I am analizing several motors Siemens 1PH7 from a metal forming machine. The motors are 4-pole, asynchronous servomotors with a squirrel cage rotor equipped with an encoder system for sensing the motor speed and indirect position. The stator is also the motor enclosure, no housing.

When analizing most of the motors have showed a severe rotor damage and most of the time assymetric sidebands, but after repeating the test several times, I realized that the same motor showed me either “severe damage of the rotor” or “excelent rotor”.

Are the pole pass frecuency considerations applicable the same way for these motors? Could these motor have a permanent magnet excited rotor? I will really apreciate your comments.

Brenda

Which MCSA instrument are you using?

Howard

Dr Penrose,

I am using an ATP OL II and its software.

Regards,

Brenda

If it is truly an ac async motor, then it is likely an induction motor. A permanent magnet motor would be eitehr ac syncronous (magnets on rotor) or dc (magnets on stator)

I'm not sure whether tradiditional current signature analysis for rotor bars rules apply to small motors such as this. We certainly don't monitor any of our motors tha small. Rotor failures are pretty rare on small motors as far as I know.

If you are seeing varying results, a few things to consider would be:
1 - is the load consistent among tests (and is it above 75%)?
2 - is the load constant during the test or are there transients and oscillations (which can cause sidebands).

Actually, it is an induction motor, but they are not small (40 to 60 kW). Load has been pretty much the same when testing (~60%). I am attaching results provided by the software for one of the motors. Thanks for your comments.

Current.doc (706 Kb, 54 downloads)

I mistakenly assumed 1PH meant 1 phase which would be a smaller motor... I see it is actually three-phase.

It looks like you have an electronic supply which gives a very noisy voltage (high THD). The current TWF is very noisy as a result.

First question is whether those 1hz sidebands on each side are pole-pass sidebands? Do you have a way to check actual machine speed? By tach? Vibration data? Also you could try to guess it from the spectrum. You can expect some small sidebands spaced about 1x on each side of supply frequency. Also if you have data up near RBPF you can calculate it from there.

If they're not pole pass, then it's not a rotor bar problem. If they are pole pass, then you've still got some more thinking to do. I am not sure how much to trust data when powered from a supply like this (the spectrum looks different than what I'm used to... very raised around 60hz). A single phase test might be an option to try if these are in fact pole pass sidebands.

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

Brenda,

Since you have an encoder as a feedback in a closed loop, current fluctuation occur rather due to this fact as the system is trying to maintain (I'd guess) constant speed as oppose to a broken rotor bar scenario. This is also evident from a mound around LF as oppose to more clear PPF SB set of peaks normally present in cases of broken bars. So, I agree with Pete that there is a good possibility that it is not a bar problem.

Also try to capture 10-15 current cycles to assess visually el. current TWF shape.

David

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

Brenda,
The current spectrum does not say too much. What is more interesting is the RMS current vs. time. There are 18 peaks in 48 seconds; it means there are 2.66 peaks per second. Your motor runs at ~ 64.5 Hz, so the synchronous speed would be roughly 1934 RPM or 32.25 revolutions per second (the actual speed would be little lower). So the peak in the current happens every ~ 12 revolution (32.25/2.66). This has nothing to do with rotor bars.
This is a good example that the programs are not that smart. Your program looked close to the main peak and found sidebands about 22 dB down. It mistakenly thought that those are the rotor bar sidebands. They are not.
Is there a gearbox somewhere down the line?
jank

Brenda

I am assuming that the load or speed is varying during the time of test. The ATPOL software is actually the Areva EMPATH software (I assume that you are using version 4.4?).

What is happening is a common issue. If you look at the line frequency on your graph, you will notice the wide base around it. This is the result of changing torque or speed on the output of the unit. The software acts pretty much the same as any of the programs, including the PdMA software and looks for peaks that exceed a threshold at the pole pass frequency on either side of line frequency. With the wide base, those thresholds are exceeded.

This is to be expected when testing servo and other variable speed/variable torque equipment. If it were actually a broken rotor bar, the peaks would be definitive ppf sideband peaks.

When I teach EMD (MCA/ESA/MCSA)diagnostics, I always recommend that you double-check the findings of the software (regardless of whether it is ATPro, EMPATH, EMax or Explorer). This is especially true with servo and spindle machines. The ATPOL and EMPATH systems are both able to test machines through fractional.

If you like, I can walk through this analysis on this string. In the folder where you saved the data there will be a file with the extension .ejr (was the original .mtr file that was converted after you opened it with the ATPOL software). Email it to howard@motordoc.net along with the nameplate information and I will analyze it tomorrow (Thursday evening) and post the analysis on this string with explanation of what you are looking at.

Sincerely,
Howard

Brenda

I have run through a quick overview of the screens from the data that you sent. I am attaching the initial analysis (pardon any grammar errors, I was at a speaking engagement all day). I have left out the buttons and most of the features of the EMPATH/ATPOL software so that others using PdMA, Baker, etc. can follow along.

What is the servo driving?

I am going to assume that there will be questions/discussion from this point.

Howard

analysis.pdf (974 Kb, 42 downloads)

Brenda

I encountered a similar challange with the same type of motor typically used in the "wire drawing" industry.
In addition to the ESA/MCSA tests, I perfomed an off-line MCA Inductance test (result see attachment).
I did the rotor test on a brand new motor and a motor in line with ESA/MCSA results similar to yours.

May be someone will comment on the outcome ?

SIEMENS1PH7_Rotor_Test1.pdf (203 Kb, 31 downloads) MCA Rotor Test

Mogli

I have marked up and attached the waveforms that you attached.

I developed the Rotor Grading System (RGS) for the first EMCAT software while I was the General Manager for ATPro and I was the original programmer. This is the EMCAT 2005, which was developed after I left ATPro in 2004, so I am not sure that the code is set up exactly the same as the original EMCAT. The first sine wave should have had a value between 5 and 10 and identified a caution for the following reasons (the numbers below explain the numbers on both attached images):

1. The RGS uses the average of the points at each rotor position. The series of averages should equate to a straight line. As it deviates from a straight line, a percentage is added across all of the test point. The numbers should add up to <5 for a good rotor, >5 and <15 for caution, and >15 for a serious issue.

2 & 3: The difference between the peaks at points 2 and 3 will cause some small variation. However, more importantly is the flat spot at point 4. This relates to casting voids in the cast aluminum rotor found in this type of machine. These casting voids, when the flat spots occur on the sides of the waveforms, such as this, they will cause some level of mechanical unbalance (taken care of through balancing after manufacturing) but will not impact the ability of the motor (or servo/spindle in this case) to produce torque.

On the second page, the findings are dramatically different. I assume that these are not identical motors due to the dramatic differences between the inductance.

First, there is a phase unbalance noted by the shift of each phase from each other. This causes the average of the points to vary significantly. At Point 5, this type of signature relates to broken rotor bars or casting voids that leave a bar open (same thing, different causes). The value on this one is a little high, but still correct as it identifies a serious issue.

Howard

SIEMENS1PH7_Rotor_Test1_revised.pdf (78 Kb, 26 downloads)

Dr. Penrose,

thanks for your comment. The rotor tests were taken on two identical motors. However, I don't know the history of the motor in use.

I am more concerned that some 70% out of 140 motors measured show the pattern of ESA/MCSA as Brenda mentioned in the original posting.

One wire drawing machine counts with a different drive allowing a clear low frequency spectra with distinct peaks plus and minus the PPF for the two out of 9 rotors which failed.

Based on my random MCA rotor test should I conduct a rotor test on all of them ?

The AT31 rotor test is not feasable as the installed motor does not turn freely (see arrangment in the attached pictur).

MOGLI

I have looked in little more detail on the Current Analysis as performed by the ATP OL II (or whatever it is called).
The main current peak is 64.5 Hz. From that we can calculate the synchronous speed of the motor: Ns= (120/2p) * f= 30 * 64.5 = 1935 rpm. From the page 2 of the data provided by Brenda’s doc file, the running speed is 1934 rpm (???) The slip seems to be quite low despite the fact that the motor is loaded to 20.754 kW or 60.7 % of the full load. Despite the speed drop of only 1 rpm they found out that the pole pass frequency is 0.366 Hz. From the speed, the pole pass frequency should be: (1935-1934)/1935 * 4 (poles)*64.5(Hz)=0.129Hz(???).
However the nameplate says something different: Speeds are given for 3 different frequencies. The closest to 1934 RPM is the 2000 rpm @ 67.4 Hz.
The synchronous speed @ 67.4 Hz is 30*67.4=2022 rpm. Hence the slip at full load is (2022-2000)/2022=0.0109. Since the motor is loaded to only 60.7%, the actual slip will be 0.0109*0.607= 0.00662. This slip would give us the frequency of the lower sideband as:
Ls=f*(1-2s)= 64.5 * (1-2*0.00662)= 63.64 Hz. If somebody wants the “pole pass frequency”, it would be 64.5-63.64= 0.86 Hz. If you look back at the current spectrum and go 0.86 Hz down from the main peak (Line Frequency), there is no chance to find anything as –24.3 dB. You can find something between 35 and 40 dB.
The point I want to make is that the assessment of the condition of the rotor was: “Multiple broken bars and end rings very likely. Severe problems throughout. Overhaul or replace ASAP. ”
It seems to me that the ANALYSIS has severe problems throughout. I personally would throw the analysis out as non-conclusive. At the same time I am convinced that there is absolutely no problem with the motor. The analysis is screwed up by the fact that there is process variation of the current. I cannot imagine a steel forming process without a gearbox. The rollers or whatever of the forming machine rumble over a manufactured product every 2.7 sec or roughly 0.366Hz (see the RMS current). It causes a current variation that is much slower than the rotor bar fault would be. Since it is so slow it widens the base of the main peak and the analysis becomes pretty foggy as a result.
I really can’t wait till Monday. I will probably buy this system and also a low voltage tester. As a result I will never experience a dull day anymore!
jank

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

Mogli:

If you have any of the ESA data on the failed motors, please feel free to send me the .ejr files for those motors and I will take a look at them. howard@motordoc.net.

On another note, the pole pass frequency is either twice the slip frequency or the number of poles times the slip (both provide the same result). It is not the slip frequency times the number of poles. The synchronous RPM is 120 times the frequency divided by the number of poles, not twice the number of poles. Also, as this is the equivalent of a variable frequency drive, you cannot calculate the RPM based upon the %load.

Howard

quote:

Originally posted by MotorDoc:
On another note, the pole pass frequency is either twice the slip frequency or the number of poles times the slip (both provide the same result). It is not the slip frequency times the number of poles. The synchronous RPM is 120 times the frequency divided by the number of poles, not twice the number of poles. Also, as this is the equivalent of a variable frequency drive, you cannot calculate the RPM based upon the %load.

Howard

I have to admit, I have a great deal of problems with “pole pass frequency”. I have never used it before and I cannot get used to it. I think it is like baseball: You have to grow with it to appreciate it. Which pole am I supposed to pass??? There are no fixed poles on the induction machine! So every time I try to use it I screw up. In no way it diminishes what I wanted to say; it means that the analysis picked up a totally wrong speed. It simply cannot be 1934 RPM with 60% load.
“The synchronous speed…”: In the agreement with the literature the symbol “p” is for the pole-pairs. The number of poles would be “P”. As an expert you should have figured it out yourself, because the calculation gives the correct result.
“You cannot calculate the RPM based upon the %load”.
Oh, really? The motor cannot care less if the power is coming directly from the power line or if it is modified in the variable speed drive. It still behaves as an asynchronous motor. There are no separate formulas (or shall I say formulae) for calculating the slip in the motor fed directly from the line or fed from the VFD. Just look up the nameplate data!
If the program took the load into consideration, it would never came up with such absurd numbers.
jank