|
Hi all,
I can't find any info on how to connect a VNA for testing dummy loads. All the set-ups I'm seeing are for thru-line measurements for DUTs like filters which of course have input and output ports. I'm not seeing any suggested set-ups where the dummy load is a termination with only an input! Is that too much to ask??
So I'd like to characterize various 50 ohm dummy loads I have using polar plots. I know how to set the controls for the bandwidth and whatnot I want to see; it's just hooking up the DUT I'm struggling with; which cables go where and is a 'tee' needed? Can some kind soul assist, please?
THanks!
|
On March 7, 2020 4:52:47 PM "Jinxie" <paul666@...> wrote: I can't find any info on how to connect a VNA for testing dummy loads. All the set-ups I'm seeing are for thru-line measurements for DUTs like filters which of course have input and output ports. I'm not seeing any suggested set-ups where the dummy load is a termination with only an input! Is that too much to ask??
So I'd like to characterize various 50 ohm dummy loads I have using polar plots. I know how to set the controls for the bandwidth and whatnot I want to see; it's just hooking up the DUT I'm struggling with; which cables go where and is a 'tee' needed? Can some kind soul assist, please? It's simplicity in itself: just connect it to port one, and measure S11. -Dave -- Dave McGuire, AK4HZ New Kensington, PA
|
I don't have a "port one". I've got one RF out and 3 inputs, R, A & B. That's all.
|
On March 7, 2020 5:03:48 PM "Jinxie" <paul666@...> wrote: I don't have a "port one". I've got one RF out and 3 inputs, R, A & B. That's all. Is this an 8505A? 8753? You need either a good splitter or (better) an S-parameter test set. THEN connect to port one. :) -Dave -- Dave McGuire, AK4HZ New Kensington, PA
|
On March 7, 2020 5:08:22 PM "Dave McGuire" <mcguire@...> wrote: On March 7, 2020 5:03:48 PM "Jinxie" <paul666@...> wrote:
I don't have a "port one". I've got one RF out and 3 inputs, R, A & B. That's all. Is this an 8505A? 8753?
You need either a good splitter or (better) an S-parameter test set. THEN connect to port one. :) Ack, sorry a directional coupler. But best to find an appropriate S-parameter test set for your VNA. -Dave -- Dave McGuire, AK4HZ New Kensington, PA
|
It's an HP 8754A. And I have a suitable phase-matched splitter (one in, three out). I was under the impression S parameter test sets were only needed for testing active devices under power. This is totally passive (obviously).
|
On March 7, 2020 5:15:49 PM "Jinxie" <paul666@...> wrote: It's an HP 8754A. And I have a suitable phase-matched splitter (one in, three out). I was under the impression S parameter test sets were only needed for testing active devices under power. This is totally passive (obviously). S-parameter test sets don't have anything to do with active vs. passive. First thing's first, please read up on S ("scattering") parameters a bit, as I'm typing on my phone and explaining it would be painful in this environment. The essence of it is it turns your source, R, A, and B ports into two ports, 1 and 2, with appropriate switching and directional couplers (or bridges) to look at transmission from port 1 to port 2 (S21) transmission from port 2 to port 1(S12), reflection of energy coming from port 1 back into port 1 (S11), and the same for port 2 (S22). For a dummy load (or "terminator") you want to measure S11. This is one of the big basic things that VNAs are built to do. -Dave -- Dave McGuire, AK4HZ New Kensington, PA
|
On Sat, Mar 7, 2020 at 01:52 PM, Jinxie wrote:
dummy load is a termination with only an input!
If its a parallel connection, yes. Why not a series connection? VNAs are intended to do "thru' measurments on systems of a characteristic impedaance (probably 50 ohms) Some are able to do "single port' i.e. connect the load across the TX port and read that. Thats not actually a 'network analysis' because a dummy load is not exactly a network (read about Fosters networks). Why cant you connect the dummy load in series between the TX and another input? (other than the obvious mechanical constraints.) That would test the resistive element by itself. Then test the rest without the load. Could reveal some interesting information such as "oh, the stray reactance is in the fixture, not the resistive load" A good RF generator would work as well... read it with a scope.
|
As Dave said, you need a coupler...probably something like this (although you probably don't need all this frequency range: .?
The basic setup is:
- Connect RF out to the RF input of the coupler (if you need to use R for phase locking, then you will need a splitter to send half to the R input and the other half to the coupler.)
- Connect A, (or B, or C) to the coupled arm
- Place a short or open on coupler test port
- Normalize (grease pencil?)
- Connect your Load.? The drop in the displayed signal is your match.
Caveats:? This is a quick and dirty way to get an approximate measurement, but it is not the most accurate and is heavily dependent upon the source match.? For more accuracy you would need to average out the open and load result and reference to that average.?? Inserting a? good-quality 3 or 6dB? pad before the coupler would help with any source match issues.? I don't remember what the 8754 is capable of.? Modern analyzers handle all the math for you and a one-port cal would consist of measuring an open, a short, and a load.? Once done, you would see basically a flat line at 0.? The load would then be connected and the resulting trace would be your match.
?
?
|
I came in to the middle of this. But I sense mumbo-jumbo.
After 40 years as a professional microwave engineer working with (and owning many) VNA's,I can say definitively that 1 port networks exist, as well as two port, even N port.
Dummy loads, one port devices, are high power terminations, easily assessed with S11 measurements, just like antennas are!
VNA's are not "only" for "thru" measure, not by any means. Thats the secret of the the directivity coupler or bridge!
Why a series connection? Shunt 1 ports are just as easy to evaluate.
A scope and generator will not provide vector measure, as such. Merely scalar.
I guess there must be something more to this thread that I missed. In that case, sorry.
If not, then the author below needs to learn a bit more about VNA's.
Jeff Kruth
In a message dated 3/7/2020 6:43:28 PM Eastern Standard Time, k_8_b_y_p@... writes:
VNAs are intended to do "thru' measurments on systems of a characteristic impedaance (probably 50 ohms) Some are able to do "single port' i.e. connect the load across the TX port and read that. Thats not actually a 'network analysis' because a dummy load is not exactly a network (read about Fosters networks). Why cant you connect the dummy load in series between the TX and another input? (other than the obvious mechanical constraints.)
|
Jinxie,
Ideally, you would use a 85044a transmission / reflection test set. Those things cost quite bit, so you could use some kind of a bridge or coupler instead. A bridge out of a 8505 VNA would work fine to 1.3 GHz and probably even to 2GHz if your 8754 has a doubler.
Vladan
|
On 3/7/20 6:43 PM, David C wrote: dummy load is a termination with only an input!
If its a parallel connection, yes. Why not a series connection?
VNAs are intended to do "thru' measurments on systems of a
characteristic impedaance (probably 50 ohms) Uhhh...wha?? No. That's one thing that they can do, but the assertion that they are "intended" to only do thru measurements is false. Very, very false. Some are able to do "single port' i.e. connect the load across the TX
port and read that. I've never seen a VNA that is unable to do "single port". And from where I'm sitting right now I can throw a wad of gum at five different VNAs. Thats not actually
a 'network analysis' because a dummy load is not exactly a network (read
about Fosters networks). In fact is *is* exactly a network, and performing an S11 analysis on one is in fact, by definition, network analysis. Wow. Please read up on what a network is and what VNAs do. -Dave -- Dave McGuire, AK4HZ New Kensington, PA
|
First, you did not state what equipment you are using; that makes a lot of difference. Some network analyzers like the NanoVNA are complet and self contained; others are not.
With those ports mentioned, you need a splitter on the output of the network analyzer. One output of the splitter will go to the R input. The other output will go to either a return loss bridge or directional coupler
.
The RF output of a RL bridge will go to the input one or two, if they have an RF input. If not, you will need a compatible detector. If you are using a directional coupler, the coupled arm will go to the detector.
Next, you will have to calibrate the bridge or directional coupler.?
There are many variations depending on the equipment that you have available. Once you have assembled the equipment and accessories, it is relatively simple to calibrate the setup and measure loads.
If you contact me directly I would be happy to provide more information.
Stuart K6YAZ
Los Angeles, USA
|
What you missed Jeff is that he didn't state what sort of test equipment he had, and what frequency range he intended to use as well as connector types.
With a modern piece of equipment like the NanoVNA (less than $50 USD) you don't need any accessories; they are all there. The only limitation is the frequency range.
A fancy network analyzer may need many accessories to make a simple S11 measurement (even if only scalar). Without more information from the user, it's a question than can't be answered.
Stuart K6YAZ
Los Angeles, USA
|
I can chime in as a ME wearing an EE hat :) I see S11 measurement as essentially a shaker table measurement where you are not allowed to touch anything but the interface plane. Imagine a shaker and a load cell + an accelerometer between the shaker and the DUT. Or think of holding a short slinky hanging down from your arm. Close your eyes - you can only feel the interface plane. Shake it in various ways and you get the slinky¡¯s S11 response.
Then stick a load cell with a test mass and another accelerometer stuck to it on the far end, and you get S12. Swap the ends and you can measure S22 and S21.
In one ¡°interpretation¡±, the load cell measures the mechanical equivalent of current. The accelerometer measures the mechanical equivalent of voltage. In such interpretation (one of many), mass is the equivalent of resistance (I=V*R vs F=a*m). If we let acceleration represent a higher derivative of voltage, you can also find mechanical equivalents for capacitance and resistance. If we let the velocity be the equivalent of voltage, then mechanical damping becomes equivalent to real impedance, and inertia is equivalent to imaginary impedance. And so on, and pardon for mathematical mistakes here. But the gist of it is: you can certainly derive mechanical analogues of electrical circuits. The reverse of it was an actual job description not too long ago, when analog computers simulated mechanical systems and were used to model such.
In an S-parameter test set - and anyone actually knowing what they talk about please correct me - directional couplers and RF switches do the job of separating the energy flow out of a port from the flow into the port.
Circulators are a particularly clever kind of a directional coupler that¡¯s connected back to itself (at least I think of them that way) - the energy can make it all the way back to the 1st port if the ports on the way reflect it ¡°back¡± - back into the respective ports, but forward in the circulator (there¡¯s an arrow on it that shows which way is ¡°forward¡±, at least on the few I got). A circulator is the RF equivalent of a lazy Susan :)
The ¡°problem¡± with bare VNAs is that the S-parameter test set is a separate device, and a costly one from what I gather, yet one without which the capability of the VNA is quite untapped. The ¡°bare¡± VNA is the RF power source and one or more fancy vector voltmeters. A test set is what connects those up in a circuit that allows a particular kind of very useful measurements (one- and two-port S-parameter measurements). It¡¯s what lets you use those load cells and accelerometers for something useful :)
Best not let the ME ramble too much, though.
Cheers, Kuba
toggle quoted message
Show quoted text
7 mars 2020 kl. 7:41 em skrev Dave McGuire <mcguire@...>:
?On 3/7/20 6:43 PM, David C wrote:
dummy load is a termination with only an input!
If its a parallel connection, yes. Why not a series connection?
VNAs are intended to do "thru' measurments on systems of a
characteristic impedaance (probably 50 ohms) Uhhh...wha?? No. That's one thing that they can do, but the
assertion that they are "intended" to only do thru measurements is
false. Very, very false.
Some are able to do "single port' i.e. connect the load across the TX
port and read that. I've never seen a VNA that is unable to do "single port". And from
where I'm sitting right now I can throw a wad of gum at five different VNAs.
Thats not actually
a 'network analysis' because a dummy load is not exactly a network (read
about Fosters networks). In fact is *is* exactly a network, and performing an S11 analysis on
one is in fact, by definition, network analysis.
Wow. Please read up on what a network is and what VNAs do.
-Dave
--
Dave McGuire, AK4HZ
New Kensington, PA
|
Leave it to the ME to mess up the Ohm¡¯s law :) I meant of course F=ma vs I=V/R, and mass is then the equivalent of conductance, not resistance. D¡¯oh. - Kuba
toggle quoted message
Show quoted text
7 mars 2020 kl. 10:29 em skrev Kuba Ober <kuba@...>:
?I can chime in as a ME wearing an EE hat :) I see S11 measurement as essentially a shaker table measurement where you are not allowed to touch anything but the interface plane. Imagine a shaker and a load cell + an accelerometer between the shaker and the DUT. Or think of holding a short slinky hanging down from your arm. Close your eyes - you can only feel the interface plane. Shake it in various ways and you get the slinky¡¯s S11 response.
Then stick a load cell with a test mass and another accelerometer stuck to it on the far end, and you get S12. Swap the ends and you can measure S22 and S21.
In one ¡°interpretation¡±, the load cell measures the mechanical equivalent of current. The accelerometer measures the mechanical equivalent of voltage. In such interpretation (one of many), mass is the equivalent of resistance (I=V*R vs F=a*m). If we let acceleration represent a higher derivative of voltage, you can also find mechanical equivalents for capacitance and resistance. If we let the velocity be the equivalent of voltage, then mechanical damping becomes equivalent to real impedance, and inertia is equivalent to imaginary impedance. And so on, and pardon for mathematical mistakes here. But the gist of it is: you can certainly derive mechanical analogues of electrical circuits. The reverse of it was an actual job description not too long ago, when analog computers simulated mechanical systems and were used to model such.
In an S-parameter test set - and anyone actually knowing what they talk about please correct me - directional couplers and RF switches do the job of separating the energy flow out of a port from the flow into the port.
Circulators are a particularly clever kind of a directional coupler that¡¯s connected back to itself (at least I think of them that way) - the energy can make it all the way back to the 1st port if the ports on the way reflect it ¡°back¡± - back into the respective ports, but forward in the circulator (there¡¯s an arrow on it that shows which way is ¡°forward¡±, at least on the few I got). A circulator is the RF equivalent of a lazy Susan :)
The ¡°problem¡± with bare VNAs is that the S-parameter test set is a separate device, and a costly one from what I gather, yet one without which the capability of the VNA is quite untapped. The ¡°bare¡± VNA is the RF power source and one or more fancy vector voltmeters. A test set is what connects those up in a circuit that allows a particular kind of very useful measurements (one- and two-port S-parameter measurements). It¡¯s what lets you use those load cells and accelerometers for something useful :)
Best not let the ME ramble too much, though.
Cheers, Kuba
7 mars 2020 kl. 7:41 em skrev Dave McGuire <mcguire@...>:
?On 3/7/20 6:43 PM, David C wrote:
dummy load is a termination with only an input!
If its a parallel connection, yes. Why not a series connection?
VNAs are intended to do "thru' measurments on systems of a
characteristic impedaance (probably 50 ohms) Uhhh...wha?? No. That's one thing that they can do, but the
assertion that they are "intended" to only do thru measurements is
false. Very, very false.
Some are able to do "single port' i.e. connect the load across the TX
port and read that. I've never seen a VNA that is unable to do "single port". And from
where I'm sitting right now I can throw a wad of gum at five different VNAs.
Thats not actually
a 'network analysis' because a dummy load is not exactly a network (read
about Fosters networks). In fact is *is* exactly a network, and performing an S11 analysis on
one is in fact, by definition, network analysis.
Wow. Please read up on what a network is and what VNAs do.
-Dave
--
Dave McGuire, AK4HZ
New Kensington, PA
|
I think that an honest question regarding measuring SWR or return loss has gone off on a strange tangent. The answer to the question is relatively easy to answer if we know what frequency range is involved, and perhaps the type of connectors will be used and what the person currently has on hand.
What we don't know is what sort of equipment is available to the person who asked the question.
The S11 measurement is a very common one and is not "black magic" to perform. Usually the results are expressed as return loss (in decibels) or as voltage standing wave ratios.
Since there are several ways to make the measurement, what is needed to describe the procedure to the person asking for help, is to know what he currently has available to make the measurement. I assume that he doesn't want to spend a lot of money buying additional equipment.?
He asked for assistance, and that is what I offered. Possibly free advice is worth what you pay for it.
Stuart K6YAZ
Los Angeles, USA
|
Hi Jinxie,
You need your existing splitter (or better a 2 resistor 1 in 2 out one) an a Return Loss Bridge (RLB) either professional or one like this
Connect the SA output to the splitter, 1 splitter output to SA R input, other splitter output to RLB input (terminate any unused splitter outputs) RLB output to SA A input. RLB "REF" to known good load. and loa under test to RLB "Test".? zero and infinate calibration use another known good load and short on RLB test post.
A Reflection test set like the HP 8502A contains these parts and an attenuator in a handy package.
Robert G8RPI.
|
Thanks, Robert (and others who've assisted here with this query).
A bit of background might be appropriate. I'm just a hobbyist and radio ham, albeit a bit more advanced than most. I bought this analyser 20 years ago when 1.3Ghz was still pretty respectable! I got it cheap because the polar plot function no longer worked. It's been in storage up until last Xmas, when I finally found time to devote to re-commissioning it. One or two other faults had arisen over the time it had not been used, so it took a while. However, I now believe it's back to full functionality again - but I needed to test it, hence the original question. I have about a dozen 50 ohm dummy loads of all sizes lying around so that seemed like the obvious thing to use for testing purposes. I just want to know if it really works! Once I've established that I've fixed the polar plotting I can then go on to learn the intricacies of vector network analysis on a proper, rigorous basis. This analyser is a keeper AFAIC. I'm not going to be selling it as 1.3Ghz is more than enough for my needs. All it is missing now is one cal knob, but that's another story. I hope this clarifies things a bit better.
Thanks again.
|
Interesing the TROLLING on this forum
Dave, you deliberately misquoted me in order to play Forum Expert and falsely contradict my statement.
Kindly read more carefully and dont put words in my mouth (keyboard) that I didnt say.
Your resposne is dishonest.
Since you didnt read my statement, Ill repeat it:
"VNAs are intended to do "thru' measurments"
You deliberately mis quoted what I said and changed it to:
"the
assertion that they are "intended" to only do thru measurements is
false. Very, very false."
You deliberately took my statement out of context to contradict , censor and bully
Wont work
I did not say "only"
You invented that.
PS youre off topic attacking the messenger and not responding to OP. Thats not how we conduct ourselves on the Internet.
|
Jinxie
Dave McGuire