to create and rate content, and to follow, bookmark, and share content with other members.
2-port TRL to measure 1-port DUT
Question asked by
on Mar 9, 2010
on Mar 11, 2010 by Dr_joel
Show 0 Likes
Is it possible to use 2-port microstrip TRL to
measure a 1-port DUT?
Thanks for your help
No one else has this question
Mark as assumed answered
This content has been marked as final.
Show 3 comments
(Required, will not be published)
Mar 10, 2010 9:08 AM
sure. No problem. do the TRL cal, attach the part, measure S11.
Show 0 Likes
Mar 10, 2010 10:10 PM
Thank you Dr. Joel for your answer.
I was wondering if you could shed some light on why is this possible. The reason for my inquiry is because there seems to be a discrepancy between theory and application.
Simple theory tells us that M=XAY, where M is the 2-port measurement, X is port 1's error box, Y is port 2's error box, and A is the desired DUT. A is simply obtained by solving the equation for A knowing X and Y. However, if only a 1-port DUT is used, the TRL algorithm would still execute the above equation, and Y would still be used in the equation, although port 2 was not used in the measurement. Hence, there lies my dilemma!
I have attached a measurement I did to test a theory. S(1,1) represents the S11 of a 2-port measurement for a simple transmission line carried out using 2-port SOLT calibration. S(2,2) represents the S11 of a 1-port measurement (with port 2 terminated by 50ohms) of the same 2-port transmission line while also employing 2-port SOLT calibration. As noticed, the results are not the same, which seems to indicate that 1-port measurement using 2-port cal doesn't reveal the correct result as in the 2-port measurement.
Can you please explain how it's possible to do a 1-port measurement using a 2-port TRL calibration?
Another question I have for TRL calibration. Which method is best:
1-Calibrate the VNA's cables using SOLT. Then use microstrip TRL to calibrate from the end of the VNA cables to the DUT.
2-Calibrate the entire setup using microstrip TRL all the way to the DUT
Thanks for your help!
Show 0 Likes
Mar 11, 2010 7:27 AM
In this case, there are two reasons why there can be a difference. But in general, all the differences are badnesses in each cal (bad standards, bad cables, bad technique). The correction for S11 can be written as S11A=f(S11m, d1, rr1, m1)+ S21*S12*F([Sm],[d1,d2, l1,l2, m1, m2, tr1, tr2, rr1, rr2), where f and F are math equations, S11m is measured s11, [Sm] is four measured s-parameters (raw), the rest are error terms. The one port cal is S11A=f(S11m, d1, rr1, m1); as you can see, if S21*S12 = 0 then the they devolve to the same math. So they must give the same answer, if d1, rr1, m1 are the same. It figures. But full 2 port relies on the proper measurment of l2 (load match of port 2) to a high degree, and if the correction is not well done, the l2 term will be poorly characterized.
1) Did the SOLT cal and the TRL cal have the same reference planes (same connector interface) as the port 2 load? If it is a microstrip test fixture, and the TRL cal used different microstrip lines, with different SMA to microstrip launches, then this is surely a large contributor to the error. Are you using such a calibration for TRL or is the TRL using coaxial lines.
Similarly, is the SOLT using coax or microstrip lines?
So, you must put the load termination at the exact same inteface as port 2 of the TRL cal.
2) when you did the SOLT measurement, what did you use for the thru line or cal method? And what did you use for the load? Coax or microstrip? A very common mistake is to use a thru and not properly account for it. Further, if you did not do the TRL and SOLT in the same connector plane, the difference will show up just as you see. The cal methods depend on the analyzer, so which analyzer are you using.
In theory, TRL is better, but in practice, because people use different lines with different sma launches, it is almost always degraded.
In answer to you question below: it depends, and it can be shown that either method has different errors and depends upon the quality of particular standards in the calkit. If the SOLT standards give a better reflection tracking by more than the transmission tracking of the second tier, then the two tier approach is better. If not, then it becomes more complicated and infact depends upon a complex function of details of the individual standards, and the quality of the fixture. This is not a very satisfactory answer, but let me explain this way:
If you have great coax standards, and terrible TRL standards, and you have a perfect raw VNA with cables, but a lousy test fixture, then TRL cal will give better results. But, if you have a perfect test fixture and a lousy VNA, then SOLT with coax will give you a better result, and you should skip the TRL cal and use port extensions instead.
Show 0 Likes
Retrieving data ...
Measuring Thermal Noise of Amplifier
HP 8594A YTO readings
Noob Graphing Question
Square law detector on spectrasys
RF simulation: beginning !