S-Parameter Measurement with Network Analyzer

[DonnyBrasko]

Hi there,

I am uncertain about a S-Parameter measurement and would like to ask someone with experiences in this field.
I want to measure the S-Parameters of an random IC. For this I use a Network-Analyzer from Agilent without the S-Parameter test set.
The IC is placed on a PCB in a RF-Case with 50Ohm-SMB connectors at the ports.
The PCB is much greater than the IC, and so I have to place microstrip-lines from the connectors to th relevant pins. Because the IC is SMD, i cannot place 50Ohm microstrip lines to the IC-Pins and therefor i will get reflections from the transmission line...not good...
Normally I would say, I need a matching network to be sure to have a 50Ohm Transmission line connected to the IC Pins.
But since I have to do a reference measurement (through, short, open) to calibrate the NA, I am not sure if i really need the matching network.
So now comes the question:
Is it possible to place an optional 0Ohm resistor on the PCB to have a through, and a second one to short the transmission line, and therefor calibrate the NA with this setup? Or is it necessary to have the matching network?
I hope i could describe the problem properly.
I know this is a kind of newbie question, but i'm uncertain and i want to be sure about this.
Any help is appreciated.

Greetings

Daniel

 

[TWW]

If you establish your reference plane at the IC pin, you can use a 50 ohm load at that point (with no IC). use three standards O-S-L at that point for the s11,s22 calibration. the thru will have some line length (from pin to pin) and some impedance that you will have to calculate for the thru. do not forget the open capacitance and the short inductance. If the intervening lines between the SMB connectors and IC pin are 25-200 ohms ZO, they should be removed from "error" two port with reasonable results. If the miss-match is much higher then your accuracy is going to suffer.

 

[buding]

Hi!

Using 0Ohm as "thru" and "short" and 50Ohm as "load" will not help You with solving this problem. If You would like to do like this You would have to modify standard definitions in Your network analyzer. There is posibility to introduced "user kit". Without exactly definitions of new standards Your measurements will have "very big" error.

The best method is to use TRL calibration. It allows to move the reference plane directly to the pins of Your IC. But this method needs extra calibration fixture.

The easiest way to do this measurement is to measure it on the edges of PCB with SMB standard. On PCB go to IC with 50OHm lines as close as You can. And then, having PCB s-parameters, and model of Your lines to IC, using ADS or MWoffice You can easily obtain s-parameters of IC.

//Buding.

 

[DonElec]

Hi buding,

thanks for this good tip - but its to late. I have already done the S-Parameter Measurement. I used 5 different PCB's, one for each standart: open, thru, short, load. The lines to the IC-Pads were the same on each PCB and so i could calibrate the NA. I know that my measurement isn't exact.
I needed the S-Parameters to model a circuit in Ansoft-Designer. The results of the simulation are satisfactory, and therefore the S-Parameters have a sufficient accuracy.

But thanks anyway for your help.

Daniel