I just had a 3-phase 415V motor tripping instantaneously on start-up. This is an 11kW motor. I did a megger test phase-to-phase, phase-to-earth of the motor and cables separately. The readings were very good,some even more than 1G-Ohm. I also measured the winding resistance and were all balanced.

However, when I dismantle the motor I found some burnt marks at the end windings and some broken strands. I'm surprised this did not show up in either the megger or resistance measurements.

Was wondering if anyone could shed some light on this? And what conditions where a failed winding would not show its face in a megger test.

It is not a very common scenario, but I have recently read a detailed report of a plant that had a very similar scenario (except no "broken strands")

One thing to consider: - there are expected to be small defects in the magnet wire insulation of a random wound motor. If there is a slot liner, then it is not much a problem unless the defect is in very close proximity to a defect from another conductor (either touching, or close enough to arc over during transient overvoltage conditions). Then we can have turn to turn fault or a phase to phase fault.

Now what happens when we start a motor - at least two things:
1 - There is movement which can cause momentary contact of two defects on two different wires
2 - There can be a voltage surge far above normal voltage during contactor closing. If we have non-simultaneous contact closure, the prestrike voltage for the 2nd pole closing can approach 2.4 pu and for 3rd pole closing can approach 2.8 pu. (1 pu = VLL*sqrt(2)/sqrt(3) = peak of nominal line to ground voltage). When that 3rd phase closes it sends a wave which when reaches the motor terminal under worst-case impedance mismatch conditions can in theory double to at the motor terminals to 5.6 pu (although will be lower within the motor).

So we can imagine that during starting, the winding movement causes contact of two defects, or the voltage surge causes arcing of two defects that are close but not touching.

Some possible followup investigation:
* Surge testing of the motor can replicate/exceed the voltage surge of starting although not the movement.
* Figure out the location of the black within the coil may shed some light. If near the line-end coil, then the voltage surge during starting is somewhat more likely to have played a role. If deep in the winding, probably didn't play a role.
** The fact that you saw broken strands but no resistive winding imbalance seems contradictory. Did you test direct at the motor? (that is more sensitive than testing from the switchgear). What type of tester and what unbalance was recorded?
* ac testing can be more sensitive in detecting turn turn faults than dc winding resistance testing. For example power frequency inductance. Becomes more evident at higher frequencies.

A few other thoughts:
- Faults can sometimes burn themselves clear. For example if there is moisture contributing to the initial fault.

This message has been edited. Last edited by: electricpete,

Calvin,

I'm new to this, so please help me out. How did you perform the phase to phase megger test? Wouldn't it show some low reistance value? Please correct me if I'm wrong.

Thanks

I have not read about this, or someone told me about this...this very simular thing has happened to me before on a 500hp motor. Rememeber one key factor the meggar is putting out a relatively small voltage (input) to evaluate if there is any resistance or lack there of in the wire and or winding of the motor. It may not have picked up on a few strands or varnish (winding coating) melted thin. Or maybe you need a better megohmmeter or insulation tester. The situation is, when full voltage is applied, and in most cases full load amps (FLA) is applied when you start up, this is enough current to jump out and ground and or trip most types of drives, thermal protection and or fault protection. The underlying problem may be a phase issue, or overheating of the motor.

Calvin,

A meggar or resistance to ground test only checks the insulation between the copper winding with reference to ground. It sounds as if you have a turn to turn problem which can only be found by the "Surge" test. The surge test is actually testing the insulation between copper to copper. These are two distincly different tests.

The surge test works on the basis of Paschen's Law which states that a voltage of 325 VDC is required to create an arc between turns .0003 inches apart.

The surge test can actually be a predictive test, giving ample time for proper maintenance scheduling to replace the motor. That's right, an electric motor failing a surge test will still run for a period of time, depending on what voltage level it failed at. As an example, I have surge tested in service 460 VAC 3 phase motors (powered down of course) using the IEEE 43-1974 standard of 2 times line voltage, plus 1000 VDC. I have documented motors that ran for over a year after failing the surge test around or close to 1900 VDC, that were still running when replaced and powered by across the line starters.

On the same note, I also have a documented case of a motor failing catastrophicaly, running only two days after it failed the surge test around 800 VDC.

I would like to add that a computerized digital advanced winding tester was used in the above examples, which steps the voltage up about 50 volts at a time to 2000 VDC. This was not an old analog unit using manual controls.

This message has been edited. Last edited by: Jim Zuidema,

Thanks for the responses. It helps me to shed light on what was probably going on in the motor. It does look like a turn-turn short.

Moejoe, I did the ph-ph megger at the motor terminals by disconnecting the cables and opening the links for 'delta' connection. I got very good readings of few hundred meg-ohms to 1 G-ohm. I've also done the megger ph-earth at the terminal & got similar readings. I've also done separate meggers on cables with both ends disconnected and got good readings.

The wdg resist was uniformly 1.0ohms for all phases and was done at directly at the motor terminals using a multimeter.

What really surprises me is that despite such good megger & resist values, the motor trips instantaneously on start-up. I do not even have the chance to capture the current reading with my clamp-meter and the motor didn't even start to turn. THis really baffles me.

It's surprising that a turn-turn fault could have caused such high current tripping. Would I be right to say if the turn-to-turn fault was in the same phase, the motor would not have tripped? Or perhaps this was a turn-turn fault involving 2 phases? But in that case how come the ph-ph megger did not show anything?

Only by opening the motor I could see the broken strands and burnt marks. (See attached picture)

quote:

The wdg resist was uniformly 1.0ohms for all phases and was done at directly at the motor terminals using a multimeter.

A more sensitive dc test would be using a bridge - should be able to get resolution to 0.001 ohm. (And of course as previously mentioned, ac tests and surge tests can provide even more info).

quote:

It's surprising that a turn-turn fault could have caused such high current tripping. Would I be right to say if the turn-to-turn fault was in the same phase, the motor would not have tripped? Or perhaps this was a turn-turn fault involving 2 phases? But in that case how come the ph-ph megger did not show anything?

The most likely type of turn to turn fault is within a single coil since there is more numerous contacts between conductors within a coil than between coils (including between phases). There is of course many contacts between coils in areas such as the endwindings, just not as many wires contacting each other as there are within a coil.

So the most likely type of fault from the perspective of proximity (ignoring issues of voltage stress) is within a coil.

What is the expected increase in current from a fault within a coil? I'd say that the starting current will increase roughly in proportion to the fraction of total series turns shorted (assuming single circuit wye for simplicity). Typical small motor might have 8 series coils and 30 turns per coil (that's a swag off the top of my head - others welcome to chime in if you have a better guess). The turn-to-turn short within a coil might short a single turn within the coil or the first/last turns within a coil. So the range of turns shorted is between 1/240 (0.5%) of the total series turns and 30/240 (12%) of the total series turns. The expected increase in starting current would be on the same order.

Is that enough to cause an instantaneous trip? Depends where you started in terms of motor starting current and breaker instantaneous setpoint. And figuring out how much margin you have to trip is not at all an easy proposition, considering the transient decaying dc offset that affects the first peak, breaker trip variability, breaker response variability. Different styles of breakers respond differently to that transient as well (even though they have the same setpoint when tested by pulse or ramp method). For example Westinghouse and Cutler Hammer type MCP breakers have a "transient suppressor" (not surge supressor) damping element built into the instantaneous trip mechanisms which is supposed to help the breaker ride through the first cycle peak without tripping instantaneously (note, that is a contradiction of the term "instantaneous"). Different motors have different L/R decay constant associated with their transient current component. Energy efficient motors have particularly long decay constant due to the low rotor resistance, so an energy efficient motor will have a higher possible first cycle peak than a non-energy efficient motor of the same exact same rating and "kva code letter" (locked rotor horsepower per kva). You can try to measure current with clamp-on probes and a digital recorder, but unless they are hall effect probes, the interpretation of that first cycle peak is dicey (ac probes have a response which allows you to catch some but not necessarily all of that transient decaying component - I have written a white paper if anyone is interested). Also even within a given style of breaker there is tremendous wide tolerance on the trip setting (something like +20 / -40% coming from manufacturers). I could go on and on because it's something I have been battling intensely lately - troubleshooting the cause of intermittent random-wound motor trips when everything under the sun tests ok.

The bottom line - a turn-to-turn short within a coil certainly could push you over the edge to tripping on instantaneous current during starting if you were close to begin with (which many motors are... breaker instantaneous trip setting is not an exact science and the approach is sometimes to increase the instantaneous setting just until no more trips occur during start.. and the National Electric Code limits how high you can put it so having a lot of margin is not always an option).

Note that the case of trip on a running turn to turn fault is a lot different than the case of trip on a starting turn to turn fault. In the case of running turn to turn fault, that 0.5% - 12% is not going to put you anywhere near instantaneous trip setpoint and probably not even near overload trip setpoint. But it produces tremendous heat and melts the conductor and causes more damage which leads to a fault between phases or to ground which trips the motor.

Summary: Turn to turn fault within a coil while running cannot cause a trip by itself... always leads to more damage (phase or ground fault which causes the trip). However for turn to turn fault within a coil while starting, it is plausible that this fault by itself can push you over the edge to instantaneous trip even if no other fault exists (which is not to say that no other fault exists in your particular scenario).

Clear as mud? That's just my opinion and I might be wrong and I probably have made it more complicated than it needs to be. But I'll be very interested to hear what others have to say on the range of subjects we have discussed. As I mentioned, it is a subject of a lot of interest to me based on similar things we are going through right now.

This message has been edited. Last edited by: electricpete,

[QUOTE]Originally posted by Calvin:
The wdg resist was uniformly 1.0ohms for all phases and was done at directly at the motor terminals using a multimeter.
QUOTE]
There are very few things to add. The previous posts provided a very good explanation of what is going on. Maybe I will touch an area of the resistance you are supposed to find out. As E-pete stated before, measuring resistance with the multimeter is rarely enough. So the question is, what resistance should you expect to find out?
Your motor is 11 kW. At that kW range, the efficiency of the motor is, let say, 92%. In other words, the losses are 8%. The 8 % of the 11 kW is 880 Watts. Divide this number by 3. Why by 3? Very roughly, 1/3 of the losses are in the rotor, 1/3 is in the iron, and 1/3 is in the copper of the stator. So the losses in the stator are roughly 880/3= 300 W. You know better than I do what the nameplate current is, but I estimate it is:
I= 11000/(1.73*415*0.92*0.9)= 18.5 Amp
The 0.9 is the power factor.
The accuracy plays a very little role. Hence, the resistance per phase is roughly:
R= (300/3phases) Watts/ 18.5^2= 0.29 Ohms.
If you measure the resistance line to line you should find roughly 2*0.29= 0.58 Ohm.
As you can see, you have to measure the resistance more carefully, the multimeter is not enough.
The efficiencies and power factors are found on the nameplates of the motors more and more often. The above calculations do not have to be accurate, they are just supposed to give you a range so that you know if your measurement of resistance was reasonably accurate. As the above calculations show, the multimeter did not give you sufficient information.
jank

Jank,

Nice work up on the problem. For more basic, handy electrical related equation formulas including a ton of other very usefull information, I'd recomend the "UGLY'S ELECTRICAL REFERENCES" handbook by George V. Hart, ISBN 0-9623229-7-0. (I bought one recently at Home Depot for less than $10.00).

Another source of related information would be the attached spreadsheet that a good friend (he happens to offline test all incoming new and repaired motors)of mine put together for the exact purpose of helping electricians determine what they should see in the field when checking phase to phase resistance values, with inductance thrown in for good measure. Granted, these are ball park values, varying with different vendors and manufactures of 460 VAC 3 Phase motors.

Resistance_Phase_to_Phase.xls (28 Kb, 31 downloads) Excel Spreadsheet

Do you have same worksheet for 3 Phase,440VAC, 50HZ System.
Tramendous learning here .thanks to all.

Thermographer,

Sorry, my friend doesn't have a spreadsheet specifically for 440 VAC 3 Phase 50 hertz motors, however the phase to phase resistance values should be comparable, or close to the same size 60 hertz motor.

I quickly scanned this thread and I am sure that the question may have been answered to some extent.

An insulation to ground test evaluates the leakage from the surfaces of the conductors nearest the groundwall and ground and converts it to resistance (hopefully in the MegOhm, GigOhm, or Terra-Ohm range). It cannot detect the leakage between individual conductors.

If you are able to break the neutral and check phase to phase insulation to ground, then you will measure only the surfaces closest to each other (or leakage paths) between the phases, not between turns in the individual coils or conductors.

If a conductor is open or there is air between damaged components, then you may not detect any issues there, either.

There are other tests that are available to detect turn-to-turn insulation values. Each technology has its own strengths and weaknesses, but they include Motor Circuit Analysis (MCA) - a low voltage testing method that detects changes in capacitance - and surge comparison testing - a high voltage test used to look for insulation weakness.

In the case of the picture above, a milli or micro-ohm test may have detected the fault as conductors appear to be open.

Sincerely,
Howard

quote:

Originally posted by MotorDoc:
There are other tests that are available to detect turn-to-turn insulation values. Each technology has its own strengths and weaknesses, but they include Motor Circuit Analysis (MCA) - a low voltage testing method that detects changes in capacitance - and surge comparison testing - a high voltage test used to look for insulation weakness.
Howard

One of the many weaknesses of the low voltage testing method is that it cannot detect the changes in capacitance. The turn-to-turn capacitance is so low that the changes of current caused by the change of capacitance is absolutely impossible to detect by the low voltage testing. The impossibility was shown in the thread “Disappointing demo from low voltage tester”.

http://maintenanceforums.com/eve/forums/a/tpc/f/7161085912/m/5731057063/p/4

Shown are calculations that prove that the changes in capacitance are totally out of reach of the low voltage testers. Yet it does not stop the low voltage proponents to repeat, that they are in fact able to detect the changes. They are not able to prove it by using the simple calculations as they were used in the above link. The only “proof” is repeating their statements over and over.
I would like to see, finally, some proof. But as we have seen before, the proofs from those guys are not coming. Including the top dogs in the low voltage testing.
jank

Sorry Jank, took a look at that mess and I am not getting involved in that one at this point in time.

In actual application the ATPro, PdMA, and Baker have more than proven themselves. The key is selecting the right tool for the job. I think that point was well made throughout the posting, that you cited, by Aditya.

The question in this thread was why an insulation to ground tester did not detect an existing inter-turn short.

Maybe the post does look little messy. I admit, that for people who do not use math in everyday life, the amount may have looked little overwhelming. Hoping, it is for benefit of everybody, this time, I will try to simplify it as much as possible, so that everybody can see there is very little science behind it. My goal is using as little math as possible. Also this time the task will be easier, because I can concentrate on the turn-to-turn low voltage testing, because the statement that low voltage testing can detect “TURN-TO-TURN INSULATION VALUES…. from… CHANGES OF THE CAPACITANCE” triggered this response.

First of all you have to find out what the capacitance turn-to-turn is.
There is a 2300V, 500 hp, 2-pole, 108 Amp, 60 slots, 6 groups of 10 coils, each coil with 5 turns… motor sitting in our shop waiting for the stator to be rewound. The length of the iron is 16” (0.4 m). Measuring with a tape measure, the length of one turn is approx. 2 meters. The width of the slot is 0.364” (9.25 mm); the width of the rectangular wire is 0.25” (6.35 mm). The thickness of the insulation (enamel) between the turns is 0.004” (0.1 mm).

CALCULATIONS:
(I know that lots of people do not like math. So feel free to jump to “CONCLUSIONS” below.)
Let’s calculate the capacitance turn-to-turn:
The capacitance per turn is Ct=(e0)* e * A/d
e0… is the permitivity of the vacuum (8.855* 10^-12 F/m)
e… is the relative permitivity of the insulation on the wire. It is roughly 4.
A… is the square area of the “capacitor” plate. It is the length of one turn (2 m) times the width of the wire (6.35 mm).
d… is the distance of the “capacitor” plates. In our case it is the equal to the double of the thickness of the wire insulation: 0.004” (0.1 mm).
Hence the Ct is:
Ct= 8.855* 10^(-12)* 4*(2*0.00635)/0.0001= 4500*10^(-12) Farads= 4500pico-Farads.
The capacitance turn to turn is 4500 pico-Farads.
The locked rotor current is about 5 times the name plate current: 5*108=540 Amps, which gives us per phase impedance:
X=2300 Volts/ (1.73*540 Amps)= 2.46 Ohms per phase. In other words the inductance is (60 Hz motor):
L= X /w= 2.46/( 2*pi*60)= 0.0065 Henry = 6.5 mH.
This inductance is created by 100 turns in series (It is clear from the winding data. You don’t have to take my word for it, but your rewind shop will confirm it). It means there is 100 turns between the beginning of the winding and the neutral (the winding is connected in wye). Of course the inductance PER TURN will be 6.5mH/100 turns= 0.065 mH.
Finally we have reached the point where we can do some comparing:
We have an inductance PER TURN of 0.065 mH and a capacitance TURN-to-TURN of 4500pF hooked up in parallel. The inductive reactance at 200 Hz (which is the ATPro default frequency) is:
Xl= w*L= 2*pi*f*L=2*pi*200*0.000065=0.082 Ohm
The capacitive reactance at 200 Hz is:
Xc=1/wC=1/(2*pi*f*C)=1/[2*pi*200*4500*10^(-12)]=
1000000000/[2*pi*200*4.5]= 177000 Ohms.

CONCLUSIONS:

Now we have 2 reactances in PARALLEL. One is 0.082 Ohm inductive reactance per turn, the other is 177000 Ohm of capacitive reactance per turn. The ratio is smaller than 1: 2 million!
Testing with low voltage tester means nothing more than measuring the current. When measuring the current, there is no part that would advertise: “I am the current due to the capacitance”. There is only one CURRENT. If you hook up the low voltage tester (10 Volt) to only one phase from the above motor, the current through the copper of the winding (inductances) will be 0.1/0.082=1.22 Amps. (The 0.1 is 10Volts divided by 100 turns). The current through the capacitance turn-to-turn will be 0.1/177000=0.000000564 Amps. It does not really matter if the capacitance changes even by the factor of 4 (by permitivity dropping to 1). The current will always be 1.22 Amp.
It takes more than unhealthy audacity to pretend that the capacitive reactance changes can be noticeable. So don’t be surprised that the champions of the low voltage testing do not hurry to get involved in a serious technical discussion. They know they would get a beating. And that is why they are silent, hoping it will somehow go away.
jank

Jank

Why thank you for summarizing the argument on the other string, especially keeping it extremely simple. I am a little worn out from reviewing papers all weekend at the IEEE DEIS CEIDP conference here in Quebec. Pretty cool stuff coming out in research on nano-composite dielectrics and low voltage partial discharge!

OK, so we will assume that you have a single layer of insulation system throughout the motor. What happens to the circuit capacitance if you have a turn to turn insulation degradation (ie: tracking) 25 turns into your 100 turns? To keep it simple, we will say it is a 1 mm carbon track with a dielectric constant of 2.2? Should we also assume that it will be a single layer of insulation between those turns?

Howard

I have already stated that no matter what the change of the capacitance turn-to-turn is, it will ALWAYS DROWN in (at least) MILLION TIMES LARGER current through the copper of the winding when testing with low voltage tester. This simple fact renders the low voltage testing useless. If the relative permitivity (of the whole winding) drops to 2.2, the original 4500 pico-Farads will drop to 2475 pico-Farads and the resulting current will still be 1.22 Amps.
I don’t really know why you talk about tracking. Tracking does not occur because of the capacitance or the capacitance change, but because of the RESISTANCE. Any NON-IDEAL capacitor is in fact an ideal capacitor with a parallel resistor. Tracking happens through the parallel resistor not through the capacitance. It also means that it is NOT FREQUENCY DEPENDENT. Hence, for the tracking, you can change frequency as you wish and the currents do not change at all.
You have changed the relative permitivity to 2.2 from my 4. The resulting change of capacitance does not mean absolutely anything, hence you cannot use it to detect “TURN-TO-TURN INSULATION VALUES…. from… CHANGES OF THE CAPACITANCE”.
Obviously, it is not the capacitance; so what is it? If the low voltage testers do not work because of the capacitance changes, but because of the leakage current (I mean the watt current) through the turn-to-turn insulation, we can discuss it further. However, the results will be just as absurd as they were for the capacitance, even worst. Capacitive currents in AC machines are always several orders higher than the watt currents (turn-to-turn, turn to ground, does not matter).
Note: Where did the 1 mm come from? I don’t understand what you are talking about. The insulation turn-to-turn is 0.1 mm. The last sentence in your posting: Will you kindly answer it yourself?
jank

Jank

Unfortunately, that is not exactly how an insulation system works. Attached is a preliminary working paper on circuit capacitance and the effects of voltage and frequency. The full paper and model is in review for publication so cannot be published here, but the gist is the same.

The 1mm was in relation to the area of the fault in altered insulation in parallel with the conductors that you have established. The good news is that I had used the simplified ideal impedance for the model and paper such that this is in line with your model.

Howard

Evaluation_of_Capacitance_in_Motor_Circuit_Analysis_Findings.pdf (204 Kb, 30 downloads)

Let’s have a look at the last page of the paper. You have to go all the way past the “About the author” to a page titled “Analysis of Winding Insulation System”.
From the dimensions we can judge that a very small motor, random wound, was modeled. You can notice, that the wire used was #18 AWG. The dia of the wire is 0.001125m = 1.1125 mm.
Then we learn that the insulation thickness was 0.000625 m = 0.625 mm. Wow, normally it would be 0.002”= 0.05 mm. It must have been a very special wire.
In the next column we can see the area of the coil 0.0121 m. We are mostly used to m^2.
Then we arrive at wire resistance: 2.01. The dimension is missing, but one would assume it is Ohm. But it cannot be Ohm, because the resistance of wire #18AWG is 0.021 Ohm per meter.
Following are the results: Good winding (L) 2.06506667 of something, I assume of mH. It is a usual dimension in low voltage testing.
If you calculate the reactance of the winding you will get:

X= 2*pi*f*L= 2*pi*400*2.06506667*10^(-3)= 5.18 Ohm.
The impedance Z should be just a little higher. And indeed it is: 203.05 Ohms. Hmm, it looks little to high! One would expect something below 10 Ohms even for such a tiny motor. So let’s calculate what the resistance is from the reactance and the impedance:
R=sqrt(Z^2-X^2)=202 Ohm. Are we going to get this resistance from the winding data? There are 36 coils total, it means 12 coils per phase. Each coil has 20 turns, the average length of the turn is 0.42 m, so the total length is:
12 coils* 20 turns * 0.42 meters per turn * 2.01 Ohms= 202.6 Ohms. It is just great! The only trouble is, the resistance per turn is actually 100 times less. And we are talking line to neutral resistance, not line to line.
But let’s go further: Shorted winding data. Everybody would expect lower inductance if the winding is shorted out. It does not seem to be the case; the inductance is constant no matter what. Similarly you would expect drop in impedance. No, impedance has grown from 203 Ohms to 8710 Ohms.
Shall I go any further? I don’t think so. I believe, for Motor Doc it should be going back to the drawing board.
There are other inaccuracies in the body of the paper. For example : “2.13*(10^-11) Farads or 2.13 pico-Farads”. Of course it should read 21.3 pico-Fards. And something what really draw my attention : “With a high AC voltage is connected between terminals, the capacitance of the insulation system dominates the current.”. I explain it to myself that 10 micro-Farads is much larger than 10 Amps.
I do not think that the paper will be published.
jank