I'm glad Steve brought this up. It's been interesting reading about the
different methods for tuning stubs. I recently built two identical sets of
automatically switched bandpass stubs for two-radio contesting. I cut open
and shorted stubs for the CW portions of 80M, 40M, 20M, 15M and 10M, based
on information at K1TTT's web site. I used various scraps of coax I had
lying around, including RG/11, RG/213 and LMR400UF, but pairs were always
made from the same material (i.e., both 1/4 wave 80M shorted stubs are made
of RG/11.) I decided to cross-check the measurements with three different
instruments. Here was the setup:
Instrument #1: Heathkit noise bridge and TS950SDX tranceiver. Bridge nulled
with a short across the Unknown port, then stub attached to Unknown port.
Reveiver null indicates frequency for which the stub is tuned.
Instrument #2: AEA CIA HF "teed" to a 50-ohm dummy load and the stub. The
unit displays the frequency at which SWR is minimum, indicating where the
stub is tuned.
Instrument #3: MFJ 259 used as an oscillator, teed to the stub and an RF DVM
and/or oscilloscope. Minimum reading on the RF DVM or scope occurs at
frequency for which the stub is tuned.
For any given stub, the three different devices showed readings that varied
by anywhere from 50 Khz to several hundred KHz. For most of the stubs, two
devices would usually show readings that were within 100 KHz. In those
cases, I used one of them to cut the stub, verified with the other and
ignored the third device. Which devices agreed and which did not varied with
the band and whether the stub was open or shorted (as you know, some of
these methods require hopping up to the next harmonic depending on whether
the stub is opened or shorted.) Frequently, measuring the same stub twice
with the same device would result in two readings that could differ by
hundreds of KHz. Very frustrating, but I did the best I could.
The type of coax used made a difference as well -- two have plastic
dielectric and one has foam dielectric. Unfortunately, I didn't keep careful
records on how this affected Q. Generally, though, it was very broad with
all three types of coax. As has been pointed out, the Q of these stubs is so
low that it's very hard to get a precise measurement with any of the methods
I used.
The oscillators and tuning in the TS950SDX and AEA CIA HF are rock solid and
very precise. Unfortunately, each has its quirks. It's tough to sweep
through a CW stub's passband with the '950 because it switches front end
components at every MHz boundary, with resulting changes in sensitivity and
audio noise floor. The CIA HF doesn't let you tune back and forth through a
frequency range, although it shows a nice graph. Unfortunately, the SWR is
flat over such a large range that it's tough to get a precise reading for
certain stubs. I think when the Q is so low the unit's computer is doing
some averaging that can be misleading. The MFJ oscillator settles down after
a brief warmup period, but the tuning is so sloppy that it's very hard to
reproduce results. The RF DVM / scope method showed promise of being the
most acurate because it showed a very definite null. However, this device
was often the odd man out, disagreeing with the other two. I suspect that
this was due to slop in the tuning or certain combinations of frequency and
open/short pulling the MFJ oscillator off frequency (or both.)
In spite of all this difficulty, the stubs work very well. They attenuate
the harmonics as expected, and get rid of all the clicks and phase noise.
Sometimes the effect is downright magical. They also work well on SSB,
although I have noticed that one or two of the combinations are not as
effective as they are on CW (probably 10M is involved, but I was too busy in
the last contest to think about it.) Obviously, they are quite broad.
Although a purist might bemoan the loss of 3, 6 or 10 dB of attenuation, the
proof is in the fact that the stubs attenuate the harmonics well enough to
be able to work the weakest signals I can hear (as long as I'm at least 1-2
Khz away from the harmonic.)
I haven't measured insertion loss, but the effect on SWR has been negligible
on most bands. By great luck, I actually get a SWR improvement on the TH-7's
worst CW band, 15M. It's up around 1.5:1 without the stubs and almost flat
with the stubs. On that band, I'm using a 1/4 wave open 80M stub to
attenuate 80M (80M transmitted on one of my radios clobbers all bands on the
other radio) and a shorted 40M stub to attenuate 20M and 10M (I punted on
trying to attenuate 40M on 15M and vice versa, but it hasn't been a
problem.)
I only had one pair of stubs that was cut wrong: the open 20M stubs designed
to suppress 20M signals on 10M. The problem wasn't harmonic attenuation --
that was fine. The tipoff was high SWR. Using the three different measuring
devices, I determined that the stubs had been cut way too short, and had to
cut a new set. Not sure how this happened. As I recall, the AEA was the most
accurate device in this case, and the noise bridge was the least accurate.
SWR on 10M is now fine and attenuation works well.
I used the noise bridge method a while back to cut the phasing lines for my
40M 4-square. At that time I didn't have the AEA and didn't know about the
RF DVM/scope method. I tried the teed dummy load method with the MFJ, but
that didn't work very well (I had some old phasing lined cut for 7.150 MHz
by Colatchco as a reference.) I cut the lines for 7.025, so I had all sorts
of problems with the rig switching receive components at the 7.000 boundary.
I was worried that the lines were not cut accurately enough, but everything
turned out OK -- the 4-square works FB.
One of these days I'd like to measure all of my stubs and phasing lines with
lab equipment and try to figure out how much error is inherent in each of
the "shack equipment" methods and why.
73, Dick, WC1M
--
FAQ on WWW: http://www.contesting.com/towertalkfaq.html
Submissions: towertalk@contesting.com
Administrative requests: towertalk-REQUEST@contesting.com
Problems: owner-towertalk@contesting.com
Search: http://www.contesting.com/km9p/search.htm
|