The following steps summarize the measurement and correction procedure used in method 1 above:

De-embedding shunt G and C parasitics:

1. Measure S-parameters of the "open" dummy device.
   
   1. Convert S-parameters to Y-parameters.
2. Measure DUT-S-parameters
   
   1. Convert S-parameters to Y-parameters.
   2. Subtract dummy Y-parameters from DUT Y-parameters. This will require software (Libra, Matlab, or WinCal).
   3. Convert back to S-parameters and store the data
3. Repeat step 2 for each DUT surrounding the dummy.
4. Plot data and/or perform further processing on data as described below.

De-embedding series R and L parasitics:

1. Measure S-parameters of the "short" dummy device.
   
   1. Convert S-parameters to Z-parameters.
2. Measure DUT S-parameters or use "corrected" S-parameter data from step 2(c) above.
   
   1. Convert S-parameters to Z-parameters.
   2. Subtract dummy Z-parameters from DUT Z-parameters. This will require software (Libra, Matlab, or Wincal).
   3. Convert back to S-parameters and store the data.
3. Repeat step 2 for each DUT surrounding the dummy.

Note that the parameters are complex quantities.

3.4.1.1 Performing De-Embedding Calculations Using Matlab

A program, called de-embedding, and some subroutines for performing de-embedding calculations have been written in Matlab with online help. (Pad parasitic removal algorithms are discussed in  [8]

This program corrects for pad parasitics when you provide the measured S-parameters from a dummy (open and/or shorted pads) and a DUT (Device Under Test) along with information about the start and stop frequencies, characteristic impedance, and file format. The program delivers the corrected S-parameters and saves all S-parameter data along with frequency in column format in the variables specified in [dummyMX,dutMX,correctMX] respectively. The option is given to input data in single column format as saved on the internal disk of the HP 8753D network analyzer or in columns with or without frequency. The user can then save the variables in files.

For help, type: help de-embedding
at the Matlab prompt once you have copied the program and subroutines to your directory. Follow Steps 1 and 2(a) below for saving and converting the measurement data into a suitable format to be used by this program. Follow step 2(c)(ii) to convert to s2p format and give the file a descriptive name with a ".s2p" extension.

3.4.1.2 Performing De-Embedding Calculations Using Libra

It will be advantageous to perform the calculations in Libra if you have used Libra to design your device. This way, the measured and simulated results can be directly compared.

The calculations for the Y-parameter correction can performed using the 2-port Negation (De-embedding) element (NEG2). This element, which represents the dummy data to be negated, is connected in parallel with a 2-port S-parameter file element (S2P), representing the measured data of the DUT (including pads), to yield the corrected S-parameters which can later be saved and plotted.

Note that the NEG2 element can only reference a circuit network name as opposed to a file containing measured data and as such, you will first need to create a new network consisting of a 2-port S-parameter file element (S2P) which points to the appropriate dummy data file, and reference this new network from the NEG2 element.

Series pad parasitics can also be removed in Libra by following the same procedure as above but connecting the NEG2 and S2P elements, representing the measured dummy and measured or "correct" DUT data respectively, in series.

Follow the steps below for saving and converting the measurement data into a suitable format for Libra.

Back to Test Guide Table of Contents