Good Day All, How can this be? The measurements below are outboard horizontal and vertical on the same machine I have been talking about for quite some time. The snap shot is from a CSI 4500 continuous online monitor. I do not understand how there can be such high Waveform P-P (red boxes) and such low overall (blue boxes) values. In fact all most all the other values are low and the machine feels smooth and runs quiet. Thanks in advance for any help.

Is this the only points on the system that is doing this?

Who setup these 2 points?

Have you taken a 2120 or 2130 reading on these points with the same setup in the continuous monitoring system?

Does the 4500 only display values per parameter or can you actually see the waveform plot that generates this value list?

What does CSI say?

Ralph,
First off, Is this possible in your experience? Can there be that much difference? Seems like it would be exponential, a huge increase in one value should certainly affect other values.
These are the only two points on this machine showing amplitudes like this.
CSI set the machine up in March of this year when we bought the 4500.
I can "demand acquisition" to look at live spectra and wave forms but only out as far out as the DCS was set up at. The g-swing in the wave form agrees with the value in the snap shot.
I don't want to involve CSI until I know what I am dealing with.
I am a "green horn" in this field but long enough to question these values when considering how the machine sounds and feels.
Thanks Ralph,

Ralph,

Comparing the red and blue boxes is like apples to oranges.

Blue = overall velocity (peak units, but based on RMS x 1.4 and possibly averaged over several spectra)
Red = waveform peak to peak acceleration (for a single block of data)

It appears that there is considerable high frequency spikes in the waveform. The Waveform p-p value is approximately equal to 2 x HFD (5.2 g-RMS) x Crest Factor (4.33) = 45 g p-p that is close to the true waveform value of 40.3 g p-p.

I assume the machine has antifriction bearings (ball/roller) and possibly the accelerometers are on a pump with some cavitation present.

Walt

Walt, the machine is an alfal laval separator/decanter. There are three bearings on each end. both ends are belt driven. The troubled end is the back drive end. We just replaced the planetary gearbox because we found large pieces of metel on the drain plug magnets and the oil had a very strong burnt oder. The low amplitudes in non-sync energy is what has me lost.

So in your opinion there can be a difference as large as this between the overall and the peak to peak readings?

Ralph,

The answer is yes depending on the F-max for the overall velocity. If the that value was 1000 Hz and most of the high frequency acceleration peaks were in the HFD band (typically above 5000 Hz), then the signal energy is in two different frequency bands and therefore not comparable.

Working the numbers backwards: take 40 g's p-p at 5000 Hz (300,000 cpm) and calculate 0.245 ips peak velocity. This does not account for the Crest Factor conversion, but it is in the ball park with 0.35 (Blue Number)

The high frequency values (HFD and Waveform p-p) may indicate poor lubrication or developing gear or bearing defects. I assume the accelerometer is stud-mounted on a rigid part of the bearing/gear housing.

Walt

This message has been edited. Last edited by: Walt Strong,

Thanks for the info Walt.

The post was by "Noisemakr" and not me, unless you are talking to me, which is quite allright.

You (Walt) are saying basically what I was "fishing" for in my questions. If the waveform is at a different "Fmax" (higher) than is the other low end data, then the Gs could be higher and appears so on the lists.
The bottom data says 140 Gs P-P. This is awfully high. You need to find out what the waveform is set to in relation to the low end data. It does, as Walt says, appear that the waveform reading is not related to the low end data. CSI is a believer in the "Special Waveform" function, so they may have set it up using a very high setting for the waveform.

I would contact them and see what they did. If they did not set it up as we think, then you would have to look somewhere else. In the meantime, I would take some data with my meter at different settings to see if I get close to what the list shows and analyze the data to see what is going on.

Only my opinion and I could be totally wrong.

Sorry for the name confusion Ralph.
The 140 g's p-p on the second sensor is way too high for a common 100 mv/g accelerometer that is normally rated for 50 g's. I would question whether the sensitivity setting matched the accelerometer or there is a bad accelerometer.
My math for the first accelerometer gives possible agreement for the displayed acceleration and velocity values, if the vibration signal is very high frequency.
Any vibration monitor system should have readily available user manual, specific setup information, and recommended actions (procedure) depending on which alarm values are exceeded. What good are alarms (vibration, fire/smoke, gas release, tornado, earth quake!)if people do not know what they mean and what actions to take when they occur?

Walt

The top one made sense to me and I can easily see that as reality. However, the bottom on is 140 g's P-P????? Something wrong if using a 0.1 V/g transducer - false data?