Check out the article 'A Low-Cost 100 MHz Vector Network Analyzer with USB
Interface' in the July/August 2004 issue of QEX, for measuring S pars.

Al, w6wqc

On Thu, 12 Aug 2004 23:38:00 GMT, "Joe Rocci" wrote:

Paul & Steve,

Steve,
I'm pretty confident that the phase shift will be constant and can be
calibrated out. If not, it can be made irrelevant by using two splitters,
one for the forward path sample and one for the return path sample.

Paul,
The idea is to use the splitter 'backwards"; drive the RF into one of the
splitter legs and feed the load through the common port. Assuming good
directivity, any RF coming out of the other splitter leg must be reflected
energy. If you put a sample of the forward energy into a scope's X input
(horizontal) and a sample of the reflected energy into the Y input
(vertical), you will get an elliptical display called a Lissajous pattern.
If X and Y are equal in magnitude and exactly 90 deg out of phase, this
will be a perfect circle. Any other phase angle will result in a elliptical
pattern whose inclination angle relative to the X axis (or Y axis) is a
function of the phase angle. The length of the ellipse is a function of the
magnitude. This is classic stuff...look it up if you're not familiar with

THanks for the explanation, but I doubt this idea has enough accuracy
for determining the parameters within any acceptable degree. Neat
concept, though!
--

"What is now proved was once only imagin'd." - William Blake, 1793.

Paul,
I agree. That's why I started out by saying it was a wacky idea.

Things like this make you think though. Sometimes those types of musings pay
off later down the road. One of my mentors once said: "A good engineer
remembers every good idea he ever heard. The only thing he forgets is who he
heard it from".

Joe
W3JDR


"Paul Burridge" wrote in message
...
On Thu, 12 Aug 2004 23:38:00 GMT, "Joe Rocci" wrote:

Paul & Steve,

Steve,
I'm pretty confident that the phase shift will be constant and can be
calibrated out. If not, it can be made irrelevant by using two splitters,
one for the forward path sample and one for the return path sample.

Paul,
The idea is to use the splitter 'backwards"; drive the RF into one of the
splitter legs and feed the load through the common port. Assuming good
directivity, any RF coming out of the other splitter leg must be
reflected
energy. If you put a sample of the forward energy into a scope's X input
(horizontal) and a sample of the reflected energy into the Y input
(vertical), you will get an elliptical display called a Lissajous
pattern.
If X and Y are equal in magnitude and exactly 90 deg out of phase, this
will be a perfect circle. Any other phase angle will result in a
elliptical
pattern whose inclination angle relative to the X axis (or Y axis) is a
function of the phase angle. The length of the ellipse is a function of
the
magnitude. This is classic stuff...look it up if you're not familiar with

THanks for the explanation, but I doubt this idea has enough accuracy
for determining the parameters within any acceptable degree. Neat
concept, though!
--

"What is now proved was once only imagin'd." - William Blake, 1793.

On Fri, 13 Aug 2004 12:02:26 GMT, "W3JDR" wrote:

Paul,
I agree. That's why I started out by saying it was a wacky idea.

Things like this make you think though. Sometimes those types of musings pay
off later down the road. One of my mentors once said: "A good engineer
remembers every good idea he ever heard. The only thing he forgets is who he
heard it from".

Too true!
Well don't forget this one; there's certainly a seed of possibility
there, bar a few inevitable hurdles to overcome.
--

"What is now proved was once only imagin'd." - William Blake, 1793.

"Joe Rocci" wrote in message
news:sbTSc.7003$BS3.1582@trndny04...
Paul & Steve,

Steve,
I'm pretty confident that the phase shift will be constant and can be
calibrated out. If not, it can be made irrelevant by using two splitters,
one for the forward path sample and one for the return path sample.

Paul,
The idea is to use the splitter 'backwards"; drive the RF into one of the
splitter legs and feed the load through the common port. Assuming good
directivity, any RF coming out of the other splitter leg must be reflected
energy. If you put a sample of the forward energy into a scope's X input
(horizontal) and a sample of the reflected energy into the Y input
(vertical), you will get an elliptical display called a Lissajous pattern.
If X and Y are equal in magnitude and exactly 90 deg out of phase, this
will be a perfect circle. Any other phase angle will result in a
elliptical
pattern whose inclination angle relative to the X axis (or Y axis) is a
function of the phase angle. The length of the ellipse is a function of
the
magnitude. This is classic stuff...look it up if you're not familiar with
it.
Joe
W3JDR


You guys seem too negative. I know I was pointing out difficulties, but
only to keep the limitations in mind. Actually, it ain't that wacky. It is
EXACTLY how the Professional equipment does it, just with really good
couplers etc... As long as the phase and magnitude can either be controlled
or known and calibrated out, it should be pretty good.

If you have a scope that can go to the frequency of interest, you have a
really good basis for success.
The one thing I couldn't get fixed in my mind is the phase considerations of
using one splitter for forward sampling and one for reverse. Two things to
consider a
1- As the freq gets higher, the phase of the FWD sample and REV samples get
"further apart" since they can not be at exactly at the same point on the
t-line, namely the load. Therefore, you have to adjust the line lengths to
'move' them to the same virtual point. Perhaps at HF this is a don't care.
I'm used to doing this @ 150 MHz.
2- If you are in a region where there is a phase change with frequency in
the splitter/coupler, are you sure that it is the same in both directions,
or does it not work that way and will it work to increase the error rather
than both of them tracking together. I just can't model this in my brain
without putting some things on paper... I guess, with the scope, you can
examine that and make adjustments for it by using "calibration loads" just
like the calibration kits used by the pros.

I still believe the Lissajous (I'm sure glad you can spell that) method is
not as good as simply measuring phase shift between the two signals directly
in normal scope mode. With the Lissajous, it is harder to get the
magnitudes equal and then measure random angles from the "tilt" of the oval.
Don't the Lissajous formulas all assume equal magnitude sine waves? You'd
have to compare amplitudes in normal mode and adjust the vert gain to set
them equal anyway, so why not just set the horizontal sweep speed for a
'nice' speed and then the shift is on a calibrated scale. 9 divisions for a
1/2 cycle and you have 20 degrees per cm and the little divisions are 4 and
you estimate from there. This way the phase measurement is almost 100%
independent of amplitude 'cuz you're looking only @ the zero crossings. I
don't think you can estimate 4 degrees on a Lissajous that easily.

I still wonder WHAT kind of transistors the OP wanted to measure...

--
Steve N, K,9;d, c. i My email has no u's.

On Fri, 13 Aug 2004 10:03:51 -0500, "Steve Nosko"
wrote:

I still wonder WHAT kind of transistors the OP wanted to measure...

Nothing toooo special: 2N5109s and that ilk.

--

"What is now proved was once only imagin'd." - William Blake, 1793.

John Miles wrote:

The fly in the ointment is that scopes typically don't have very high
bandwidth (more than a few MHz) in X-Y mode. The X-axis amplifier
section of even a relatively-fast analog scope like the 2467 is only
rated at 3 MHz, and you'll probably get significant phase shift long
before that point is reached

-- jm

For sure, you shouldn't use an analog scope for this, for the reason
John gives. In an analog scope, both the vertical amplifiers and the
vertical CRT deflection system are optimized for speed, but the
horizontal aren't. Even if you use identical amplifiers for X and Y, the
horizontal CRT deflection structure will limit the X bandwidth, and give
you considerable phase shift well below the cutoff frequency.

Modern digital scopes, on the other hand, should work equally well on
the two axes.

Roy Lewallen, W7EL

Use a direction coupler for forward and reverse measurement.
Use a 2 Chan. Scope great bandwith not nessesary, it is the amplitude that
is to be measured.
reflection coefficient = Vreflected/Vincident = polar magn. and angle
Chris Bowick Book page 104, reflection coefficient may be plotted directly
in the Smith Chart, and the corresponding impedance read off the chart.
Or did I missed something?


"Joe Rocci" schreef in bericht
news:T0eSc.4297$Kv2.4186@trndny09...
Steve,
A decent (not the dollar store variety) CATV splitter has directivity of
about 30 dB from 5Mhz to over 500 MHZ. I think this would do the job.

Joe
W3JDR


Steve Nosko wrote in message
...

"Paul Burridge" wrote in message
...
On Tue, 10 Aug 2004 01:53:44 GMT, "Joe Rocci" wrote:

Here's an idea that's just whacky enough that it might work...has
anyone
tried it?

If you have a dual-trace scope with enough bandwidth, you might be
able
use
Lissajous patterns. The idea would be to drive the complex load
through
a
directional coupler with fairly good directivity. A sample of the
incident
signal would go into the scope X input and a sample of the reflected
signal
would go into the Y input. With a pure reactance as a calibration
load,
adjust the X and Y scope gain for a perfect circle on the display.
Replace
the test load with a complex load, and the magnitude and inclination
of
the
line/oval display can tell you amplitude and phase of the reflected
signal,
from which any other metric can be calculated.

Nice idea, Joe! If it could only be made to work, you'd be a genius.
--


OK Actually, this looks good as long as you have a directional coupler
for
the frequency of interest! Paul didn't say. (SWR Bridge, actual
bridge,
for lower frequencies) You'll also have to allow for the difference in
location of the two samples by adjusting the line lengths to the two
scope
probes intil they are both at the same "distance" from the load.
Judging
phase shift from a circle can't be as good as simply looking at the two
waveforms. The scope horiz gain can be adjusted for some nice number of
divisions for each half cycle, say 9...
I've got a 100MHz. dual trace storage scope, but any directional
couplers around here only will be down to 130 Mhz at best... I do also
have
a good bridge for down to 5 MHz., (also about 1 meter worth of stretch
line...so it is possible as long as the power level is within the bridge
capability.
--
Steve N, K,9;d, c. i My email has no u's.

Read the whole thread....

Joe
W3JDR



bviel wrote in message
...
Use a direction coupler for forward and reverse measurement.
Use a 2 Chan. Scope great bandwith not nessesary, it is the amplitude that
is to be measured.
reflection coefficient = Vreflected/Vincident = polar magn. and angle
Chris Bowick Book page 104, reflection coefficient may be plotted directly
in the Smith Chart, and the corresponding impedance read off the chart.
Or did I missed something?


"Joe Rocci" schreef in bericht
news:T0eSc.4297$Kv2.4186@trndny09...
Steve,
A decent (not the dollar store variety) CATV splitter has directivity of
about 30 dB from 5Mhz to over 500 MHZ. I think this would do the job.

Joe
W3JDR


Steve Nosko wrote in message
...

"Paul Burridge" wrote in message
...
On Tue, 10 Aug 2004 01:53:44 GMT, "Joe Rocci"
wrote:

Here's an idea that's just whacky enough that it might work...has
anyone
tried it?

If you have a dual-trace scope with enough bandwidth, you might be
able
use
Lissajous patterns. The idea would be to drive the complex load
through
a
directional coupler with fairly good directivity. A sample of the
incident
signal would go into the scope X input and a sample of the
reflected
signal
would go into the Y input. With a pure reactance as a calibration
load,
adjust the X and Y scope gain for a perfect circle on the display.
Replace
the test load with a complex load, and the magnitude and
inclination
of
the
line/oval display can tell you amplitude and phase of the reflected
signal,
from which any other metric can be calculated.

Nice idea, Joe! If it could only be made to work, you'd be a genius.
--


OK Actually, this looks good as long as you have a directional
coupler
for
the frequency of interest! Paul didn't say. (SWR Bridge, actual
bridge,
for lower frequencies) You'll also have to allow for the difference
in
location of the two samples by adjusting the line lengths to the two
scope
probes intil they are both at the same "distance" from the load.
Judging
phase shift from a circle can't be as good as simply looking at the
two
waveforms. The scope horiz gain can be adjusted for some nice number
of
divisions for each half cycle, say 9...
I've got a 100MHz. dual trace storage scope, but any directional
couplers around here only will be down to 130 Mhz at best... I do
also
have
a good bridge for down to 5 MHz., (also about 1 meter worth of stretch
line...so it is possible as long as the power level is within the
bridge
capability.
--
Steve N, K,9;d, c. i My email has no u's.

"bviel" wrote in message
...
Use a direction coupler for forward and reverse measurement.
Use a 2 Chan. Scope great bandwith not nessesary, it is the amplitude that
is to be measured.
reflection coefficient = Vreflected/Vincident = polar magn. and angle
Chris Bowick Book page 104, reflection coefficient may be plotted directly
in the Smith Chart, and the corresponding impedance read off the chart.
Or did I missed something?

Yup, much of what was said. You simply restated it, to some degree. We
were MUCH deeper...way past the basics.

The scope has to work reasonably well at the frequency of interest. They
can die pretty fast abofe their useful frequency.

You also have to measure *phase* between the two signals.
--
Steve N, K,9;d, c. i My email has no u's.





"Joe Rocci" schreef in bericht
news:T0eSc.4297$Kv2.4186@trndny09...
Steve,
A decent (not the dollar store variety) CATV splitter has directivity of
about 30 dB from 5Mhz to over 500 MHZ. I think this would do the job.

Joe
W3JDR


Steve Nosko wrote in message
...

"Paul Burridge" wrote in message
...
On Tue, 10 Aug 2004 01:53:44 GMT, "Joe Rocci"
wrote:

Here's an idea that's just whacky enough that it might work...has
anyone
tried it?

If you have a dual-trace scope with enough bandwidth, you might be
able
use
Lissajous patterns. The idea would be to drive the complex load
through
a
directional coupler with fairly good directivity. A sample of the
incident
signal would go into the scope X input and a sample of the
reflected
signal
would go into the Y input. With a pure reactance as a calibration
load,
adjust the X and Y scope gain for a perfect circle on the display.
Replace
the test load with a complex load, and the magnitude and
inclination
of
the
line/oval display can tell you amplitude and phase of the reflected
signal,
from which any other metric can be calculated.

Nice idea, Joe! If it could only be made to work, you'd be a genius.
--


OK Actually, this looks good as long as you have a directional
coupler
for
the frequency of interest! Paul didn't say. (SWR Bridge, actual
bridge,
for lower frequencies) You'll also have to allow for the difference
in
location of the two samples by adjusting the line lengths to the two
scope
probes intil they are both at the same "distance" from the load.
Judging
phase shift from a circle can't be as good as simply looking at the
two
waveforms. The scope horiz gain can be adjusted for some nice number
of
divisions for each half cycle, say 9...
I've got a 100MHz. dual trace storage scope, but any directional
couplers around here only will be down to 130 Mhz at best... I do
also
have
a good bridge for down to 5 MHz., (also about 1 meter worth of stretch
line...so it is possible as long as the power level is within the
bridge
capability.
--
Steve N, K,9;d, c. i My email has no u's.