Interestingly, there's an IEEE paper on designing filters using transmission
line segments and stubs (and coupled lines, too). I can look for the reference.
But, in general, yes, you want to put the stubs at the "right" distances from
each other. In theory, of course, if your antenna and Tx were broadband
matched, then where the collection stubs and series sections are along the line
wouldn't make much difference, but, in practice, what you are doing is trying
to filter out something at a frequency OTHER than the one where the antenna
(and perhaps the output of the PA) is resonant/matched.
That being the case, the distance from the Tx to the first stub becomes
important, because you're looking into a mismatch. Ditto for the antenna.
I don't know what the output Z of a tuned output amplifier (e.g. tubes) away
from the desired frequency would be off hand, so I couldn't comment on Tom's
general assertions, but he's probably right.
When I was contemplating this kind of thing, I whipped out the XLZIZL
spreadsheet (which can do the stub Z calculations for both series and shunt, at
least for a few ladder rungs), loaded in the Z of the antenna from the model or
measurement, and started fooling around with the length of the series segments.
A modeling program is probably the easiest way (after just doing "rule of
thumb" approaches), because computationally, it's pretty complex (quite similar
to doing ladder filter synthesis from scratch). (well, actually, you can
approach it as a filter problem, the problem is that the "rungs" of the ladder
aren't a nice RLC, except in the narrow band case)
>From: Jim Brown <firstname.lastname@example.org>
>Sent: Feb 2, 2010 11:31 AM
>To: Tower Talk List <email@example.com>
>Subject: [TowerTalk] Using Stubs to Reject Harmonics
>Last night, I built a 160M stub using very low loss coax, carefully tuned it
>with an HP generator and spectum analzyer, and stuck it at the output of my
>Titan Amp, whose output stage is a Tee network (inductor output), and listened
>on another radio to the 2nd harmonic. The stub didn't do much to reduce the
>harmonic. As noted below, Tom strongly recommends staying 1/4 wavelength away
>from the amp.
>My question to the list: Have others seen this, tried it? Measured the result
>in a meaningful way?
>= = = = = = = =
>>Question 1): it would seem that placing the stub as close to the linear is
>>the best place for it
>W8JI: The best place to put a hi-reject stub is exactly a 1/4 wl from the
>source, if the source has a low pass filter in the output (like an
>>Question 2): Okay I 'm gonna put the stub/trap outside. Is there a length of
>>coax running to the rig from the stub that would be better or worse? I know
>>some have talked about using two stubs and a crtical spacing between the two
>>exists (is it 1/4 wave or 1/2 wave ??).
>W8JI: 1/4 WL at the harmonic fy, always! Unless the source is high Z at the
>harmonic, like a series "C" T network!
TowerTalk mailing list