Sign up Calendar Latest Topics Donate

  Author   Comment  

Sr. Member / Supporter
Posts: 310
Reply with quote  #1 
High Resolution

In your work, what do you consider a normal and high resolution spectrum?

I think based on my discussions with others that this term is subjective. If I use 1600 lines, will my spectrum be high-resolution? If the width of the bin is 0.25 Hz or smaller, would that be high-resolution? Or maybe if the new spectrum is 5 times higher in resolution?

Regards- Ali M. Al-Shurafa

Avatar / Picture

Sr. Member
Posts: 1,121
Reply with quote  #2 
I mainly take all velocity on 1600 lines and acceleration at 3200 lines but it's a legacy from the guy who instructed me originally.
I do take 800 lines velocity on some fans in a plant we have almost 10 years of data from and but it's mainly due to the set up which was done by the guy before me and from a time when data storage And collection speed was a lot slower on our Analyzers.

To add to the above and which I didn't outline this is with monthly/bi-monthly route type stuff.

When I do troubleshooting or one-off work I crank the resolution up a lot higher as we all know you probably have only one shot to get as much data as possible.

Sr. Member
Posts: 76
Reply with quote  #3 
I think that it is quite subjective too.
Generally speaking I would consider a high resolution spectrum one that allows you to distinguish 2x shaft speed from 2x LF.
Where this does not apply (e.g. a turbo generator) it may be anything with a bin width below 1 Hz.
I hear sometimes that we need a high resolution spectrum to verify if that subsync is a rub at 0,5x or oil whirl.
Danny Harvey

Sr. Member / Moderator / Supporter
Posts: 1,402
Reply with quote  #4 
I think the term "high-resolution" as we use it in the field, would be more accurate if it were termed "higher than normal resolution".

For most applications, resolution doesn't need to be that high. But there are things that we would like to get a better look at like pole pass sidebands around harmonics of running speed so we do a lower fmax, higher resolution reading. Typically we will go from something like 120,000 cpm fmax and 1600 or 3200 lines of resolution to 12,000 with 1600 lines.

In my standard AP sets that I made for CSI software, I made incremental fmax's with two resolutions and then picked whichever one was appropriate. So I guess the answer in practice for me is really that high resolution is twice normal resolution. Unless I want more.


Avatar / Picture

Sr. Member / Supporter
Posts: 599
Reply with quote  #5 
Originally Posted by vogel
Generally speaking I would consider a high resolution spectrum one that allows you to distinguish 2x shaft speed from 2x LF.

I am 100% on board with this definition.

The last bridge I did modal analysis on, we took 66min recordings, 6 overlapping 1000second averages. That gave us a delta f of 0.001 Hz. It was critical when you are setting the baseline of a 0.253 Hz mode. I guess that is high resolution. Our hardware's lowest sampling rate was 2kS/s, so we ended up with 100,000 lines of resolution.

Does nobody run into the original analog definition of resolution anymore? Where resolution is the smallest signal that can reliably be resolved from the noise floor? High resolution in that case would have nothing to do with the recording parameters, but rather just a low noise floor.

Avatar / Picture

Posts: 1,810
Reply with quote  #6 
My collection parameters for typical pumps, fans, etc with rolling-element bearings is 2000 Hz and 1600 lines. If I see/hear/feel something of interest or concern during a data collection I will usually take an extra spectrum at 400 Hz and 3200 lines, giving me 10X my original resolution. That allows me to resolve the most typical problems such as 3X vs 3.09 (BPFO), 2X vs 60 Hz, etc. Beyond that, I use enough resolution to resolve 1x side bands around any peak of concern.

But 99% of my work is routine surveys of run-of-the-mill equipment. Your mileage may vary.

"The trend is your friend"
Previous Topic | Next Topic

Quick Navigation:

Easily create a Forum Website with Website Toolbox.