A tale of two feeders

I presently have a doublet, 33 feet overall, and fed by 300 ohm twin feeder.

The distance from the wall of my first floor flat, to the pole at the other end of the garden, is 72 feet.

Next week, I intend to replace the existing doublet with a 66 foot version, which will shift the feed point further away, and involves replacing the feeder with a piece that is a bit longer.

I’ve already odered it, and the piece that I will replace, will be used for portable operations.

Today, I found a piece of 450 ohm ladderline, which I’d forgotten about, and, all day, I’ve been wondering if two pieces of twin feeder/ladderline of different impedances, each of which they say is virtually invisible to RF, could be joined together, without introducing a mismatch?


1 Like

I think you’ll invariably get reflections at such a junction since you have an impedance discontinuity

That was more or less one of my arguments, but it’s been on my mind all day, and I can argue it for either answer!

It’s academic anyway, because I won’t be going down that road.

Thanks, John


I think you meant that these transmission lines have low loss even with high SWR. They certainly are not invisible at any frequency.

As a general rule you join unequal impedance transmission lines when you are making some sort of a match. For example a quarter wave length of 75 ohm line transforms a 100 ohm load to a 50 ohm load.
Otherwise joining unequal impedance lines isn’t a good idea.


Thanks Ron,

I knew what meaning I’d intended to convey, but managed to pick a bad choice of word for it.


Taking these two points together suggests that impedance transformations using unequal impedance lines is a source of loss!


Except in the special case of a transmission line transformer, where at the design frequency there will be no impedance discontinuity. (The impedance changes evenly along the length, so there is no sudden step to cause a reflection)

A. Pedant. :wink:


Yes. But the losses are low and an acceptable price for the match.

There are losses everywhere. That’s life.

A perfectly matched transmission line has losses that are directly proportional to the length of the line.

For the above line the losses increase as a function of frequency. (Not linear).

We are normally only interested in steady state conditions - a measure of these is the SWR.

Any change in impedance causes a reflection and when everything settles to the steady state this is seen as a change in the SWR. If the SWR increases the loss increases. We would be talking fractions of a dB in most instances. So yes a discontinuity will increase losses in most cases. Mostly we don’t worry. For example the PL259 SO259 combination represents a non-50 ohm impedance. That’s why type N or BNC connectors are favoured above 50 MHz. On HF the extra loss is not noticed.

So yes Brian, including an impedance change is likely to increase losses. Sometimes it might improve the VSWR and then the overall losses would be reduced.

If we had a 450 ohm line and connected to a 300 ohm line the losses would likely be higher than using either line for the whole run. It is complicated by having a high SWR on the line . We chose to connect to a multiband doublet with high Z line to minimise the losses. It is possible that on one or more bands a cunning arrangement of mixed impedance lines could be used for a better impedance for out ATU and overall the losses might be less. But it may well make the situation on other bands worse. Usually simplest is best with the adding of some more line being a remedy if the ATU protests at what it sees.

If I did not have enough 450 or 300 ohm line but the combination were long enough and I was not wanting to spend any money then and only then would I join them.

A quarter wave matching section uses a line that is not matched and has a non unity SWR. It will have higher losses than if it were matched. But being a short length of line and usually of low loss to start with the matching section loss is small and acceptable. Even when a phasing harness with a half wave included the SWR is usually moderate, less than 3:1 in the cases that I can think of.

In a current balun using transmission lines, for a matched load the SWR on each element is 1.0 :1 if the winding impedance is correct. Losses are therefore very low,

Sorry Mr Pedant, I know of no case where your tapered impedance exists. Maybe it does on PCB at microwave frequencies. Can you point to other examples? It would require smoothly changing the LC values. Collins once used several different impedance lines in a stepped Z configuration in an attempt to provide a wide match. Not sure if it worked as hoped but it was an approximation to a tapered impedance line. I don’t see it in the shops nowadays.

So we live with losses and try to minimise them and don’t do things that create more loss.


1 Like

ISTM that some of these posts are mixing up two very different concepts:

  • The impedance measured at a specific point on a transmission line at a particular frequency

  • The characteristic impedance of a length of uniform transmission line.


Indeed. So it would appear that the answer to the original question is that two pieces of twin feeder of different impedances can be joined as long as the second piece presents the same impedance as the first piece at the joint. Thus knowing the frequency and the velocity factor the length that the second piece should be trimmed to can be found. An untrimmed random length will probably be a source of loss. As the dam’ meercat says, “simples!”


I think you can do it. What it does mean is that the impedance presented to the ATU (or AMU if you prefer) would be different when using a length of mixed feeders than it would be if using a full length of either 450 or 300 Ohm feeder. If the ATU can match to that impedance, it’ll work. The SWR will likely be different on the two sections but that doen’t realy matter.


Hi Ron, maybe it is my misunderstanding!

My logic goes like this:

Take the example of an electrical quarter wave of 75 ohm line. One end is terminated with 100 ohms, and the other end with 50 ohms.

If you were to cut this line at some random point, and measure the impedance (at the design frequency) looking towards the 100 ohm termination, it would be somewhere between 50 and 100 ohms, depending on where you cut it.
Measuring towards the 50 ohm termination would give the same reading, I think?
If this is correct, then at no point on the line is there a step change in impedance, so there should be no reflections…

I’m very open to correction :o)


I would suggest not relying on intuition to guess what happens along a transmission line with mismatched termination. If you don’t want to do the maths (and I wouldn’t blame you) draw it out on a Smith chart.


1 Like

Sorry I did misunderstand you.

Yes Mr Smith’s charts are very helpful but I could not get a really sharp copy to print at A4 size. So I wrote an EXCEL routine to do the heavy lifting. It all helps with understanding what happens.

EZNEC also allows for not just antenna analysis but for the effect of connecting more than one transmission line. eg a length of 450 ohm line and then 50 ohms.


1 Like


cutting the quarter wave line you would find the impedance at any point along the line is reactive. It will have two components, ie. a resistance and a reactance. There will be no point at which the impedance matches any other line unless it has the corresponding impedance with the opposite reactance. Which the bit of the quarter wave you cut off, did have.

To connect your two transmission lines together - no problem. But solving the resulting impedance transformations will be a multi step process. And will vary on each band. The standing waves will vary on each part of the feedline, but as long as you can sort out the lengths and the impedances to work correctly, it will transfer power. Simpler not to do it, but if you have to, it certainly can work.

The input impedance of the antenna system will clearly vary greatly with frequency. On some bands it may be an impedance that your ATU can deal with. On others it may have a very low or very high impedance that your tuner will adjust for, but will be very inefficient due to the high circulating current or other factors that make your ATU into a liability rather than an asset. The vendors of commercial ATUs like to suggest they can tune a bit of wet string, but in doing so, they omit to tell us the real cost (in losses) of using an ATU outside a reasonable operating range, usually defined (by them) as an SWR of less than 3. Whereas the impedance of a centre fed full wave antenna would produce an SWR of more than 20, for example, if fed with a 50 ohm line.

That’s why most of us plump for OCFs, fan dipoles or known designs of multiband non-resonant antennas like the zs6bkw, which has a specific non-resonant feedline length to compensate for the reactive load of the antenna.

Good luck,


1 Like

Look at it this way: If you add the ladder at the house end it will have no effect what so ever on losses in the twin lead + antenna combination. The question reduces to whether the ladder would fare better or worse than the twin lead feeding the particular impedance at the joining point. My bet is that the ladder will always have lower losses due to its open construction, even in those cases (depending on band and exact dimensions) where the SWR would be higher on the ladder.

Besides, the reflection due to a change in Zo from 450 ohm to 300 ohm seems trivial compared to the reflection coming back from the antenna itself on any band.

73, Villi TF3DX

I actually use a combination of 300-ohm line and 450-ohm window line to feed my primary SOTA chasing antenna. This loop antenna is approximately a full wave at 14.06 MHz - about 70 feet of insulated wire. It’s mounted in the vertical plane - it hangs from a tree limb about 40 feet above ground. The feed is at the bottom center of the kite-shaped loop, so the polarization is horizontal. The feedline length is reduced by having the feed point down low - about 11 feet off the ground. The antenna is supported with elastic cords, and it has been through more than a year with many big winds and snow events.

This type of antenna is inherently more durable than a dipole with the high feed suspended at the middle of the span. The pattern is dipole-like, with a wave angle suitable for most of my contacts here in the states.

I have about 45 feet of 450-ohm line from the antenna to the entry into the house, and then there’s about 25 feet of 300-ohm line to the tuner in the shack. I decided to use the 300-ohm section - after much deliberation - because the feedline runs close to my computer, suspended about 3 inches below the ceiling, and I wanted to minimize radiation inside the house.

This is ham radio, and this is a compromise for sure! I ran Eznec models of my system, including the two feedlines, and I saw the losses add up on some of the bands. I recommend modelling these kinds of systems with Eznec, because the models allow trying out different solutions in order to evaluate the various trade-offs.

I use homebrew balanced tuners to feed my antennas. These use old designs from the 1930’s, with link coupling, series-tuned-primary, parallel-tuned-secondary, balanced split-stator variable capacitors, and large air-core balanced inductors. I usually run only 100 watts, so the losses in the lines are not high enough to cause real trouble. I don’t really care much about the actual impedances on various bands, the line SWR, or the Smith Chart - even though these determine the loss on the feedlines.

This antenna has exceeded my expectations, which were based on Eznec models. The plane of the loop is N-S, so it generally radiates E-W. Since I’m in Colorado that’s fine for many activations. The loop tunes broadly on 20M and gets out well. On 30M it also loads and performs well - I often work activators on the East Coast. On 17M and 15M it tunes nicely as well. The bipolar pattern holds from 30M through 15M.

Initially I swore never to load up this antenna on 40M, because it’s a half wave, and the feed impedance at the loop is many thousands of ohms! However, I tried it, and my tuner was able to match the feed easily - apparently the mismatched feedline transforms the crazy feed impedance to something reasonable. The loss in the 300 ohm and 450 ohm lines is so high at 7 MHz - several db - that there’s no danger of the loop feed flashing over. Some time ago I learned the hard way that a half-wave loop can be a tricky devil, if fed with real power!

As long as you have a good tuner, the additional mismatch of the 300-ohm-to-450-ohm line transition is no big deal. Most of the “450 ohm” window lines are really closer to 400 ohms anyway. When feeding a single wire or loop on several bands, you will have much greater mismatches and losses to consider.

When a loop is fed on non-resonant frequencies, the predicted SWR’s on the feedlines seem high on some bands, but the losses are reasonable, if the feedlines have relatively low loss and are short. Just don’t use coax, and don’t expect miracles when feeding a half-wave loop.

I’ve achieved Super Shack Sloth mostly with this vertical loop in my cottonwood tree. For some reason it picks up less noise on most bands than my other wire antennas. The ability to jump from band to band with one tuner is helpful when chasing. The cost of the system was - and is - so low that it’s not worth thinking about. The wire in the tree is almost invisible.

I encourage doing these kinds of experiments. Think seriously about loops in various configurations! The 20M full-wave vertical loop needs very little real estate.

Every time I make a SOTA contact on 30M or 40M with this mismatched system with the 300 ohm and 400 ohm lines soldered together I smile!

Of course, if the tuner was located at the loop feed, the system would provide higher efficiency - maybe next year!




A good account George!

The idea that a mismatch will invariably result from adding a line to one of different Zo is a misconception if the original line has a reflection on it. Bear in mind that an extension with the same kind of line will be mismatched at the joining point (any point along the line for that matter), hugely so if the SWR is high. Depending on system dimensions in terms of wavelength, the new Zo may give a better match or worse. Steady state is assumed here, which is enough for our slowly varying modulation modes.

As for the half wave loop, it is essentially a quarter wave transmission line shorted at the far end. Save for loading by radiation and losses, that short will be transformed into an open circuit at the line input. The half wave loop is resonant, but resembles a parallel LC circuit (Hi-Z) in contrast to a series LC (Lo-Z) for the more common feed at current maximum.

73, Villi TF3DX

Hi Villi,

I think you meant that if you have a feed line with a high SWR adding in an extension of different impedance will mostly still result in a high SWR, so not much point fussing. If you are talking about 450 ohms to 300 ohms and vice versa that is probably correct.

HOWEVER, any change in impedance in a feed line system produces reflections and can change the SWR as seen at the TX end. It is inevitable, a fact and not a misconception.

You note that a better match may be obtained by changing the feed line impedance. Correct, especially if the impedances and lengths are chosen after checking the numbers. Otherwise, mostly it will be still a high SWR.

The G5RV and it’s derivative the ZS6BKW are two of many instances where the feed line impedance is changed and a better match is obtained.


A fact about the G5RV which seems to be generally forgotten:

In his Radcom article of July 1984 (pp. 572-575) Louis Varney discussed feeding the G5RV. The section on the feeder begins with the words “The antenna can be fed by any convenient type of feeder provided always that a suitable type of matching network is used.” Later in the section he says: By far the most efficient feeder is the open wire type… If this form is employed, almost any length may be used from the centre of the antenna to the matching network (balanced) output terminals."

What makes an antenna a G5RV is not the matching section to coax (which is just one of the options that he describes), but the 102-foot top section which he selected as being particularly effective as a 3/2 lambda 20 metre antenna. In other words, the antenna that he regarded as the best form of his design is simply a 102 foot doublet. Incidentally, in the form with the matching section to coax that is now immortalised as “the G5RV” it should be noted that the coax that he recommended was not 50 ohms but 75 ohms.

I can vouch for the fact that if you replace the coax by an extension of the “matching section” to the tuner, the antenna will tune and perform well on all bands 160 - 6 metres, since I have used it for 14 years. On 160 metres it may not be a DX magnet but I have made a number of successful chases on that band!


1 Like