nanovna user post provides an interesting example for study #1

At https://groups.io/g/nanovna-users/message/9185 a user posted a measurement made with his nanovna of a length of coax with termination.

Above is his initial reported measurement of an approximate 350′ length of coax with a known good dummy load on the opposite end. 350′ is 106.7m. Whilst this chart is less value than a Smith chart rendering, understanding the nature of things allows us to infer the Smith chart.

A first point to be made is that the measurement is probably of 101 points in a single scan, so the accuracy of narrow features, the minima, is probably affected and less reliable.

This is a classic pattern of an approximately correct load on an approximately correct long low loss transmission line.

We might assume that at very low frequencies, there will be negligible impedance transformation and that the “good dummy load” would result in |s11|dB being a very small number… and a local minimum.

Lets take the local minimum around 29.2MHz, we can count minima, visible and missing to infer that at 29.2MHz, there will be 14 complete trips around the Smith chart from load to source, or 7λ, so 71.86m electrical length at 29.2MHz.

The electrical length is way less that the reported physical length of 107m, so does it have an internal fault? It is unlikely that an internal fault will report a generally low value of |s11|… so lets proceed on the assumption that approximately is more like unknown.

An important detail of the original plot is the slope of a line across the top of the maxima, it rises slowly from left to right. This hints that if all else is good, the line’s Zo is not quite 50+j0Ω.

Now if the load is 50+j0Ω (as claimed) and the VNA is properly calibrated, there are a number of practical questions that arise:

  1. how was the nanovna SMA connector transitioned to the large line under test; and
  2. how was the termination transitioned to the large line under test?

For further discussion of this issue see Antenna analyser – what if the device under test does not have a coax plug on it?.

Lets assume that in both cases an adapter of two short wires is used, and that each can be modelled as a short transmission line segment of higher Zo.

Let’s build a simulation of these assumptions in Simsmith and plot the magnitude of s11.

Above is the simulation circuit. The detail of the end sections T2 and T3 are guesses as discussed above. They are not a total guess, they are hinted by the |s11| value at 7MHz.

The characteristics of T1 were varied to obtain the same character of minima, and a similar character to the maxima.

Above is the tweaked response of |s11| for comparison with the original measurements. Again, do not focus on the value of the minimums as they are affected greatly by scan resolution.

So, my conclusion on the limited information given initially is that the line probably has an electrical length of around 73.5m, and nominal Zo around 48Ω and does not have an obvious serious internal fault. Apparent low Zo may be a result of drum storage of cable with foam dielectric and possible migration of the centre conductor.

The results raise questions about the test fixtures that need to be resolved to make more certain conclusions. Fixture issues are at the heart of many if not most measurement failures, and posters almost never show pics of their measurement setup.