530 IEEETRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES, VOL. 51, NO. 2, FEBRUARY 2003 Calibrated Waveform Measurement With High-Impedance Probes Pavel Kabos, Senior Member, IEEE, Howard Charles Reader, Uwe Arz, Member, IEEE, and Dylan F. Williams, Fellow, IEEE Abstract—We develop an on-wafer waveform calibration tech- nique that combines a frequency-domain mismatch correction to account for the effects of the probe on the measurement with an oscilloscope calibration. The mismatch correction is general and can be applied to any type of microwave probe, including scan- ning and internal-node probes for noninvasive integrated-circuit tests. We show that, for the commercial high-impedance probe we used, this calibration approach allows accurate on-wafer waveform reconstruction for a variety of probe ground configurations. Index Terms—Calibration, high-impedance probe, on-wafer measurement, waveform measurement. I. INTRODUCTION W E HAVE developed a calibration for on-wafer wave- form measurements performed at the internal nodes of an integrated circuit with a high-impedance probe and an oscilloscope. The calibration relies on a mismatch correction based on the broad-band frequency-domain probe-characteri- zation method introduced in [1] and on a separate oscilloscope calibration. The calibrations account for the finite impulse response of the oscilloscope, signal distortion in the probe, and multiple reflections between the probe and oscilloscope. The goal of internal-node probing and field mapping is to measure voltage and/or current waveforms at different locations within digital or microwave integrated circuits. A number of broad-band probes with low invasiveness have been developed for this task. Common approaches include electrooptic probes [2], photoconductive probes [3]–[5], microwave microelec- tromechanical system (MEMS) probes [6], and commercial microwave probes [7], [8]. Most optical, and some micromechanical probes, incorporate waveform measurement systems directly in the probes and cannot be characterized directly with the method developed here. However, a broad class of passive and active microwave probes, including those described in [6]–[8], can be treated as standard two-port microwave devices, and can be characterized with the method of [1]. Manuscript received March 22, 2002. P. Kabos and D. F. Williams are with the National Institute of Standards and Technology, Boulder, CO 80305 USA (e-mail: kabos@boulder.nist.gov; dylan@boulder.nist.gov). H. C. Reader was with the National Institute of Standards and Technology, Boulder, CO 80305 USA. He is now with the Electrical and Electronic Engineering Department, University of Stellenbosch, Stellenbosch, Matieland 7602, South Africa. U. Arz was with the University of Hanover, 30060 Hanover, Germany. He is now with the Physikalisch-Technische Bundesanstalt, 38116 Braunschweig, Germany. Digital Object Identifier 10.1109/TMTT.2002.807842 (a) (b) Fig. 1. Our experimental apparatus, on-wafer CPW transmission lines, and the equivalent circuit used to develop our mismatch corrections. Here, we develop an on-wafer waveform calibration method for measurement systems consisting of an oscilloscope and a passive or active microwave probe characterizable with the method of [1]. The calibration combines a novel mismatch correction with previously developed oscilloscope calibrations. We show that the calibration works well even when the probes have a ground contact that can either be arbitrarily positioned on the circuit or not used at all [7]. II. WAVEFORM MEASUREMENTS Fig. 1(a) shows a sketch of our experimental apparatus. The source on the left-hand side drives the input of a conventional ground–signal–ground (GSG) microwave probe. This probe has a low loss and a nominal impedance of 50 . While the method is applicable to any type of signal source, in our experiments, we used a comb generator that creates a distorted 800 MHz sine wave rich in harmonics at its nominally 50- coaxial output. The GSG probe just to the right-hand side of the source injects the signal from the source’s coaxial output port into a coplanar waveguide (CPW) transmission line printed on a gallium ar- senide (GaAs) substrate. The goal of the experiment is to ac- curately measure the waveform at the right-most end of the CPW transmission line, as illustrated in Fig. 1(a), with the high-impedance probe and oscilloscope. Here, the subscript “ ” indicates that the voltage in the CPW we wish to measure is near the tip of the high-impedance probe. To measure the waveform , we contacted the CPW center conductor near its end with the high-impedance probe. The probe transmitted the signal at 0018-9480/03$17.00 © 2003 IEEE