This topic has been discussed a few times in the past. But anyway...

Attached is an example on how resolution may change the appearance of a spectrum creating an impression to suspect looseness due to subsync harmonics and raised floor (Fig. 1). In reality it looks that nothing is there to be afraid of (Fig. 2).

In real life though seldomly do we acquire data in routine surveys with such a high resolution.

Dave

Resolution_floor.doc (56 Kb, 115 downloads)

David-G
The times on the data are close together. Was averaging use and if so how many. Also was overlapping used and if so how much.

What do you have? Sleeve brgs? Mass and/or HP or a photo also may help.

I would not write this off as having no problem. Depending on circumstances, it may have a serious problem. Fundamental, 0.5X and w/broad banded across the base could mean serious trouble.

More? I would want much more info to base a decission.

Dave,
It looks like the low res spectrum has more averaging, and it looks like a resonance just below 2x. The high res plot also shows it, but it looks like just one average. There wasn't enough time (energy) to clearly delineate the resonant frequency.
Howdy Barry!

David,
You seem to be misinterpreting the effect of finer resolution.
As you go to finer resolution, components that are at discrete and nearly-constant frequencies will be essentially unaffected. But components that are more broad-banded will be pushed down toward the background.
If you made the resolution fine enough, you could essentially make them disappear. But that doesn't mean that they are not real or not significant.
If you went in the other direction, to coarser resolution, say 300 rpm, you'd find that those broad non-synchronous peaks would come up relative to the integer orders of machine speed. And that would tell you more about their relative importance in the overall vibration picture.
If it's not something purely mechanical, could be off-optimum rough hydraulics etc.

{There wasn't enough time (energy) to clearly delineate the resonant frequency.}

Ron, how much time would be needed to do this. I did the math and the high resolution reading would take 8 seconds. That is a little over 230 revolutions of the shaft if it is only 1 average. How many revolutions would be needed to clearly delineate the resonant frequency?

What exactly does 150 rpm or 7.5 rpm resolution mean? How do you arrive at those numbers? What I see on the first plot is 40K cpm (666 Hz). 40K/150 = 233 lines? On the 2nd plot I see 10K cpm (166 Hz), so 10K/7.5 = 1333 lines?

My "standard" setup for 1800 rpm machines is 2000 Hz & 1600 lines. Does that translate to a "resolution" of 75 rpm?

If you are only using an Fmax of 40K cpm, that is too low in my opinion.... you are missing a lot of information at that Fmax. Most rolling-element bearings have "ring tones" in the 1200 - 1600 Hz range, in my experience.

Rusty, I think he means CPM. With your setup you have a binwidth of 75 cycles per minute. I was assuming that the 1st Spectrum was really 60,000 cpm and 400 lines and the 2nd was 12,000 cpm and 1600 lines. That is CSI data so I know that he could not have anything between 800 and 1600 lines.

Technically, is "bin width" the same as resolution"resolution"? (Just trying to get my terminology correct)

Good question. I think that Lines of Resolution are the actual number of lines that the Fmax is divided into. Bin-width is the spacing in frequency units between each line.