gear noise analysis in a production environment 1

[ColinScowen]

Hi, there's a lot of preamble to this post, but it will help to understand my problem,
we make powertools, with the usual universal electric motor with a pinion gear driving up into a gearcase where the rotational force is converted to lower speed axial and rotational forces.

We test the tools for 'noise' at the end of the production line. Our operators are using the oldest form of sound analysis equipment, their ears. But, whilst this is fine if we have a problem that can be attributed to a single part, be it, rotor, crank etc, it becomes a much more complicated issue to locate multiple causes that cannot be easily isolated. (we have several sub assemblies that cannot be tested independently, and assembly can also be an issue)

We have toyed with noise measurements in our EOL testers, but this has never really worked for a number of reasons (ambient noise, test booths having adjoining walls).Up to now, all of this has been thought through with an acoustic microphone in mind, mounted to the tester. I have all the data on the speed of the armature, the transmission ratios and gear data.

What I am curious about is if it is possible to measure the sound that is coming from the running unit with some sort of contact device, rather than a microphone set at a distance from the tool. (I had a look at the microphone download from the FAQ page, but that didn't answer my question.)

I think that vibration analysisi would also give me most of the same data, but I am not sure it would catch excessive backlash or tight mesh (please correct me if I am wrong), it is also not feasible to use magnetic mounts for accelerometers, as the housings are plastic or zinc/magnesium.

Regards,
Colin.
(electronic engineer in a mechanical world)

 

[GregLocock]

What sort of frequency are you interested in?

Accelerometers generally need a good contact with the device being measured. You might be surprised by how effective plasticene (modelling clay) is. Or devise a mechanical clamp.

Non contacting methods include lasers, and capacitance. or you could try acoustic intensity.



Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[ColinScowen]

Armature spins at 19,300 rpm, has a 6 tooth helical gear on the top, meshing with a 51 tooth gear. At the moment I am interested in the audible range, as I can easily corelate this with the EOL operators.

The laser method would seem to suit best, of the non contact methods, as I think I could mount that to one of the end testers in the bank of 4 that we have, and would seem to suffer least from ambient noise, as I could focus directly on the housing.

Do you know of any suppliers for this type of measurement equipment? I tried Google and Yahoo, and most sites were for using a laser to bug a window. Interesting, but not really relevant.

Colin

 

[GregLocock]

That's the right technology. Problems you'll have - 2 kHz is a bit high in frequency.

I don't know who makes them these days - B&K might still do one. It used to be limited to 400 Hz.

Oh, you might want to look at SPATE (thermal emissions). I doubt that it would be cost effective.

A soundproof booth might be the best cheapest alternative!



Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[SomptingGuy]

When I was in the NVH business, we had a huge stock of really expensive B&K accelerometers (I think they still do). B&K accelerometers are virtually bomb proof. Then we spotted that Endevco made much cheaper versions that were not as rugged, but certainly rugged enough for us. And because they were a lot lighter, fixing methods were simpler - cyano cryalite would hold them to most things, firmly enough to measure up to your sort of frequencies.

 

[GregLocock]

You need to fasten the accelerometer to the test item without marking it.

Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[TPL]

How about having a test table made up?

The starting point could be a simple plate, sitting on springs underneath and supported horizontally by springs on each of the 4 edges. To one or more edges of the table, you could permanently fix (bolt) an accelerometer (good ones go up to 20kHz easily - more than adequate for what you need). The table will be fitted with a quick release mechanism designed to allow each power tool to be quickly fixed and fired up for testing - the acclerometers will pick up all defects from unbalance, to gear mesh and bearing problems. You can view the accelerometer output on an RTA (Real Time Analyser). Based on your experience of the signature that a good unit would produce, you could then produce an overlay for the RTA screen display allowing go/no go decisions to be easily made.

Having said that, you could also do the same thing with the ac output signal from any decent Sound Level Meter - viewing the noise spectrum on an RTA (again with an overlay) will allow you to discriminate between background noise and noise from the power tool.

Theres a company in the UK called Picotech which make little devices for turning a PC into a spectrum analyser (do a google search) - really cost effective. I'm not affiliated with them, I just bought one of their instruments for our offshore people to use and am really impressed with its capability

 

[Rob45]

If the test parts are not hot, you can stick an accelerometer to the part with beeswax, which cleans up readily.
A PCB 353B18 accelerometer is relatively inexpensive, good to 10KHz, and weighs only 0.06 oz.

 

[GregLocock]

Just thinking about it, maybe this is a valid excuse to use acoustic intensity which uses two microphones to creat adirectional meaurement of sound power.

I still think an acoustic enclosure, either of the entire workstation or at least a box to put the test item in, is likely to be the best solution.


Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[gearman1234]

We made a measurement of gear noise using a microphone. This sound was recorded using a notebook computer. Then we analysed this noise using softwares available on the internet ( evaluation vesrions available) . These softwares give a frequency spectrum. The frequecy of gear mesh is usually high so you see a peak of sound pressure at that frequency or x1, x2 or x3 (so on) harmonics of that frequency. Then you can say that that noise is coming from a gear mesh. You need to hold the micrpohone near to the gearbox as possible. This has worked for us . You maight want to try it. That has led us to take corrective actions on a particular gear pair to reduce noise. You really need to correalte the data with the calculated rotational or mesh frequency.

 

[heavenly]

Hi,
If you suspect gears problem,better to dis-assemble gears & gear housing parts,bcoz after several revolutions, there may be possiblities in the tolerance variations
1.measure the centre distance in gear housings
2.teeth wear also may be the problem,give these gears for
metrology to check the profiles of helical gears,i
suppose you will be using bearings at bothe end shafts,in
case if you are using PB such stuffs better to avoid by
replacing bearings

 

[McVibe]

Colin,
Take a look here:

http://www.DLIengineering.com

- we can set you up with a vibration analysis test stand and can use our automated diagnostic system to tell you what specific problem the unit has (i.e. gears Vs. bearings vs. shaft looseness etc.) - this is our specialty and we have been doing it since 1966.
It is easy to separate mechanical components by frequency (as noted by another contributor to this thread) - you can take this a step further and qualify mechanical faults and severities. If you would like more information, email me at:

afriedman@dliengineering dot com