766 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 34, NO. 4, JULY/AUGUST 1998 Effect of Surge Voltage Risetime on the Insulation of Low-Voltage Machines Fed by PWM Converters Mike Melfi, Member, IEEE, A. M. Jason Sung, Sidney Bell, Member, IEEE, and Gary L. Skibinski, Member, IEEE Abstract—This paper investigates the repetitive surge-voltage withstand of low-voltage mush-wound machines operated on adjustable-speed drives (ASD’s) using insulated gate bipolar transistor (IGBT) semiconductor technology. Historical work on surge testing of motor insulation has focused on one or more of the following aspects: 1) large horsepower motors; 2) medium-voltage form-wound motors; 3) single-shot impulse- type transients; or 4) low-voltage machines with surge risetimes 200 ns. IGBT drives can have risetimes of 50–200 ns. Thus, a new study on electrical stress of insulation systems due to the nonlinear voltage distribution of mush-wound motors when sub- jected to repetitive steep square-pulse waveforms (rather than impulse wave testing) is presented. Magnitude and risetime of the repetitive ASD surge-voltage transient induced on the machine terminals is reviewed first. Next, surge propagation into the winding was investigated to identify maximum voltage stress points on the conductor insulation. Potential failure mechanisms observed at these points are then discussed. The significance of decreasing surge risetime and increasing cable lengths on internal nonlinear voltage distribution is studied with experimental results from a 7.5-hp motor with a tapped stator winding. Index Terms— AC motor, insulation, pulsewidth modulated inverter, rise time. I. INTRODUCTION A. Motivation for Adjustable-Speed Drive (ASD)—Motor Surge Voltage Study T HERE IS A LONG and rich history of studies regarding the effect of surge voltages on machine insulation. Re- searchers in the 1920’s recognized the presence of abnormal surge voltages and proposed preventive measures [1]. A rig- orous analysis of surge voltage propagation and distribution on internal voltage stress in windings was done in the 1940’s [2]. There have been many studies done (too numerous to accurately acknowledge them all) on surge voltage distribution Paper IPCSD 98–20, presented at the1997 Industry Applications Society Annual Meeting, New Orleans, LA, October 5–9, and approved for publication in the IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS by the Electric Machines Committee of the IEEE Industry Applications Society. Manuscript released for publication March 3, 1998. M. Melfi is with the Research Center, Rockwell Automation/Reliance Electric, Euclid, OH 44117 USA. A. M. J. Sung was with the Research Center, Rockwell Automa- tion/Reliance Electric, Euclid, OH 44117 USA. He is now with Becton Dickinson Company, Franklin Lakes, NJ 07417 USA. S. Bell is with the Athens Motor Plant, Rockwell Automation/Reliance Electric, Athens, GA 30603 USA. G. L. Skibinski is with the Standard Drives Business, Rockwell Automation–Allen Bradley Company, Mequon, WI 53092. Publisher Item Identifier S 0093-9994(98)05185-8. Fig. 1. “Lightning surge” impulse magnitude, risetime, and duration. in windings. For the most part, they have been done on large horsepower form-wound medium-voltage motors with the IEEE 1.2/50- s surge impulse voltage wave of Fig. 1 applied to the terminals [3]–[9]. Investigation of nonrepetitive surge-impulse-type withstand on sinewave-powered low-voltage machines was done in the 1970’s. The data were summarized in [11] and officially recorded by an IEEE working group in [12]. Investigation of repetitive surge-voltage withstand on ASD- powered low-voltage mush-wound machines was done in the 1980’s with 480-V current-source inverter (CSI) drive technology [13]. This drive produced a 1200-V spike with a 400- s risetime at a rate of six times per 60-Hz cycle. This slow risetime allowed a linear distribution of voltage across the windings which, although it increased the volts per turn stress, was far below destructive levels. Laboratory life tests and field experience has shown this to be the case. Pulsewidth modulated (PWM) ASD’s using bipolar junction transistors (BJT’s) with risetimes of 1–2 s were popular in the late 1980’s. These drives applied steep-fronted voltages to the motor with in the 500–1200 V/ s range. The effect of this risetime on the nonlinear voltage distribution within mush-wound motor windings was initially discussed in [14] and [15]. NEMA MG-1, Part 30, specifies 1000 Vpk with a 2- s risetime for general purpose motors. PWM BJT drives had few failures over the past ten years, although the voltage stress was greater than for CSI technology. Section II- B analysis shows that a 480-V drive with 300 ft of output cable and BJT’s of 1- s risetime will have only a 30% overshoot transient above the dc-bus level, or 1000 Vpk. The reason for the work described in this paper is to in- vestigate the repetitive surge-voltage withstand of low-voltage mush-wound machines powered by the 1990’s preferred drive of choice, a PWM ASD using IGBT technology with rise- times in the 50–200-ns range. Steep-fronted voltage wave- fronts applied to the 460-V motor now have ’s in the 0093–9994/98$10.00  1998 IEEE