Why am I getting sign inversion with s2rlgc?

Hi,

I am using the s2rlgc function to calculate attenuation and phase constants across a frequency range from a .s2p file I measured on a 50-ft coaxial cable from 2 to 30 MHz with a calibrated Network Analyzers . When I then apply the calculated attenuation and phase constants to determine s-parameters of a cable of a longer length, such as 142-ft, in the plots below I get a sign inversion on the s21 imaginary component around 9 and 28 MHz, as well as, a slope inversion on the s21 real component at the same frequencies. Which then cause a jump on the smith chart below.

Is this a known issue for s2rlgc? Or, does anyone have an explanation as to why this is occuring? Or, a way to fix it? Is the length of cable measured the issue? Should a shorter cable be measured?

Thanks for any help!

[David Goodmanson]

Hello Markus,

For the 50 ft measurement, what kind of coax cable are you using? The coax speed factor comes out somewhere around .92, which is pretty fast. Aside from that, the 50 ft measurement makes sense and I think you will probably come out with a good result if you do a 142 ft measurement. Which leaves, how are you doing the analytical calculation for the 142 ft cable?

[Marcus]

The s2rlgc function gives the attenuation and phase constants. Then, for a cable of length z, with ports terminated with matched loads, S21= e^(-γz)=e^(-αz) e^(-jβz).

Answer 1

[Mark]

When s2rlgc is given sparam data at multiple frequences, it uses "unwrap" to track the phase and try to stay on the correct branch of a complex sqrt. Instead of handing s2rlgc single-frequency slices, you should build the 3d array of sparam data and give it s2rlgc once.

Best,

Mark