Measuring Impedance of a filter

[jf1020]

I want to know the input and output impedance of a bandpass filter on a 50 ohm NETWORK analyzer. I have measured it and have the s2p file but I am not certain and I am looking at the results correctly.

How do I go about knowing the filter's impedance at a given frequency?

Here is the circuit :



Here is the smith chart of measured s2p data:


Thanks!

 

[albbg]

You can't do it with a spectrum analyzer unless you build a bench with a generator and directional coupler, but you'll only have the modulus of Z . For a complete (and precise) you need a network analyzer.

How did you extract the S2P file ? From your description I suppose it's a result of a simulation.

 

[jf1020]

Oops that was a type-o. The s2p is measured from a network analyzer. I have edited the type-o

 

[FvM]

Most VNAs can be switched to impedance display for the markers. But you can also convert s11/s22 into impedance https://en.wikipedia.org/wiki/Impedance_parameters

Obviously the filter impedance is far from 50 ohm matching, which would be quite plausible if it's designed for 1 ohm impedance as suggested in the schematic.

I presume you know that you need to calibrate the VNA for a suitable reference plane or at least perform port extension to the filter terminals.

 

[BigBoss]

The impedance is obvious..
Change marker type Normalized to 50 Ohm, you will see Input and Output Impedance..
I didn't understand what the problem is ???

 

[albbg]

I'm not sure, but I think jf1020 wants to measure the input and output matching of a Zo=1 ohm filter using a standard 50 ohm VNA. So he's asking how to convert the S2P parameter measured in a 50 ohm system into the same measurement as if it was kept with a 1 ohm VNA.
Or it could be the filter in the sketch is normalized for a Zo= 1 ohm so it has to be de-normalize to 50 ohm.

jf1020 could you please clarify ?

 

[jf1020]

What is the normalized option called?
I dont recall seeing something like that?

The impedance is obvious..
Change marker type Normalized to 50 Ohm, you will see Input and Output Impedance..
I didn't understand what the problem is ???

- - - Updated - - -

Yes measure 1 ohm filter with
50 ohm analyzer

I'm not sure, but I think jf1020 wants to measure the input and output matching of a Zo=1 ohm filter using a standard 50 ohm VNA. So he's asking how to convert the S2P parameter measured in a 50 ohm system into the same measurement as if it was kept with a 1 ohm VNA.
Or it could be the filter in the sketch is normalized for a Zo= 1 ohm so it has to be de-normalize to 50 ohm.

jf1020 could you please clarify ?

 

[E Kafeman]

You can try to connect AnTune to you network analyzer. It can be set for 1 Ohm normalized system impedance and will then show all parameters correctly and live, such that 1+0i Ohm will be in center of the Smith chart.
If you have the measurement as a 50 Ohm S11 file, can it be imported in AnTune and normalized to any system-impedance.
You can find AnTune here.

 

[biff44]

books list one ohm filter component values as an "example" for people to use. 99.99999% of the people will modify the values to work in another impedance system, such as 50 ohms. You are the very first person I actually saw that built the 1 ohm impedance terminated filter. Good for you! But, as you can tell, a 1 ohm impedance is hard to measure on a 50 ohm network analyzer....that damned bilinear transform gets you every time.

 

[albbg]

The input and output impedance of a two port network are related to S-parameter as:

Zin=Zo[(1+S11)/(1-S11)]
Zout=Zo[(1+S22)/(1-S2)]

Be careful not to confuse Zin and Zout (input and output impedance) with Z11 and Z22 that are elements of the impedance matrix. They are different concept.

These two Zin and Zout are measured from the VNA and don't vary changing the characteristic impedance of the system, so if you just want to calculate the return loss you can just use them referred to the Zo impedance you want (f.i. 1 ohm).

Then if you have measured Zin and Zout in a 50 ohm system (or any other impedance: doesen't matter) and you want calculate the return loss in any Zo system (1 ohm or what you prefer), simply apply:

RLin=20*log(|Zin-Zo|/|Zin+Zo|)
RLout=20*log(|Zout-Zo|/|Zout+Zo|)

If you want, instead, recalculate the S-parameter (that depends from the value of Zo) you have to apply a little math.
Starting from the two previous equation (omitting the subscript numbers) if we perform the measurement under two different system characteristic impedance, let say Zoa and Zob we can write:

Z=Zoa[(1+Sa)/(1-Sa)]
Z=Zob[(1+Sb)/(1-Sb)]

where Z will be Zin or Zout depending from which you want to convert, Sa and Sb will be accordingly S11 or S22 measured in a Zoa and Zob systems. Then

Zoa[(1+Sa)/(1-Sa)]=Zob[(1+Sb)/(1-Sb)]

Let's suppose you performed the measurement in a system with Zoa and you want to convert to Zob, then Sa is known and Sb is unknown.
Solving then by Sb We have:

Sb=[Zoa-Zob+Sa*(Zoa+Zob)]/[Zoa+Zob+Sa(Zoa-Zob)]

Sa and Sb are complex numbers.