Hi,

I was wondering if I could get some ideas from people as to what might be wrong with the setup of a microstrip TRL cal kit (10MHz to 20 GHz).

Basically, all I have is 4 standards and I am implementing this TRL cal on the E8363B PNA:

1.- Match (Two 100 Ohm resistors in parallel. These resistors are rated up to 20 GHz with 0.1% tolerance)

2.- Thru (0.222" Length. Delay set to zero to establish reference plane at the middle, i.e. 0.111")

3.- Reflect (0.111" section of Transmission Line. Set up as an open and all non-linear terms for the shunt capacitor model set equal to zero)

4.- Line (0.318" Length and Delay of 20.8 psec). This line was check to meet the required 20 to 160 degree of insertion phase. It is 15 degrees at 2 GHz and 150 degrees at 20 GHz, but it is fine since I am far away from 0 and 180 degrees)

For the low frequencies (10Mhz to 2 GHz) I have specified the Match standard as my reference impedance and from 2 GHz to 20 GHz the "line" is the reference impedance.

After I perform a TRL calibration I do some confidence checks to see whether or not the calibration is adequate.

I first measure the thru and check for S11, S22 on the smith chart. I get a perfect response of 1. I then check S21 & S12 for phase and the variation across the band is +/- 0.2 degrees which is also good. These results tell me that the reference plane has been established at the center of the thru.

To test the actual response of the cal kit I have a different section of transmission line not used during the calibration. This is a line of 0.260" length. When I do check for S11 in dB I get a sharp transition in the 2GHz region. This is due to the change of reference impedance from the match to the line standard. I expect this abrupt change but the magnitude of this change is 6 dB which I feel is a large number. This tells me that the match standard impedance and the line impedance are not equal. I read in the literature that since this impedance difference between "match" and "line" is fixed or known it can be corrected by using the "complex impedance" option for the PNA (not available on 8510). I have tried different values on this section of the program (48-1j ohms, 10+0j ohms, and even -50 ohms). I then perform the same steps as explained before but I see no difference in the abrupt response in the 2 GHz region. I am suspecting there may be something wrong with my cal standard definition setup or something is odd.

I tell the computer to use the "thru" and not the "reflect" to set my reference plane and to use "line" as my reference impedance. Everything else is done as specified in the application notes.

Does anyone know what might be wrong? Does anyone have had this problem before? Any help would be greatly appreciated.

Thanks for reading,

Dan

I was wondering if I could get some ideas from people as to what might be wrong with the setup of a microstrip TRL cal kit (10MHz to 20 GHz).

Basically, all I have is 4 standards and I am implementing this TRL cal on the E8363B PNA:

1.- Match (Two 100 Ohm resistors in parallel. These resistors are rated up to 20 GHz with 0.1% tolerance)

2.- Thru (0.222" Length. Delay set to zero to establish reference plane at the middle, i.e. 0.111")

3.- Reflect (0.111" section of Transmission Line. Set up as an open and all non-linear terms for the shunt capacitor model set equal to zero)

4.- Line (0.318" Length and Delay of 20.8 psec). This line was check to meet the required 20 to 160 degree of insertion phase. It is 15 degrees at 2 GHz and 150 degrees at 20 GHz, but it is fine since I am far away from 0 and 180 degrees)

For the low frequencies (10Mhz to 2 GHz) I have specified the Match standard as my reference impedance and from 2 GHz to 20 GHz the "line" is the reference impedance.

After I perform a TRL calibration I do some confidence checks to see whether or not the calibration is adequate.

I first measure the thru and check for S11, S22 on the smith chart. I get a perfect response of 1. I then check S21 & S12 for phase and the variation across the band is +/- 0.2 degrees which is also good. These results tell me that the reference plane has been established at the center of the thru.

To test the actual response of the cal kit I have a different section of transmission line not used during the calibration. This is a line of 0.260" length. When I do check for S11 in dB I get a sharp transition in the 2GHz region. This is due to the change of reference impedance from the match to the line standard. I expect this abrupt change but the magnitude of this change is 6 dB which I feel is a large number. This tells me that the match standard impedance and the line impedance are not equal. I read in the literature that since this impedance difference between "match" and "line" is fixed or known it can be corrected by using the "complex impedance" option for the PNA (not available on 8510). I have tried different values on this section of the program (48-1j ohms, 10+0j ohms, and even -50 ohms). I then perform the same steps as explained before but I see no difference in the abrupt response in the 2 GHz region. I am suspecting there may be something wrong with my cal standard definition setup or something is odd.

I tell the computer to use the "thru" and not the "reflect" to set my reference plane and to use "line" as my reference impedance. Everything else is done as specified in the application notes.

Does anyone know what might be wrong? Does anyone have had this problem before? Any help would be greatly appreciated.

Thanks for reading,

Dan

You may also try unchecking the "LRL Autocharacterization" box in the calibration kit's TRL class assignment to help address this. This setting is discussed in more detail in the PNA Help file.