Interpreting vibration data?

Measuring real-world vibration can be tricky. The very mounts and plates on which they are mounted can give stronger measurements, and vehicles are filled with various and sundry natural frequencies.

Where was the sensor attached? It likely felt whatever was the strongest there, which may not have been engine rpm. ("diaphraming" fo surfaces, and appliances-on-brackets vibrating as a spring-mass are fairly common.)

And, look to instrument reading errors when considering vibration and harmonics, since 4000 is a harmonic of 2000, within many ranges of instruments. e.g.,

4000 +/- 5% = 4200-3800 1900 +/- 5% = 1800-2000 (x 2 = 3600-4000)

2) and/or look closely at the setup to see what you were reading.

I would first look for a device near your sensor that is connected to the engine by pulley so as to be turning at a bit less than half the speed. I might even check "belt slap" being put into the engine through a mount. And a dozen other things.

fwiw

Do you know what the frequency response is of your accelerometer?

Is the instrument mounted on the engine block? If not, why would you expect it to have the engine frequency as the largest component? A moving vehicle has a whole pile of other sources of vibration, and if the engine is properly mounted, anything not on the engine will pick up those vibrations more than the engine's.

-Paul

I'll try: a four stroke 4 cylinder engine gets two power pulses per rev, that's 8000 pulses per min or 133 Hz

But you saw a fundamental at 1900rpm = 32 Hertz Different source: probably not one bad cylinder in four. Perhaps a tire? say 20 in diam at 60 MPH = 88 fps / piX1.7 = 17 Hz

That looked promising: two unbalanced tires possibly??

Brian Whatcott

Brian Whatcott wrote in news: snipped-for-privacy@4ax.com:

They don't really add like that. One bad cylinder will generate every half order (where an order is a multiple of crankshaft speed), form half order up.

Two sources at 17 Hz do not create 34 Hz, in a linear system, they interfere constructively or destructively, depending on phase, to generate another 17 Hz signal.

Cheers

Greg Locock

If your sensor wiring runs anywhere near an HT line its likely that you are just seeing EMI on your sensor lines. Do you get the same fundamental frequency ( and phase) and roughly the same magnitude on all three axes?? around 4000 rpm held _really_steady_ for your full 10 minutes would give you about 30 milliseconds between pulses assuming coupling from one HT line was dominant and a 4-stroke.

Best of Luck Mike

Thank you for the reply everyone! I also decided to check out a Vibration and Shock handbook to see if I can find anything else. The accelerometer has the following characteristics:

Frequency Response (Nominal, 3dB): 0 - 600 Hz Sensitivity (FCLK = 250kHz): 25.0 mg/pulse/sec Power Supply Rejection Ratio: 40 dB

I don't think anyone's mentioned this yet, but remember cylinders fire every other RPM. So that means your cylinders are firing at about

2000RPM (pretty close to the 1900RPM you measured). You might be picking up the shocks from the nearest cylinder with your sensor.

As an aside (and I think it was mentioned before), just because your engine is operating at 4000 RPM doesn't mean everything is vibrating at that same frequency. Think of the 4000 RPM as an excitation frequency (input). That doesn't nessesarily mean that's what you'll get as an output.

Dave

Brian Whatcott wrote in news: snipped-for-privacy@4ax.com:

Yes, that's it, but also 4000, 6000 etc etc

Ah no, in signal analysis we talk about sinewaves. Impacts generate many frequencies, but, oddly, the frequencies won't double if there are twice as any impacts per second (in general).

Cheers

Greg

Hmmmm..I see you are unconvinced. I have this idea: spectral analysis with FFTs is not rocket sci. And I notice that if I analyze a waveform with one one millisec pulse per sec, my fundamental is 1 Hz.

If I analyze a waveform with TWO one millisec pulses per sec (though they may not be equally spaced), I see a fundamental of 2 Hz

What could I be doing wrong, do you think? :-)

Brian W

Brian Whatcott wrote in news: snipped-for-privacy@4ax.com:

I don't think you are doing anything wrong.

Agreed, the two impulse result gives a 2 Hz picket fence as expected. If we use 6 pulses, we get a 6 Hz fundamental. etc etc.

None the less, adding two 17 Hz sinewaves does not produce one 34 Hz sinewave. In rotating machine analysis when we talk about frequencies we are talking about sinewaves.

I can't say I like weaseling out of it like that, but am pretty confident about both those paragraphs.

Cheers

Greg

Greg Locock wrote in news:Xns971442174DB8677777777777777777777@211.29.133.50:

Incidentally, what happens to your two times fundamental if the two impulses are reasonably close together? It is no longer the highest amplitude. At around 400 ms separation the next 'harmonic' is just as high, and below that 3 Hz starts to dominate.

So, two pulses per rev does not generate a strong fundamental at 2/rev, unless the pulses are reasonably out of phase.

Cheers

Greg

[snip]

two frequencies close together will cause a beat frequency--see

or google on "beat frequency"