Tuning electrical line length using phase of measured s21 – nanoVNA

The nanoVNA has put a quite capable tool in the hands of many hams who do not (yet) understand transmission lines.

A recent online posting asked why phase of s21 of a desired 40° section of 75Ω matching / phasing line did not reconcile with other estimates of its electrical length.


Let’s firstly review the meaning of s21.

Considering the two port network above, \(s_{21}=\frac{b_2}{a_1}\) where a and b are the voltages associated with incident and reflected travelling wave components. Implicit in the meaning of s parameters are the port reference impedances which in the case of the nanoVNA are nominally 50+j0Ω.

So, if b1 and a2 are both zero, s21 gives the propagation coefficient for the network at the frequency of measurement, and the phase of s21 is the propagation phase change for the network and should imply the electrical length of a transmission line if that was the subject network.

So this suggests that if the line Zo is not 50+j0Ω, there may be non-zero b1 and / or a2, and the phase of s21 will not reconcile with the phase length of the line.

Let’s consider a special case where the network is a lossless half wave of line (of any Zo). a2=0 by virtue of the 50+j0Ω load provided by the VNA port, and b1=0 by virtue of transformation of that 50+j0Ω to 50+j0Ω looking into the transmission line… so we should expect that phase of s21 will be -180°.

An exercise for the reader is to consider the special cases of odd and even number of electrical quarter waves of lossless line.

Line loss changes things a little for those special line lengths, there is no longer ideal impedance transformation and there will be some small difference between phase of s21 and electrical length.


The OP required 40° at 1.83MHz of 75Ω line. Lets work the case for Belden 9212, a copper RG11A/U coax with manufacturer specs from1-1000MHz.

Using TLLC to solve for s21 (wrt 50Ω) of a 40° length, we get s21=7.069e-1-j6.379e-1.

Converting the cartesian format to polar we have s11=0.9522∠-42.06°, so the phase of s21 for the 40° length is -42.06°. The error is not great in this case, it is only 5%… but you could do better using the datasheet, a four function calculator and a tape rule.

A small complication at HF and below

Velocity factor is not independent of frequency for most practical transmission lines, significantly below say 10MHz. The error is small for low loss lines, and is worse for very lossy lines.

The error is insignificant for the case worked above, but might be significant for say CCS RG6 at 1.83MHz.


Whilst phase of s21 is equal to electrical line length for some special cases, the phase of s21 is not an accurate measure of electrical line length in many cases.

There are better ways to cut a section of line to a specified electrical length.