Guys, there are reasons these low tech low buck schemes are not widely used.
- The fundamental principle in instrumentation of any type is to figure out what you need to know the details on, and then sense that or something very closely related. If we are concerned with torsional vibe, you measure torsional accelerations or displacements, not linear accels of a housing that may or many not have anything to do with torsionals;
- Sensors used must be at least four times faster than the fastest vibe response you may have, with ten times faster being better still. Why? If you can not see the waveform of the signal, you are getting huge noise and maybe no signal at all. You can even draw a "wrong" conclusion about what is happening, which is worse than not being able to understand what is happening;
- Power the sensor adequately to make a usable and clean signal;
- Get the signal out of the sensor and to your recording equipment;
- Make sense of the signals.
The predominant equipment on this stuff grew out of fundamentals and real needs.
Rotec and its competitors use either optical or metallic tooth targets sensed with laser tachs, mag pickups, or reluctance pickups. They measure the rotating speed of each wheel with lots of teeth and way fast sensors. Data taking equipment is really fast - 10GHz to 100Ghz clock on the time stamp for every tooth passage. No linear pickups of housings are normally done.
Then we get to processing, because we are not concerned with wave form shape (sinusoidal is usually assumed) we want to know what is vibrating with what and what mode is big at what frequencies. Fast Fourier Transforms of each signal, FFT on one signal vs another signal, and so on. Now maybe there is software out there that does all this once you have stored the data, but if it gets cumbersome, we are unlikely to get much from it.
Is the data useable? Sure our 4x rotation (2x firing) on a 6000 rpm 4 banger is only 400 Hz, but if you want to know that you have a stiff system, you had better KNOW that your lowest mode freq is well above 400 Hz. You have to be able to sense vibrations to at least 800 Hz. Go with a 7000 rpm V8 and and now you need more like a 1kHz. OK, maybe that is not hard with sensors designed and marketed for this purpose, but what is the frequency response of your stereo voice coil when mechanically driven? What is its signal rejection in other axes? You do not know, and so you will have to characterize your sensors in some way, and select ones that are fast enough.
Now if you have the ability to get data and process it, can you analyze for A vibrating relative to B, A to C, A to D, B to C, B to D, and C to D? Frequently we find that we need to sense further afield than we started out, or worse, we find that some internal assembly has all sorts of vibe content and need details from within.
This drives the need for expensive sophisticated systems... If your repurposed speakers and strain gages and oscilloscopes can extract usable results, fabulous. Somehow I do not see most of them rotating with the parts, and the strain gages end up needing telemetry to get the signals out. the exercise ends up being about getting the telemetry to work, not about gathering info on what is happening in the machine. Somehow pointing a laser tach at a paper target seems way more reliable and useful.
I shall remain skeptical until shown differently. Please, prove me wrong. I would love to know there is a system out there that is cheap, easy to use, and covers the bases...
Billski
