AnsweredAssumed Answered

De-embedding

Question asked by lipchyuan on Jan 10, 2011
Latest reply on Jan 11, 2011 by Dr_joel
Hi,
    I would like to measure the bare die IC capacitance (or S parameters) without IC package, this die has a differential transmiter channel TxP and TxN. To extract the capacitance/S-para value of the die Tx channel, I have to measure the the S-parameter of the package (Spkg s4p) with PC board and package with PC board + die (Spkag+die s2p).  Since PC board has two SMA connector with trace that connects to IC packagae TxP and TXN pads, I did a 4 port calibration and connect two probes for TxP and PxN pad probing with probe extension. After the calibration and port extension, i got the Spkg S4p file. The next step  is to get the Spkg +die s2p file with 2 port calibration. After getting this two data file, I can use ADS to de-embedded the die S-parameters.
   My question are :

1) Although it is differential line, can I measure and de-embeded as what I mentioned above (i.e treat it as single endeded)? If not, should I convert to mix mode S-para then only do de-embedding?
2) My Spkg S4p has bad match  -5-10dB return loss from 3Ghz to 8.5GHz. Will it affect my de-embedding accuracy? If yes, how to estimate that?
3) Since it's differential line, the coupling between the trace TxP and TxN in the package  is about -20dB, will it affect my calibration? Am I right it will affect the calibrationdirectivity?
4) Side question, does port extension compensate for mismatches? I think the port extension only compesate for loss and delay.

Thanks in advance,
Lip Chyuan  

Outcomes