I agree with what you're doing. The problem is the distortion plot. It will only show you non-linear distortion. You may see resonances on response plots, but it has been shown to be difficult to measure cabinet wall contributions.
Interesting thought... for example if the cabinet is acting like a passive radiator, and vibrating linearly across all frequencies it may not necessarily show up as distortion, it would just increase the output of the driver(s) linearly across the full bandwidth measured. (Perhaps it would show up as a long decay time, on a waterfall plot?) That is definitely somehing to keep in mind. (and maybe that kind of result would not really be noticed except as a smearing in time of a given frequency)
My initial hypothesis was that a panel resonance would be limited to a specific frequency(ies), whatever is inherent to the material and thickness being used, such a resonance(s) would not square up with the input signal and therefore would show up as one or more non-linear distortion peak(s) somewhere along the bandwidth being evaluated. (if it is audible at all) To be clear, I have not done this comparison myself, I am theorizing based on other posts I have read, and my limited experience with REW.
Oh, for the days where wood, a hand-saw, some glue and a Full Range driver were enough!
Six - Mpls.
My initial hypothesis was that a panel resonance would be limited to a specific frequency(ies), whatever is inherent to the material and thickness being used, such a resonance(s) would not square up with the input signal and therefore would show up as one or more non-linear distortion peak(s) somewhere along the bandwidth being evaluated. (if it is audible at all) To be clear, I have not done this comparison myself, I am theorizing based on other posts I have read, and my limited experience with REW.
Oh, for the days where wood, a hand-saw, some glue and a Full Range driver were enough!
Six - Mpls.
Last edited:
In what way does the cabinet produce distortion and what should it indicate?
I think it can be reasonably taken that any measurable output from anything other than the drive unit itself can be classed and quantified as distortion. I would suggest that the output from the driver suspended in free space could be compared with its output when fixed into an enclosure and the results compared. Analysing the sources and magnitudes of the 'unwanted' output would likely be the result of months' or years' research! I could fairly safely speculate about the origin of some of the sources, but as for the relative magnitude compared with the cone's output...
Last edited:
Yes, distortion and there are different kinds but often the word is used to describe harmonic distortion, and this can cause confusion. The rew measurement described earlier is designed to look at harmonic distortion. Primarily, the cabinet wall distortion is not harmonic distortion..
When a response plot doesn't look linear, this is not non-linear distortion. Non-linear distortion is an issue with level vs time, not level vs frequency.
When a response plot doesn't look linear, this is not non-linear distortion. Non-linear distortion is an issue with level vs time, not level vs frequency.

Allright, this is the final (outside) measurement at 60 cm between mid and tweeter.
At home I measured in DIRAC and the final result is absolutely stunning.
This is the inroom DIRAC response and correcte curve.

Thanks for all your help and advice, could not have done it without you.
Even if I had to frequently redirect attention from theoretics to practical advice ;-).