help! the "error" in 2 ports transmission measurem

I am just starting to use network analyzer (agilent 8722ES) to measure the S parameters of some materials in coaxial line. I put the ring sample inside the coaxial line and measure S11 and S21. Usually I have several peaks in S-parameter amplitude graph because of the resonance inside the sample. Then I disconnect the coaxial line. reinstall the sample and repeat the measurement again. I found out that the peak position shift about 20MHz (forgot to say, the frequency range I used is 0.5-18GHz). I repeat lots of times, almost all of them have a random small peak shift(either positive or negative shift).
The question is : Is this normal in S parameter measurement?
If it is not, is this because of the system error in network analyzer? Or is it caused by my falsed connection between cable and coaxial line?
I don't have too much experience with network analyzer, so any suggestion and advice will be very helpful. Thanks

you can try to fix the coaxial cable connecting the VNA and the probe. We got same problem too. The location,shape of the this connecting cable changes, you can see the S-paramter got from VNA changes also, quite annoying.
Best Regards,

The answer may be splitted into 2 topics:
- The measuring method
- The tip & trips

The measuring method is mainly related to the calibration method, SOLT, TRL, E-Cal, etc.
Generally the U(s11) (uncertainty related to s11) is higher if the observed value is near the external Smith-Chart's circle.
A TRL Calibration is one of the most accurate.
SOLT is less accurate than TRL but still remain very accurate.
Agilent TRL* has poor accuracy for high values of s11.
To be accurate when you observe high values of s11, try a SOLT and a
S, offset S,L,T and compare the results.
To be accurate when you observe low values of s11, try a S,O,slidingL,T,
...

tip&trips
instrument coax cables and connectors should be in exellent conditions.
Also the Standards should be in exellent conditions.
Don't move the coax cables. Fix it in many points with scotch tape. Dont' make any torsion to the coax. Be the contact metals cleans. Etc. Etc.

Set sweep mode to step, set power > -10 dBm,
Read Agilent AN 1287 series
Read Agilent AN "Principles of connectors care"
Read Agilent PN 8510 series

Hi sergio mariotti,

Quote: "@gilent TRL* has poor accuracy for high values of s11"
Please explain shortly why is it so.
Thanks
g579

can you give some comments about the fluctuation of VNA reading when cable connecting the VNA and probe moving?
Best Regards,

hi mariotti,
thanks for ur advice. I already fixed the cable, and the cal. kit and connector are in very good condition. I tryed SOLT, but the situation is not getting better, there still has like 20MHz uncertainty.
Do u have more suggestions?

Rinji

To: asdfaaa:
coax cable: quality grade: the best for measurement. If I remember well the values... a smooth 90° bend may vary |s21| of a quantity LESS THAN 0.01dB*GHZ+0.05 : Of course this value is too high. But in practice the |s21| variation is very smaller than the guarantee.
As the cable is "good" as the |s21| variation is smaller than guarantee.
I.e. a never used GoreTex cable, not bent from Cal to Measure, show a |s21| variation less than 0.03 at 22 GHz after 10 minutes.

To: rinjic
If you have a 3 samplers (or mixers) VNA, like hp 87xx whit NO opt 400 installed, a ripple may be given by the instrument. That ripple may by in the range of +/-0.1 dB around the |s21| and/or +/- 0.03 (linear) around |s11| (look the s11 on Smith Chart )

But... are you sure that the ripple is generated by the VNA? May be the real response of the DUT?
To undertand better, coul'd you post a plot of your measurement?

rinjic,
As you said the peaks come from resonances of the ring. If the peaks move it means the resonance frequency is moved. I suppose the exact resonance frequency is a function of the material and its dimensions and also the edges condition around it. When you took off and put back the ring, it might moved a litte bit toword one side and it changed its resonant frequency.

Yes, a little bit position change of the ring will influence the result. But it is just for the phase. The magnitude of S parameters should still be the same since the length of the ring doesn't change, right? Correct me if my understanding is wrong. Thanks!

rinjic

I think it is not due to the DUT because I tried using other materials. The ripple is always there. Next time I measure them, I will post a plot here. But do u think it is possible the cable has a problem? if it does, how can I check it? Thanks for ur help!

rinjic

I think it is not due to the DUT because I tried using other materials. The ripple is always there. Next time I measure them, I will post a plot here. But do u think it is possible the cable has a problem? if it does, how can I check it? Thanks for ur help!

rinjic