Hi all,
Steve, thanks for posting this response. It permits elucidation
of the final fallacy that has been hanging us up. This subject
is also well covered in Maxwell's book. Take a look at figure
7.1 on page 7-2 in "Reflections". Notice the region shown as
"Overall Impedance Transformer T". This is a good image to keep
in mind when thinking about mismatched systems with a conjugate
matcher in line.
Remainder of comments interspersed in reply as below.
Eric N7CL
>From: "Steve Thomas" <steve@n6st.org>
To: <towertalk@contesting.com>
>Date: Sat, 5 Sep 1998 12:12:52 -0700
>
>Hello all,
>
>I'll comment at various places in the message.
>
>Steve N6ST
>
Snip... (for brevity)
>I realize that the 20:1 SWR example is merely an extreme case
>for illustration purposes. You are correct that SWR doesn't
>(precisely) cause loss, it causes reflection. The magnitude and
>phase of the reflection depends on the impedances at the
>mismatched junction. I know this from many measurements with
>professional quality equipment (particularly an HP 8510B Vector
>Network Analyzer among others).
>
>If the reflected signal doesn't see another mismatch on its way
>back to the source, it will continue traveling until it hits a
>matched load, where it will be absorbed. If it sees another
>mismatch, a portion will be re-reflected and again travel toward
>the load (an antenna in the discussions that have occurred on
>TowerTalk.) Line loss takes its toll every time the signal
>passes through and eventually the signal will disappear into the
>noise.
Yes. But as we have already seen, if the line loss is very low
to begin with, all this additional line loss due to reflections
still adds up to an insignificant amount. This amount in the
current (relatively extreme) example is 0.01 dB. Not enough to
get exercised over.
>>
>> He offered an example of a broadcast station operating an
>> antenna at a 20:1 SWR successfully. At 600 kHz (or wherever
>> they were in the BC band), the losses in the airline they were
>> using probably amounted to very nearly zero. A reasonable
>> guess might be on the order of .001 dB/100ft.
>
>Think about the receive signal. When it hits a mismatch, some
>portion will be reflected back to the antenna, to be re-radiated
>and forever lost.
Very little will be reflected back to the antenna. What little
is reflected back to the antenna has its current and voltage out
of phase and encounters a 20:1 mismatch. So almost all of it is
converted back to forward (away from antenna towards transciever)
power and returned down the line to the transciever. But yes, it
is true that 0.01 dB will be forever lost to increased line loss.
>>
>> So for that particular case, would you rather have your
>> receiver tied to a tuner that looses 0.5 dB plus the 0.001 dB
>> (we'll assume 100 ft distance here) airline and a 20:1
>> mismatched antenna or to 100 ft of 1dB/100ft coax and the same
>> antenna with a perfect 1:1 match?
>
>I'll take the 1dB/100 ft case any time with a matched antenna
>system. If I've lost 7.4 dB due to reflection at the
>antenna/feedline junction, I'll never get it back no matter how
>low the loss of the feedline is.
This is a very appealing way to look at it. The only problem is
that physics doesn't work this way. And this is why I have been
sticking this discussion out. This fallacy has caused a lot of
people to do a lot of sometimes expensive, sometimes dangerous
and usually unnecessary work.
There is no loss of 7.4 dB due to the mismatch between the
antenna and the transmission line. This 7.4 dB is the mythical
loss due to SWR that does not exist.
Remember the tuner? If it is properly constructed and adjusted,
it _and_ the transmission line form a matching network for the
antenna. The tuner is usually composed of lumped elements and
the transmission line is a distributed element. But combined,
they are the total matching system for the antenna. There is no
reflection at the antenna terminals because the system appears to
the antenna to be properly matched at that point. The total
system loss going either way through the system is 0.51 dB.
Consider this. Would you be happier if I moved the same tuner up
to the antenna? Would the receive signal be better off now? The
antenna is now matched at its terminals (as it was before). OK.
Here we go.
An antenna that produces a 20:1 SWR for 600 ohm airline doesn't
by itself produce a 1:1 match to 50 ohms. If it is a current fed
antenna it presents an impedance of 30 ohms for a raw SWR of
1.67:1.
So to get a perfect match to your 1dB total loss 50 ohm coax,
some tuning at the antenna will be required. Since it is the
same tuner, we'll assign it the same loss (0.5dB) as before.
Now the total system loss is 1dB for the matched coax plus 0.5 dB
for the matching network. Total system loss between antenna and
tuner is now 1.5 dB. This is true for signals going either way.
Reciprocity applies here as it does when the tuner is in the
shack where it is much easier to reach.
Too much loss in the matching network? OK the SWR isn't too
bad. Lets just leave the tuner out. Now the total system loss
is only 1.106 dB. This is still worse than the tuner with
airline by 0.596 dB. _And_ the receiver is now looking at a
1.67:1 mismatch at its input terminals. This might actually
improve its NF. But that is yet another discussion. Better not
count on it.
>>
>> I'll bet that you couldn't tell the difference.
>
>If the signal is large, you're right, it wouldn't make a
>significant difference in the ability to copy a signal. But if,
>on the other hand, I'm trying to work a weak signal on 160
>meters where 1/f noise dominates the receiver, losing 7.4 dB
>could make the difference between being able to copy the signal
>and just hearing noise.
External natural and manmade noise dominates my receiver at 160
meters. If 1/f noise is the problem for you at 160, you _really_
should replace that coax ;-).
But that doesn't change the point. 7.4 dB would be easily
noticeable. It represents more than a full S-unit by anyone's
definition. But I still don't think you could perceive the
approximately half dB _increase_ in signal level that the airline
system would actually provide over the 1dB loss coax.
>>
>> For the above example:
>>
>> Coax & perfect match: Total system loss = 1.0 dB
>>
>>
>Tuner with airline and 20:1 mismatch: Total system loss = 0.51 dB
>
>plus 7.4 dB in reflected loss at the point of the 20:1 SWR.
As I said before, this loss is nonexistant.
>So, give me 1.0 dB loss instead of 7.91 dB loss any time!
>
So, we'll give you 1.0 to 1.5 dB loss (depending on network
topology) instead of 0.51 dB loss and you'll still be happy. (?)
Unless, of course, that was actualy a voltage fed antenna (240:1
SWR to 50 ohms)... But that is another story.
Snip... (brevity)
>
>73, Steve N6ST
>
>
If you are having a hard time accepting this, there is a simple
experiment that you can do since you have access to a network
analyzer.
1. Make some center fed dipoles 13 inches long from 1/4 inch
copper tubing or stiff wire. Make 3 of them.
2. Do what is necessary to match two of the three to 50 ohm coax.
at 146 MHz. This will probably require some stubs or
possibly a hairpin. Lumped elements can be troublesome and
lossy at this frequency. Also do what ever is necesary (bead
balun, etc) to decouple the feedline shield from the antenna
currents. Use 3 meters of feedline on each one.
3. Connect the third one to 1/2 wavelength of airline and match
the end of the airline to 50 ohms. This should require
essentially the same network as for the first two antennas.
But you have a network analyzer, so it shouldn't be difficult
to get everything all matched up correctly. Also take care
to keep antenna currents from flowing on the 50 ohm side of
the match point (bead balun, etc). Use 2 meters of 50 ohm
feedline (same type as for the first two antennas) from the
match point to the analyzer for this antenna. This will make
the total feedline also 3 meters including the airline.
4. Set the first two antennas up in a fixured arrangement
parallel to one another, at the same height, vertically
oriented, and separated by one meter. Connect one to each
port of your network analyzer.
5. Measure S21, and S12 at 146 MHz (should be the same, no?).
Record the values.
6. It might be nice to record S11 and S22 as well while you are
at it just to be sure everything is matched.
7. Now substitute the airline fed antenna for one of the
others. Keep the airline clear of the ground or other
obstructions that would interfere with its proper operation.
8. Measure S21, and S12 at 146 MHz (S21 should be many dB
different than S12 now if your theory is correct, yes?).
Record these and the S11 and S22 values.
Let us know how the measured values compare.
Have I built any bias one way or the other into the experiment
setup? I don't think so. But please let me know if you think I
have. I'm sure that such an experiment can be structured fairly.
73, Eric N7CL
--
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
|