Waterfalls for "early reflection" investigation

[bjs]

Can REW do this?

I wanted to generate a waterfall of the first 25ms (or so) and examine internal and early reflections of my full range speakers. But REW seems to limit me to a window length minimum of 50ms and a time span minimum of 200ms (!). Is there a reason why these limits exist versus letting the user select values of their choice? Assuming REW generates the analytic function correctly I wouldn't think so (?).

Also (since I'm here...) does the Waterfall plot do it's own windowing on a unwindowed impulse? Or is the Waterfall windowing done after the impulse response windowing as been applied?

Thanks.

[Bob_99]

Sorry! Should have read the question more carefully and my answer obviously was not relative to your question.



Bob

[JohnM]

bjs wrote:

I wanted to generate a waterfall of the first 25ms (or so) and examine internal and early reflections of my full range speakers. But REW seems to limit me to a window length minimum of 50ms and a time span minimum of 200ms (!). Is there a reason why these limits exist versus letting the user select values of their choice? Assuming REW generates the analytic function correctly I wouldn't think so (?).

Those limits are from when the waterfall was restricted to low frequencies, I've since reduced them to 10ms min window and 10ms min span for the next release. Will those limits be suitable?

Also (since I'm here...) does the Waterfall plot do it's own windowing on a unwindowed impulse? Or is the Waterfall windowing done after the impulse response windowing as been applied?

The waterfall does its own windowing.

[bjs]

JohnM wrote:

...I've since reduced them to 10ms min window and 10ms min span for the next release...

Cool!

10ms minimum time span works for me.

Suggest going all the way down to 1ms window for the tweeter resonances....still plenty of frequency resolution for the high stuff and things decay much quicker up there. Also 10ms width will pickup room reflections too soon which may or may not be desirable depending on what one was trying to do.

Windowing rise times may become an issue...especially if the windows were all designed with subwoofers in mind...i dunno...

[bjs]

JohnM wrote:

...The waterfall does its own windowing.

OK.

Since much mid/high frequency testing of tweeters, xovers, compensation networks etc is anechoic it would be nice to also be able to pre-window the impulse to remove the room in those cases. Then do the waterfall as is done now.

This approach would also open up a second waterfall type for free. A user could initially remove the room and then set a wide waterfall window spanning all the data. The result is a "cumulative spectral decay" waterfall in addition to REW's existing "moving window" waterfall display. Both are useful in my experience.

From a user viewpoint, a little check box on the waterfall display to "Use the pre-windowed impulse" as the basis would add a lot of flexibility and power to the existing capability.

[JohnM]

I'll add that to the feature requests list.