Spectrometer bands

Posted by: maczrool

Spectrometer bands - 05/09/2002 19:29

What frequency does each band of the various spectrometer visuals represent? Anyone know?

Stu
Posted by: tfabris

Re: Spectrometer bands - 06/09/2002 09:03

This isn't much information, but what I do know is: Some are linear divisions across the audible frequency spectrum (32 bands=audible spectrum divided by 32), others are logarithmically weighted (the ones with "Log" in their name).
Posted by: maczrool

Re: Spectrometer bands - 12/09/2002 15:22

Thanks, I kind of figured that much . I guess then I would need to make a series of test tones at the low and high end to determine what the bandwidth of the spectrometer is and then divide by the number of bands.

Stu
Posted by: schofiel

Re: Spectrometer bands - 16/09/2002 13:24

Well, I ran a 20 - 20 kHz sweep test track on the spectro visuals, and it showed a pretty sharp peak across the whole display. I then used CoolEdit to create a track which stepped at known frequencies up the spectrum and marked them on a paper overlay. They were pretty good: if only we could create a background scale image (a la AnnaVU) to allow us to annotate the spectrum diagram so it could be used for reasonable calibration purposes. Whaddaya think, Toby?
Posted by: maczrool

Re: Spectrometer bands - 16/09/2002 13:50

So would you say the first band starts at 20 Hz and the last finishes at 20 kHz? I have sweep tones, but I never know quite when they start and finish, so I was considering using discrete tones instead.

Stu
Posted by: schofiel

Re: Spectrometer bands - 17/09/2002 01:54

Well, the sweep used to start at 20, and held for one second before carrying on, and I could clearly see it down at the bottom of the scale. However, I don't really know what the scale is meant to be displaying. Toby indicated that it was not really a serious display, just for fun, and you couldn't count on the calibration of the scale. Hence, you would be better off by determining which band centres you would actually like to use, then creating a sequence of tones of say 1-2 second duration. This would let you calibrate the frequency scale, at least. This was my intent, so that if we could place an image behind the scale, we could see specific frequency markers.

It's interesting to note that the HSX109 has a spectrograph visual, with a calibrated frequency scale. It's supposed to use a common code base with the car - the only major difference being the display resolution (which would make the text graphics difficult). How about it, Toby?