Wheel bearing failure

Ask questions relating specifically to the PicoDiagnostics NVH equipment and software here.

Re: Wheel bearing failure

Postby Steve Smith » Fri Sep 22, 2017 5:15 pm

Thank you all for the continued posts and for the feedback on the Verso repair.

What is interesting here was the clear 174 Hz peak attributed to the 14 ball rollers and the single pit on the outer race.
This reinforces the challenges faced with bearing diagnosis as I estimate the T1 in your capture to be around 10 Hz:
10 Hz x 14 Rollers hitting 1 pit = 140 Hz but it’s not that predictable.

The good news is that your bearing would emit a “ping” for every event of a roller falling into the pit or “spall”.
As wheel speed increases these “ping” events increase in proportion so generating noise. Add into the mix non spherical or pitted ball rollers and it’s no surprise we arrive at a frequency above the “expected” when viewing the visible damage.

Here you were able to compare one hub against another and use “amplitude” as a true indicator of the source of the noise.

This is the exact same technique we have used with the following Subaru Front Left Wheel Bearing, however, there was no clear increase in amplitude at a specific frequency in which to determine bearing failure. Instead we have a predominately higher amplitude from the accelerometer attached to the LH Front Hub assembly.

WHEEL BEARING 4 X ACCELEROMETER.jpg
NVH Capture


Ironically we also obtained a similar result to you Martin when attaching additional accelerometers to each Shocker Tower. The RH Shocker tower indicates a higher amplitude which is the opposite side to the offending bearing!
For this reason we must be aware of this characteristic when using this technique for bearing noise detection. I do wonder if the drive train arrangement can impact on the transfer of bearing noise and vibration.

For example: Both the Verso and the Subaru have a direct lateral transfer path via their front differentials. Could the transfer of wheel bearing vibration be minimised at the opposite shocker tower where there is no link via the drivetrain? I guess we need to find a RWD vehicle with a failed front wheel bearing. (I am sure there is a link)

Unfortunately not all bearing components were recovered from our Subaru as one of the inner bearing races was not returned after replacement. With all that said, we have enough evidence to condemn the bearing and link to our obtained NVH results.

OUTER RACE .jpg
Outer Race


ROLLERS -1.jpg
Rollers


INNER RACE-1.jpg
Inner Race


We are all still learning in this new field of diagnosis and the more we can share the easier this will become, for example:

Single pit/Spall = Single offending frequency when the accelerometer is placed in key location (hub)

Lubrication failure (no pit) = No definable frequency but a general increase in amplitude across the entire spectrum

No pits or spalling to the bearing race meant no defined frequency from this style of bearing wear. Instead we have a low frequency, cyclic “rumble” which does not lend itself to clear detection within the frequency spectrum

So, in order to compile an NVH Test Plan for wheel bearing diagnosis we ideally require an accelerometer and microphone measuring both noise and vibration simultaneously
In a perfect world, multiple accelerometers are a real help for simultaneous comparison

1. Mount the NVH mic inside the cabin as this will be used as your sync signal or reference during playback. We need to capture exactly what the customer can hear on the NVH Mic.
Avoid using the NVH Mic adjacent to the offending Hub.

The audio files contained in the link below demonstrate how mounting the NVH Mic inside the cabin will provide improved recognition of bearing noise and natural shielding from ambient noise
https://drive.google.com/file/d/0Byzrif ... sp=sharing

2. Mount an accelerometer adjacent to the suspect wheel hub. Be aware of cable runs about rotating and steering components

3. Drive the vehicle at the optimum speed to obtain the loudest bearing noise, try to utilise a smooth road surface (avoid textured road surfaces)

4. Maintain road speed for a minimum of 2.5 seconds (longer if possible) for the noise to appear in the frequency display

5. At the appropriate road-speed select neutral and allow the vehicle to coast whilst maintaining road speed (engine speed returns to near idle) Here we confirm the noise to be road speed related and not linked to engine speed if the noise persists

6. Once you have confirmed a “road speed” related noise drive the vehicle on a variety of road surfaces. If the noise alters with a transition in road surfaces it is more likely to be tyre noise.

7. Repeat steps 2, 3 and 4 with the accelerometer positioned adjacent to the opposite wheel hub and compare amplitudes (as with the Subaru) and any unknown frequencies (as with the Verso)

Whilst this is not the definitive guide to wheel bearing diagnosis (I am sure collectively we can modify these steps) there are a number of pointers above we can adopt to assist with identification of offending wheel bearings regardless of their wear pattern

There is a fantastic roller bearing tutorial here during the first 13 minutes of this video from the great team at the Mobius Institute: https://youtu.be/67Et4vbKhOM

A final word re the Stethoscope approach.
Whilst it is subjective you may find with bearing wear such as the Subaru above (where the bearing is clearly audible without applying load) a stethoscope is a valid solution (There is a stethoscope recording in the link below)
https://drive.google.com/file/d/0Byzrif ... sp=sharing

Be aware: Using the stethoscope may lead you to falsely condemn components based on the additional characteristic mechanical noises you can hear given the sensitivity and proximity of the stethoscope sensing tip.
Listening to a rotating transmission gear train via a stethoscope will have you thinking there is a major fault given all the audible meshing noise when in fact this is perfectly normal and inaudible via the cabin

I hope this helps, take care……..Steve
Steve Smith
Pico Staff Member
Pico Staff Member
 
Posts: 481
Joined: Sun Aug 25, 2013 9:22 am

Re: Wheel bearing failure

Postby Steve Smith » Fri Sep 22, 2017 5:16 pm

Hi Mark sorry, I will feedback ASAP re your FFT settings rest assured.

Take care......Steve
Steve Smith
Pico Staff Member
Pico Staff Member
 
Posts: 481
Joined: Sun Aug 25, 2013 9:22 am

Re: Wheel bearing failure

Postby Mark Dalton » Sun Sep 24, 2017 8:01 am

Thanks Steve, it's greatly appreciated.
User avatar
Mark Dalton
TwoWaves
TwoWaves
 
Posts: 81
Joined: Mon Dec 01, 2014 11:35 pm
Location: Melbourne

Re: Wheel bearing failure

Postby Steve Smith » Thu Oct 19, 2017 5:20 pm

Hello and sorry for such a late reply.

With regards to your question Mark and the FFT options:
Looking at the “Response Slider” under Options>Advanced Options>FFT by default this is set to position 4 which allows for a balance between accuracy, resolution and processing time.

In the example below our FFT Response Slider is set to position 5 (More Accurate)
Notice how well defined our T1 vibration appears in the Frequency view with approx. 4 seconds of signal history data transferred into the FFT for processing

T1 MAX ACCURATE 1.jpg
More Accurate


Moving the Response Slider to More Responsive (position 1) has an adverse effect on resolution with approximately 0.5 seconds of signal history data transferred into the FFT for processing

T1 MORE RESPONSIVE.jpg
More Responsive


Whilst the response is great, when playing back the recorded data the Frequency view appears very “jittery” with an almost unrealistic formation of our offending T1. This may well be ideal for repetitive momentary noises or for pin-pointing specific events within the signal history.

For continual vibrations such as T1 and bearing noises using either “More Accurate” Position 4 (default) or 5 allows for a more “fluid” display of events in the Frequency view with excellent resolution. This is of real value if we want to isolate the commencement of vibration and track its progress from start, through to peak and finish.

Response slider position 4 (default) obtains approx. 2.5 seconds of signal history data transferred into the FFT for processing. This provides sufficient resolution in the Frequency view, with a realistic update speed, requiring only 2.5 seconds of data from the signal history.

T1 DEFAULT 1.jpg
Deafult


Coming back to your question Mark “for noises like a wheel bearing are you better with a 50 second capture or a 5 second capture and the slider all the way one end or the other?”

The settings above are all about personal preferences dependent upon what it is you wish to capture. Thinking about the complexities of noises from wheel bearing and their illusive frequencies I think it best to leave the Response Slider in the default position as this should have all avenues covered.

If we then discover an offending frequency (which we have proven is not always possible) then adjusting the Response slider in order to sacrifice accuracy for response may be helpful.

The good news here is we can experiment post capture to find out what works and what doesn’t

I hope this helps and I hope I have not answered a question with a question, take care……Steve
Steve Smith
Pico Staff Member
Pico Staff Member
 
Posts: 481
Joined: Sun Aug 25, 2013 9:22 am

Re: Wheel bearing failure

Postby Mark Dalton » Thu Nov 02, 2017 7:26 am

Hi Steve,
Thanks for that info. Every little bit of knowledge is Gold. I'll take two things away from this that I can use.
1. I've had a couple of cases where a vibration was only present under WOT from take off and the vibration only lasted 1-1.5 seconds. Moving the slider all the way to the responsive end should help those vibrations show up much clearer and not be averaged out of the FFT.
2 A GM Field Engineer I was talking to said he though short captures were better than long captures. Your info suggests to me that the capture length shouldn't matter because only the data in the shaded area is being analysed by the FFT.
So now I don't have to worry about trying to stop and start captures and concentrate on the road at the same time.
User avatar
Mark Dalton
TwoWaves
TwoWaves
 
Posts: 81
Joined: Mon Dec 01, 2014 11:35 pm
Location: Melbourne

Previous

Return to PicoDiagnostics NVH

Who is online

Users browsing this forum: No registered users and 0 guests