Browse Prior Art Database

High Frequency On-Wafer Calibration Method with Enhanced Accuracy

IP.com Disclosure Number: IPCOM000015308D
Original Publication Date: 2002-Jan-11
Included in the Prior Art Database: 2003-Jun-20
Document File: 3 page(s) / 32K

Publishing Venue

IBM

Abstract

The proposed solution describes a method for on-wafer calibration of high frequency two-port measurements. In particular, it is well suited for use in Silicon processes (CMOS and/or SiGe) and describes a way to determine the accuracy of a high frequency measurement by controlling the accuracy of a DC resistance measurement. Advantages are that this method allows the acurate de-embedding of active and passive CMOS and SiGe devices to very high frequencies. It is also very beneficial that no well controlled on-wafer resistors or transmission lines with known impedances have to be available. It is especially suited for calibrations on lossy substrates, a situation typically found when using CMOS processes. The problem is solved by applying a TRL calibration of unknown reference impedance to the measurement of a passive device (nominal resistor). This measurement is extrapolated to DC frequency and compared with the value of a DC resistance measurement. The comparison of the extrapolated value with the value measured at DC allow the determination of the reference impedance and the complex impedance over frequency of the passive device (resistor). The dielectric loss of the transmission line is estimated for low frequencies and for high frequencies based on known principle behaviour of a transmission line and the measured loss. This loss estimation significantly enhances the accuracy of the calibration at very high frequences. Based on the knowledge of the TRL calibration coefficients, a LRM calibration my be performed. The following is a high level step-by-step algorithmic desctiption .

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 48% of the total text.

Page 1 of 3

High Frequency On-Wafer Calibration Method with Enhanced Accuracy

    The proposed solution describes a method for on-wafer
calibration of high frequency two-port measurements. In
particular, it is well suited for use in Silicon processes (CMOS
and/or SiGe) and describes a way to determine the accuracy of a
high frequency measurement by controlling the accuracy of a DC
resistance measurement. Advantages are that this method allows
the acurate de-embedding of active and passive CMOS and SiGe
devices to very high frequencies. It is also very beneficial that
no well controlled on-wafer resistors or transmission lines with
known impedances have to be available. It is especially suited
for calibrations on lossy substrates, a situation typically found
when using CMOS processes.

The problem is solved by applying a TRL calibration of unknown
reference impedance to the measurement of a passive device
(nominal resistor). This measurement is extrapolated to DC
frequency and compared with the value of a DC resistance
measurement. The comparison of the extrapolated value with the
value measured at DC allow the determination of the reference
impedance and the complex impedance over frequency of the passive
device (resistor). The dielectric loss of the transmission line
is estimated for low frequencies and for high frequencies based
on known principle behaviour of a transmission line and the
measured loss. This loss estimation significantly enhances the
accuracy of the calibration at very high frequences.

Based on the knowledge of the TRL calibration coefficients, a LRM
calibration my be performed.

The following is a high level step-by-step algorithmic
desctiption .

//----------------------------------------------

// (1) determine M12 = line*(thru)^-1
//----------------------------------------------

// (2) determine the eigenvalues lambda1 and lmbda2 of system M12
//--------------------------------------------------------

// (3) derive gamma from eigenvalues
//

// Now gamma is calculated from lambda = exp(-gamma*Length) -> gamma =
-Ln(lambda)/length
//--------------------------------------------------------

// (4) derive the error boxes A and B at port 1 and port 2
//----------------------------------------------------------------------------

---------

// (5) calculate effective reflections with respect to the (yet unknown) load
impedance based on the measured short and load
//----------------------------------------------------------------------------

---------------------------------

// (6) calculate a first estimate of Zw based on the DC resistance and the
refelection coefficients of the load

1

Page 2 of 3

//

//Zw1 = Rdc1 * (eins - rload1[i]) / (eins + rload1[i]); (Rdc1 is the measured
DC value of the load)
//----------------------------------------------------------------------------

------------------------------------

// (7) Prepare estimates of C0 = real(gamma/Zw/(j*omega)) and fit a 4nd order
polynom over the specified data range (the constant part of the
polynom is the first g...