Reply
 
LinkBack Thread Tools Search this Thread Display Modes
  #21   Report Post  
Old August 11th 03, 10:24 PM
Dr. Slick
 
Posts: n/a
Default

"Jimmy" wrote in message . com...
This means your shield has RF on it .
Also triming cable has nothing to do with SWR.


In theory, yes. It would be on the constant VSWR circles, assuming
50 Ohm transmission line!

Trimming the cable WILL have something to do with the series
reactance that the PA will see, and may affect the impedance match.


Slick
  #23   Report Post  
Old August 11th 03, 11:42 PM
 
Posts: n/a
Default

My 2 cents (wont even buy a lolipop any more!) From the old VHF,er
magazine, couple of thoughts: 1) Very lossey coax will show a prefect
termination impedence even if open, or shorted, if sufficiant length is used
(for the coax's attenuation ) . For that reason, still keep a coil of RG-8
size coax (for power dissipation), with RUBBER DIALECTRIC (was origianlly
made for attenuators at the Microwave Region) Even at 80 meters, with no
load , has very low SWR! At 10 meters, and up, with a UHF connector on one
end, and a N connector, on the other, with NO TERMINATION, makes a great
dummy load to 2-300 watts!. and, 2) Just because you think you have a great
match, can be a random event. to test , place a 1/4 wavelength piece of
coax (figuring with velocity factor). If the SWR stays flat, you can assume
a good match, if, howerer, the SWR climbs, then you need to check for a
proper match ( this will throw the worst condition into your coax , with
this extra length, kinda like the opposite of trimming coax for the best
match legends of old CB lore)! as info, Jim NN7K


"Tom Bruhns" wrote in message
m...
(Dr. Slick) wrote in message

. com...
...
What do you mean by "calibrated to the line"?


The SWR meter should read zero reflected power when connected to a
load whose impedance is equal to the line's. Does it? If not, it's
not properly calibrated. Putting it another way, what's the
directionality of the bridge?

...

Not so much a surprise as a disappointment! A difference of 70
watts incident power is totally unacceptable with only 8 feet of coax
length added.


But the load presented to the amplifier is totally different in the
two cases, most likely. Only if the amplifier's output impedance were
the complex conjugate of the line's, and the line were lossless, and
the amplifier behaved as a linear time-invariant system would you
(should you) expect the power to remain unchanged.

The transmitter does not have a mismatch sensor on it for reducing
power at high SWR. It naturally does output less wattage as the
transconductance is reduced with higher temperatures, but my
measurements were done in close proximity to each other.

Exactly how would you measure the output impedance (S22) of a PA?
Drive it with a small signal (class A, linear mode) in one frequency
(F1), and then inject another frequency (F2) into the output, and then
measure the reflected F2 power? If i remember correctly, that's how
someone was trying to measure S22 at my former place of employment.
But all bets are off as soon as you go non-linear, or non-class A.
S-parameters are supposed to be all small signal.

I don't think you can properly measure the S22 of a class C PA.

Tam brought up the suggestion that i try a really long piece of
RG-58 from meter to dummy load, to make the Cantenna more like 50
Ohms. Gonna try it.


Or just tune the load to zero reflected power after you're sure the
meter is properly calibrated...but that can be a catch-22 situation
that the really long RG-58 can help with. Be aware, though, that
"50-ohm" line seldom is -- it can be off 5 ohms or more. It's a cause
of some consternation to those of us involved in calibration of
precision RF test equipment.

Load-pull techniques are commonly used to characterize RF source
impedances. You make known incremental changes to the load, and
deduce from the change in output power what the source impedance is,
assuming it's a linear time-invariant system. Note that adding length
to a mismatched line is one way to make an incremental change to the
load... I wouldn't necessarily say that the output impedance of a
class C amplifier is meaningless, but it may well not be constant for
all loads. and may depend on parameters you'd have trouble controlling
from day to day.

Cheers,
Tom




  #24   Report Post  
Old August 12th 03, 05:07 AM
Tarmo Tammaru
 
Posts: n/a
Default

Tom

He is trying to measure the output power of an amplifier. His meter is
calibrated for 50 Ohms; so, he needs a 50 Ohm load.

Tam/WB2TT


  #25   Report Post  
Old August 12th 03, 09:05 AM
Dr. Slick
 
Posts: n/a
Default

(Tom Bruhns) wrote in message om...

I'm not sure i agree with your statement as you have written it.

The power would remain the same only if the load and the line
were perfect 50 Ohms and also lossless, then the length of the coax
should not matter at all.


It's true that if the load and line are the same impedance (e.g. 50
ohms), no matter what the line length is, you'll present that
impedance (e.g. 50 ohms) to the source and it will deliver constant
power, independent of line length. If the line is lossless, then the
power delivered to the load will be independent of line length. Fine.
But your OP suggested that the load is NOT the same as the line
impedance (indicated SWR not 1:1), so another case is perhaps equally
or even more interesting.

If you require that the source impedance match the line impedance (and
because the line is lossless, it's impedance must be purely
resistive), then the net power delivered to the line for a given SWR
on the line is invariant with changes in line length (changes in
reflection coefficient angle in the load presented at the source).
(Straightforward though perhaps tedious to prove, or just convince
yourself with some examples.) Since the line is lossless, the power
to the load is invariant. That's the scenario I suggested above.


You're right about this, and it reminds us that if there is any
loss at all, we theoritically move away from a purely resistive
characteristic impedance into a complex one. This furthers the
complexity on the problem, as we must expect reactance in our coax.

We'll have to wait for the room temperature super-conductor that
is also flexible to come closer to the ideal 50 Ohm test setup!



So with a lossless line, and EITHER the load OR the source matched to
the line impedance, the power delivered to the load will be invariant
with line length.


Agreed, as the impedance will not move on the Smith chart.


Slick


  #27   Report Post  
Old August 12th 03, 10:48 AM
Roy Lewallen
 
Posts: n/a
Default

Almost correct.

The transmitter output impedance has no effect whatsoever on the line's SWR.

Roy Lewallen, W7EL

Dr. Slick wrote:
W5DXP wrote in message ...

Jimmy wrote:

I hope this is a wifes tale that has long been discredited. Trimming cable
changes the impedance seen by the transmitter with no change to SWR.


However, a 50 ohm SWR meter reading will change as one changes the
length of the ladder-line. When the 50 ohm SWR meter reads 1:1,
the transmitter is seeing 50 ohms. That's how I tune my dipole.




This is very interesting, as it may indicate that the "wife's
tale" may have some validity in _some_ cases.

For example, if your antenna happens to have an impedance of say,
60 Ohms, and your transmission line happens to NOT be a perfect 50
Ohms (usually the case), and happens to be around 54.8 Ohms or so...
then at odd integer multiples of 1/4 wavelength (velocity factor
included, of course), you will be right on 50 Ohms.

Additionally, as we have mentioned before, PAs rarely have 50 Ohm
output impedances anyways (who actually measures this??). So an
impedance transform via perfect 50 ohm coax length may indeed get you
closer to a conjugate match and higher incident power, which may
improve your VSWR (which is based upon the incident and reflected, of
course).


Slick


  #28   Report Post  
Old August 12th 03, 04:36 PM
Tom Bruhns
 
Posts: n/a
Default

(Dr. Slick) wrote in message . com...
(Tom Bruhns) wrote in message om...

This could be a whole 'nuther thread. For a reference, I don't have
any difficulty making a 50 ohm load with 40dB return loss out to a GHz
or so for less than $10, and most of that is the connector. Does
someone need to document how to do that? But I submit that if you
want accurate SWR measurements on a particular line, you should
calibrate your SWR meter to that line, and that doesn't take any
reference except the line itself.

Cheers,
Tom



But can you make a 50 Ohm dummy load with those specs that can
handle
300 Watts? The stays at 50 Ohms out to at least 200 Megs or so?


50 +/- what? What return loss are you shooting for in this 300W dummy
load? Do you really need 40dB, or is 30dB good enough? I believe
it's possible to bootstrap yourself into measurements that are far
more accurate than you'll need for what you are doing, and do it quite
economically if you don't count your time. But you ought to first ask
yourself just what accuracy you really need, and understand why.

Probably one can make a quite reasonable broadband power load, at
least to your 200MHz limit. There have been some good construction
articles on making tapered shields for power load resistors, for
example, to get good high frequency performance. But if you can make
just a good low-power one, you can bootstrap your way to accurate
measurements at high power. Use the low power one to insure your
directional coupler is good to some tolerance, and use that to tune
your load at whatever frequency you wish to check. Even your cantenna
should be low enough Q when tuned with an L network that it would be
acceptable over the whole of the 2-meter ham band. It's tedious, but
you can re-tune for any spot frequency inside or outside the ham band.
And given one accurate load, you can determine what impedance your
long, lossy coax is and then use that as a dummy load (probably quite
broadband). For example, if your cantenna is good enough through
30MHz, and pretty good at 54MHz, and under 2:1 SWR at 150MHz, then
perhaps enough RG-58 to give you 10dB loss at 150MHz (150 feet or so),
feeding that cantenna, would work fine from 1MHz to 1GHz. Just beware
of power dissipation in the line itself at higher frequencies. You
can even cascade large coax, small coax and the cantenna, to insure
power handling.

Beware of harmonics messing up your readings!

Also, you'd probably do well to consider how small a change in power
results from a modest load change, for various source impedances.
What's the worst case? What's the best? Is there a reason it might
be nice if an amplifier output was "reasonably close" to 50 ohms, or
doesn't it matter at all? If you think all this through, you may
realize that if your dummy load is even only 20dB return loss, it will
be just fine for the measurements you need to make. But YOU should
convince YOURSELF of that, or of what you really do need.

And i'm still not sure what you mean by "calibrate your SWR meter
to the line". All the SWR meters i have seen are all for 50 Ohms.

Could you tell us the exact procedure?


Of course not; I know nothing about YOUR SWR meter. If you understand
how yours works, you should be able to see how to adjust it, though it
may not be worthwhile. To a close approximation, practically all of
them work by sampling the line current and voltage at a point. The
current is somehow turned into a voltage, and if you adjust either the
voltage sampling percentage or the voltage produced by a given
current, you will have adjusted the calibration impedance. At the
high frequency end, you may need to worry about reactive or
phase-shift effects. If it's not adjustable, don't you at least want
to know WHAT impedance it's calibrated for? Or at VERY least, what it
reads with a 50 ohm load?

Cheers,
Tom
  #29   Report Post  
Old August 12th 03, 06:41 PM
Richard Clark
 
Posts: n/a
Default

On Tue, 12 Aug 2003 02:48:49 -0700, Roy Lewallen
wrote:

Almost correct.

The transmitter output impedance has no effect whatsoever on the line's SWR.

Roy Lewallen, W7EL


Hi Roy,

Entirely incorrect.

Transmitter output impedance that does not conform to transmission
line Z, when presented with a mismatched load through that line, adds
mismatch uncertainty in the form of an indeterminate SWR and
indeterminate Power to the load.

This has already been demonstrated twice. This has long been
documented with NBS/NIST references going back 4 decades. There is
nothing mysterious about it at all, and it conforms to the rather
simple principles of wave interference so poorly presented by Cecil in
months past.

The authoritative site:
http://www.boulder.nist.gov/div813/index.html

Direct reference:
"Juroshek, J. R.; A Direct Calibration Method for Measuring
Equivalent Source Mismatch; Microwave J., pp. 106-118;
October 1997

Obscure references:
http://www.boulder.nist.gov/div813/r...00S_n2nNet.pdf
"With vector measurements of the generator and meter reflection
coefficients Ãg and Ãm, respectively, the power of the incident
signal am can be related to the power of the source."

http://www.boulder.nist.gov/div813/r...FRad_ARFTG.pdf
which describes radiometer calibration (perhaps too exotic for this
group)
"tests are based on two assumptions. First, the network responds
linearly to our signal ( no power compression), and second, the
radiometer is sufficiently isolated from the source impedance."
...
"One of the assumptions made in deriving eq. (2) was that the
output from the radiometer is not dependent on the source
impedance. In the construction of the radiometer, two isolators
are inserted at the input of the radiometer to isolate the
radiometer from the source."
...
"The mismatch uncertainty depends strongly on the poorly known
correlation between uncertainties in the measurements of different
reflection coefficients, and so we use the maximum of the
uncertainties obtained by assuming either complete correlation or
no correlation whatsoever."

"Forthcoming Paper: Influence of Impedance Mismatch Effects on
Measurements of Unloaded Q Factors of Transmission Mode Dielectric
Resonators"
IEEE Transaction on Applied Superconductivity

"Analysis of Interconnection Network and Mismatch in the
Nose-to-Nose Calibration
Automatic RF Techniques Group , June 15-16, 2000 , Boston, MA -
June 01, 2000
"We analyze the input networks of the samplers used in the
nose-to-nose calibration method. Our model demonstrates that the
required input network conditions are satisfied in this method and
shows the interconnection errors are limited to measurement
uncertainties of input reflection coefficients and adapter
S-parameters utilized during the calibration procedure. Further,
the input network model fully includes the effects of mismatch
reflections, and we use the model to reconcile nose-to-nose
waveform correction methods with traditional signal power
measurement techniques."

As I mentioned, obscure references. However, given the impetus of
their discussion is long known (and that I have already provided the
original references they rely on), NIST presumes the investigators
already have that basis of knowledge.

73's
Richard Clark, KB7QHC
  #30   Report Post  
Old August 12th 03, 07:02 PM
JDer8745
 
Posts: n/a
Default

He sed:

"The transmitter output impedance has no effect whatsoever on the line's SWR."

---------

THIS IS ABSOLUTELY CORRECT!

73 Jack
Reply
Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules

Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are On
Pingbacks are On
Refbacks are On


Similar Threads
Thread Thread Starter Forum Replies Last Post
50 Ohms "Real Resistive" impedance a Misnomer? Dr. Slick Antenna 255 July 29th 03 11:24 PM
The Cecilian Gambit, a variation on the Galilean Defense revisited Richard Clark Antenna 11 July 24th 03 07:30 PM


All times are GMT +1. The time now is 04:05 PM.

Powered by vBulletin® Copyright ©2000 - 2024, Jelsoft Enterprises Ltd.
Copyright ©2004-2024 RadioBanter.
The comments are property of their posters.
 

About Us

"It's about Radio"

 

Copyright © 2017