Hello again,

I would like to know if when analyzing data, if it is common to adjust your 1X rpm up or down to see if bearing freq will line up? In the attached file, this 100 hp dc motor is making some noise and vibrating more than normal. There is a cooling unit bolted to the top of this motor, and I believe most the excessive vibration is coming from it. But back to the original question. When I captured data, rpm of the motor at that time come out to 1693 rpm. I was unable to verify with a strobe or tach at the time, but pane view reading were close to 1690.

Now at first, the bearing fault freq do not exactly line up to the larger peaks. If I go and change the 1X rpm value in my software, I can get the fault freq to line up with the peaks. By increasing the 1X rpm value in the software to 1722, the faults start to line up.

This motor has belts on it driving a gearbox. Is adjusting rpm up and down a bit a good thing or bad? Any/all advice/tips would be great.

X

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

Speeds.pdf (46 Kb, 76 downloads)

1693 RPM w/6214 brg = 14614 cpm @8.6324 BPFO.

Set your RPM to the machine's RPM. The bearing isn't harmonically related.

Xracer,
Instead of changing the rpm, you can do the same thing by nudging fault frequencies. Place the cursor on top of a fault frequency, hold down the left mouse button and drag left or right. The percentage of change will be displayed. I occasionally have to do this.
Is this needed because the geometry of the bearing changes, due to worn or damaged parts, etc.?

quote:

Originally posted by Sam Pickens:
1693 RPM w/6214 brg = 14614 cpm @8.6324 BPFO.

Set your RPM to the machine's RPM. The bearing isn't harmonically related.

Hmmm when I look at the bearing info that is loaded with my software, I show the following

When using the nudge option, what would be the limit % in either direction? Below is before nudge

Here is after nudge of + 1.64%

Xracer,
Limit %? Well, probably the further the nudge, the more doubt it is the bearing frequency. Look very carefully at what else could cause these harmonics. I hope we get more help with this question.

Hi Xracer,

When software displays bearing numbers like this it leads the user to think (IMO) that that is the bearing fitted without doubt. This is often not the case. You may not have a bearing number at all, the wrong bearing number might be entered, a version of the bearing with a different number of rolling elements might be fitted, a different manufacturer bearing might be fitted (you can’t control purchasing and their drive to buy from cheaper sources) or in the case of motor drive end, an NU*** series bearing might be fitted instead of the 63** series shown on the nameplate if the motor has been overhauled or vice versa.

To talk about percentage nudging limits, or even % nudging full stop is very dangerous in my opinion. For the software provider even to attribute a measured value to this leads people to think it is important, there is enough to worry about without introducing useless metrics which are probably sold as “features” instead it just leads people astray. What will they do next, introduce alarms on maximum nudge value deviations. Boy what a waste of time.

The idea that, if this set of non-synchronous peaks does not quite match what software tells you (and therefore conclusion that bearing is OK)will lead to disaster eventually.

You need to learn how to identity bearing faults without having fault frequencies at your finger tips and treat the bearing fault frequency as a bonus when in the end it confirms diagnosis.

The bottom line is that if you encounter what appear to be non-synchronous frequencies as displayed in your attached plots they are 99% certain bearing related (please be kind to me with this since I do not know what other machines are around and what this motor drives). (I assume the 1x is actual motor running speed and not 1x of the cooling unit)

Typical ball bearing types 62** with 10, 11, 12 & 63** series with 7,8,9 rolling elements all have BPFO & BPFI families that are very close across a broad range of dimensions. So do NU3** & NU2**, you’ll get an idea of this if you do a bearing search for a bearing and show the results as a list as you can do in CSI (not individually) I’ll try to attach an example. You simply must start to get “a feel” for the frequencies different bearing families generate.

Finally, always display your frequency axis in orders; you’ll make a life a lot easier for yourself since it will be instantly recognisable if a set of frequencies is non-synchronous (relatively difficult to do on your plots), the most important characteristic pointing to a bearing fault. The first step is of course to beyond doubt ensure that the spectral speed is exactly the same as the running speed of the machine at time of data collection.

Hope this helps. Kind regards.

p.s you'll notice that for each additional rolling element the BPFO increases by approximately a FTF and BPFI increases by (1-FTF). This holds true for the majority of bearings I have encountered and helps you when you do not have FF details for a bearing that is not listed in the database. BPFO + BPFI = #B/R of course.

This message has been edited. Last edited by: Vibe-Rater,

I was using CSI's freq program but could have made a mistake; but here it is again:
8.6324 = BPFO = 60 RPM or 1 Hz - that's the CSI listing.

But my 30 y/o book for N214 = 6.93 6214 deep groove = 4.06 MRC 214-s = 3.865

And gotta

Thanks, Vibe-Rater. If I can nudge a smidgen
to make the fault frequencies line up with the peaks, I will, to make my report more clear. Rarely does the nudging go past .03 percent.
Hey, thanks for the posting. I have not been to vibration school , just had good coaching for awhile, and have a huge appetite for learning.

I went back and changed the bottom scale to display orders and have the first pointer selected on running speed of the motor and second pointer on the suspected brg fault freq. Cooling unit motor(s) run at approx 3550 rpm. There is a peak between 4 and 5 orders. When I place my pointer on that peak, (4.63) order, its 2x-4x are also shown. Not knowing exactly what brand bearings are actually installed, it does look to me like it could be a bearing related issue. Or am I off track? I guess I'm the odd ball with a commtest setup

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

Hi Xracer,

IMO it's bearing related. In particular an 11 rolling element version. See attached.

Now the question of severity. I think it's not bad for a shaft running at 4 pole speed. Depends entirely on historical data. If it has come out of nowhere you could change it and possibly find light damage. I would probably monitor it for a week and see. Then 2 weeks, etc. give it a bit of grease to start with and monitor again.

Displayed in orders you can immediately see it is non synchronous (provided you have made sure speed is correct) @4.6x it's unlikely to be transmitted rotational speed of anything since it is close to 8000 rpm.

Sam, you lost me with the 8.63.

See ff's for 6214 below, you'll need to disregard the Dodge. Best regards.

Historical data I do not have since we are starting our database from scratch. I have yet to search the maintenance log(s) to see when the last time this motor was replaced. This I will do when I get back to the office. I do know that this motor/machine was sitting idle/off for a few months, and then when started back up, the operator asked to have it looked at. So the vibrations could have been there before. There is a slight noise coming from the drive end area of the motor. This was my first indication that something many be going on. I'll keep monitoring it to see if things change after a longer time running. As for the "actual true speed", I did not have a chance to strobe it, and not sure if enough of motor shaft is exposed to get a good reading. I would say that the displayed rpm is within 2-3%. Thanks again for everyones help.

quote:

I would like to know if when analyzing data, if it is common to adjust your 1X rpm up or down to see if bearing freq will line up?

IMO, never. Adjust the running speed only if it is not correct for the point at which the data was taken. Unless of course you are using more than one shaft speed and analyzing the same point for the possibility of a problem on one of the other shaft speeds which you might not have a point and data on.

example
Input of a gearbox turning 29.90 Hz , then changing the speed to 8.4 Hz for the output shaft in an effort to determine if the "defect" is from the output rather than the input.
But as a general rule, I would not adjust my true running speed for the sake of making the BPFO align with the calculated frequency. UYATTCSFSSRTTIADSITS.

The bearing geometry (dimensions/no of rolling elements etc) is used as the basis of calculating fault frequencies.

The calculations are based on the assumption that the relative motion of all the components results only from rolling: if there is any slipping then the calculated and actual fault frequencies will not line up, so I would say that it is worth nudging the cursor.

For identifying a problem, I would suggest that looking at the frequency differences between the peaks in the spectrum is as useful as the absolute frequencies.

Vibe Rater: That is the number that the frequency calculator in the CSI software displays. Then I gave a list of numbers that other sources give. I went back and looked at the cSI program again and re-entered the 6214 brg and got the same frequency listing.

But and regardless: I don't relate RPM to brg n.nnn in multiples of RPM as it is not harmonically related to RPM in whole numbers.

Hi Sam,

Then there must be something wrong with the information entered into the CSI calculator. 8.63 = BPFO of a 20 rolling element bearing 213** family.

See the following link, click on bearing database, a very handy fault frequency tool. There’s a bearing calculator as well.

http://www.fis-services.de/site/en/t3.Bearing_Database/Bearing_Database.html

I entered the physical dimensions for a 6214, 11 balls in the CSI calculator and it does not come up with 8.63x.

quote:

But and regardless: I don't relate RPM to brg n.nnn in multiples of RPM as it is not harmonically related to RPM in whole numbers.

I’m not exactly sure what you mean but I think that’s right and implies – fault frequencies are 99% never integer multiples of running speed.

You should look at frequency in orders since a 6214, 11 ball will always generate 4.10x (sorry made mistake should be 4.56x) regardless of whether 123456 rpm, 123456 Hz, 2 pole, 4 pole, n pole, 60 or 50 Hz LF, variable speed or whatever. It is very easy to remember BPFO 4.10 and 3.08x (63** 8 ball), 5.69x (223** #b/r 15) etc etc etc. becomes a hell of a lot easier to deal with machines for which you have no bearing info at all and not miss faults because peaks don't align with info. Imagine trying to remember the above values in cpm or Hz?? Nightmare. Regards.

This message has been edited. Last edited by: Vibe-Rater,

I do frequency analyses and never really use order analyses. Order analyses seems to make the job harder to me. I also use Hz instead of cpm and work in g's from accelerometers but will do the head math for velocity conversion. My old software would display the FFT and list the frequencies plus amplitudes in g's, IPS & mils.

I have a solid working program that allows the producing of planned, scheduled maintenance. A field proven model.

Vibe-Rater
"You should look at frequency in orders since a 6214, 11 ball will always generate 4.10x"
Do you mean that you should always see 4.1 orders? I am also more familiar with HZ.
Thanks

Hi rnl,

Yes, that's correct. Regardless of running speed a 6214 #11 b/r will always show 4.56x. There might be very slight differences between manufacturers or if physical dimensions have changed due to damage but in 99% of cases this is true. If you are looking at a spectrum and you see a 4.10x (10x0.41)then a version with 10 rolling elements is fitted. (4.56x-0.41x(FTF)=4.15x)very close to 4.1 and good enough as AS A RULE OF THUMB. This rule also holds true when predicting BPFO for multiple rollers more or less. I.e take a SKF 22312 13 #b/r. BPFO = 5.31. + (3x0.41) = 6.54x. The bearing data base shows SKF 22312ECC as having BPFO = 6.584x. Pretty close. You can see that there is 1.23x difference between the fundamentals. Enough to make some people think it must be coming from somewhere else. The rule is also true for BPFI, just add / subtract (1-FTF) for each rolling element more or less. Thinking in orders is much easier than trying to remember these rules when thinking Hz / CPM, saves a lot of time on the calculator. I'm sure the non believers will come back with arguments against (like line frequency issues should be looked at in Hz) etc. etc. but in summary frequency in orders is a far easier and superior way of working in the vast majority of cases. Best regards.

This message has been edited. Last edited by: Vibe-Rater,