Measuring impedance with network analyzers, possible?

[mtxx]

Hi guys,

i know the main function of a network analyzer is not to measure impedance, but i still have to rely on it since i dont have a impedance analyzer.

i read in many notes and some text book, that from a 2 port measurement,
the B parameter is the series impedance.

https://www.google.com.my/url?sa=t&...uKIGfD3znSpBiyBeg&sig2=EuvqaxUuefAlMXgqaSyUOg

unfortunately most books describes the relationship between abcd parameters and impedance very briefly.
This note above, in slide 20, it says that the B parameter is obtained when port 2 is shorted. this is not mentioned in other notes.

i would like to know more in detail about ABCD parameters and its relationship with impedance. and is port 2 really needs to be shorted?
if you know more about topic, please enlighten me.

or more directly to solve my problem,
what are the most accurate way to measure impedance with VNA?

thank you very much.

 

[drkirkby]

Hi guys,

i know the main function of a network analyzer is not to measure impedance, but i still have to rely on it since i dont have a impedance analyzer.

WRONG. One of the main functions of a network IS to measure impedance. (At least if we are talking about a vector network analyser)

i read in many notes and some text book, that from a 2 port measurement,
the B parameter is the series impedance.

https://www.google.com.my/url?sa=t&...uKIGfD3znSpBiyBeg&sig2=EuvqaxUuefAlMXgqaSyUOg

unfortunately most books describes the relationship between abcd parameters and impedance very briefly.
This note above, in slide 20, it says that the B parameter is obtained when port 2 is shorted. this is not mentioned in other notes.

i would like to know more in detail about ABCD parameters and its relationship with impedance. and is port 2 really needs to be shorted?
if you know more about topic, please enlighten me.

Forget about ABCD paramters. Look up S parameters, which are easily measured with a network analyser. Once you know them, the impedance is easily calcuated. In fact, on the HP vector network analyser I use, there is a menu where you can get impedane values.

If you network analyser is not a vector one, but a scaler one, then I don't think you can measure impedance.

 

[volker_muehlhaus]

One of the main functions of a network IS to measure impedance.

Yes, but .... The network analyzer measures reflection and transmission. From there, we can calculate impedance.

If we connect the DUT at one of the ports, we can calculate the impedance from the measured reflection ρ=(Z-50Ohm)/(Z+50Ohm). We can easily see that results from reflection measurement are most accurate for DUT impedance near 50 Ohm, and less accurate for DUT with impedance that is much different from 50 ohm. It is not very accurate if you try to measure mOhm or kOhm value from network analyzer reflection measurements.

To measure a DUT impedance that is very small or very large, we might use a 2-port measurement and connect the DUT in a different way, but that is limited in precision as well.

So it really dependends on the DUT if measurement with a network analyzer gives useful accuracy.

 

[mtxx]

Yes, but .... The network analyzer measures reflection and transmission. From there, we can calculate impedance.

If we connect the DUT at one of the ports, we can calculate the impedance from the measured reflection ρ=(Z-50Ohm)/(Z+50Ohm). We can easily see that results from reflection measurement are most accurate for DUT impedance near 50 Ohm, and less accurate for DUT with impedance that is much different from 50 ohm. It is not very accurate if you try to measure mOhm or kOhm value from network analyzer reflection measurements.

To measure a DUT impedance that is very small or very large, we might use a 2-port measurement and connect the DUT in a different way, but that is limited in precision as well.

So it really dependends on the DUT if measurement with a network analyzer gives useful accuracy.

less accurate for DUT with impedance that is much different from 50 ohm.

the above statement is for the magnitude of the impedance?
or the real part of the impedance?

 

[volker_muehlhaus]

less accurate for DUT with impedance that is much different from 50 ohm.

the above statement is for the magnitude of the impedance?
or the real part of the impedance?

Magnitude.

 

[FvM]

For a close estimation of impedance measurement from S11 accuracy, you should refer to the VNA performance charts. S11 uncertainty can be converted to impedance measurement error. To get a rough idea, I would look at the impedance mapping in a Smith Chart.

 

[godfreyl]

or more directly to solve my problem,
what are the most accurate way to measure impedance with VNA?

Is this for your parallel plate capacitor experiments?

I don't know anything about VNAs, but here's an easy way to measure the impedance. The basic idea is to make a potential divider with the cap under test and 50 Ohms of resistance, then measure the attenuation at, say 100MHz.

The pic below gives an idea of the physical layout. There's two important points to notice:

• Both pieces of coax are properly terminated, so there's no problem with either reflections or lumped impedance of the coax.
• There's no big unnecessary loop of wire around the DUT to add inductance and mess up the measurements.

The measuring device can be a scope or anything else that can measure the amplitude of the signal.

 

[mtxx]

Is this for your parallel plate capacitor experiments?

Hi, yup.
still working on it.

i'm thinking of improving the fixture, reducing any unwanted fixture effects, i just realized that the coaxial cable that connects the plates are absolutely unnecessary. How silly of me to think i need to ground them, they are already grounded. *facepalm.

now need to concentrate on getting good impedance measurement. seems like vna is not a good device, buts its the only one i have. will work on it while i try to think of better alternatives.

- - - Updated - - -

The pic below gives an idea of the physical layout. There's two important points to notice:

• Both pieces of coax are properly terminated, so there's no problem with either reflections or lumped impedance of the coax.
• There's no big unnecessary loop of wire around the DUT to add inductance and mess up the measurements.

The measuring device can be a scope or anything else that can measure the amplitude of the signal.

I've look around my lab, fortunately i found that i do have a signal generator.
agilent one that is able to generate 250k-1GHz.
and i have several models of oscilloscope as well.

making the fixture in your diagram is not hard,
i'll just use my existing one, with a little modification.

I'm totally new to signal generator, never use them before, but i can read the manual and learn its basic operation.

but i need help in the Z calculations
can you elaborate in more detail how what kind of signal i should generate, and how should i process the displayed output to get the Z?
If its too much trouble, posting links on this method, or the keywords to search would be sufficient too. thank you very much for your great idea!

 

[FvM]

I don't see any advantage of the signal generator + oscilloscopa setup compared to a two port measurement suggested by Volker. The basic problems adressed in your previous threads related to the same measurement problem will stay, however:
- you need to understand the relation of impedance and S parameters
- with increasing frequency, the distributed nature of the test fixture and the impedance under test becomes effective.

Calibrating the test setup with small chip resistors respectively capacitors seems reasonable.

 

[mtxx]

I don't see any advantage of the signal generator + oscilloscopa setup compared to a two port measurement suggested by Volker. The basic problems adressed in your previous threads related to the same measurement problem will stay, however:
- you need to understand the relation of impedance and S parameters
- with increasing frequency, the distributed nature of the test fixture and the impedance under test becomes effective.

Calibrating the test setup with small chip resistors respectively capacitors seems reasonable.

indeed the S parameter and Z relationship is important.
i've look through alot to find a good relationship, currently trying the one written in this note
https://www.google.com.my/url?sa=t&...R3614SjyY-HKCG5GQ&sig2=JaZ5VvzY1Xlh8oGIqkM24A

at the last part there is a mention of how S parameter can be used to calculate impedance.
testing to see wether this works or not

 

[Element_115]

Hi guys,

i know the main function of a network analyzer is not to measure impedance, but i still have to rely on it since i dont have a impedance analyzer.

i read in many notes and some text book, that from a 2 port measurement,
the B parameter is the series impedance.

https://www.google.com.my/url?sa=t&...uKIGfD3znSpBiyBeg&sig2=EuvqaxUuefAlMXgqaSyUOg

unfortunately most books describes the relationship between abcd parameters and impedance very briefly.
This note above, in slide 20, it says that the B parameter is obtained when port 2 is shorted. this is not mentioned in other notes.

i would like to know more in detail about ABCD parameters and its relationship with impedance. and is port 2 really needs to be shorted?
if you know more about topic, please enlighten me.

or more directly to solve my problem,
what are the most accurate way to measure impedance with VNA?

thank you very much.

Your Network analyzer should have an option in the "Display" or "Format" menu to
use the "Smith Chart" . With this you can see the input impedance S11, and the Output impedance S22.