NETA Insulation resistance test min resistance

[W@ttson]

NETA says the following:

1. IR 1 min = kV + 1 for most windings made before 1970, all field windings, and others not described in 2.2 and 2.3. (kV is the rated machine terminal-to-terminal voltage in rms kV)
2. IR 1 min = 100 megohms for most dc armature and ac windings built after 1970 (form-wound coils).
3. IR 1 min = 5 megohms for most machines and random-wound stator coils and form-wound coils rated below 1 kV.


If I am reading the above correctly, the intent of #2 is 100MOhms minimum for DC armatures and AC windings built after 1970 (form-wound coils) ABOVE 1 kV.

Why don't they list the voltage rating? #3 gives the below 1KV rating.

 

[paulengr]

NETA says the following:

1. IR 1 min = kV + 1 for most windings made before 1970, all field windings, and others not described in 2.2 and 2.3. (kV is the rated machine terminal-to-terminal voltage in rms kV)
2. IR 1 min = 100 megohms for most dc armature and ac windings built after 1970 (form-wound coils).
3. IR 1 min = 5 megohms for most machines and random-wound stator coils and form-wound coils rated below 1 kV.


If I am reading the above correctly, the intent of #2 is 100MOhms minimum for DC armatures and AC windings built after 1970 (form-wound coils) ABOVE 1 kV.

Why don't they list the voltage rating? #3 gives the below 1KV rating.

First off be careful with NETA. This is a standard written by testing-only companies for testing-only companies. There is little to no engineering or understanding of repair or maintenance requirements. They never met a test they didn’t like. The same test standards recommend regularly unwiring and running insulation resistance tests on control wiring! The failure rate just from miswiring has to exceed the tiny amount of detected failures especially with a test with little predictive capability.

NETA is getting this text from IEEE standard 43 which is similarly poorly written. It helps to understand what you are testing to put it in context. I’ll put it this way: nonpolymeric insulated coils (asphalt and mica) leaked pretty bad so the 1 Megaohms per kV plus 1 Megaohms was the rule of thumb for years. The leakage was just plain terrible. Recognizing that polymer insulation’s were far better EASA and IEEE changed both the test voltages and the acceptance criteria in the 1970s. There are still some issues. It is easier to read this backwards. Above 1 kV the physical insulation types and thicknesses are such that 100 Megaohms is a comfortable minimum. In practice most MV coils measure into to Gigaohms unless they are severely contaminated or damaged. Similarly the thick slot insulation used in form wound coils meets that same requirement. Generally speaking form wound coils are used on 200 HP or larger machines where the number of turns is roughly 1 turn with many parallel conductors. Flat wire form wound coils work much better at higher HP than random wound. Everything else under 1 kV except rotors will meet the 5 Megaohms specification. With rotors there is little need for thick insulation and insulation resistance is generally terrible. As noted in IEEE standard 43 and EASA 1 Megaohms per kilovolt is enough. In practice I often see rotors that are pretty bad but most of the time this has more to do with carbon contamination at the slip ring or commutator assembly than the rotor itself when I’ve been able to disconnect them.

So this is pretty easy to understand if you have a solid understanding of motors in general. Either way most of the time except for spot testing I mostly use an MCE tester that includes a graphical PI test recommended by the latest IEEE standard 43. You can readily see if there are issues with this result irrespective of the numerical results. If it doesn’t look right (roughly a logarithmic chart) something is wrong with the motor and it’s just a question of whether or not it is bad enough to repair it. NETA mentions PI as an optional test but fails to mention graphical PI. As an example of why this is superior if you see a stair-stepped pattern this is an indication of severely crazed insulation that is breaking down. It will pass insulation resistance if it’s dry but it is failing and rewinding is warranted. PI (numerical) is also useless above 5 Megaohms but graphical PI is valid anyway.

I work for a large motor manufacturer and I test and troubleshoot hundreds of motors a year. Mostly these are field tests on less than perfect equipment as well as some brand new or recently rebuilt motors.

 

[W@ttson]

wow what a thorough answer.

I like that you gave the history of where the 1 MOhm came from and why the 1970s was the tipping point.

I didn't like that #2 and #3 both listed form wound motors but one had below 1KV.

I'll have to look into regular PI test vs graphical PI test.

 

[paulengr]

wow what a thorough answer.

I like that you gave the history of where the 1 MOhm came from and why the 1970s was the tipping point.

I didn't like that #2 and #3 both listed form wound motors but one had below 1KV.

I'll have to look into regular PI test vs graphical PI test.

The problem with the way it is written is that it’s a bit ambiguous. They would be better with some kind of chart.

The problem with the standard IR (insulation resistance) is that it is highly temperature sensitive and even after the suggested temperature correction not very repeatable….the temperature correction doesn’t actually work that well. There is a much more extended list of temperature corrections for different insulation types but it too has the same issue.

So conceptually when we run an insulation test there are roughly 3 capacitances to overcome. The first one is just the regular dielectric impedance. We use a higher voltage to charge this capacitance quickly. If is usually fully charged in about 10-20 seconds. After that what you will see is the insulation molecules slowly aligning themselves to the applied electric field. Insulation resistance slowly climbs after that point but gradually almost levels off. Eventually in theory all we are left with is the leakage to the motor frame which is the important thing we are trying to measure.

The IR test uses a reading at 1 minute then attempts to temperature correct it (if you do the test by the book). At this point polarization is just starting so it’s somewhat of an approximate value but it’s quick and easy.

The PI or polarization test takes this a step further by attempting to use the polarization effect as an advantage. With the polarization index test you continue to run the insulation test for 600 seconds or 10 minutes. At the end of 10 minutes divide the 10 minute reading by the 1 minute reading. Since both tests are at the same temperature PI is immune to temperature variation. It basically measures the slope of the polarizing curve. Anything in the range of 2-6 is considered good. In the range of around 1.25-2,0 is considered questionable. High can also be bad. The numerical PI test is considered invalid if IR exceeds 5,000 Megaohms. The standard does not say if this is 5 G temperature corrected or not so there is always some question.

The problem is some motors simply never reach full polarization even in ten minutes and I have my suspicions they mess up the 1 minute reading, too. These motors have ridiculously high PI scores. I’ve seen over 10. With low scores it’s not necessarily bad either or not indicating a problem. Many motors in tire plants are contaminated with carbon black, a semiconductor. I get all kinds of strange results from these and interpretation is difficult such as a quick rise to an acceptable IR that flat lines after that resulting in a PI near 1.0. Does it pass or not? Experience says it’s good enough (still running one year later). Or sometimes I get a large “dip” in the middle of the PI test a couple minutes in that causes PI to fail but IR passes. It’s not an anomaly…I can rerun the test after fully discharging and get a similar result.

There are lots of other examples. That is why the numerical PI test has fallen out of favor. There are simply too many exceptions to the rule.

A similar test with short lived popularity is the DAR test, Dielectric Absorption Ratio. Think of it as PI on steroids. Nobody likes waiting 10 minutes for PI to finish. So the DAR test takes the ratio of a 30 second reading to the 60 second reading. Quite often nothing has settled down in 30 seconds so DAR was quickly rejected because it passed too many clearly defective motors. Even NETA dropped DAR years ago.

This is getting way off into technical details about the test which is so utterly simple that PI kind of over complicates it.

Another variation that has its proponents for years was the capacitance to ground test. That initial capacitance though really tells you very little about insulation condition. PdMA claims it can measure level of contamination but I haven’t seen much evidence of that.