Mike Howell
EASA Technical Support Specialist

Most service centers perform some form of surge comparison testing, though terminology and methodology may vary. In simple terms, two winding responses or waveforms from a fast rise-time surge are compared and if there is an excessive difference, the unit under test may have a defect. The waveform that is produced by the pulse is unique to the unit under test, which for example, could be a stator winding. The waveform will be a function of the resistance, capacitance and inductance of the test circuit and quite a few variables can affect those characteristics.

One difficulty or challenge with surge comparison testing has been its subjectivity. That is, it is not always easy for operators to reach the same conclusion when comparing two waveforms. Within the last few decades, several equipment manufacturers have begun to utilize analytical methods to evaluate the surge comparison test results. The goal is to remove as much subjectivity as possible so that disposition of the unit under test is a simple decision for the operator. The analytical method that has become most popular, in various forms, is use of the Error Area Ratio (EAR).

Do you have to have EAR capabilities in order to perform surge comparison testing satisfactorily? No, but if you have the capability, a basic understanding of the data reported by the equipment can help you make an informed decision.

What is the EAR?
For the purposes of this article, we will just look at the problem graphically as explained in some literature [1]. In Figure 1, the two waveforms shown are representative of a typical surge comparison test. Common practice is to plot voltage on the vertical axis and time on the horizontal axis. So, at time “A” in Figure 1, a pulse is applied that has a very fast rise as seen by the steep slope. Let’s call waveform “1” in Figure 1 the reference waveform and waveform “2” in Figure 1 the test waveform. If the two waveforms were of test circuits with the same resistance, inductance and capacitance, we would expect them to look identical, i.e., they would look like one waveform when superimposed. In Figure 1, you can see that waveform “1” and waveform “2” are not identical; waveform “2” has different time dependency (frequency) and different voltage levels (amplitude). The question is, how different is too different? Early automated testing systems used voltage comparisons where at some points in time the difference in voltage between the two waveforms would be calculated as shown by “3” in Figure 1. Most literature suggests that the EAR is a more effective method.

If we expected the two waveforms in Figure 1 to be identical, we could consider any difference between them to be error. Figure 2 shows those same two waveforms with the area between them shaded. This area will be referred to as the Error Area. It is evident that as the waveforms become more different, the area between them will become larger. To develop a common set of acceptance criteria applicable to windings of different size, shape, configuration, etc., we need to normalize our result to something simple for comparison – like a percentage. Common practice is to compare the Error Area shown in Figure 2 to the area under the reference, which is shown in Figure 3. So, if the area in Figure 2 was 3 volt-seconds and the area in Figure 3 was 30 volt-seconds, the EAR would be 3/30 = 0.1 per unit = 10%.

Now that we’ve covered what the EAR is, let’s look at the two most common applications that are the Pulse-to-Pulse EAR (P-P EAR) and the Line-to-Line EAR (L-L EAR). These are two different tests and should be treated as such. A third use, not addressed here in detail, would be to compare an individual coil to a reference “golden coil” waveform stored in the memory of the test set. 

Pulse-to-Pulse (P-P EAR)
When performing a surge test in the past, you may remember finding a short or fault that only occurred above a certain voltage. That is, when bringing the voltage up slowly, you may have noticed a sudden change in the waveform, likely an increase in frequency (shift to the left) and maybe a noticeable change in amplitude. The P-P EAR test looks for this type of change – a pulse is applied to a winding, let’s say phase A, and then another pulse is applied to phase A, but at a slightly higher voltage. Because the second pulse is at a higher voltage, there is a reasonable and predictable difference between the two waveforms that is accounted for by the computer program. But, if the change becomes significantly larger than expected, so will the calculated P-P EAR, indicating to the operator that a potential defect has been identified. Because the P-P EAR test is only testing one winding or coil at different voltages, there are no concerns about differences in winding configuration or the magnetic circuit path, allowing for testing of assembled machines. This isn’t the case with the Line-to-Line EAR.

Line-to-Line (L-L EAR)
The L-L EAR is similar to what we traditionally think about with the surge comparison test – comparing two different windings or coils that we believe should be the same. For example, with a three-phase stator you’d get a value of L-L EAR for A-B, B-C and A-C. It isn’t unusual for L-L EAR values to exceed P-P EAR values for a given unit under test. There are a several factors that cause this with some examples being:

• Winding configuration
  • Various concentric winding patterns
  • Lap winding over the span (lazy-lapping)
  • Testing one path only of a part-winding-start connection
• Core iron condition
  • Shorted lams
  • Ground out pockets from failures
  • Geometric dissymmetry, e.g. varying back iron dimension
• Rotor position, if installed
  • Stator-rotor mutual coupling is a function of position

Since the L-L EAR is a comparison of two windings, it is valuable for finding differences between two windings – reversed coils, missing or extra turns, connection issues, etc. However, any of the conditions above can result in false negatives. Winding resistance, small (dummy) rotor testing and phase balance testing can provide useful information in these cases.

But, how high is too high?
After all the previous discussion on removing subjectivity from the surge comparison test and using an analytical method to determine acceptability, well – there is still some subjectivity. This means the same pass/fail criteria for L-L EAR and P-P EAR won’t necessarily be appropriate for every unit under test. At the time of this article, some of the manufacturers producing surge testers with specifications indicating EAR capabilities are not publishing pass/fail criteria within those specifications. The values you get depend not only on the unit under test but also on the equipment itself and how the data are calculated. So, operators should lean heavily on the manufacturer for assistance with determining acceptable starting points. If you are considering purchasing a test set with these capabilities – discuss this with the manufacturer in detail before placing the purchase order. Typically, P-P EAR values for three phase stators (with or without rotor installed) are below 10%. For L-L EAR values, typical results for acceptable three phase stators (without rotor installed) are below 15%.

Bibliography
[1] J. Lebesch, "Method and Apparatus for Automatically Calculating the Integrity of an Electrical Coil". USA Patent 5,111,149, 5 May 1992. 

Categories: AC motors, Testing, Motor testing

Tags: Surge comparison test

Rate this article:

4.1

Print