¿ªÔÆÌåÓý

ctrl + shift + ? for shortcuts
© 2025 Groups.io

Advice sought on specific measurement.


 

I want to use my NanoVNA to determine if the cables I use in my lab are good enough, like the correct Z, no sharp bends or crushed dielectric or bad connectors, and so on.

My first vague idea was to

1. Calibrate solt for a plane at the end of a short cable.
2. Connect the cable under test, the CUT, to the end of the short cable in 1.
3. Connect something, (short, open, load?) at the end of the CUT.
And then know what a perfect CUT with Zc= the load test R would look like on the NanoVNA display.
And that I don't know.

Can you advice on this?


 

For evaluating cables I normally just do a S21 log measurement, which is somewhat of a "catch all" test for losses due to impedance problems, cable defects etc. I only calibrate with a "through" measurement (with a union connector in place where the CUT will eventually be placed). Then I make the S21 log measurement with the cable being tested in place (where the union connector was during the through calibration). The frequency range depends on the needs...but higher frequencies near the nanoVNA upper limit will certainly be more sensitive in terms of detecting differences in cable performance.


 

Hans,
The tdr function, available standalone in the recent firmware upgrades and connected to a pc via software program, is good for evaluating cables. It can give you info about discontinuities, cable length and impedance bumps. I measured a 25 ft length of RG58 cable using the tdr option of Rune's nanoVNA Saver program and it found a short at 5 ft from the end of one of the connectors. Saved me the grief of installing a defective cable in my ham shack.


 

I'm playing with rather more exotic HW, a Tek 11801 & SD-24 TDR head which is a ~26 ps rise time TDR evaluating the quality of Chinese RF connectors. But the principles are the same and I'm very interested in using the nanoVNA for such things. I'm a retired seismic research guy, so I've been having a blast. Most fun I've ever had with a piece of test gear and all I'm doing is playing.

I look at the time domain S11 and the time domain S21 response. The S11 TDR shows any discontinuities and the rise time of the S21 TDR gives me an estimate of the frequency limit. In my case I'm putting the connectors between a couple of pieces of RG402 terminated with SMA-M connectors.

If you do an SOLT on a short known high quality cable, then substitute the CUT and repeat the SOLT at the end of the CUT you should get a very accurate picture of cable and connector quality. In the frequency domain there should be a linear phase shift difference between the reference cable and the CUT. The ratio of the S21 values should give you the loss.

I can't keep up with who's done what on this topic in the nanoVNA FW & SW, but the MATLAB clone, Octave is an ideal tool for taking data via the nanoVNA console and analyzing it. The frequency spacing determines how long a cable you can test. 1 MHz spacing will give you the ability to test to 100 m. The resolution is dependent upon the maximum frequency. So 901 points from 1 to 900 MHz is probably a good choice of sweep parameters. You'll need to use the reference cable results to compensate for amplitude errors in the nanoVNA. If you pad lots of zeros on to the end of the frequency domain measurements you will get very fine sampling in the time domain.

I recommend doing this with bare uncorrected data at least once using Octave just for the educational experience. It's actually *very* simple. And there are plenty of people here able to help out.

I used the 11801/SD-24 to test another APC-7/N-F adapter yesterday from the same seller as in the photos here:



Comments on second sample:



I've still got tests of BNCs and other stuff to do with the 11801 and then I'll be duplicating some of the tests on the nanoVNA and my 8753B.

Have Fun!
Reg


 

I did something similar a couple of years ago when type N "connector savers" were offered from China. These are male to female adapters. The first batch wasn't very useful about about 1.5 GHz. The second batch, from the same vendor were much better, working well to about 4 GHz. No idea why that was. They looked identical. I use APC-7 connectors, so your research is interesting to me.
Stuart K6YAZLos Angeles, USA

-----Original Message-----
From: Reginald Beardsley via Groups.Io <pulaskite@...>
To: nanovna-users <[email protected]>
Sent: Sun, Sep 22, 2019 8:25 am
Subject: Re: [nanovna-users] Advice sought on specific measurement.

I'm playing with rather more exotic HW, a Tek 11801 & SD-24 TDR head which is a ~26 ps rise time TDR evaluating the quality of Chinese RF connectors.? But the principles are the same and I'm very interested in using the nanoVNA for such things.? I'm a retired seismic research guy, so I've been having a blast. Most fun I've ever had with a piece of test gear and all I'm doing is playing.

I look at the time domain S11 and the time domain S21 response.? The S11 TDR shows any discontinuities and the rise time of the S21 TDR gives me an estimate of the frequency limit.? In my case I'm putting the connectors between a couple of pieces of RG402 terminated with SMA-M connectors.

If you do an SOLT on a short known high quality cable, then substitute the CUT and repeat the SOLT at the end of the CUT you should get a very accurate picture of cable and connector quality.? In the frequency domain there should be a linear phase shift difference between the reference cable and the CUT.? The ratio of the S21 values should give you the loss.

I can't keep up with who's done what on this topic in the nanoVNA FW & SW, but the MATLAB clone, Octave is an ideal tool for taking data via the nanoVNA console and analyzing it.? The frequency spacing determines how long a cable you can test.? 1 MHz spacing will give you the ability to test to 100 m.? The resolution is dependent upon the maximum frequency. So 901 points from 1 to 900 MHz is probably a good choice of sweep parameters.? You'll need to use the reference cable results to compensate for amplitude errors in the nanoVNA.? If you pad lots of zeros on to the end of the frequency domain measurements you will get very fine sampling in the time domain.

I recommend doing this with bare uncorrected data at least once using Octave just for the educational experience.? It's actually *very* simple.? And there are plenty of people here able to help out.

I used? the 11801/SD-24 to test? another APC-7/N-F adapter yesterday from the same seller as in the photos here:



Comments on second sample:



I've still got tests of BNCs and other stuff to do with the 11801 and then I'll be duplicating some of the tests on the nanoVNA and my 8753B.

Have Fun!
Reg