Performance of the Nasa Digitizing Core-loss Instrumentation Performance of the Nasa Digitizing Core-loss Instrumentation

The ‘standard method’ of magnetic core loss measurement was implemented on a high frequency digitizing oscilloscope in order to explore the limits to accuracy when characterizing high Q cores at frequencies up to 1 MHz. This method computes core loss from the cycle mean of the product of the exciting current in a primary winding and induced voltage in a separate flux sensing winding. It is pointed out that just 20% accuracy for a Q of 100 core material requires a phase angle accuracy of 0.1° between the voltage and current measurements. Experiment shows that at 1 MHz, even high quality, high frequency current sensing transformers can introduce phase errors of a degree or more. Due to the fact that the Q of some quasilinear core materials can exceed 300 at frequencies below 100 kHz, phase angle errors can be a problem even at 50 kHz. Hence great care is necessary with current sensing and ground loops when measuring high Q cores. Best high frequency current sensing accuracy was obtained from a fabricated 0.1-ohm coaxial resistor, differentially sensed. Sample high frequency core loss data taken with the setup for a permeability-14 MPP core is presented. MAGNETIC MATERIALS DEVELOPMENT AND CORE-LOSS INSTRUMENTATION Magnetically soft materials having very low core loss at frequencies up to a MHz are of great value in aerospace power conversion applications, where bulk and mass are critical. At high frequencies, lower specific losses enable the reduction of bulk and mass of magnetic components, while remaining within specified temperature rise limits. Presently, solid-state device (transistors, thyristors, diodes) switching losses in kilowatt level converters limit these frequencies to a few hundred kHz, but a rise in frequency is sure to follow the arrival of better switches. A significant, if slow, response in the past 2 decades has been the development of amorphous and partially recrystallized magnetic ribbons based on iron and cobalt, whose improvement continues to this day. And the future may bring bulk new materials based on a composition of nanometer-sized particles that are electrically isolated, but magnetically ‘exchange field’ coupled, to reduce losses. Progress in such materials development is difficult and depends both on innovative materials nanostructuring and the ability to accurately characterize incremental improvements. Accurate measurement of high frequency core loss at B-field levels found in real applications is fraught with a number of spurious influences that are fundamental and difficult to control, yet can induce hidden errors of the order of tens of percent. This paper presents the author’s experience with a well-established measurement method set up on a high-speed digitizing oscilloscope and discusses the performance of several current probes at high frequency, with reference to characterizing a high Q core. THE ‘STANDARD METHOD’ OF CORE LOSS MEASUREMENT AND ITS LIMITATIONS In this widespread and relatively easy to set up method of taking magnetic core loss [1, 2], the magnetic induction in the core under test is driven to a desired amplitude by current in a primary winding and the value of this induction is measured by integrating the output of a secondary winding that senses the induced voltage. The average core loss is simply the timeaverage of the instantaneous product of the primary current i(t) and a voltage v(t), which is the secondary voltage multiplied by the winding turns ratio N1/N2. Winding (N1) losses are automatically excluded. The hysteresis loop is correctly shown by the data. And robust, bench-top power amplifiers are available that, with proper tuning of the N1 circuit, can drive test sized cores (~ an inch OD) to high flux levels at even a megahertz. These are great advantages of this ‘standard’ method, as compared to the various commercial impedance analyzers and bridges, which can put out at most a few volts and lump both core and winding losses. Phase shift errors due to high frequency reactances False time shifts between signals representing the exciting current and induced voltage introduce an increasing fractional error in the measured time-average loss power ) t ( i ) t ( v P = , as P decreases. In the language of linear circuits, a low power loss can be described by