Immediately following WW II, the high demand for goods permitted the sale of almost anything, even if it had serious flaws. The low quality provided an opportunity for other countries to sell material that appeared to be of higher quality than that produced in the U.S. Soon, U.S. industry became increasingly aware of the need to assure that all of its products functioned properly. There was a great increase in testing the finished products. It was the time when many of the present magnet wire test standards were created. In 1951, a number of professors, under the leadership of Richard D. Dermer at the Indiana Institute of Technology, built test equipment to meet those standards. Product quality improved.
issue: May 2003 APPLIANCE Magazine
Electrical Manufacturing & Coil Winding
Why Test Magnet Wire?
Email this Article
by John A. Whitney, consultant, A/Z-Tech, Inc.
It has been widely recognized that continuous testing during product manufacture has many advantages over sample testing after product completion. This is a further discussion of accomplishing this with magnet wire.
In recent years, the concept of "zero defects" was born. To assure this, testing was needed during manufacture. If a defect remained undiscovered until after the product had been completed, not only did it have to be scrapped, but its scheduled delivery to the customer was not met. Industry decided there had to be a better way. Thus, the in-line testing concept was born.
Particularly in the magnet wire industry, the product was being delivered in larger quantities. The manufacturing conditions when the first of the spool was made could be quite different than those that existed when the last 100 ft (30.48 m) were put on the spool. A test that assured that the last 100 ft were of high quality might not necessarily mean that same quality was present during manufacture of the first 100 ft, or even the middle 100 ft. If it took an entire shift to manufacture the spool, testing the last 100 ft might mean only considering the last 30 sec of the process. This certainly did not assure the quality of the entire product, which, in some instances, manifested itself in consumer-oriented products.
In an article entitled "Helping the Bottom Line," which appeared in the June 2002 issue of Appliance Magazine, Eng. Renato Romagoli of AEA s.r.l. (Angeli di Rosora, Italy) was quoted as saying: "√Čcustomer satisfaction strongly depends on appliance performance at start up and appliance reliability in time. Manufacturers need to check 100 percent of production." For example, the author says that 24 percent of washing machines and 15 percent of dishwashers sold in the European Community need to be repaired during the first year. Of course, all of these repairs cannot be attributed to poor magnet wire. The potential causes are myriad.
Nonetheless, the importance of in-line testing, of testing all of the product, is now widely accepted. In the magnet wire industry, a number of methods have been devised and tried. The results were often disappointing. It was necessary for the test results to be valid initially and to continue to be valid. It was essential that the testing in no way deteriorated the quality of the wire.
A major advantage of magnet wire in-line testing was that a process that produced wire with less than optimum quality might be discovered and corrected in time to improve or correct problems so that the wire quality improved. This was far better than just scrapping wire that had been found to be inferior.
Even when a satisfactory test method appeared to have been found, it often turned out to produce different results than the long accepted National Electrical Manufacturers Association (NEMA) MW 1000 specified sampling test results. There was the problem of which to believe. The customer may have become quite familiar with the NEMA standard sampling tests and may have acquired equipment to perform those tests. How do you argue with a good customer who believes he has discovered that the quality of your wire is less than he has expected? Exchanging rejected wire does not entirely solve the problem.
What was needed was in-line testing that produced the same results as the well- established NEMA sampling tests. A worthy endeavor, but not easily accomplished.
There were attempts to use the NEMA specified equipment for in-line testing, such as the two v-grooved sheave electrodes. However, problems were discovered. Wire manufacturers found that they counted about 17 percent of the defects twice. Even though it was realized that the wire quality was better than the sampling tests showed, the wire quality was often substantially better than the NEMA-required minimum, so there wasn't a great deal of concern about the test producing a few more fault indications per 100 ft than actually existed. However, with continuous in-line testing, the test pessimism of 17 percent created by the dual sheaves did attract some attention.
Another problem with the V-grooved sheaves used in sample testing was that they could produce work hardening in the wire when it passed over the 4-diam pulleys that the NEMA standards specified. While this was not a problem in sample testing where the test samples were discarded, it was a concern with 100-percent in-line testing.
One approach was to use larger diameter pulleys. However, this produced a wire to pulley contact length of more than the 1-in length that had been required by the NEMA tests since the early 1950s. NEMA required that defects substantially longer than 1-in were to produce more than one fault count. A large pulley that contacted several inches of wire would either produce more than one count for a very tiny fault, or it would count a long bare area the same as the single tiny fault. The large pulley could not produce in-line test results compatible with the NEMA specified sampling tests. Since this compatibility was desirable, this type of in-line test equipment became quite a disappointment.
An electrode using a conductive carbon foam sponge was also tried. It seemed to have great promise. It eliminated double counting and contacted the wire for only the desired 1 in. Unfortunately, an uncorrectable deficiency was soon found. The material had a plastic memory. The presence of the wire could deform the sponge, so that there was no longer electrical contact with the entire periphery of the wire. Some of the wire faults were not detected and counted. Again, it proved to be not compatible with the NEMA specified sampling tests.
Carbon fiber brush electrodes were proposed. A demonstrator was built that passed wire through several successive test electrodes. There were defect counters for each of the electrodes. In addition, there was a computer display of the test results for each of the detector heads.
The close and predictable spacing of the double counts from the two v-grooved sheaves became glaringly apparent so they could be easily distinguished. It also became apparent that the carbon fiber brush electrodes did not double count any of the faults. Two sequential carbon fiber brush electrodes were provided. Any difference in their counts would become immediately apparent. There were none. However, defects missed by the carbon foam electrode did become apparent.
There were still challenges. The NEMA standards specified a wire speed of 60 ft (18.288 m) per min. Clearly, wire manufacturing and use involves both higher and lower wire speeds. Still further, the NEMA standard specified that long faults or bare spots were to be counted at a rate of 450 counts per min. The result was that at a wire speed of 60 ft per min the counts would be accumulated at a rate of one for each 1.6 in (4.064 cm) of bare wire, rather than the desired one per inch of bare wire.
The time rather than length specification was a holdover from the early 1950s when the maximum repetitive count rate of the mechanical counters was the limiting factor.
There was an additional time-based parameter in the NEMA standard. It was that faults that presented signals shorter than 4 -6 msec, noise pulses generated by other equipment, should not be counted as faults. Since even a very short fault would produce a signal 0.083 sec long, it appeared that signals 4-6 msec long could not represent true faults. However, we had discovered that on small diameter wire with very thin build insulation, faults could produce a phenomenon known as "relaxation oscillation" and could produce a series of pulses each only 1-msec long. Contrary to previous thinking, these were demonstrated to be valid faults.
We have not found "relaxation oscillation" to take place with carbon fiber brush electrodes and that noise pulses were all very short, of the order of microseconds. For these reasons, it appears that the 4 -6 msec requirement could be disregarded with no detrimental effects. This was necessary if the system was to operate with very fast wire speeds, when the wire moved 1-in in 4-6 msec, the equivalent of 830-1,250 ft (252.984-381 m) per sec.
The NEMA 6-MW technical committee has recently proposed a number of changes in the HVC sampling test specification as a result of their studies and research. In particular, Mr. Don Barta, chairman of that committee, has attempted to increase the 450 per min repetitive count rate for long faults or bare spots to be closer to the equivalent of one count per inch of fault. With the specification stated as counts per time, the tolerances on wire speed and electrode length limit the count rate that can be specified. With the specification changed to fault length per count, rather than time per count, the standard will be able to come much closer to what is desired. An additional advantage of such a change is that it would be applicable to any wire speed, not just the 60 ft per minute specified for the sampling test.
The study and research by the NEMA 6-MW technical committee has also disclosed that the carbon fiber brush electrode eliminates the double counting that is characteristic of the dual v-grooved sheaves as well as the under counting characteristic of the carbon foam electrodes. They have proposed changes in the standard to use carbon fiber brush electrodes instead of the dual v-grooved sheaves. Not only will this provide a better and more repeatable sampling test, it will facilitate NEMA compatible in-line testing systems.
Some users of the early in-line HVC systems discovered that they left an electric charge on the wire after it had passed through the test electrode. This accumulated electric charge on the wire spool was the cause of some difficulties.
A system has been developed that will discharge the wire after it has passed through the carbon fiber brush test electrode. As a consequence of this study, we also discovered that at very high wire speeds the capacitance of the wire to the electrode was such that the NEMA specified series resistor prevented the wire from being charged sufficiently during the short time it was in the 1-in electrode. This prevented the achievement of the NEMA specified test voltage. This discovery was discouraging. It appeared to prevent NEMA compatible HVC testing at high wire speeds. Fortunately, we soon developed a cure for this problem. We have provided a crosswise carbon fiber brush ahead of the 1-in test brush electrode. This preceding brush charges the wire so that the proper test voltage is achieved, permitting the achievement of the NEMA test standards at very high in-line wire speeds.
We think it is very important to attain full compatibility between the manufacturers' sampling tests, which have been used for decades, the manufacturers' in-line testing of all of the wire rather than just small samples, and the magnet wire users' sampling and in-line testing. But even beyond the benefits from such compatibility, potentially the greatest benefit to accrue from in-line testing is that it will immediately reveal any process deficiencies in time to correct them and eliminate making wire that must later be scrapped. The result should be better wire at lower cost, an unbeatable combination.
As these benefits of in-line testing are widely recognized and accepted there will be increasing interest in in-line testing for other wire parameters such as surface defect detection and more exotic applications such as dielectric strength testing and dissipation factor testing.
Recognition of the value of in-line testing stimulated the development of several test systems. Unfortunately, they were quite expensive and did not conform to the NEMA test standards nor did they produce results compatible with the NEMA sampling tests that the industry had been familiar with for many years.
Now that gap has been closed.
ABOUT THE AUTHOR
John A. Whitney earned a B.S. in Electrical Engineering from the Indiana Institute of Technology, and an M.S. in Electrical Engineering from Purdue University. He taught electrical engineering classes for 17 years while developing test equipment in Indiana Tech's R & D Laboratories, now known as A/Z Tech, Inc. (Fort Wayne, IN, U.S.) He also served as manager, Engineering, Advanced Development, for 25 years at the Franklin Electric Co., in Bluffton, IN, U.S.