Sign up Calendar Latest Topics Donate
 
 
 


Reply
  Author   Comment  
Noknroll

Avatar / Picture

Sr. Member / Supporter
Registered:
Posts: 843
Reply with quote  #1 
Any idea why Velocity is showing peaks at 63Hz/6.6X while pkv has (sort of harmonic humps) peaks at approx 53Hz/5.5X
Measurements were grouped so pretty much taken at the exact same time.
pkv V vel.PNG 

Vibe-Rater

Avatar / Picture

Sr. Member / Moderator / Supporter
Registered:
Posts: 1,397
Reply with quote  #2 
The 6.58x looks like outer race from a 16 rolling element 223** SRB And the 5.546x rolling element marking for the same bearing. You might find 5.546x = 2 BSF and cage rotational sidebands in PkV. Do you have the bearing number?? rgds  And LOL you are still persisting with Hz!!
Noknroll

Avatar / Picture

Sr. Member / Supporter
Registered:
Posts: 843
Reply with quote  #3 
I only added orders for you mate.
I don't have any bearing details....that'd be too easy [smile]
it is belt driven, the low freq peaks in pkv are 0.34X and have asked client to check the belts
Was hoping there was a high tech explanation why both measurements weren't focusing on the same frequency(s)
Vibe-Rater

Avatar / Picture

Sr. Member / Moderator / Supporter
Registered:
Posts: 1,397
Reply with quote  #4 
Hi Nok, thanks mate. I think everything you see is "real" Peakvue is picking up 2xbsf since it looks beyond your HP filter whilst the normal data is picking up BPFO for a 16 rolling element 223**. And 0.34x would be ftf. Whether that is a problem or not, history will tell. In my opinion I think people jump at FTF in Pkv too quickly. It really isn't a cage fault. It is merely the rolling elements travelling about the bearing which they do in time with the cage.  So BPFO levels are low in my books but have a look at the history.  I would be interested to see if my guess at 223** dimension series is correct. Almost all bearing dimension series have quite a tight set of fault frequencies so once you think in orders it all falls into place. You can pick the bearing type easy. Ask Sinsk.  Not very high tech explanation but hopefully it makes sense.
Danny Harvey

Sr. Member / Moderator / Supporter
Registered:
Posts: 1,402
Reply with quote  #5 
Nok,

You might see something more in the PeakVue by setting the cursor on what looks like a bunch of little harmonics of running speed and hitting the d key to remove those peaks.

You might try the same on the velocity spectrum with the biggest harmonics.

I would also auto-correlate the PeakVue TWF as well as looking at it before auto-correlation.



Ralph Stewart

Avatar / Picture

Sr. Member
Registered:
Posts: 235
Reply with quote  #6 
Noknroll,

This is not a very "high tech" answer, but it might be something to consider.[smile]

There appears to be a large resolution difference in the two plots.

This lack of resolution might make one (especially the velocity one) or maybe both, not to really have the order numbers showing on the current 2 spectrums, 5.546x and 6.581x, but maybe, in reality, be more closer to the real sum of the 2 fundamental frequency orders, to equal 12 for a 12 roller or ball bearing. Even may be a remote possibility equal a sum of 13 or 11 with better resolution.

An equal resolution and fmax might show 1 spectrum to be the inner race at maybe 6.4xx orders and the other to be showing the outer race of 5.5xx orders.

The Peakvue data appears to be only showing one of the two races (possibly the outer race, 5.546x) with signs of an early stage defect, while using the 1k HP Filter in the Peakvue Data, that is, if the Peakvue cursors are actually on the correct defect frequency position. Both frequencies could possibly be there in the Peakvue data with closer analysis, IMO.

I would take an equal resolution and equal fmax spectrum of both Peakvue and conventional data to see what they show, as far as the order factors go, in each new equally designed plot.

Just my opinion and I could be totally wrong.

Thanks and Have a Great Day,
Ralph

__________________
Ralph Stewart
AlertAnalytical.com
Sinski

Avatar / Picture

Sr. Member
Registered:
Posts: 394
Reply with quote  #7 
Nok

Pretty much as Vibe says although 0.34x is a bit low for ftf on a 223** series bearing and BSF on most are around 2.6x to 2.7x (BSF*2 = 5.2x to 5.4x) so that is out a little as well. Would really need to find out what the bearing is to confirm but they do characteristically look like that. The different frequencies in each would simply be to do with the HP filter in peakvue looking above 1kHz and the velocity spectrum looking below. I see this a fair bit on paper machine rolls and usually it can indicate that there is some slight corrosion marking on the outer race and a roller or 2. Still quite low levels though so could be left and monitored at this stage.
Noknroll

Avatar / Picture

Sr. Member / Supporter
Registered:
Posts: 843
Reply with quote  #8 
Thanks for the replies fellas
Yes, I agree 0.34X is too low for ftf but not out of the realms of possibility for belt frequency.
I tried the D key but nothing really stood out.
Ralph I just zoomed in on the velocity spectrum to give it the same 400Hz Fmax as pkv to sort of compare apples with apples, that's why the resolution looks low.
wasn't really concerned with such low levels was more curious than anything.
From what you blokes are saying it appears it just the difference between velocity and peakvue with its filters. I suppose if it were a more advanced/pronounced fault the 2 (vel & pkv)  would "lock" onto the same frequency?
OLi

Sr. Member
Registered:
Posts: 1,915
Reply with quote  #9 
It maybe would be a result of the PV processing? If you take the peak in each frame of data as you do in PV and they are by chance all to the "left" or another extreme and then label it to the in theory center freq. for the frame it may be shifted by this offset? That is the "peak" part in PV as I understand it. Normally you don't compare so much maybe and the precision in where brg. faults end up IRL are not that good so it doesn't matter and normally it is maybe not the same offset from frame to frame so it may in average compensate but this may be a special case like this. It may be better to compare a wide range accel FFT to the PV so the comparison would be in a similar amplitude scale, not that it would change the actual freq. just size of what you see. Only my 1 SEK and may be wrong. 
__________________
Good Vibrations since early 1950's, first patented vibrometer 1956 in the US.
http://www.vtab.se
Previous Topic | Next Topic
Print
Reply

Quick Navigation:

Easily create a Forum Website with Website Toolbox.