Data analysis software for L.D.Vibrometer results.

[Warrk]

Does anybody know what software to analyse output of dual channel Laser Doppler Vibrometer measurement(details of setup:

http://www.eng-tips.com/viewthread.cfm?qid=176437&page=1).

Requirements:
determination of PEAK velocity values(not RMS) and its variations live during measurement,
measurement of phase difference between channels also live,
FFT analysis of signals (# of frequency lines ~40k for 0-40kHz range or ~80k for 0-80kHz range)prefered live for both channels.
Measurements are fatigue tests of microalectronical components (duration-hours) so amount of data is HUGE.

Already tested:

http://www.sinusmess.de/e/produkte/samurai.htm

FFT properties not sufficient (number of lines 25601 for 0-20kHz and 12801 for 0-40kHz)

Next for tests:

http://www.lds-group.com/docs/display_product.php?product_item_id=65

Any help/suggestions would be appreciated especially if You have ever used Perception or have experience in similar aplication of another soft.

Regards.

 

[GregLocock]

I can't imagine any /display/ that could display 80000 spectral lines per second, in any useful form.

Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[Warrk]

At the moment I'm using postprocesing soft that calculates 131072 frequency lines every 1,5 sec for spectrum 0-51,2kHz. Unfotunately it can't be used live during measurement.
In this case do You Gentlemen know soft that can record data AND calculate live at least 25k lines for spectrum 0-40kHz ?

Huge thanks for any help.

 

[GregLocock]

Haven't tried it but matlab should give you a decent shot at that processing speed... but trying to do that while taking the data is probably a bit too hard. What are you going to do with these numbers once you have calculated them? write them to disk? If so you may be able to do it on a multi cpu PC, but I bet you run into sync/timing problems.





Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[Warrk]

Frequency spectrum is only needed to monitor tests, so it doesn't have to be saved during measurement(timing problems can be eliminated by usage of 3 HDD's 1-system+soft,2-acqured data,3(optional)-FFT results). As You mentioned multi (4)CPU PC is reqiured for this purpose (already got it).
Best compromise I've found as 65536 spectral lines that works live during measurement(SpectraLAB 4.32-supports also post procesing).

Instead of using dedicated data aqusition unit Creative EMU 1616M sound card (192kHz/24bit) can be used(about500$), never tried it before (vibrometer voltage range is +-10 and sound card -+0,4V) but with voltage converter should work fine.... (I hope).

Huge thanks for advices and remarks Greg.

 

[GregLocock]

In my experience the most useful thing is real time waveforms, but it is very nice to watch an exponentially averaged spectrum at the same time. What I'd actually do is stream the data to disc, unprocessed, and then in a separate process read back the latest data and analyse that. You might like to contact Prosig, just thinking about it as that is what we use for this sort of job.



Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.