TLR calibration for infixture deembedding
using TLR calibration for infixture deembedding, I dont know how far the
reference planes can be together.
If my DUT is only lambda/10 or lambda/20 ...(50 to 500 um)...is there a possibilty to get
the s-matrix.
Is there a geometrical distance restriction for the Vector Network Analyzer in general.
Thanks for any reply
Your LINE standard must be not less than 20 degrees long at the lowest frequency and not more than 160 degrees long at the highest frequency. Be sure the analyzer is in STEPPED mode, not SWEPT. You can use multiple LINEs if you need to cover more bandwidth.
@madengr
Thanks for reply.
My DUT is very small, therefore I need to know if I can get the S-Parameters of it.
Of course there is an uncertainty in placing the reference planes. I mean, if my line standards are plus,minus 50 per cent in width the reference plane are not placed exactly. But is this the only restriction. Or can we say, that a 40 GHz VNA can only measure DUTS larger or longer than lambda/100 or so...
Maybe you mean 50% length variation in the LINE, not the width. The width of your LINE has to be very precise since this determines the system impedance for your measurement. The length of the LINE does not have to be accurate, just the width (really the impedance). The postition of the REFLECT with respect to the THRU will determine your phase accuracy.
The size of you DUT does not matter. It can be small yet very reactive or lossy.
The TRL itself doesn't require prior knowledge of line impedance to work. so the width can be any reasonable value.
The length of Line standard doesn't have to be the same as or similar to that of DUT. Once calibrated with good standards (with appropriate lengths etc.), the DUT can be less than 1/100 Lamda (say).
If the DUT is really so short, I would verify its connection to the "calibrated port"/reference plane is in fact the same as for the calibration standards, and there is no new transmission signal "path" (e.g. coupling through higher order mode). The latter can't be calibrated out with TRL.
Sure about that? The VNA needs an impedance standard. TRL it gets it from the LINE. For SOLT it would be the load, usually sliding. The advantage of TRL for coaxial is that airlines can be honed to a very precise impedance at high frequencies, much better than a resistive load.
http://www.vnahelp.com/tip11.html
The "Line Z0" in the above link is actually computed by the TRL calibration process. This doesn't contradict the earlier statement that "TRL" doesn't require prior knowledge of Z0.
OK, I see what you mean. The user still has to know the line impedance. If I think I am using a 50 ohm line, but it is actually 45 ohms, then the S-parameters will be normalized to 45 ohms, and 45 ohms must be given in the S2P file header If I tell the VNA the line is 45 ohms it is capable of re-normalizing to 50.
madengr, your statement above is correct. It was an error for me to say that the actual line impedance was computed in TRL calibration. The line impedance is not computed, but rather implicitly defined--the TRL-calibrated s-parameters should be referenced to such line impedance--whatever the value is. If a user wants a specific reference impedance value (such as 50ohm) then re-normalization has to occur. And that re-normalization requires user to specify the value of actual line impedance. However the accuracy of the user-specified value doesn't affect the TRL calibration procedure itself.
In above link to the tip, "Line Z0" should be understood as the actual line characteristic impedance. Its value doesn't enter into the equation of the TRL calibration. "System Z0" is teh desired reference line impedance, "Offset Z0" is what the user thinks the "Line Z0" is. "System Z0" and "Offset Z0" are used in the re-normalization of the S-parameters.
Yeah, I think we both agree now. Just a mis-interperatation of words.
FYI if anyone reading this has an HP 8510B VNA the TRL Z0 re-normalization is screwed up. It can be fixed in a firmware upgrade however it's now obsoleted so you may be SOL. I pulled my hair out for 2 days trying to figure it out until I tried an 8510C and got the correct result.