I keep being offered Youtube videos showing how to measure velocity factor of coaxial line.
I did indulge one this afternoon. The author explains that measuring s21 phase is the basis of his measurement.
The DUT for the demonstration is 3.76m of coax, no mention of where it was measured from and to.
No mention of the calibration details, so we might assume that a short jumper was used to connect Port 1 to Port 2 for the through test, perhaps the very one shown in the pic below.
Above is the test jig, one end of the coax (UHF plug) attaches to a UHF-F to SMA-M adapter which is attached to the VNA. The other end of the coax appears to connect via a UHF-M to a UHF-F to SMA-F adapter, and the 100mm long jumper cable.
In this pic, he is measuring the length of the coax jumper so as to add it to the length of the coax for the purpose of determining the DUT physical length and comparing that to the electrical length worked out from s21 phase.
But there is no mention of the cascade of adapters and plugs at both ends of the coax.
So what happens with the simplest calibration of the NanoVNA?
Above, last step of the calibration is the through step configured as shown above. When the calibration is complete, a measurement of the calibration through fixture should show zero phase at all frequencies, the display above shows the phase trace on the zero line, and a bit hard to read, phase at the marker is 0.01° (45MHz).
So, the time delay of the coax jumper shown above is actually part of the calibrated VNA, and unless you take steps to remove that time from the operating state (using e-delay), delay is zero from Port 1 connector to the end of the coax jumper connected to the Port 1 connector.
When you take one end of the jumper off and insert the DUT coax cable with adapters in between, you are measuring the phase delay of the DUT and it’s adapters.
So, in the video, ignoring the adapters and incorrectly adding 100mm for the jumper miraculously results in a measurement that reconciles with 0.5% of the datasheet (though it is doubtful that the adapter lengths were near 100mm, so unlikely to offset the false inclusion of the jumper length)! Another factor is that the adapters may not be, and are probably not the same velocity factor as the DUT.
This was obviously a foamed dielectric cable, and the manufacturing tolerance is not as good as solid dielectric cable… so for critical application, accurate measurement of velocity factor may be wise. The operative word was “accurate”.
I note that lots of people give their measurements of solid PE dielectric cables and come up with values well away from 0.67. Unless the cable is defective, it is almost certain their measurement technique is deeply flawed… whether it is not counting adapters, effects of UHF Zo, failure to calibrate properly or fundamental misunderstanding of the VNA (as above)… and they would be better relying upon the datasheet.
Be the devil’s advocate of your own measurements.
Question everything!
On a positive / solution note, if you want to measure velocity factor to better than 1% accuracy, see Velocity factor solver.