Owen Duffy wrote:
Ian White GM3SEK wrote in
:

...

No problem. It's a standard technique for matching yagis, for example:

http://www.mydarc.de/dk7zb/start1.htm
(click the third item in menu list)

Ian,

The DK7ZB match (http://www.mydarc.de/dk7zb/images/match-12+28.gif) is
interesting. Ideally, the transformation sections (the inside of the coax
TL) are an electrical quarter wave, and the balun (the TL formed by the
outside of the coaxes in parallel and the boom etc) is a quarter wave.

I would be surprised if constructions typically shown achieve that
optimisation, due at least to the use of solid PE dielectric inside the
line, and a air / vinyl dilectric outside.

The construction at
http://www.mydarc.de/dk7zb/DK7ZB-Mat..._Isolation.jpg is quite
complex, containing what might be seen as as an electrical quarter wave
transformer and the balun formed of a U shaped stub of physical length
0.5*0.25*0.66=0.08 wavelengths shunting one side of the DE, but this stub
lies with its hot side against the cold boom forming another o/c TL
element. If the balun shunts the DE with a high impedance, it will be due
to not just the loop length, nor just the s/c stub length, but parallel
resonance with the boom / half loop o/c stub.

If that is the case, the construction is not simply explained by the
first diagram that I referenced above.

I am not suggesting that this implementation doesn't work, but that there
are design parameters beyond those implied by the electrical circuit
given.

Thoughts?

There are two separate issues here. The impedance transformation with
paralleled coax is no problem (which answers the original question).

The other issue is the way the feedline is looped to form a balun... or
so it is claimed. Like you, I am quite dubious about that.

It is highly reminiscent of a system used 20-30 years ago on the yagis
made by Antennes Tonna (F9FT). The coax exited vertically downward from
the feedpoint, and the instructions were to leave a quarter-wavelength
hanging loop before the coax came back up to the boom. There was no
electrical connection to the boom at this point - the coax was taped
tightly down, and then continued along the boom as normal.

The quarter-wavelength loop was about 19in for 144MHz; but for 432MHz it
should only have been one-third of that, which was an impossibly tight
bend for heavy low-loss coax.

No doubt this arrangement would *somewhat* increase the common-mode
impedance at the feedpoint, and thus act as a sort of balun, but it
never seemed like a properly engineered solution. Many people replaced
it with a home-made sleeve (bazooka) balun, and eventually the Tonna
company did the same (though again, only sort of).

I completely agree that the velocity factor for the common-mode current
on the outside of the coax will be different from the velocity factor
inside the cable. This applies to both single and paralleled cable. Once
again this makes a "quarter-wave" loop a less than satisfactory solution
for a balun.

50MHz is much easier. For that band I have built several yagis using
paralleled 75-ohm satellite TV coax to match a nominal 28-ohm feedpoint
to 50 ohms [1]. An electrical quarter-wavelength is about 1.0m so there
is no problem about coiling the paralleled coax to make a reasonably
effective choke balun (the problem with velocity factors does not arise
because the choking impedance is predominantly inductive). First tape
the two cables together along their entire length, and then coil them
into three turns.

However, the effect of any choke balun can be spoiled by the capacitance
between the coax shield and the boom, which appears in parallel with
only one side of the feedpoint. To minimise this effect, the entire coil
should be spaced away from the boom by a *much* wider insulating gap
than DK7ZB recommends. If I'm feeling cautious, I sometimes add a few
ferrite beads at the start of the main 50 ohm feedline.



[1] All credit to DK7ZB for making the point that it's much easier to
design high-performance yagis with a feedpoint impedance in the region
of 25-30ohms. He feels that increasing the feedpoint impedance to 50
ohms (eg by adding a closely-spaced first director) makes optimization
more difficult, especially when 28 ohms is such an easy impedance to
match. But that is yet another separate issue...


--

73 from Ian GM3SEK 'In Practice' columnist for RadCom (RSGB)
http://www.ifwtech.co.uk/g3sek