[Sylberg]

From my understanding and readings, the inpulse response is use to calculate all the other measurements made by REW. So quality needs to be good.
Is in normal to get a impulse at -30db FS max while my measuring level was showing a -5dB headroom during the test.
The %FS becomes a poor +/-1% signal that I cannot be sure I can trust.
Is there any full bandwidth typical measurement that someone can share with that I can compare with?

 

[glaufman]

Posting your graphs will make it easier to tell. Here's some full-range I posed with impulse and waterfall a few years ago.

http://www.hometheatershack.com/forums/rew-forum/9469-foray-into-rew-questions-galore.html

 

[JohnM]

Is in normal to get a impulse at -30db FS max while my measuring level was showing a -5dB headroom during the test.

You cannot directly compare the levels during acquisition with the level of the derived impulse response. To see the quality of the impulse response make sure you plot it normalised and look at where that places the noise floor.

 

[Sylberg]

Thanks Greg. This helps a lot to compare with my own measurements. I finally found the 1/3 octave-smoothing filter.
here are a sample of my graph.
Other than optimizing the frequency response, I want to understand where reflections are coming from and the associated delays. I would also like to measure the pseudo- anechoic frequency response of the speaker (the impulse components before the first reflection). To make this I need to be sure that I have the Impulse Response measurements right.

First is the frequency response at 75dB on SPL (system calibrated with a SPL)
Second is the Impulse response in dB-FS. It gives a -30dB signal
Third is the %-FS with a poor 5% span.

 

[JohnM]

That measurement is fine, the noise is down almost 60dB. Select the check box to plot the IR normalised.