When you are referring to g's for acceleration, are you talking rms, peak, or peak to peak.

I will have some where the rms values will be very similar, but the pk-pk values will be much different.

I.E.

1.5 g's rms on two different bearings
Pk-pk will be 9 g's on one
pk-pk will be 19 g's on the other

Can anyone explain this or am just missing something?

I've been in the vibration business for 25 years, and Sam is the only person I've ever heard of that only uses acceleration to do analysis. I can't imagine not using velocity to detect, analyze, and diagnose mechanical faults. I think Sam is just being a "purist" when he insists you can't use an accelerometer to generate a faithful velocity spectrum. It might not be perfect, but it's plenty good enough.

Thanks for the compliment Rusty. But I use the four basic or normal parameters that most use. In my own software the FFT is normally displayed in g's 0-P while there is a digitized listing below of:
Freq (Hz) g's IPS mils.

I said four -- don't rule out sound measurements.

While velocity may be OK and you may very well be using IPS in and have a successful program; but, working in g's is so much easier to me and give much more info that is quicker to analyze and accuracy is high. I started using acceleration in 1980 for cap data exclusively and by 1983 worked in acceleration and it (IMO) gives much more information. A picture is worth a thousand words - no picture, no words. Acceleraton gives a much larger picture. IPS data that are in the dirt are clearly displayed in g's.

THMotorMan; You have 1.5 g rms on two bearings. And on one bearing you have 9 g's assuming TWF and double on the other. One transducer? Three grossly different readings! Check your system.

Rocking Rod & Buzz: I do machines over a wide frequency range. Some very slow RPM but some high. An Elliott compressor for example with 3576 RPM input and calculated out on all gears and impellers (hi-speed side 51,300) still give fundaments of the highest component <20,000 Hz.

Now, lumping the rest of normal together in a sense -- I will do a 1200 RPM A-F machine at ~2 kHz while a 3600 RPM will also require 2 kHz I will also get 5 kHz data (remember all along the machine dictates frequency range; it will tell you what you need to look at). At day one, view your data and see how your FMAX should be set; the spectra will tell you.

So, most all FMAX will be 5 kHz or below. I basically never look out to 100 kHz. Bear in mind is look at frequency in Hz.

Phone is a'ringin'

[QUOTE]Originally posted by Sam Pickens:

<So, most all FMAX will be 5 kHz or below. I <basically never look out to 100 kHz. Bear in <mind is look at frequency in Hz.

Not too familiar with khtz. What is the equivalent to 5khz? I'm used to cpm or rpm...

I know of khz in terms of radio waves..

I also like velocity because I can focus down where the low freq machine faults are around running speed up through about 50K (50,000 CPM)for looseness, misalignment, balance and other issues. Even though most of our machines FMAX's are set at 120K (120,000)

I guess it's just a matter of what you are used to. I was trained to use both VEL & ACC at all the schools I attended. Not saying Sam's approach is wrong, I'm just speaking for myself. It's good to hear about others experiences who have been in this business a lot longer than I.

I'm about to start my 7th year,I'm a level 1 almost a 2. And still I feel I have not scratched the surface. I started late, I'm 55 and will retire in less than 7 years.

********************************************

quote:

Originally posted by Sam Pickens:
Rocking Rod & Buzz: I do machines over a wide frequency range. Some very slow RPM but some high. An Elliott compressor for example with 3576 RPM input and calculated out on all gears and impellers (hi-speed side 51,300) still give fundaments of the highest component <20,000 Hz.

Now, lumping the rest of normal together in a sense -- I will do a 1200 RPM A-F machine at ~2 kHz while a 3600 RPM will also require 2 kHz I will also get 5 kHz data (remember all along the machine dictates frequency range; it will tell you what you need to look at). At day one, view your data and see how your FMAX should be set; the spectra will tell you.

So, most all FMAX will be 5 kHz or below. I basically never look out to 100 kHz. Bear in mind is look at frequency in Hz.

Phone is a'ringin'

Let me clarify my last post, that was 120K for velocity. And 250 to 300K on ACC..

We also use HGE3 readings out to 300K
************************************


[QUOTE]Originally posted by Rockin' Rod:
I also like velocity because I can focus down where the low freq machine faults are around running speed up through about 50K (50,000 CPM)for looseness, misalignment, balance and other issues. Even though most of our machines FMAX's are set at 120K (120,000)

I guess it's just a matter of what you are used to. I was trained to use both VEL & ACC at all the schools I attended. Not saying Sam's approach is wrong, I'm just speaking for myself. It's good to hear about others experiences who have been in this business a lot longer than I.

I'm about to start my 7th year,I'm a level 1 almost a 2. And still I feel I have not scratched the surface. I started late, I'm 55 and will retire in less than 7 years.

********************************************

This thread seems to attract a large crowd. I guess not surprising when it's about the $ sign!

quote:

IPS data that are in the dirt are clearly displayed in g's.

What about low-frequency events, like belt-induced vibration, imbalance, 1/2x (looseness), low-frequency beats, bearing cage defects, etc. I'm not buying that acceleration gives you a better picture of these type of events.

If you are using an accelerometer, acceleration will provide you with data as good as it gets. You must be using the right transducer. Your signal must be above the noise floor and above the 'dirt'. It displays a voltage and your EU conversion then displays g's, IPS or Disp (mils). The transducers resolution must be such that the frequency of interest will have a sufficient magnitude that it is producing voltage at that frequency to rise above the noise floor. Your acceleration signal doesn't magically become velocity or displacement. It is a voltage signal in g's acceleration and you may elect to integrate to velocity or displacement. Note that you are using a voltage signal from an accelerometer and merely changing display to a different EU. If acceleration doesn't display it then it isn't magically produced in IPS.

However, if you are wanting IPS and using a velocity transducer, then your signal enhancement in IPS or display may be clearer on those signals above the linerarity cut-off - say 450 RPM or above. To get a clearly difined amplitude below than then one must use the curve generally provided by the OEM.

Gotta love the shared information. I would have to say I agree with most. One thing that really determines what you do and call is what the customer needs. And education is the tough part. Trusting the data and knowing what to back up your calls with is what really gets the buy in. The way I see it the more advanced the notice the better the chance you have of not only making a good call but also of helping improve the overall equipment reliability. MTBF increase of 1 to 5 years is pretty impressive. That is money in the bank regardless of how you put it. But when it comes to any analyzing I think you gotta show them dollar for dollar that it works or when budgets get cut so does the program. Also strongly believe it isn't the tool you use but how you use it that really matters. And your ability to make what you say tangible to those who see the squiggly lines as hocus pocus, black box magic. Seeing is always believing and sometimes they just gotta get burnt to stay outta the fire.
Keep on providing the Best and then go home to the family and enjoy 'em while you can!!! Cuz that is really the reason to do this in the long run!

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

So Sam, are you saying that you 'collect' (and store) your data as acceleration (spectra and waveforms), but that whey you do your analysis, you are converting it (in your software) to display as velocity (or displacement)? So you are not viewing only an acceleration spectrum when you are diagnosing lower-order mechanical defects? Is that what you are saying?

It is very interesting reading these quotes but I have read and (I am a beginner) you should go outside and work the machine on any bad signals that are produced to verify problems. Can you make a call just on the route readings? This extra analysis would add to the cost or time I would think. Am I right or wrong. Please advise

quote:

Can you make a call just on the route readings?

Welcome smokinjo! In answer to your question, Yes and No. I take all my own route data. When I walk away from the machine, I am 80% done with it, because I know if there is any reason to suspect a problem. If so, I have collected enough "extra data" to make a call back at the office. As a consultant, I don't have the luxury of running out and checking a machine that is suspect.

There are companies that use data collectors (technicians) to just collect data, which is then uploaded and reviewed by an "analyst" somewhere. Personally, I think this is dangerous and I would never endorse it. Nor would I ever be comfortable analyzing data and making calls based on someone else's data.

But that's just me. Obviously it works well enough, or people wouldn't be getting paid to do it that way. Of course, if they could get a "dedicated" analyst to collect the data and do the analysis, for the same money, I'd think that they'd prefer that.

But there's all kinds of ways to skin a cat. You just have to find a way that works for YOU.(that's just an expression: no animals were harmed in the writing of this paragraph)