As long as your head is locked in a vise.
this is true...though that expression can be exaggerated in some situations
here is my truth 2031a with the IR window out too wide. you can see the ripples in the excess phase plot caused by me opening the window up. its much smoother (but not perfect ) if I window the measurement properly
Just because something is objectively true in isolation does not mean that you can extrapolate it out to determine that more of a good thing must be better. This is why it is an over simplification in the context of the perception of an entire audio system.
Isn't the minimum phase more smoothed, compared to the measured phase? It seems odd to me that it should be that smooth, looking at the amplitude response. That would also cause all the small ripples in the excess phase.
Last edited:
Is it like shown in this lab/demo? Showing cut-in of the modes.
Evanescent Modes in a Waveguide - YouTube
/örjan
Last edited:
yeah sorry you're correct. didn't recalculate the min phase response
here are the before and after. 3ms gate and 20ms I suppose this means that everything pretty much in this measurement is minphase? If i do the same thing with a room measurement of the same speaker then the excess phase has MASSIVE wraps in it
Last edited:
There's the excess phase due to the crossover (what makes the central "wrap") but that's about all I can see.
BTW, in your measured phase the pure delay is over-compensated. The right amount of compensation should render the phase curve flat above the pass band of the device (i.e. group delay should approach zero for infinite frequency). If it's bending upward, it's clearly over-compensated. It doesn't really matter for much of the analysis but it looks a bit weird.
Maybe for a better readability it makes sense to show the excess group delay, i.e. taking group delay of the excess phase.
- Now I'll have to think more about this because obviously it shows nothing interesting. It's not intuitive to me but it seems this is how it is.
BTW, in your measured phase the pure delay is over-compensated. The right amount of compensation should render the phase curve flat above the pass band of the device (i.e. group delay should approach zero for infinite frequency). If it's bending upward, it's clearly over-compensated. It doesn't really matter for much of the analysis but it looks a bit weird.
Maybe for a better readability it makes sense to show the excess group delay, i.e. taking group delay of the excess phase.
- Now I'll have to think more about this because obviously it shows nothing interesting. It's not intuitive to me but it seems this is how it is.
Last edited:
Do you mean a source that excites all the modes? The modes always exist, but get excited to a variable degree depending on the nature of the wavefront. Below cut-in they will be evanescent and decay rapidly, they will only reach the mouth above cut-in unless the device is very short.
But the contrary view to what you say is to look for more frequency content above cut-in. It's easier to find something than to look for nothing.
In this thread, since the intent is to minimize these modes, they will be very hard to find - that's the intent!
I noted the depletion of delayed signals in the impulse when foam was added to the waveguide. This could be HOMs, but it could also be other things - a reflection off of the diaphragm for example. Since all of this gets melded together into a single response, it is difficult to sort out.
What do you mean by "repeat"? You mean copy paste at different times?
Yes, good idea! These kind of software (Matlab, Octave, Mathematica, various Python packages, etc.) are just great when fiddling with physical concepts. A good option is Python + Numpy + Jupyter! That way I can share my horn code with somebody...
It looks like that for the second peak of half the amplitude:
and phase:
with the impulses given by:
I haven't properly scaled the frequencies to anything but the behavior will be qualitatively the same.
Edit: Forgot to put the phase.
Last edited: