Some related comments, if you don't mind.
Temperature changes seem to be the biggest "detuner" of largely
mechanical devices like cavity duplexers. We often send our repeaters
off to live in less-than-ideal environments, then expect cavity
input/output impedances to remain as we measured them in the shop?
Don't think so.
IMHO, we're making the same mistake I made in a post the other day,
saying "VSWR" when what we really mean is "reflected power" as indicated
on a meter.
Jeff is correct, VSWR along a transmission line doesn't change if
source, load and line impedances are stable, the ratio remains the
same. What does change, and what is affected by line length, are actual
impedances along the line under not-so-perfect-or-stable conditions; the
actual impedances along the line change but the ratio does not. For
example, 100+j0, 25+j0, 40+j30, and 40-j30, are different impedances yet
all exhibit a VSWR of 2:1 in a 50-ohm impedance system.
Voltage is proportional to impedance. We can't really have a voltage
standing wave ratio greater than 1:1 without a voltage differential, and
that really can't happen if impedances along the line remain the same.
Our friends at Agilent have put together a Java applet demonstrating
what happens along a transmission line. Maybe you're aware of it, it's
really kind of cool. The applet allows you to change the load impedance
of the model and see the changes, so have fun with it.
http://education.tm.agilent.com/index.cgi?CONTENT_ID=6
Our in-line power meters, like our trusted Bird 43, do not directly
measure power. They're really voltage meters calibrated in watts at a
specific impedance. That's why they can be fooled into displaying an
erroneous reflected power reading, perhaps lulling us into a sense of
security that the VSWR on the line is acceptable when it may not be.
With most transmitters I'm familiar with, a "high VSWR" condition is
detected from a reflected RF sample from a directional coupler at the
transmitter's output, so it's not a "real" VSWR measurement per se, it's
a voltage measurement. Worse, these couplers tend not to be very
selective, so out-of-channel and even out-of-band energy can cause "high
VSWR trips" even when our measurements indicate all is well on our
frequency of interest.
Great discussion, keep it going. If I repeated what was already
mentioned, my apologies.
73, Russ WB8ZCC
On 8/14/2010 12:53 PM, Kevin Custer wrote:
Jeff DePolo wrote:
Because the impedance is not matched between the transmitter
and duplexer, the 'apparent' loss of the duplexer is greater
than the manufacturers stated loss of the duplexer. Changing
the cable length is not changing the loss of the duplexer,
it's changing the power that is accepted at the transmitter
port of the duplexer by matching the output impedance of the
transmitter to the input impedance of the transmitter port of
the duplexer.
But if the duplexer is tuned to 50 ohms, and the cable is 50 ohms, varying
the cable length isn't going to change the Z seen by the transmitter. Or
are you suggesting the duplexer is purposely de-tuned from 50 ohms?
Purposely, accidentally, by lack of good design - people not having
the right equipment to tune it correctly - whatever.
And also that by varying the cable length between the
transmitter and the duplexer that you can vary the reflected
power on that same line?
Yes.
With all due respect, that's not possible, regardless of what the Z is of
the duplexer. The only time it could have an effect on the reflected power
would be if the transmitter/PA were spurious, and the amplitude/frequency of
the spurs changed with varying load Z, and I think we can both agree that if
that's the case, we have bigger fish to fry.
And this is where I believe the duplexer manufacturers are covering
their butt. They don't want the problem with complex reactance
presented by the duplexer to be their problem. Not that I don't
agree, because it's usually the transmitter that is really at fault.
Joe Ham buys a new duplexer and hooks it up to his 110 Watt MASTR II
repeater and gets 50 watts out the antenna port. He does his homework
and realizes that he should only be loosing 29% with the 1.5 dB of
insertion loss stated in the paperwork - but he's loosing over 50%.
The duplexer manufacturer supposedly engineered and tuned it for a 50
Ohm system. He knows that the cable he connected to the transmitter
is good, because when he disconnects the end going to the transmitter
port of the duplexer and connects it to his Bird 43 terminated with a
good load - it reads 110 watts.
Now, is the transmitter becoming spurious and the cable length being
changed in length satisfies the match between the duplexer and
transmitter - I don't know... All I can tell you is I have followed
the suggestions written in the WACOM manual and it has worked. I had
one instance of a ham radio club loosing PA's left and right on their
2M machine. They told me of the situation and I offered to do a
little testing. The 110 watt PA would put out 110 watts into a Bird
and dummy, but only 45 watts was coming out the antenna port of the
duplexer. At the time I didn't own a spectrum analyzer. The repeater
wouldn't duplex without desense. I changed the length of the line
between the PA and duplexer until I got the power to read about 75
Watts as I remember. That was 13 years and they still have the same
PA - no desense either.
Not to belabor the point, but whatever the VSWR is on a length of
transmission line, that's the VSWR that's on the line *regardless of
length*. You can't change the VSWR by changing the length of the line. As
you vary the length, you go round n' round the Smith Chart in a constant
VSWR circle, with the Z repeating cyclicly every half-wavelength, but you've
still got a complex Z that nets a 1:5:1 VSWR relative to 50 ohms at the end
of whatever length of line you choose (cable loss effects notwithstanding).
There are an infinite number of complex Z's that yield a 1.5:1 VSWR - cut
the line to any random length and you'll hit one of them.
In a situation where the duplexer and transmitter have
differing impedances, and a cable optimized in length matches
these impedances, the mismatch at the duplexer is minimized,
therefore the power reflected by the duplexer is minimized.
I think what you're really saying is that the mismatch at the *input to the
matching section* (i.e. the cable between the PA and the duplexer), NOT the
mismatch at the duplexer, is minimized.
Sorry - that is what I meant to say. Many of us use converted
commercial gear in the ham band. Many don't take the time to properly
convert the receiver and especially the transmitter to properly
operate in the adjacent ham band. So, when you run a 150.8 to 174 MHz
amplifier in the 2M ham band or a 450 to 470 MHz amplifier in the UHF
ham band is it going to represent a good 50 Ohm impedance? Likely not...
We need to realize that most duplexer manufacturers know what they are
doing and their products are presenting a 50 ohm match on its intended
frequencies - unless somebody has adjusted on it. But, because the
duplexer is not a perfect load, it creates reactance and the
transmitter/PA may not like it. If it doesn't like it, it may become
spurious. If it becomes spurious, it isn't putting out all of its
power on the intended transmitter frequency. The spurious emissions
might be getting reflected and if the transmitter/PA has a power
sampling circuit (read power set control of some kind) it might be
shutting the PA down because the off-frequency RF being reflected by
the duplexer is being sensed as high VSWR.
The duplexer's input Z isn't
changing; you can't change that unless you re-tune the cavities or change
the load at the antenna port. Whether or that the transmitter
likes/dislikes the different Z it sees as you change cable lengths is, I
guess, what's up for debate...
I have found that when you get a transmitter that is 'picky'
about the length of interconnecting cable, power being read
at the output port of the duplexer is low and you cannot
alter the tuning of the cavity closest to the transmitter to
make things right. In other words, the place where lowest
VSWR and maximum power transfer occurs is at two completely
different places, and power transfer is not up where it
should be (transmitter makes 100 watts into a dummy load but
only shows 50 watts on the output port of the duplexer that
has a stated 1.5 dB loss (29 %)).
That would imply that either duplexer is presenting a load Z substantially
far removed from 50+j0, OR the transmitter doesn't like a 50 ohm load, or
something inbetween, would it not?
Yes.
As you get close to the 'optimum' cable length, the lowest VSWR and maximum
power
transfer occur near the same place when tuning the cavity
closest to the transmitter.
But again, *you're NOT changing the VSWR*! You can't change the VSWR by
varying the length of the line! I just want to make sure we're on the same
page - the VSWR on a transmission line doesn't vary with length (loss
notwithstanding).
You are changing the VSWR when tuning the cavity closest to the
transmitter. I realize that impedance transformation cannot occur
when you have a 50 Ohm cable (of any length) and a perfect 50 Ohm load
- but I think you will agree that a duplexer doesn't, in any way shape
or form, present a nice 50 Ohm load. Some transmitters just cannot
deal with it without some form of matching after the fact - like a
Z-Matcher, Isolator, Circulator, or even a critical cable length.
I usually pay more attention to what is coming out the
antenna port of the duplexer - first. Then, when things are
right, comparing forward power going to the duplexer and
power going to a good dummy load will be very close the same,
since matching the impedance of the transmitter to the
impedance of the duplexer was accomplished by some means.
Can you give me some real-world examples of what combinations of duplexers
and transmitters you've run across that just didn't want to "play nice"
without having to resort to changing cable lengths? Like a highband Micor
110 watt H split paired with a Q2220E or whatever. I'm just curious if I've
done any of the same combinations.
GE MASTR II 110 watt 150.8 to 174 MHz PA and WACOM WP-641. Motorola
MICOR 150.8 to 162 MHz PA and WACOM WP-641. Hamtronics 45 Watt 2M PA
and Sinclair Q-202.
I think you know me well enough by now Kevin that I'm not looking to pick a
fight, I'm just a hard-ass when it comes to basing technique on solid
engineering foundation. I can't say I've ever had to play with cable
lengths to either get a transmitter/PA to make rated power, or to get the
"apparent" loss of a duplexer to meet spec. Have I just been lucky? Maybe.
But if I'm *that* lucky, I'm in the wrong business, I shouldn't be sitting a
hotel room in Harrisburg on a Saturday waiting for a tower crew to show up,
I should be living the good life in Vegas making a living playing
blackjack...
Well, I cannot believe that I'm the only person on this list that has
had success with optimizing the length of cable between the duplexer
and transmitter/PA.
If there are others that have - I'd like to hear from you - and would
you include your equipment / situation.
I'll get us some tickets for Vegas - Jeff.
Kevin
