Thermographer,

I have already mentioned in an older posting on this thread (see Feb 29) that All-Test's assessment of winding contamination is unreliable, so we are in agreement on this point.

Now, regarding your readings: Those recommendations are coming as impedance is not equal between the phases. Please confirm whether you moved the rotor to the max. impedance position for each phase before taking those readings? If not, the test was not performed correctly. I have seen in most cases that impedance tends to be equal when taken at max. position (turn rotor keeping an eye on impedance for any one phase till it reaches max. value, repeat for other phases). Please inform later about your findings.

You seem to have bought the instrument without evaluating it first & are now frustrated?? Either use it where it does work (rotor bars, interturn shorts) or discard it & blow your money. I do know that I spent $ 20,000 buying their products & am not in a hurry to throw them away.

Only AT-PRO can "confirm" about phase angle is , not I. Perhaps "MotorDoc" if he's still around?

Regards,

Aditya

Hello Mr Aditya
The method u are suggesting is for using the ATP 31. Yes i have done the same testing with ATP 31 also and with same procedure that moving a rotor upto getting max impedence and then evaluating the phase balance . i got it max 4 % unbalance with ATP 31. but what are the ATP IV results those i have already put here. which says winding contamination.
One more thing today i check one motor with ATP 31 which results are as below. i request you if you can help me to data intepreation. i knw you have wide experience with this and u can do it better than me. please

ATP IV
T1-T2 T1-T3 T2-T3 DIFF
R 2.84 3.00 2.85 3.67
Z 98 99 50 39.27
L 19 19 20 3.45
PHI 77 76 76 1
I/F 46 46 46

Recommendations are same as previous one.
( I have evaluating the results with moving the rotor and results are same regarding Impedence Unbalance)

ATP 31 WITH 200 HZ

Z 28.4 27.9 27.9
PHI 68 67 67
I/F -45 -44 -44

PH BALANCE AFTER ROTOR POSITIONED ARE:
P.B 0 1 4

I have done rotor test finally which showing abnormal data . ( No sinewave )

which results i should follow and whats your opinion regarding motor health?

thanks for reading and for your info i m not frustrated but would like to know a better way of testing if i could not know.

thanks & regards

i must say that its an excellent report by you for GTG.
Little bit confusion regarding Circuit Analysis you have done . can you confirm in test no 5 , 7 , 8 & 9 you mentioned ALL TEST IV ( 400 HZ). Can you tell me How it is possible to change the testing frequency in ATP IV? What i know it is possible in ATP 31.

thanks & regards

quote:

Originally posted by Aditya:
A) AC resistance (not impedance) & DC resistance values do not necessarily have to be equal (skin effect, eddy current & hysteresis, as you pointed out). So, comparing the two as you have done may be a wrong approach. Then again, I can't argue with Pythagoras. So I looked at some of my data (see attachment, Page 13, Tables 4 & 5):
B)The DC resistance as measured by a rather precision micro-ohmmeter & that given by the All-Test IV are nearly equal. Does All-Test give AC or DC resistance? - I don't know.

The XL value as calculated from (Z x sine phi) does not match that calculated as (2 x pi x f x L).
As seen from the last report's comparison with Jeumont, All-Test & Jeumont's impedance values are almost identical.
C)So, now I'm really confused. I have data which shows that All-Test resistance & impedance values match those from other instruments, yet the impedance, inductance & phase angle numbers don't make sense. Is capacitance playing some role here?

This is not about defending numbers or instruments, its about understanding whats going on. I have used All-Test instruments to find faults successfully so I do trust them, but many things have never been explained.

D)I had sent a mail to All-Test to join the discussion, as well as taken up the matter with the PdMA Regional Manager. Both companies responded that they want to be neutral, so its left to us users to figure out the theory.
Regards,
Aditya

Responses:
A) In real life the increase of resistance in the stator is kept to minimum, well below 10%. You cannot expect increase of the resistance by 2600% (as seen above). You may expect something like 6%.
B) I am pretty sure that the resistance measured by All-Test is DC. You have to measure watts while AC testing to get the AC resistance.
C) In order to assess the impact of the capacitance I have attached an equivalent diagram for a 2000 Hp Reliance motor in a pdf file. (You calculate the actual Ohms from the “per unit” values by multiplying by BASE OHMS). The first page is the diagram as provided by Reliance, I have just added the 2300V (Line to Neutral voltage). In the second page I have added the capacitance to ground and calculated the elements for 60 Hz. Since Reliance does not provide this information I used capacitance per phase calculated for 2500 hp motor that was rewound in our shop some time ago. Hence the capacitance may be a little bit bigger than the actual one: 0.054 microfarads. The third page is for 200 Hz.
You may just wander why is the turn-to-turn capacitance neglected. I have also calculated the turn-to-turn for the 2500 Hp: 47 picofarads. Totally negligible! (If you want details I will provide them). Having the equivalent diagram you can compare the currents through the capacitance with the current through the rest of the circuit at 10 Volts (low voltage tester voltage):
1) At 60 Hz: Current through the capacitance: 10/49147= 0.000203 Amp
2) At 60 Hz: through the rest of the circuit: 10/1.36= 7.32 Amp
3) At 200 Hz: Current through the capacitance: 10/14744= 0.000678 Amps
4) At 200 Hz through the rest of the circuit: 10/4.552= 2.197 Amp.
So, for example, at 200 Hz you have to measure current: 2.197-0.000678=2.196313 Amps. Then, lets say, that the capacitive current of 0.000678 Amps has changed by few percent, lets say, by 10% to 0.000618 Amps (the increase as huge as 10% is very unrealistic-more discussion would be needed). So after the motor overheated you no longer measure the current 2.196313 but 2.196382 Amp. The difference in % is 0.003%!!!
The resolution required to catch those differences is phenomenal, completely out of reach by technology bought for few bucks. So if you wander why the testers do not work as insulation testers, the numbers above should provide a hint.
D) It is a truly unique way to keep customers (and potential customers) happy and informed.
jank

reliance_eq_diagram.pdf (147 Kb, 18 downloads)

Thermographer,

The motor you are testing is absolutely fine in my opinion. To answer all your points, I take readings with both the AT-31 & the AT-IV, AT-IV readings are taken after confirming max. impedance position with AT-31. Also, the AT-IV test frequency is determined by changing the AT-31 test frequency till the impedance readings are identical for both instruments.

The logic for the AT-IV contamination prediction is as follows: See if impedance is equal for all phases or not. If not, see if inductance has a similar variation. If both change similarly, the variation is considered to be due to rotor position. If only impedance varies, then is is assumed to be due to a change in capacitance (as contamination would change the capacitance).

While I agree that contamination would change the winding capacitance, its effect on impedance would be too nominal to consider. I simply ignore such recommendations from the software. There is no available answer (with me at least) on why the test frequency changes at times for one of the phases. I would suggest that you stick with PI & DD for contamination assessment.

Jank,

Your numbers are in line with what I imagined, viz. that the effect of capacitance on the impedance is negligible.

I'm off this thread now guys, its like flogging a dead horse. No responses forthcoming from PdMA & All-Test, some things work, some don't, strange numbers; eventually its left to our individual opinions.

Regards,

Aditya

I think that this “dead horse” needs few more kicks. (Otherwise, I am an animal lower kicking neither dead nor alive ones).

If you read the manual for the All-Test Pro carefully, (just Google the All-Test Pro and you can’t miss it) you will find out that the testing voltage supplied by the low voltage tester is not referenced to the ground of the motor. The manual clearly states, that connecting the yellow lead is only necessary for high voltage testing. For the low voltage testing, you just hook up two leads from the tester to two leads from the motor. If it were referenced to ground there would have to be connection to ground of the motor. It, of course, means that the calculations from my last post on this topic were significantly in error.

If the tester is hooked up just to each side of the winding, without any reference to ground (such as All Test), the energy in the capacitor to ground will be:

W=1/2(Cg/12)V^2

It means 12x lower than if the input voltage were connected between one side of the winding and ground. The Cg is the capacitance of the winding to ground. You can roughly measure such a capacitance with a Fluke. The other capacitor involved is the capacitance turn-to-turn.

We will now go back to the equivalent diagram of the 2000 Hp motor from the last posting:

The total turn-to-turn capacitance calculated from the total length of the wire is 0.3 microF. Hence the capacitance per turn is Ctt=0.3/N=0.3/80=0.00375 microF= 3750picoF. Those 80 capacitors (for 80 turns) are connected in series, so the capacitor equivalent to the 80 capacitors in series will be 3750/80=47pF (as said last time).
We have said that the capacitance to ground Cg=0.054microF. The equivalent (when not referenced to ground) will be 54000/12=4500 picoF.
We finally have the total capacitance that is seen by the low voltage tester: 4500+47=4547picoF.
The reactance of this capacitor at 200Hz:

Xc=1/(2*pi*f*C)=10^12/(2*3.14*200*4547)= 175100 Ohm

Current at 10 Volts and 200 Hz: 10/175100=0.0000571Amp.
Last time we have said that the current through the rest of the circuit at 200 Hz is 2.197 Amps. Those currents are 180 degrees out of phase, so the current measured by the low voltage tester is 2.197- 0.0000571=2.1969429 Amps.
And, as before, let’s “overheat” the winding.
What happens to the insulation when it overheats? I had a chance to measure capacitance on a good part of a failed large motor. The winding was then burned out for 20 hours at 700 deg F. I would say, that the winding was overheated after that. Yet, the capacitance dropped by mere 30%!!! This may answer questions about the function of capacitance in the assessment of the condition of the winding. There are smart ways to measure capacitance and there are st… I would say, not smart ways. If you let the capacitive micro-amps drown in the Amps of the inductive current you obviously picked up the wrong approach.
The 10% drop of capacitance in an overheated winding (as estimated last time) seems to be justified. The smaller current (after decreasing the capacitance) will be 0.0000514 Amps, new total current: 2.196486 Amp. The percent change: 0.00026%. I was wrong by the factor of 10 just because the tester is not referenced to ground (last time I calculated 0.003%). What is remarkable, that engineers from All-Test can catch those differences by measuring to 1 to 2 even zero decimal places (see the phase angle). I needed 7. Just imagine how many decimal places would one need if only one coil or one turn overheats.
The calculations behind the numbers presented are not obvious. I will provide details to anybody who asks. I want to keep this post short.
As we have heard above, we have to figure the theory ourselves, the low voltage tester manufacturers will not help us. And I am not surprised. Their equipment and testing methods would need a brand new set of natural laws and some substantial changes in math.
But now I am dead serious: The people who developed IEEE 1415 must have gone through the problems described many times in the years of development of the standard. They should provide the explanation! A document read around the world should be based on a solid scientific ground. There are enough reasons for serious doubts only in this little posting. I would like to see the details that convinced the working group to include the low voltage testing in the IEEE 1415-2006.
jank

Jank,

I am not sure where you are heading on the previous post.

If the point was that All-Test cannot assess ground insulation condition based on capacitance measurements, you are absolutely correct. The AT-IV can measure capacitance but they do not do any diagnostics with that number & I do not think they make claims regarding ground insulation assessment (other than taking an IR reading).

The All-Test instrument (as I see it) is useful for broken rotor bars & shorted turns (NOT weak insulation). I am yet to see anything in your argument against those points. We can do an experiment though: I will make a form wound coil, take high frequency impedance measurements, purposely short 2 or 3 turns & then take readings again. Lets see if the impedance values change or not. If I find some junk motor, will cut bars & see for impedance waveform variations also. Give me a couple of weeks as I am travelling right now.

IEEE 1415 talks about phase angle & phase balance, we have already discussed the mystery about the phase angle numbers earlier. I too am hesitant about those now. I have seen the phase angle variation correlate with impedance variations but the theory is not clear at the moment.

I can see your favourite bogeymen ‘Noah Bethel’ & ‘Howard Penrose’ on the IEEE 1415 Working Group. Do let us know if you get a response from them.

Regards,

Aditya

quote:

Originally posted by Aditya:
Jank,

I am not sure where you are heading on the previous post.

If the point was that All-Test cannot assess ground insulation condition based on capacitance measurements, you are absolutely correct.

Aditya

Maybe we should take it one thing at a time. In my posting, I am talking about overheated winding. Not only me, also other people have doubts that All Test can detect conditions such as contaminated or overheated winding. Read for example what one gentleman has to say:
----------------------------
Quote Aditya:
Negatives:
1. All-Test claims that it can detect winding contamination & overheating by comparing the impedance & inductance values of the three phases. Does not work! We have often found severe contamination using IR, PI, DD & tan delta which was confirmed physically; but AT-IV never picked it up.

End quote
-----------------------------------
The quote was slightly different before it was edited. I am sorry I missed the original version, because it described the real situation much better.

In my post, I am basically saying the same as the above post. The difference is, that I explain in some detail why All Test cannot achieve what it claims.
If All Test is able to measure capacitance to ground, they should look at those numbers and judge for themselves if it is possible that the capacitance change will affect the readings of the impedance in such a way, that can be detected. Before you start protesting what I am now talking about, take into consideration the following: If you want to asses the condition of the winding (contamination, overheating) there are no other entities that can modify the readings, but CAPACITANCE of the winding and insulation resistance (both turn to turn and to ground). The Mega ohms of resistance are totally negligible, even more negligible than capacitances. No need wasting time on resistances.
That is why I went to great lengths to include the capacitance to ground and even capacitance turn to turn (as low as 47 pF) in my calculations. From that, there is one inescapable conclusion: “"The impedance of the winding, as measured by All Test, cannot give you ANY information about the condition of the winding, unless there is a hard short". Also all the talk about the phase angle and doubling the frequency does not mean a thing. No “mystery” is involved. The method cannot work at 60 Hz or 600 Hz. The capacitances are simply too small to be noticeable at such a low frequencies. If you have impedance and inductance (plus frequency), anybody can calculate the phase angle.
I have to admit that I have never had an urge to shut down a 4000 Volt motor for a routine checkup (or any well running motor in fact) to test it with 10 volts for shorts. It is simply preposterous. If there were a problem, it would show itself as a hole in the winding. Just see how the low voltage testing “helped” to Mogli.
And it is essential to point out that it is not important what YOU say that All Test is good for, what counts is what THEY claim it can do. I feel that it is not fair to those who were let to believe that they are actually testing the winding. It seems to me, that you are the only person who can get some mileage from All Test. Some others just promised success stories and ended up with deleting their own posts.
---------------------
Quote Aditya:
I am yet to see anything in your argument against those points.
---------------------------
I do not believe that anybody on this board wrote more about rotor bar testing than I did. I would suggest, perform a search for “rotor bar, jank” or something like that. The fact is, that the low voltage tests for rotor bar problem did not impress me for a good reason. The main reason is that the majority of the rotors are with closed slots. I would say at least 95%. The effect of the closed rotor slots makes the low voltage testing virtually useless. I have been trying to point out the difference between the rotor with open and closed slots for a long time. The low voltage tester simply does not have a chance TO SEE the rotor bars, broken or not. Anybody who has ever done low voltage test on a rotor with closed slots can look back and see that the impedance measured by low voltage tester is simply nonsensical (too high). I do not want to post the link to PdMA 3500 hp motor with closed slots and 22 broken bars again. I have done it at least twice before.

If you manage to get the junk motor, I hope it will be one with closed rotor slots. That is when the testing becomes difficult. On the other hand, the result of the test on a single form coil is quite predictable. However, I would certainly encourage everybody to perform experiments. A well-performed experiment is certainly much more convincing than any 100 posts. We have seen it before, and it would be good to see it again.

Hope this is taken as it was intended: Just another contribution to a good discussion.

jank

Jank,

I think we are both saying the same thing, different words.

I have stated time & again that in “my opinion”; All-Test (or PdMA or any MCA for that matter) cannot assess ground insulation, including contamination or overheating. If All-Test claims it can, that is their affair. Everyone is free to decide what is important for themselves, that does not stop me from expressing my view.

Coming to inter-turn insulation: All-Test can pick up a hard short, as can surge comparison testing. I am yet to see a confirmed “weak” inter-turn insulation case detected by any method to date. I have heard stories that you get fuzzy peaks in the surge waveform if the inter-turn is weak, never seen it.

A high voltage motor will fail instantly in case of a hard short, but not so for DC armatures & synchronous motors’ rotors. This is where “I” would use All-Test. Also useful when you have to trace out the failed coil, this is not always visible. If someone uses All-Test as a condition monitoring tool for high voltage motor insulation, then they are on the wrong track.

Regarding the “mystery” about phase angle: The resistance, inductance, capacitance & impedance values as measured by All-Test match those taken by other instruments. I also have a NIST traceable calibration certificate for my AT-IV unit. I assume you are not doubting this point (else, please advise. We can verify this point very easily). If you agree with me, then those numbers still don’t work out with the phase angle value. Mystery!!

Broken rotor bar detection: Yes, I have often seen that small die-cast aluminium rotors do not give a proper impedance pattern. However, I don’t agree that 95 % of rotors have closed slots. Almost all of the motors that come for repairs (typically 200 kW & above) have open slots. MCA works just fine for such motors. If someone were to run a condition monitoring job on motors, they would anyways first target such larger size motors; not the FHP ones.

The problem with most electrical “analyzers” is that they are over-marketed. No one sells a vibration analyzer with a guaranteed spec for unbalance or bad bearing detection. PdMA, All-Test, etc. should sell their instruments based on measurement capabilities, not on analytical grounds. Most of the controversy arises due to this.

I’ll test those coils on Wednesday & post the results.

Regards,

Aditya

Surge insulation can pick up insulation that is initially weak, but not initially shorted. It does this by applying a steep front waveform that gives a much higher turn to turn voltage than in normal operation on the critical first few turns. Similar to a hi-pot of a winding or hydrostatic test of a pressure vessel, you can find flaws more effectively by increasing the stress during the test. And yes, it is a potentially-destructive test.

EPete,

Did you actually pick up 'weak insulation' motors with the surge test & were able to verify the 'weakness' in any way? I have been using the surge tester for almost 15 years & have never found one.

One of the reasons for this could be that IEEE 522-2004 specifies the applied voltage should be 3.5 times the rated voltage, with a rise time of 0.1 microseconds (or 5 times with a rise time of 1.2 microseconds). Clients get shivers with the idea of applying that level of voltage & will allow 1.5 times the rated voltage at max.

Secondly, the stress is mostly on the line coils as you've pointed out. Weak insulation of the other coils is probably not getting stressed. So, the test is not as reliable as when we are testing individual coils.

I would really appreciate it if you could attach some surge waveforms of weak insulation cases for my reference.

Regards,

Aditya

We indeed do agree on most of the technical part. Since you mentioned the NIST, I am kind of ignorant what it really means. I do not suppose that the NIST guarantees that the unit measures for example capacitance or impedance correctly. Since there is undoubtedly a mystery something must be measured incorrectly. Looking for example on the impedance measurement of the winding (the 5.1 MW motor), the 5.17 Ohms per phase at 50 Hz seems to be way too high. (I assume it is 50 Hz motor). I would not be surprised if the impedance were 1 Ohm. If I followed the report correctly this was a reading with the rotor out and the neutral point of the winding open. When I let the 461 Amps (nameplate current) flow through the 5.17 Ohms it results in voltage drop of 2383 Volts. The voltage line to neutral is 6600/1.73=3815 Volts, you know what I mean? Somehow the doubt starts right there. I simply do not think that the unit should be able to measure impedance on 5.1 MW motor at 50 Hz, never mind at 25 Hz.
Similarly as Pete, I have also seen week insulation. It was a 4000 Volt motor with a turn to turn short. But it was not exactly a hard , copper to copper short. The bare copper of two turns was separated by thickness of the turn to turn insulation. The current however could cross the tiny gap over the surface, not through the insulation. The surge test consistently indicated a perfect short. However the motor ran perfectly at low voltage (less than 600 Volts). I will try to recreate the situation sometimes in the future.
jank

Tested two identical coils of a 700 kW, 3.3 kV, 990 RPM slip-ring induction motor on Wednesday. I had to run out midway as a client’s generator developed a rotor earth fault. Will be running additional (& more systematic) tests again next week, but the initial readings were an eye-opener:

All-Test Parameter: Coil 1, Coil 2
R (ohms):0.047, 0.043 (AT-IV accurate range is 1 Ω plus)
Z (ohms):1, 1
L (mH):0, 0
Phase angle:23, 23
I/F:-39, -39
R (ohms, by μΩ meter):0.190, 0.192

Surge comparsion test was performed on these coils, congruent waveforms recorded.

Next, stripped ground insulation of Coil 1. Total no. of turns are 13, turn insulation was enamel only. We scrapped off the enamel layer between the top two turns. Note that the turns were not yet in physical contact. Readings were as under:

All-Test Parameter: Coil 1
R (ohms):0.047
Z (ohms):1
L (mH):0
Phase angle:20
I/F:-36
R (ohms, by μΩ meter):0.188
(note the significant change in phase angle & I/F)

Surge test was again performed, the waveforms remained congruent.

Finally, we inserted a hacksaw blade between the two turns to short them physically. The surge test detected the short immediately. However, as soon as the blade was removed, the waveforms were congruent again. This was even though the surge voltage crossed 8 kV.

My interpretation of this – The AT-IV was actually more sensitive to the “capacitive short or weak interturn insulation” than the surge test.

The attached video shows the surge patterns initially with the hacksaw blade between the turns & afterwards with the blade removed.

I will repeat these tests next week & add the following:
1. measurement of inductance with another instrument
2. measurement of impedance by applying AC voltage at 50 Hz
3. measurement of all parameters by AT-IV with the hacksaw blade in position
4. recording the video of the surge test before “weakening the turn insulation”

As Jank says, a well-performed experiment will be much more convincing than any 100 posts!

Regards,

Aditya

P.S. I took that video with my cell phone, so it will be grainy on a large screen.

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

Surge_Test.mp4 (3,026 Kb, 19 downloads)

Forgive me my skepticism. I am looking at the results and I can see that the unit All Test has no problem with Z=1 Ohm, L=0 and R=0.047 Ohm. We have talked about Pythagoras, and I bet he would not agree. Also the ratio of the reactance and the resistance w*L/R=w*0/0.047 =0= tg (phi)= tg 20 degrees ??????, the phase angle. tg of 20 degrees is 0.36, not zero as your unit says.
Why do you think the impedance changed by 10% just by striping the ground insulation? The phase angle and the I/f changed similarly. The resistance is wrong by the factor of 4. Shall I take it for granted that the phase angle is correct?
If you stripped the insulation from between two turns you have decreased the capacitance, because you have replaced the permitivity of the enamel (e= 4 to 5) by permitivity 1 of the air. The capacitance decreased- the capacitive current must have decreased- this current was decreasing the inductive current- if the inductive current increased (because the capacitive current decreased) than the phase angle must have increased (draw a vector diagram).Your results say the opposite.
I would say that both the All Test and the surge tester worked outside the operation limits. Too much loading.
I wish you a good luck with the second phase of the test. If I can suggest something, maybe series the experimental coil with a good motor winding. I bet there will be a lively discussion about the test results. I will check it when I come back in about 3 weeks.
jank

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

I think a no. of points are being overlooked rather conveniently:

1. Those were actual good coils taken out from the motor as it was being rewound, not experimental. So, are we to now understand that testing the form-coils of a 700 kW, 3.3 kV motor is outside the range of a 10 kV, 660 A, 5 joule Surge Tester? I wonder what Baker thinks of that!

2. The AT-IV readings changed with the weak insulation, the surge test did not.

3. The case of the 5.1 MW sync motor’s rotor coil defect is documented & accepted by both the client & Jeumont. So, we have 4 possibilities:
• The report is fake & I’m lying
• There was no fault but Jeumont & Reliance believed me at face value & rejected those coils without any confirmatory check
• The instrument worked magically at that time but is always wrong otherwise
• The instrument works well for inter-turn shorts
As said before, we are free to make our choice.

4. Why didn’t the surge test detect the weak insulation for the 4500 kW & 2250 kW motors (Page 2)?

5. As far as broken bars in open slots are concerned, the AT-31 has found them time & again. We find them using ESA & re-confirm with the AT-31 before opening the motor. Will post pictures of the graphs & the rotor next time I catch one. Old ones may not be believed.


In the experiment, the impedance did not change by 10 %, in fact did not change at all. In reality, there may have been some change, but nominal. BTW; if turns are shorted, the inductance (& thus the phase angle) will decrease, not increase.

The ‘R’ value is below the accuracy range of the AT-IV, also it will not display an inductance value below 1 mH. So, I will not use these numbers to calculate the phase angle. Does AT-IV derive phase angle from these numbers or by some other method? – I don’t know, ask them.

I will add the good winding to the coils next time. Would you like this to be the same stator or will any motor do? Please advise, we will conduct it accordingly.

Regards,

Aditya

P.S. I am really enjoying this particular thread. It is forcing me to think about issues that were ignored earler. I hope we all get something positive out of it eventually.

My post 26 May 2008 06:58 PM was stating that surge test can find weak insulation in the first few coils by increasing the turn voltage to the point that it begins to fail. It is the same principle as a hi-pot, except applied to the turn insulation. I don't think anyone can disagree with that, except to question at what level test voltage corresponds to insulation that is really "weak". I did observe a case where after a failure of one of our 4kv motors and evident turn short, surge test confirmed weak turn insulation on a non-failed portions of the winding. This was a form-wound machine that had no dedicated turn insulation tape, just relied instead on the magnet wire enamel (not even glass over enamel) for the turn insulation function - a weak turn insulation system design IMO. I have my written notes from that inspection including winding resistance readings. And photo's of the coils as we pulled apart the point of failure and the other coils. I don't think I have the acutal traces, but they looked like any other surge test failure. Since the machine was segregated (connections cut) and not symmetric to allow comparison between phases, we slowly increased the test voltage and observed changes in waveform occuring somewhere below 10kv indicating relatively weak insulation. On at least two 13.2kv motors during failure analysis, we have tested healthy non-failed portions of the winding all the way up to 40kv and seen no change. I can dig up the details (photos and test notes) of that 4kv machine if they are of interest.

I appreciate you providing the results of that shorted turn experiment. The results are very interesting. I agree that at first look, it certainly seems to show weak turn insulation evident on the BJM and not the surge tester. I would be curious to know - what test voltage was used for the surge test? Do you happen to have a photo of the scraped insulation.


I can tell you there has never been presented here a coherent theoretical explanation of the principle by which that type of detection can occur. In an effort to explore possible explanations:

http://www.ameu.co.za/events/convention2006/speakerspap...endran%20Moodley.pdf

Figure 3a and 3b on page 5/9 at the above link shows "Frequency Response Analysis" (impedance vs frequency) of transformer with shorted turns. In figure 3a, the phase with shorted turns is significantly different throughout the frequency range, including low frequencies down to 20hz (close to dc) and power frequency. But the spacing between the shorted and unshorted phases does seem to grow as you get out into the region where the resonances are 1000hz and beyond.... apparently because the resonant frequency of the shorted-trun phase is altered slightly. Perhaps even if the phase magnitudes were very close together at low frequencies, this shifting resonance effect might help to show a difference in phases at higher frequencies? Just thinking out loud. But of course some cautions about this theory: 1 - Transformer of course has some big differences from motor 2 - this was an obvious hard-short, since the phase difference was clearly evident even at low frequencies. Would be more relevant if we had results that did not appear as difference at low frequencies (dc and power frequency) but showed up at higher frequencies.

Your point 3 – Agreed - the sync motor rotor was evident in resistance, pole drop, and BJM tester. But the changes were larger magnitude with BJM tester and seem to provide a good confirmation.

Your point 4 - None of the instruments identified any problem on those other 2 machines and the nature of the failure is not clear.

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

Item 1.
The coils were on the table not inside machine, L is way down. Loading may be excessive. That what I am talking about.

Item 3.
I am talking about stator coils, 5.17 Ohms @ 50 Hz. This impedance does not seem to be possible. No need talking about lying.

Item 5.
I am talking about closed slots, always.

Now I really have to go.

jank

EPete,

I would appreciate it if you could share your 4 KV motor data.

The surge test voltage in the experiment was a little over 8 KV; the motor itself was rated for 3.3 KV. The experiment was done in a rush, I did not take photographs. I will be repeating the exercise again, will be more systematic and take photos & videos this time. At the moment, both the surge testers have gone for site work, so this will be done after 8-10 days.

I think a short experiment like this cannot be the basis to state that All-Test can detect weak insulation, that was not my point. The point rather was that the surge test pattern did not change in spite of applying nearly 3 times the rated voltage. Let us see what happens when we repeat the exercise the way Jank suggested.

Jank,

I did not understand your point initially, just went through IEEE 522 & saw what you mean about the coils being inside the machine. It does seem like a painful exercise though. We make coils all the time & have always tested them on a table. All coils are tested with surge test, hi-pot & tan delta prior to insertion. The surge test is done after the turn press as well as during the final acceptance testing in front of the client.

Do you normally test coils only after inserting them in the stator? I spoke with our testing guys; the only times they have come across shorted turns is when coils are being re-insulated & a fibreglass sleeve is used as turn insulation.

Secondly, it has never occurred to us to take impedance & inductance readings on coils; nor did any client ask for it. Is this part of routine testing over there? In fact, I don’t have a separate inductance meter, will have to get one.

I still don’t get your issue with the impedance readings at 50 Hz for the 5.1 MW motor. If you see the on-line data on this motor, we get an impedance of 24 Ω phase-to-phase. This would anyways change with the motor load. Where is the 1 Ω coming from?

I sincerely apologize for the “lying” comment. It came out in a moment of irritation & was uncalled for.

Regards,

Aditya

11-KBX-203.doc (119 Kb, 14 downloads)