I have a question on mounting accels. I have a saw that is going down this week because of peakvue amplitudes in the spectrum and the twf are heading for the moon. I have been keeping a close eye on this particular peice of equipment for the last 6 months because of installation issues that I need not get into. Last week I was on vacation for a week and a half and when I returned the amplitudes really took quite a jump. Not wanting everyone here to FREAK OUT I went about my own business got a new housing with shaft bearings etc togather and ready for install so if we were to lose it down time would be minimal. With that said on to my real question. These particular peices of equipment have mounting pads(glued) with accels perminantly mounted due to saftey issues. After getting everything ready that I needed to in case of a breakdown I went back down to the machine to do some more digging around. I took a reading with my rare earth two pole magnet mount accel and the amplitudes dropped off the chart compared to before. I went back to the office so I could dump and keep that data. Next I started changing my mounted wires around withe my coioled wire to see if that was the difference same thing magnet mount readings way below my glued mounting pad readings. Next I thought maybe a bad accel. changed them around same thing magnet mount way below mounted pad. I geuss my question is has anyone else ran into this problem or seen this. I only have a handful of eqipment that are mounted perminantly all the rest is using the magnet and I am starting to realy think that the difference between the two methods are quite drastic as far as taking Peakvue readings go.

Any Thoughts?

My first though would be settling time for the transducer. Often there will be a spike after hooking up to a stand alone accel. Monitor the waveform next time before you collect data, once the waveform settles you’re ready to start collecting data. A bad connection breaking the circuit will cause the same thing to happen. However, there is always the possibility the solid mount transducer has a higher frequency range than the rare earth magnet and the signal you are trying to capture and process with peakvue is not getting picked up.

Mounting methods will make a huge difference. The better the contact, the better the frequency response. The other plus with the mounted accels is they're in the same place every time (you hope).

Waylon brings up a good point, too, about settling time. Make sure the waveform has stabilized when you hook up the BNC.

I don't think that I have a spike problem beacuse I am used to waiting about five seconds before aquiring data and I also take a regular spectrum and waveform data before aquiring the peakvue reading.

Thanks

Jim,

I agree, permanent vs. dual rail magnet will make a big difference. Two related questions in this regard:
1. what is this difference numberwise in your case? (if possible post plots)
2. what is your HP setting? For example, a
HP=5 KHZ may attenuate one mounting much more then another.

I was unclear from your email...Have you removed the permanent accel and mounted the magnet directly onto the pad?

Is this a Bipole or flat rare earth magnet?

All high frequency signals are highly directional, although I think you have already covered that if you have taken magnet readings right next to the pad.
More likely I think is either the mounting method, or the frequency range of the accel.

THe other thing I was unclear on was whether or not you were getting this signal from just 1 accel or multiple...I am assuming the high readings are from just 1.

I have used two different accels and they both give the same readings both have low amplitudes when rare earth magnet is used for mpounting and both give high readings when screwed into the glued mounting disc that I have installed. I never would have imagined that there would be such a drastic difference between the two but in this case I have convinced myself otherwise

Jim,
You can get a large difference when going from a stud mount to a magnet mount with PeakVue. It depends on what frequency the energy is being generated. If it is a lubrication problem it is usually very high frequency energy, with a mag mount the accel can't see the energy being generated. Can you post the spectrum and waveform from the stud mount? If it has high g's in the waveform and little or no frequency content in the spectrum other than cage frequency I would suspect a lube problem. Most bearing defects are low enough in frequency for it not to make a big difference in waveform amplitude as long as you are using a flat magnet on a prepared surface. If you can monitor the PeakVue waveform and give the bearing a few shots of grease and see if the amplitude drops way off. I have seen the waveform amplitude from a bearing defect double just by scraping the paint off when using a magnet mount.

I'm not sure if this is going to work or not its been a while since I have attached any data and I can't seem to recall how I did but the first spectrum is with a magnet mount and the second is with a stud mount accel.

Doc2.doc (68 Kb, 86 downloads)

Sorry this was the magnet mount the first on is stud mounted

Doc1.doc (66 Kb, 71 downloads)

Jim,
Can you post the autocorrelated waveform?

Here is the corelated peakvue

Peakvue_Corelated.doc (28 Kb, 71 downloads)

Is the spindle speed actually 3600 rpm?

Jim G.,

How can you shut machine down for maintenance based on this PeakVue data? Neither spectrum, nor waveform, indicates a specific fault is present. I would not assume that the stud-mounted (glue base) accelerometer is giving better data than the other accelerometer. I have seen cases when an adhesive base caused a natural frequency below 10 kHz, and the only correction was to remove and remount it. The F-max for a PeakVue measurement is totally uncontrolled, so the door is open for goofy high frequency data.

Jim Crowe,

What is use or fault interpretation value of autocorrelation of PeakVue waveform?

Walt

quote:

Originally posted by Walt Strong:
... The F-max for a PeakVue measurement is totally uncontrolled, so the door is open for goofy high frequency data.

This is a valid question. As raw data in PeakVue is sampled at very high rate (I believe 100 KHZ or maybe 256 KHZ) chances are accelerometer mounted resonance will be excited regardless of mounting type. Although the envelope may carry valid fault information, the amplitude will be greately overstated.

Same question could be applied toward the HFD method which measures analog overall up to 20 KHZ.

David

Walt,
I have done a lot of research on lube problems (due to a decision once by someone to go to only one grease for everything) and I also use a lot of stud mount accels. I have found that without proper lube a stud mount accel will often give over 40g's waveform amplitude in peakvue. If you have a lube problem and the waveform is autocorrelated you will either see a very low correlation value or you will see the period of cage rotation.

Jim C.,

You are saying that Jim G. saw machine has a lubrication problem based on high PeakVue G's and low autocorrelation, correct? If an accelerometer on flat rare earth magnet was attached to a steel disk attached to machine surface with adhesive, would you expect adequate frequeny response for detecting poor lubrication?

Jim G.,

Can you improve lubrication without taking machine out of service to see if PeakVue levels will drop and to prevent an actual bearing fault from developing?

Walt

Walt,
Exactly, in fact when I say I use stud mount accels they are in fact attached to a machined disc that is epoxied on. That is why I asked for the autocorrelated waveform and if the speed was actually 3600 rpm. His autocorrelated waveform looked similar to some of mine with lube problems but the period shown was too high frequency to be cage. I would have expected to see the period a little less than 1/2 order. You don't always see cage with lube problems though. His best bet is to shoot a couple shots of grease in while monitoring the waveform and see what happens. I had bearings run several months with this problem so I would never call a bearing to be changed without determining if it was a friction problem because of lube, but with his amplitudes I would want to be sure it is lube. The fact that it has no frequency content in the spectrum says either it probably has a lube problem ot the bearing is so far gone it is not generating the bearing frequencies anymore, just random impacting.

I have already taken the machine down and replaced the bearings on it. It is running as I am typing. As for the question on speed it's actually running around 5400 rpm. I was tempted to just remove the retainng caps and add some grease and monitor it but these particular saws are greased on initial assembly and the manufacturer recomends that the covers be removed yearly and repacked. Note the manufacturer recomends this and we realy haven't had any issues with this practice. I have attached a picture of the spall on IR of the OB outer bearing that I found after I dismantled them. The OB inner bearing also had what looked like some dulling but under the scope it was real fine spalling starting to show up. I'm glad now that I didn't try adding grease and monitoring it for another week but I almost did!!!

Jim G.,

You made a good call in that bearing clearly had faults present. At 5400 rpm the bearing may not have lasted very long. Is there any way and to convert bearing for external grease application to extend bearing life? A hand packed grease bearing doesn't seem like a great application for a 5400 rpm service speed.

Jim C.,

Thanks for the info on PeakVue.

Walt