Dear all

I am using a probing station for measurement of LTCC based circuit, The transition i am using is shown in figure,

I ll be using Multiline TRL calibration;

Now obviously transition impedence is not 50 Ohm it is 72 Ohms now as i am using THRU with zero length should the impedence of line be 50 Ohms or 72 Ohms according to Agilent application note (8510-8A calibration for non coxial measurements pp15 table 1) it is said that if you are using non zero length THRU then the impedence of THRU and LINE should be same, (mine is zero length THRU?)

2. It is also said that insertion phase of LINE should be 90 w.r.t insertion phase of THRU, at the same time it is said that the optimum length is 90 w.r.t center frequency, in my case both would be different so what should i do, frequency is 10-20GHz

3. In WinCalXE when you define probe type and pitch it ask for ISS to be specified and it will be used along with NIST TRL calibration, now in my case i am not using any predefine ISS but my own set of TRL kit on LTCC how do i change that option as it wont let me have custom ISS, if i click load button then it asks for *.wci file where would i get that?

Your help is really appreciated.

I am using a probing station for measurement of LTCC based circuit, The transition i am using is shown in figure,

I ll be using Multiline TRL calibration;

Now obviously transition impedence is not 50 Ohm it is 72 Ohms now as i am using THRU with zero length should the impedence of line be 50 Ohms or 72 Ohms according to Agilent application note (8510-8A calibration for non coxial measurements pp15 table 1) it is said that if you are using non zero length THRU then the impedence of THRU and LINE should be same, (mine is zero length THRU?)

2. It is also said that insertion phase of LINE should be 90 w.r.t insertion phase of THRU, at the same time it is said that the optimum length is 90 w.r.t center frequency, in my case both would be different so what should i do, frequency is 10-20GHz

3. In WinCalXE when you define probe type and pitch it ask for ISS to be specified and it will be used along with NIST TRL calibration, now in my case i am not using any predefine ISS but my own set of TRL kit on LTCC how do i change that option as it wont let me have custom ISS, if i click load button then it asks for *.wci file where would i get that?

Your help is really appreciated.

VNAs don't measure impedance, they measure refleciton coefficient. The reason we set impedances is to tell the VNA what a particular reflection means. In the case of TRL, the THUR and the LINE must be of the same impedance (theorectically, they can be of any known impedance, but the TRL algorithm in most VNAs take the shortcut of assuming they are the same), but if the THRU is of zero length, the actual value of it's impedance has no effect as the zero length removes it's influence. S11=0, S21=1 regardless of impedance if the length is zero. You miust tell the analyzer what the impedance of the line is, or it won't know what the reflections from the line mean. In some cases, the line is assumed to be the same as the system impedance, but some allow the line to be of a different impedance. In your case, you must decide what you WANT the reference impedance to be. If you want it to be the line impedance, then you set Z0= 72 and line Z=Z0, if you want it to be 50 ohms, the set the system to 50 and the line to 72.

No, both are the same. If the thru is zero length, and the center frequency is 15 GHz, compute the physical length that would give 90 degrees phase shift at 15 GHz. Since the thru is zero, it has zero phase shift at 15 GHz, and the line will have 90 degrees at 15 GHz, thus it will have 90 w.r.t. THRU and 90 w.r.t. center frequency (15GHZ)

Not sure about the workings of WinCalXE wrt loading the iss, but if you are using TRL, the only thing it needs to know is the Z of the line, unless you are trying to do LRRM. For TRL, you can probably skip Wincal and just use the VNA cal. If I were trying this with WinCal myself, I might load an ISS file, modifiy to add the correct line Z and resave it. Since the reflect model does not matter in TRL, it should work just fine.