Hi
For an industrial application to build a rotor analyzer I want to write the corresponding SW starting from scratch.
A tool I would like to implement is the spectral analysis of the vibration signals for later use in an in-situ balancer.
I started implementing the FFT algorithm to analyze the signals. Right now I don't have the hardware (accelerometer and AD-aquisition) to capture the signals. For this reason I am currently working with the signal from my laptop microphone. :icon_idea:
It's still a work in progress and I'll write the various modules in test applications first.
If anyone is interested in such topics, I will post them here. :biggrin:
Biterider
my dad did something similar in the early 1970's to minimise the oscillation on the drum of the Xerox photocopiers .. the crucial bit of evidence is the mass of the item ...
regards mike b
Hi Mike
You are right. This technique is not new and has been around for a number of years.
You can buy some equipment to get the job done, but if you want a specific job done, you'll have to pay a lot to hire a specialist.
On-site balancing isn't easy, but it pays well.
Biterider
Hi
While waiting for the right hardware, I will continue to develop the software.
At the moment I have the first try of the (i)FFT object. It only has basic streaming capabilities and can only handle REAL4 as it uses SSE-SIMD technology.
The code is based on the work of Siekmanski that he posted here a while ago.
I've tested it so far but haven't done any performance testing, but I think it should be fine. :rolleyes:
Biterider
Hi
Here is the next building block. It is intended to provide a signal source for the development phase.
This file handles microphone input from various possible sources connected to your PC.
The signal is manipulated in memory to prepare it for playback by the PlaySound API.
It's not finished yet and I haven't spent too much time on the documentation as it will be removed at the end. :biggrin:
Biterider
Hi
Besides the (i)FFT algorithm, I planned to add additional tools.
Some of these are the correlation and cross-correlation algorithms. Has anyone already implemented something in this area?
https://en.wikipedia.org/wiki/Cross-correlation (https://en.wikipedia.org/wiki/Cross-correlation)
Biterider
Hi
Nobody has ever used the correlation or cross-correlation algorithms? :dazzled:
Biterider
Hi,
Well yes. But nothing using assembly that I recall. And not
very efficient, mostly simple algorithms in FORTRAN.
Regards,
Steve N.
You could consider audio echo cancellation methods, which I sure with a slight modification can be used for balancing.
Both use FFTs and the balancer Inverse FFT would naturally be dependent on the rotational velocity.. etc..
:icon_idea: