|
Hello Group:
I have an 8753C and am wondering what the upper limit on cable length would be if I place the cal standards at the end of a cable to do remote measurement of S11.?? In other words I am using SOL, with the cal standards placed at the end of a 100 foot length of RG-213.? Theoretically placing the measurement plane at the end of the cable.? I have already done this and the results seem to be accurate relative to an actual measurement at the that physical place.
I hope I have explained this issue sufficiently.?
Regards,
Karin Johnson??
BTW: I have had an account on the Keysight forums in the past and have recently tried to login to that site, but It won't let me.? Possibly password have changed, and I cannot create a new account.? I have posted with Keysight support and hope they return the email.?
|
You mention port extensions but then in the post you refer to calibration ,? I just want to point out that those are different things !
When you do a calibration on a VNA you move the reference plane to the reference plane of your connector/cal standard no other adjustments are needed.
Port extensions are a software correction that are normally used to move the reference plane AFTER a calibration has been performed , this is typically done when you work with fixtures or other situations where a calibration is not practical
or possible
?
toggle quoted message
Show quoted text
From: [email protected] < [email protected]>
On Behalf Of Karin Johnson via groups.io
Sent: Thursday, May 26, 2022 11:46 AM
To: [email protected]
Subject: [HP-Agilent-Keysight-equipment] Limits of the Port Extension on the Cal screen for calibrating at end of cable
?
Hello Group:
I have an 8753C and am wondering what the upper limit on cable length would be if I place the cal standards at the end of a cable to do remote measurement of S11.?? In other words I am using SOL, with the cal standards placed at the end of a 100 foot length
of RG-213.? Theoretically placing the measurement plane at the end of the cable.? I have already done this and the results seem to be accurate relative to an actual measurement at the that physical place.
I hope I have explained this issue sufficiently.?
Regards,
Karin Johnson??
BTW: I have had an account on the Keysight forums in the past and have recently tried to login to that site, but It won't let me.? Possibly password have changed, and I cannot create a new account.? I have posted with Keysight support and hope they return the
email.?
|
Hi Karin,
"Upper limit" is not as hard a cliff as your question presupposes.
Whether or not a given length of cable is too much depends on your
acceptance criteria. What are you trying to measure, and to what
accuracy do you need to measure it?
When you add cable, you get loss, so that loss subtracts directly
from your dynamic range. For example, if you are making meaurements
at a gigahertz or so, with RG-213 you can expect to lose about 10dB
in each direction, roughly speaking. If you aren't trying to measure
small deviations from an exquisitely well-matched load, you'll be
fine. If you're trying to resolve small changes in the vicinity of a
-60dB S11 null, then you may find the measurement challenging.
The other factor to consider with a long length of cable is that it
isn't rigid, so stability of your measurement can be affected. And
if you are doing a lengthy set of measurements, temperature effects
can come into play as well. Whether these are "don't cares" or
showstoppers depends on what you can tolerate. If you can tolerate
large error bars, your measurement problems become much less severe.
If you're trying to make publication-quality measurements, you may
find that difficult.
Good luck!
--Cheers,
Tom
--
Prof. Thomas H. Lee
Allen Ctr., Rm. 205
350 Jane Stanford Way
Stanford University
Stanford, CA 94305-4070
On 5/26/2022 09:45, Karin Johnson
wrote:
toggle quoted message
Show quoted text
Hello Group:
I have an 8753C and am wondering what the upper limit on cable
length would be if I place the cal standards at the end of a cable
to do remote measurement of S11.?? In other words I am using SOL,
with the cal standards placed at the end of a 100 foot length of
RG-213.? Theoretically placing the measurement plane at the end of
the cable.? I have already done this and the results seem to be
accurate relative to an actual measurement at the that physical
place.
I hope I have explained this issue sufficiently.?
Regards,
Karin Johnson??
BTW: I have had an account on the Keysight forums in the past and
have recently tried to login to that site, but It won't let me.?
Possibly password have changed, and I cannot create a new
account.? I have posted with Keysight support and hope they return
the email.?
|
One thing to consider is that RG213 is not phase stable so if you flex the cable after calibration it may affect your measurement
?
toggle quoted message
Show quoted text
From: [email protected] < [email protected]>
On Behalf Of Tom Lee via groups.io
Sent: Thursday, May 26, 2022 1:10 PM
To: [email protected]
Subject: Re: [HP-Agilent-Keysight-equipment] Limits of the Port Extension on the Cal screen for calibrating at end of cable
?
Hi Karin,
"Upper limit" is not as hard a cliff as your question presupposes. Whether or not a given length of cable is too much depends on your acceptance criteria. What are you trying to measure, and to what accuracy do you need to measure it?
When you add cable, you get loss, so that loss subtracts directly from your dynamic range. For example, if you are making meaurements at a gigahertz or so, with RG-213 you can expect to lose about 10dB in each direction, roughly speaking. If you aren't trying
to measure small deviations from an exquisitely well-matched load, you'll be fine. If you're trying to resolve small changes in the vicinity of a -60dB S11 null, then you may find the measurement challenging.
The other factor to consider with a long length of cable is that it isn't rigid, so stability of your measurement can be affected. And if you are doing a lengthy set of measurements, temperature effects can come into play as well. Whether these are "don't cares"
or showstoppers depends on what you can tolerate. If you can tolerate large error bars, your measurement problems become much less severe. If you're trying to make publication-quality measurements, you may find that difficult.
Good luck!
--Cheers,
Tom
--
Prof. Thomas H. Lee
Allen Ctr., Rm. 205
350 Jane Stanford Way
Stanford University
Stanford, CA 94305-4070
On 5/26/2022 09:45, Karin Johnson wrote:
Hello Group:
I have an 8753C and am wondering what the upper limit on cable length would be if I place the cal standards at the end of a cable to do remote measurement of S11.?? In other words I am using SOL, with the cal standards placed at the end of a 100 foot length
of RG-213.? Theoretically placing the measurement plane at the end of the cable.? I have already done this and the results seem to be accurate relative to an actual measurement at the that physical place.
I hope I have explained this issue sufficiently.?
Regards,
Karin Johnson??
BTW: I have had an account on the Keysight forums in the past and have recently tried to login to that site, but It won't let me.? Possibly password have changed, and I cannot create a new account.? I have posted with Keysight support and hope they return the
email.?
?
|
Hello,
There is one more issue to consider. Long lines give delays. Once
I wanted to measure the loss of a coaxial cable on a vna. Was an
interesting experience: no response. Until I realized the receiver
had already traveled higher up in frequency at the time the
response was back. You need to consider a delay in between
frequency steps, long enough so as to give the receiver time to
read the response. This is a documented feature (of course).
Good luck,
Harke, PA0HRK
On 26/05/2022 20:09, Tom Lee wrote:
toggle quoted message
Show quoted text
Hi Karin,
"Upper limit" is not as hard a cliff as your question presupposes.
Whether or not a given length of cable is too much depends on your
acceptance criteria. What are you trying to measure, and to what
accuracy do you need to measure it?
When you add cable, you get loss, so that loss subtracts directly
from your dynamic range. For example, if you are making
meaurements at a gigahertz or so, with RG-213 you can expect to
lose about 10dB in each direction, roughly speaking. If you aren't
trying to measure small deviations from an exquisitely
well-matched load, you'll be fine. If you're trying to resolve
small changes in the vicinity of a -60dB S11 null, then you may
find the measurement challenging.
The other factor to consider with a long length of cable is that
it isn't rigid, so stability of your measurement can be affected.
And if you are doing a lengthy set of measurements, temperature
effects can come into play as well. Whether these are "don't
cares" or showstoppers depends on what you can tolerate. If you
can tolerate large error bars, your measurement problems become
much less severe. If you're trying to make publication-quality
measurements, you may find that difficult.
Good luck!
--Cheers,
Tom
--
Prof. Thomas H. Lee
Allen Ctr., Rm. 205
350 Jane Stanford Way
Stanford University
Stanford, CA 94305-4070
On 5/26/2022 09:45, Karin Johnson
wrote:
Hello Group:
I have an 8753C and am wondering what the upper limit on cable
length would be if I place the cal standards at the end of a
cable to do remote measurement of S11.?? In other words I am
using SOL, with the cal standards placed at the end of a 100
foot length of RG-213.? Theoretically placing the measurement
plane at the end of the cable.? I have already done this and the
results seem to be accurate relative to an actual measurement at
the that physical place.
I hope I have explained this issue sufficiently.?
Regards,
Karin Johnson??
BTW: I have had an account on the Keysight forums in the past
and have recently tried to login to that site, but It won't let
me.? Possibly password have changed, and I cannot create a new
account.? I have posted with Keysight support and hope they
return the email.?
|
Could you have used a single point frequency measurement n the VNA? I know the 8510 will do that.
But I am sure the experience was educational - thanks for the education - it would have taken me a bit of time to catch on to that.
Cheers!
Bruce
Quoting "Harke Smits via groups.io" <yrrah@...>:
toggle quoted message
Show quoted text
Hello,
There is one more issue to consider. Long lines give delays. Once I wanted to measure the loss of a coaxial cable on a vna. Was an interesting experience: no response. Until I realized the receiver had already traveled higher up in frequency at the time the response was back. You need to consider a delay in between frequency steps, long enough so as to give the receiver time to read the response. This is a documented feature (of course).
Good luck,
Harke, PA0HRK
On 26/05/2022 20:09, Tom Lee wrote:
Hi Karin,
"Upper limit" is not as hard a cliff as your question presupposes. Whether or not a given length of cable is too much depends on your acceptance criteria. What are you trying to measure, and to what accuracy do you need to measure it?
When you add cable, you get loss, so that loss subtracts directly from your dynamic range. For example, if you are making meaurements at a gigahertz or so, with RG-213 you can expect to lose about 10dB in each direction, roughly speaking. If you aren't trying to measure small deviations from an exquisitely well-matched load, you'll be fine. If you're trying to resolve small changes in the vicinity of a -60dB S11 null, then you may find the measurement challenging.
The other factor to consider with a long length of cable is that it isn't rigid, so stability of your measurement can be affected. And if you are doing a lengthy set of measurements, temperature effects can come into play as well. Whether these are "don't cares" or showstoppers depends on what you can tolerate. If you can tolerate large error bars, your measurement problems become much less severe. If you're trying to make publication-quality measurements, you may find that difficult.
Good luck!
--Cheers,
Tom
--
Prof. Thomas H. Lee
Allen Ctr., Rm. 205
350 Jane Stanford Way
Stanford University
Stanford, CA 94305-4070
On 5/26/2022 09:45, Karin Johnson wrote:
Hello Group:
I have an 8753C and am wondering what the upper limit on cable length would be if I place the cal standards at the end of a cable to do remote measurement of S11.?? In other words I am using SOL, with the cal standards placed at the end of a 100 foot length of RG-213.? Theoretically placing the measurement plane at the end of the cable.? I have already done this and the results seem to be accurate relative to an actual measurement at the that physical place.
I hope I have explained this issue sufficiently.
Regards,
Karin Johnson
BTW: I have had an account on the Keysight forums in the past and have recently tried to login to that site, but It won't let me.? Possibly password have changed, and I cannot create a new account.? I have posted with Keysight support and hope they return the email.
|
I realized a couple of hours after sending the initial post, that I was incorrect with associating Calibration with the Port Extensions.
Here is, I hope a better explanation of what I am trying to do.
First off the measurement frequency is 14 MHz.? So not too beyond the capability of the 8753C.
I want to place the measurement plane at the end of a 100 foot length of RG213.? So I place the SOL standards, one at a time at the
end of the cable.? Supposedly the 8753C measurement plane is now at the end of the cable.
Now I can take S11 measurements relative to the measurement plane at the end of the cable.? My question should have been
" Is there a limit in terms of how long a cable can be before this calibration technique fails."
This is probably related to the storage of the cal correction factors in the memory of the 8753C.? Is it via floating point numbers, or integer numbers?
How large or small can the calibration correction values be???
I seem to be able to obtain valid data using this technique, but wanted to ask on the Keysight forum with some of the folks who actually designed
the instrument.? I think Dr. Joel it still active there.?
My rant about not being able to login to the Keysight forums still exists.? It seems their login/validation screen seems broken.
At one time I had an account over there but they have seemed to have changed the supporting vendor for the forum sites.
As always,
Regards,
Karin Johnson
|
On 5/27/22 19:50, Karin Johnson wrote: This is probably related to the storage of the cal correction factors in the memory of the 8753C.? Is it via floating point numbers, or integer numbers? I don't know about the 8753 in particular, but for the 8510, for example, the calibration correction factors are floating point numbers that are the coefficients for the third-order polynomial that describes the curve (C or L) in the frequency domain for the calibration standard in question. I'd be very surprised if the 8753 did it any differently. Dave Kirkby should be able to chime in on this with more information. -Dave -- Dave McGuire, AK4HZ New Kensington, PA
|
As far as I am aware, the port extension feature only adds pure phase. You enter a delay parameter, or a physical length + velocity factor. I don't recall ever seeing any option to dial in any other parameters (e.g., loss), but that may be due to my rarely using port extensions, and then only to shift the reference plane by tiny amounts.
--Tom
--
Prof. Thomas H. Lee
Allen Ctr., Rm. 205
350 Jane Stanford Way
Stanford University
Stanford, CA 94305-4070
toggle quoted message
Show quoted text
On 5/27/2022 18:12, Dave McGuire wrote: On 5/27/22 19:50, Karin Johnson wrote:
This is probably related to the storage of the cal correction factors in the memory of the 8753C.? Is it via floating point numbers, or integer numbers? ? I don't know about the 8753 in particular, but for the 8510, for example, the calibration correction factors are floating point numbers that are the coefficients for the third-order polynomial that describes the curve (C or L) in the frequency domain for the calibration standard in question.? I'd be very surprised if the 8753 did it any differently.
? Dave Kirkby should be able to chime in on this with more information.
??????????? -Dave
|
Karin,
As I understand your question, you are concerned that the calibration data for a long cable requires larger numerical values due to the many 360 degree repeats of the phase.
The analyzer determines the “N” in N*360 by looking a the series of points rather than at a single frequency value. If you were to measure phase at one frequency only, the analyzer would not know if N=1 or N=10000000. It would merely give you a value within one 360 degree range.
The main thing to be aware of is the delay, as Harke mentioned. I would make this measurement using the “LIST” (or "stepped")? sweep and I would slow the sweep speed down as much as necessary to get to a point of diminishing returns. For the same reason, using a narrow IF bandwidth makes things worse.
The other problem is what Tom mentioned, the cable loss reduces your dynamic range.
Vladan
|
One more comment regarding maximum cable length. If you have option 010 (time domain), there is a limit to the time interval you can display, but I think that's tied to the math for the transform. Just to make sure that there is no such problem in the frequency domain, compare your measurement with a small number of points with a measurement with the maximum number of points. If you pick some low number of points (e.g. 11) and the analyzer no longer gives good results, it's because it couldn't make sense of "N".
Vladan
|
On Thu, 26 May 2022 at 17:45, Karin Johnson < karinann@...> wrote: Hello Group:
I have an 8753C and am wondering what the upper limit on cable length would be if I place the cal standards at the end of a cable to do remote measurement of S11.?? In other words I am using SOL, with the cal standards placed at the end of a 100 foot length of RG-213.? Theoretically placing the measurement plane at the end of the cable.? I have already done this and the results seem to be accurate relative to an actual measurement at the that physical place.
I hope I have explained this issue sufficiently.?
Regards,
Karin Johnson??
In addition to the points raised by others, there is another factor to consider. You will need to slow the sweep speed down, below that automatically chosen for you. Otherwise the frequency of the oscillator will have shifted outside the passband of the receiver by the time it gets back to the receiver. This is well documented in one of the HP Application notes.
Dave
|
On Sat, 28 May 2022 at 02:22, Tom Lee < tomlee@...> wrote: As far as I am aware, the port extension feature only adds pure phase.
You enter a delay parameter, or a physical length + velocity factor. I
don't recall ever seeing any option to dial in any other parameters
(e.g., loss), but that may be due to my rarely using port extensions,
and then only to shift the reference plane by tiny amounts.
--Tom
--
Prof. Thomas H. Lee
That's true on the 8753, but the PNA-X, and probably some lesser instruments, allow loss to be added too. But no port extension is needed here.
Dave
|
On Sat, 28 May 2022 at 00:50, Karin Johnson < karinann@...> wrote: I realized a couple of hours after sending the initial post, that I was incorrect with associating Calibration with the Port Extensions.
Here is, I hope a better explanation of what I am trying to do.
First off the measurement frequency is 14 MHz.? So not too beyond the capability of the 8753C.
? A wavelength of roughly 20 m.
I want to place the measurement plane at the end of a 100 foot length of RG213.? ? Em, wish you would work in metric! Let's call that 30 m. So your cable
is 1.5 wavelengths long, which is far from unusual for a VNA.? My 20 GHz
VNA has cables that are 600 mm long, so 40 wavelengths. It calibrates okay, with cables 40 wavelengths long.
So I place the SOL standards, one at a time at the
end of the cable.? Supposedly the 8753C measurement plane is now at the end of the cable.
? More precisely, at the reference plane of the connector on the end of your cable. Depending on the connector, the reference plane may or may not be near the end of the connector. The reference plane of the connector is the outer conductor mating plane. In the case of a female N, it is well inside the connector - roughly 8 mm, as I measured with a bit of wire and ruler next to me. ????
Now I can take S11 measurements relative to the measurement plane at the end of the cable.? My question should have been
" Is there a limit in terms of how long a cable can be before this calibration technique fails."
This is probably related to the storage of the cal correction factors in the memory of the 8753C.? Is it via floating point numbers, or integer numbers?
How large or small can the calibration correction values be???
? The calibration factors to be entered into the VNA include the offset delays (lengths) of the calibration standards. They are only of the order of mm. The fact your cable is 100 m long does not change those numbers.
I seem to be able to obtain valid data using this technique, but wanted to ask on the Keysight forum with some of the folks who actually designed
the instrument.? I think Dr. Joel it still active there.? ?
I would certainly slow the sweep speed down, and see if that changes your results. If it does, then you have a problem.
?
My rant about not being able to login to the Keysight forums still exists.? It seems their login/validation screen seems broken.
At one time I had an account over there but they have seemed to have changed the supporting vendor for the forum sites.
I wish Keysight would just use some open-source forum software, instead of all this proprietary crap, which they keep needing to change.?
As always,
Regards,
Karin Johnson
? Dave
|
Dr. Kirkby: " You will need to slow the sweep speed down, below that automatically chosen for you. Otherwise the frequency of the oscillator will have shifted outside the passband of the receiver by the time it gets back to the receiver. This is well documented in one of the HP Application notes"
-----------
Attached is an example.
|
Thanks to all who replied.
I used to design fiber optic systems, among my other system design tasks, and am aware of the fact the receiver cannot sometimes track the source because of long delays in the cable or fiber.? I have slowed the sweep and have not noticed any appreciable discrepancies with the data.? So my conclusion is that using a 30 Meter (my tip of the hat to Dr, Kirkby) cable should enact the calibration, and would not be an issue for the 8753C.? I agree with Tom about only the phase being a factor in the port extensions.? But I am not using port extensions.? I have a piece of software that will allow me to enter data for measurements taken at various points along a fixed length of cable, and can "BACk OUT" the cable to reflect the actual data at the calibration plane.
All of that experimentation with that software and real world measurements seem to correlate correctly.??? This question from the very beginning, excepting my original mistake, was somewhat academic.? I have been using this method for a number of years and it does work.?
I guess recently I've become a bit bored looking for some new project to undertake.
I truly respect the groups knowledge and time putting up with my questions.
Regards,
Karin
|
Not sure if this has been mentioned, and it is entirely academic, but the attenuation of the cable could degrade accuracy if the cables have significant enough loss.? However with your cable (RG213), length (100ft) and freq (14MHz) this loss is small. and?will not make a significant difference.
If you pushed the frequency up into the GHz, then it could be significant. Cable loss will eat into the available dynamic range of the return loss calculations, and if there is enough loss then the mathematical precision and instrument stability will start to limit measurement accuracy.
i.e. with enough loss, then the signal reflected back to the VNA by the open/short and the zero reflection for the load will start to look the same.
Cheers, Roger
On Sat, 28 May 2022, 11:50 AM Karin Johnson, < karinann@...> wrote: I realized a couple of hours after sending the initial post, that I was incorrect with associating Calibration with the Port Extensions.
Here is, I hope a better explanation of what I am trying to do.
First off the measurement frequency is 14 MHz.? So not too beyond the capability of the 8753C.
I want to place the measurement plane at the end of a 100 foot length of RG213.? So I place the SOL standards, one at a time at the
end of the cable.? Supposedly the 8753C measurement plane is now at the end of the cable.
Now I can take S11 measurements relative to the measurement plane at the end of the cable.? My question should have been
" Is there a limit in terms of how long a cable can be before this calibration technique fails."
This is probably related to the storage of the cal correction factors in the memory of the 8753C.? Is it via floating point numbers, or integer numbers?
How large or small can the calibration correction values be???
I seem to be able to obtain valid data using this technique, but wanted to ask on the Keysight forum with some of the folks who actually designed
the instrument.? I think Dr. Joel it still active there.?
My rant about not being able to login to the Keysight forums still exists.? It seems their login/validation screen seems broken.
At one time I had an account over there but they have seemed to have changed the supporting vendor for the forum sites.
As always,
Regards,
Karin Johnson
|
Dave McGuire
Dr. David Kirkby, Kirkby Microwave Ltd