Pawsey stub velocity

It's an interesting question, but there is an inherent problem. Published velocity factor may contain an error as great as 10%, so what does that mean to the calculation of stub length? Seems to me that you are forced into empirical measurements either way to determine the "proper" stub length - where "proper" is whatever your design specs are. In other words, it sure would be nice to be able to compute "the" answer, but I don't see how that is possible without measuring the velocity factor - or measuring the stub performance at the design frequency and over the design frequency range. :-(

Ed

Why not use a transformer?

I have measured it by simpy measuring a physical length of cable with my trusty dressmaking tape and then comparing that with the Time Domain Reflectometer measurement using the VNA. The VNA can be set to any velocity factor, so I chose the one that gave the result to the best approximation of the physical length; It came out as 67%.

I have one on order but I am looking at other solutions too. Balancing the centre point of the vertical diople is only part of the problem, getting the co-ax to it from underneath is much more difficult if you don't want standing waves on the braid. A choke is a distinct possibility and very cheap to make.

As I understand it:

- The current flow on the _inside_ of the feeder coax is subject to the cable's velocity factor, because the electrical fields are applied across the cable dielectric.

- The current flow back down the _outside_ of the feeder coax (which is what you want to choke off, in order to force balance in the antenna) is not subject to the cable's velocity factor, because the electrical field on the outside isn't going through the cable dielectric. It's going only through the outer insulation and then out into space.

Again, if I understand it correctly, the presence of the outer insulation (on both the feeder, and the choke section) does cause current flow here to have a velocity factor of somewhat less than 1.0 (as you would see in a bare wire). However, the velocity change is much less than what occurs inside the cable (the VF here might be .98 rather than .67 as it might be inside the coax), and most opinions I've read say that it can generally be neglected when figuring out the length of the choke section (and thus the point at which the bottom of the choke is soldered to the feeder).

I don't believe it matters significantly whether you remove the center conductor from the choke section, or simply trim it off flush at both ends and don't connect it.

Yes I think that’s right. The quarter wave transmission line we want is formed between the stub shield and the feedline shield. These fields are in the thin outer jacket insulation and air so velocity factor will be high and near one. Spacing between the two shields should be minimised?

I agree, that seems like the correct explanation - mystery solved!

I guess for precision you could make a test piece of two lengths of the intended coax taped together and measure the velocity factor of the shield to shield transmission line thus created. I imagine the spacing is critical.

I have attempted to thread the co-ax inside the bottom element of a dipole and use the tubular element as the Pawsey stub, but the spacing between the co-ax outer and the inside of the tube is uncontrolled and the SWR is unstable.

That's the usual approach for this sort of stub, for several reasons.

It will help minimize radiation from the transmission line (by keeping the distance between two sides of the TL down to a tiny fraction of one wavelength). It keeps the length of the connections _between_ the coaxes (at the top and bottom) to a minimum, and minimizes any excess inductance introduced by the connection. It's also physically easy and convenient - just use plastic zip-ties to fasten the two pieces of coax together, and glob some good weather-sealant over the connections at each end, and you're done.

Spacing the two sides of the TL apart will probably make a tiny difference in the velocity factor, but I believe it'll be negligible (since all you're doing is adding air dielectric).

It'll also change the characteristic impedance of this TL, but again, that really won't make a significant difference in behavior. No matter what the impedance of a quarter-wave TL is, it'll transform a short at the bottom into an open circuit at the top.

Now, this TL will only be a "perfect" quarter-wave length at a single frequency. On either side of that frequency its transformation will be imperfect, and the impedance looking down into it from the top won't be "infinite". Over a limited bandwidth (say, 2 MHz of the 2-meter band) that's only a percent or two and it'll look enough like an open circuit to provide good choking performance over the whole band. An error in the cable length of a percent or so, or getting the VF wrong by a percent or so, will have a similar (and probably neglible) impact on the actual choking performance.

So, I'd say the pragmatic thing to do is pick your center operating frequency, assume a VF of .98 or so, calculate the length, cut, strip ends of jacket, solder, weatherproof, and be happy!

The fancy thing to do is actually measure the actual VF, by making a test line out of two pieces of the coax closely strapped together and shorted at the far end. A longer test cable set, and a lower test frequency (one at which this cable set is theoretically 1/4 wavelength long) might make this easier.

This could be done using a NanoVNA, and a coax jumper a few feet long ending in a choke (run the coax through a ferrite core or toroid a few times) and short clip leads. Calibrate the NanoVNA with this jumper in place (open, short, and a good 50-ohm resistor) at your chosen test frequency. Then, sweep the cable set around that frequency, looking for the point at which the S11 impedance is as high as possible (resistive) and has no reactive component. At that frequency, the test cable is a quarter-wavelength... so, knowing its length you can compute the VF, and you can cut a similar stub which will be exactly

The NanoVNA firmware (or its host software) may have a "distance to fault" feature - a pseudo-TDR which sweeps the frequency and looks for this high-resistive-impedance peak. Or, there may be a "compute cable velocity factor" feature... same idea, different user interface.

If you have access to an actual TDR (one which sends a sharp pulse down the TL and lets you look for the actual reflection) you can get a very precise VF measurement that way.

All of this is, very probably, just gilding the lily and going for "perfect precision" that doesn't matter at all in actual operating practice. That's something I've never been guilty of myself... no never!... well, hardly ever :-)