Spectrum Analysis of Vibration

It would really help to know what kind of machine, and what the frequency pattern looks like (amplitude and frequency), what kind of analyzer you are using, and what kind of transducer you are using to measure the vibration.

But I suspect that if it is not changing with speed then you have an instrument problem somewhere. Is the primary frequency in the spectrum electrical line frequency for the country you are in ( 50 Hz or 60Hz)? If so then you have a grounding problem and you are not measuring vibration, but rather electrical noise.

Is it a piston engine or a gas turbine engine?

Thank you for your suggesion. I give the answer of your query .....

1.)Type of sensor is Accelerometer of Allen Bradley make.

2.) Spectrum is Amplitude Vs Frequency.

3.) Machine is surface minner.

Please suggest what kinds of fault I can do?

No I mean precisely what amplitudes and frequencies are you seeing in the Spectrum?

(5 gs at 50 Hz, 20 gs at 100 Hz for example)

Again I am pretty sure you have an instrument problem and you are not measuring true vibration.

Thank you for your valuable suggestive comment. The frequency band of our analysis is .2 to 4200 Hz. I see amplitude present all over the frequency range. The amplitude is very less for some frequency but amplitude present in all frequency. Actually the sensor gives directly 0 to 5 volt. The sensor is connected to a signal conditioner which gives 4 to 20 mA corresponds to 0 to 5 V. we do spectrum analysis on this 4 to20 mA. Is this is the fault? Whether we have to do spectrum analysis on 0 to 5 V signal? please help me.

We use rockwell automation vibration module. The sensor directly gives voltage signal. The sensitivity of this sensor is 495 mV/g. We connect this to XM-121 dynamic module made by allen Bradly. This module gives BNC output. So which signal i can take for spectrum analysis.

If the BNC output has 495 mv/g use that. That should be the vibration signal.

This is the sensor's sensitivity. I don't have any data regarding buffer out put. I see at the time of open circuit the voltage of BNC o/p is 8.138 V but it becomes 0.38 volt when it connect to spectrum analyzer. Why?

Apologies if I'm off-base, but, as per Steve's post (#5), sounds as though you've waded-into this "hip-deep" wearing only ankle-boots.

For what it's worth: ASNT (American Society for Nondestructive Testing) offers certification in Vibration Analysis, and courses in preparation for same.

Likewise, there is some good information online ... however (at my last search), the *best* info remains in papers for which you must pay. As far as we've come with VA (my father was somewhat involved decades ago, working for Gilbert Commonwealth), the science remains in its infancy, compared to some others...

Wishing you the best of success ~

There are experts in this field, why not hire one and get it sorted out? Most of the CM (condition Monitoring) people do it like SKF, INA-FAG, to name a few.

The accel signal is ac, but it rides on a dc carrier. If you put a volt meter on the signal you should get about 8 volts DC that is the carrier. When you measure the AC it should be a mililvolt signal, although if you are see a few gs it might be a few volts AC.

If you are only seeing 0.38 volts, then either the vibration is a bit less than a g (very good for your machine) or you still have a signal connection problem.

Some transmitter modules require a special adapter be installed on the BNC output to read it properly, especially if they are wired in current loop configuration. I suggest that you contact Rockwell for some help.

Thank you for your suggestion. I got my result after doing spectrum analysis on the BNC signal .

Recently I had taken the vibration (velocity) of a mining equipment. In the vibration spectrum I found that spectrum lines present at 0.5X harmonics (upto 20X ) of engine speed. The amplitude at 0.5X is low compare to other spectrum lines. please give me the some hints regarding this type of spectrum pattern.

Please suggest me which standard I have to follow for vibration measurement of mining machinery.

Spectrum alaysis has value only when you (1) systematically record the signal values against chosen frequencies in x, y and z axes for each part of the machine, (2) compare values over time, analyze the vaiance and (3) identify what parts in the machine are generating excessive signals, (4) that part is probabily malfunctioning and risking a failure.

It may require 2-3 years for a fault to develop but data must be recorded at least once a week. The drill may protect you from more expensive noises.