AnsweredAssumed Answered

Strange thing in 8720D calibration

Question asked by avith on Aug 28, 2009
Latest reply on Aug 28, 2009 by Dr_joel
Hello,

I use the HP 8720D NWA for the filter measurement.
If the NWA was calibrated by the calkit 85052D in combination with the home made SMA calibration standard, I saw the ripple in the passband (about 0.9dB pk - pk).
If the NWA was calibrated by the calkit 85033C in combination with the home made SMA calibration standard, I didn't see the ripple in the passband.
If the NWA was calibrated by the calkit 85052D in combination with the Agilent's calibration standard, I still see the ripple in the passband.

The ripple should not be from the filter itself. This has been verified in another accurate measurement setup and no ripple found.

The results really confused me.
1. Is the ripple from the residual calibration error because the home made calibration standard does not fit the preset calkit   model well?
    If so --> why I also see the ripple when the offical calstd and calkit are used.
2. Is the calikt 85052D suitable for the 8720D NWA?  I saw the difference in the calkit definition between the 85052D kit and 85033C kit.
3. Is there anything related to the firmware of NWA. My NWA has never been updated for the firmware since 1998
4. The 8720D NWA is the 3-receiver model. Both ports share the same reference channel. Does that mean the switch error can not been canceled out? How Agilent deal with the switch error in a 3-receiver NWA? How if the internal switch has been replaced.
5. Is the SOLT algorithm used by 8720D is actually the SOLR(reciprocal)? --> The thru only provide the information of S12=S21.
6. If the SOLR algorithm is used, the assumption is e'22 = e00. Is this assumption still valid with an imperfect switch? I saw some discussion about it on a paper (I forget the name).
7. Back to the start point: If the ripple is related to the calibration, how to explain that the ripple was mysteriously gone if the 85033Ckit + home made calstd was applied?

Thank you very much for the help.  

Outcomes