Keith E. Jones

LRM and LRRM Calibrations with Automatic Determination of Load Inductance

Two new techniques are presented in an effort to achieve greater accuracy and better repeatability in the on-wafer calibration of vector network analyzers. The first is a method of determining an inductance value for the match standard during the calibration process so that only its resistance, not its reactance, needs to be known. The second is a new type of calibration, LRRM (line-reflect-reflect-match), which is a variation of LRM with several possible advantages. Also, a simple series resistance-inductance model for a coplanar load is experimentally investigated to 40 GHz and found to provide a good description of the load behavior.

Where Are My On-Wafer Reference Planes?

In order to characterize the accuracy of on-wafer measurements, there is the need to understand microwave waver probe calibrations over a matrix of planar transmission line types, sizes, and substrate materials. One of the calibration issues which arise is the electrical lengths of the planar calibration standards relative to each other and relative to the device under test. In this paper, three techniques for verifition of planar calibration standards are presented and shown to agree within ±0.1 ps of electrical length.

Accurate measurement of high-speed package and interconnect parasitics

The feasibility of using high-speed wafer probes to measure accurately the parasitics associated with packages and interconnects used in conjunction with high-speed integrated circuits is demonstrated. A way of calibrating out the fixture-related errors while providing a reconfigurable mechanical interface to specific points of interest is demonstrated. These points can be at the perimeter and also internal. Calibration at the point of interface to the structure under test and a well-controlled mechanical interface enable large numbers of highly repeatable measurements to be made easily and accurately. The effect is to extend the full capability of test instrumentation to the physical points of interest, with accuracy and repeatability from site to site. The ability to calibrate at the probe tips is necessary to eliminate distortions from cables and probes. Some sources of errors, such as crosstalk, should be minimized in the probe, since they are difficult to correct in most cases.<<ETX>>

Reducing Non-Systematic Errors in Microstrip Measurements

Accuracy of error-corrected network analyzer measurements is often limited by connection repeatability errors to the device under test. An improved coax to microstrip adapteris described which provides -46 dB repeatability to 26.5 GHz. Microstrip calibration standards were developed to permit adapter de-embedding. Data from a wafer-probed MESFET is compared to measurements of the same device bonded into a microstrip carrier.