I have personally experienced and measured some of the effects that Norm
When I first modeled and built a four element, five band quad using
separate feeds for each band via an Ameritron switch, I was quite
disappointed with the results. The actual quad's characteristics,
particularly SWR, bore little resemblance to those predicted by the
model. Driven element lengths had to be significantly adjusted in order
to get the SWR minima within each band. Even after these adjustments,
the performance seemed poorer than expected and the bandwidths and SWR
minima were very sensitive to the position of the driven elements
relative to each other.
In discussing my poor results with L.B. Cebik (W4RNL), he accurately
suggested that I had probably failed to use (electrical) 1/4 wave coax
links between the remote switch terminals and each driven loop's feed
point and to set up the switch so that switch end of the coax for the
inactive bands was left open. The objective was to (electrically) short
the feed points of the inactive bands. I was indeed using random
lengths of coax and was grounding the switch end terminals of the
inactive bands. Aside from the fact that I had modeled the inactive
bands' feed points as closed and the actual antenna's characteristics
were randomly somewhere between open and closed depending on the lengths
of the links, one of the best kept secrets in multiband quad
construction is the use of 1/4 wave links. The issue doesn't arise in
the context of two element quads because the feedpoint impedance is
around 100 ohms and, typically, 72 ohm, 1/4 wave coax matching
transformers are routinely used. Feed point impedances for four element
quads are closer to 50 ohms.
When I remodeled and rebuilt my four element, five bander at a new QTH,
I set up the completed quad on a 12 ft. step ladder prior to installing
it on the tower and checked the SWRs with the inactive bands opened and
shorted. There were indeed major differences on most, but not all,
bands between the two conditions. (I have the comparative data in my
notes, if anyone is interested.) The actual quad conformed to the
model's predictions quite faithfully and only a smidgen of adjustment
was required to nudge the resonant frequency of the 10M driven element
into the desired part of the band. I have recently modeled and built a
two element 30/40M quad which, of course, used 1/4 wave 72 ohm matching
sections, and the model was again quite reliable.
The bottom line here is to use electrical 1/4 wave sections for the
links and leave the switch ends open for multiband quads. It does
work. OTOH, what bothers me and I believe this is what Norm's referring
to, is that those 1/4 wave sections are only 1/4 waves for the active
band, so the feedpoints of the inactive bands are really not shorted.
Maybe someone has a clear explanation for exactly what's going on here.
>In a message dated 3/13/2006 4:02:02 PM Eastern Standard Time,
>From: "firstname.lastname@example.org" <email@example.com>
>Subject: Re: [TowerTalk] remote antenna switch
>What are the cases for a remote coax switch with unused antennas being
>Grounded, or UNgrounded? For example, I have a 5-band quad with an RCS-8V
>coax switch. So there are four unused bands on the quad with one active, since
>have five separate feedlines.
>In general it is best to have the unused quad loops appear "OPEN" to minimize
>some of the potential interaction. Regards "static" or EMP dissipation as
>long as you are feeding direct with coaxial cable the shield will always
>a DC path.
>Now the problem is achieving the "OPEN" condition because it is a complex
>relationship between the length of the feed lines, VF and the frequency of the
>band in use. Each unused feed line will have a different "electrical" length
>depending on those three variables, Cable Length, VF and TX frequency.
>As you know a "quarterwave" cable terminated in a SHORT reflects an OPEN at
>the other end. On the other hand a HALF WAVE cable length would reflect a
>SHORT at the other end. Given the harmonic relationship of some bands the feed
>line could look like a Half wave, or some fraction which will transform some
>different complex Z to the antenna termination.
>The answer is not easy to find - first try it (shorting jumper), observe the
>results, compared to being open. Antenna patterns (forward lobe shape and
>F/B) can be effected by the different cases). If there is an obvious "bad"
>effect, then change the length of the feed line to that loop.
>There is some analytical work being done on just what are the optimum
>individual feed line lengths would be to minimize interaction. First drafts
>been completed, but it still needs some refinement before it can be published.
>"The Power of a Cubex Quad"
>TowerTalk mailing list
TowerTalk mailing list