over $100/bbl. So I have no hope of ever working again. But if life hands you a lemon, it's time to look for lemonade recipes. So time to buy the bench of my dreams and get serious about electronics.

I'd thought of doing the TDR VNA program before I bought the 11801, but once I got that I was absolutely bewitched. The software is trivial to write, but there is a huge amount of stuff you can do just looking at the TDR trace. I'm reluctant to make pronouncements until I've actually verified the derivations. That will be painful. I was looking at a text by Robinson and Treitel, the stars of Norbert Wiener's GAG and rather befuddled by it. The depressing part is all the way through I checked off the equations as I verified them. So I *really* did understand that notation once upon a time. But over time you start using other notation for various reasons. The concepts don't change, but it's hard to double check something if the notation is not familiar.

Joel Dunsmore's book on VNA measurements impressed upon me how foreign the time domain is to many EEs when working in RF.

You really can't go past the BW of a DSO the way you could with an analog instrument. They've already gone too far. Keysight is selling 750 MHz DSOs as 1 GHz DSOs as is pretty much everyone else so far as I have seen. That and all the bugs is the motivation for the DSO FW project.

Spectrum flattening of the system response is the cornerstone of the work by GAG, the Wiener inverse filter, aka prediction error filter. Sadly in seismic processing shops it is often done improperly. When I started with Amoco I got concerned that what we were doing was not correct. So I called Sven Treitel at the Tulsa research lab and asked him. He said that my understanding was correct. What we were doing was wrong. And remarked, "Sometimes the prophet is not heeded in his own country."

For the sake of the wider audience I'm going to do the posts on EEVblog, but this and the HPAK list are much more congenial and more the sort of people I'm accustomed to working with. So I'd like you to hold my feet to the fire if I explain something badly.

I've worked for 7 oil companies, 6 of which no longer exist. I swear it's not my fault. Every place I worked it took about 6 months to learn the local jargon. When you're switching from time to frequency with each sentence it gets messy and everyone develops a local shorthand to save time.

The formulation for going from TDR to VNA is:

Collect multiple reflection events from an impulsive or square wave source

Window the reflection segments

Phase shift each window so the phases match

Sum the time domain traces

Fourier transform

Compute magnitude and phase

QED sort of...

You have to account for losses due to reflections from previous interfaces. Dunsmore calls this "the masking effect". There are a number of other details which one can run afoul of. As I have the 8753B and 11801 to compare the results from a 200 MHz Instek 8 bit DSO and a 100 MHz Owon 12 bit DSO using Leo's signal sources I'll be able to do good demonstrations of the finer details. There are a lot of retired EEs with top grade skills on this and the HPAK list. You are my referees of choice.

I'm not big on cyber space. I much preferred having the Stanford PhD across the hall wander in my office and start scribbling on my white board. Not because they were better company. I just like face to face better. But I live in rural Arkansas. There aren't many people like that to be found in a town of 7,000. So yu are all very important to me.

So again, thank you and...

Have Fun!
Reg

Join TekScopes@groups.io to automatically receive all group messages.