I've suspected this for some time. It's reassuring to hear you say it. If one goes freestanding, go round, if in rectangular baffle, go rectangular.
Honestly, even down there I would expect an error of less than a dB or so, it's really not a major effect because it's a small load change in comparison with the huge EM damping. But still this is not the region where driver resonances enter the picture and cause several dB errors in the predictions. You could compensate for this is you wanted to, but I don't see the point.
The bigger issue is the plane wave assumption.
For example, the load on a direct radiator changes with frequency. If you model this effect, which I have, you will find a miniscule difference in the results when compared to a model that does not take this load into consideration (like ALL T-S modeling.). It's just not a big effect.
If this is the point, then I disagree. Rectangular offers nothing that elliptical doesn't and elliptical doesn't have the issues with the corners.
I think that what I am saying is that this is quite possible and certainly better than not using PWT data.
To go further you would need a model of the driver, which I am guessing you are reluctant to do. I did an AES paper on this decades ago, it's not brain surgery.
And, you will need to develop your PWT to measure the wavefront curvature.
With these last two things, you could create a very accurate model of any driver (once measured on the PWT) on any waveguide.
Why "abrupt"? Change in curvature matters a lot, but why must a flat baffle be more "abrupt" than free standing?
Or are we talking about different things again?
Last edited:
I think what Mabat is saying, is that any non-rectangular shape in a (ultimately rectangular) baffle will have flat surfaces between the curvature of the waveguide and the roundover of the baffle. Not sure what that would mean acoustically though.
Seems this was cleared up while I was typing
Last edited:
Quite recently I was actually informed that the brain surgery is not that hard
But thanks, this is some more food for thought. I'm only not sure it is all worth the effort.
Sure, got that.
Last edited:
Yes, exactly. These are two additional places prone to diffraction, in my opinion. You would have to gradually decrease the curvature to zero and then slowly increase it again. This would take a lot of space.
Are you mixing up "curvature" with "slope"?
The slope has to go to zero, yes. But a smooth blend curve can be created for any size baffle that goes from the design angle to zero - plane of the baffle. The bigger this curve the more gradual, the less the diffraction and reflection, so yes, bigger is better, but "huge" is not required. Sometimes you say things that are more extreme than reality.
Am I mistaken or did Fluid show that the difference between free standing and baffle were not very great?
The slope has to go to zero, yes. But a smooth blend curve can be created for any size baffle that goes from the design angle to zero - plane of the baffle. The bigger this curve the more gradual, the less the diffraction and reflection, so yes, bigger is better, but "huge" is not required. Sometimes you say things that are more extreme than reality.
Am I mistaken or did Fluid show that the difference between free standing and baffle were not very great?
Didn't Fluid basically show that a rectangular waveguide in a baffle with large roundovers behaves similarly as a free standing rectangular waveguide with rollback termination? That's not the same thing as a round waveguide in a rectangular baffle. It does seem desirable to me to design the shape of the waveguide according to how / where it will be used. I.e. rectangular if it's to end up as part of a baffle and reserve round / elliptical for free standing.
I don't have a strong opinion on this at all. But I also don't see why one should not design a waveguide to have the best custom curveature in the corners of the baffle (which effectively makes it rectangular) rather than a standard circular one with these flat surfaces in-between the two curves of the wg and baffle respectively. I don't know whether it would make much of a difference, but are you saying that in your opinion that would actually be detrimental?
No that is not my conclusion, putting a roundover straight on the edge of the horn can be better than using a baffle but if you choose good sizes for the width and depth of the enclosure the overall effect can be similar. The depth of the enclosure can be quite interactive with the width where the waveguide starts to lose control.
This is where the freestanding horn starts to pull ahead again it loses control more gracefully.
Optimizing the curve from start to finish does seem to work better than anything else.
I agree if the baffle is kept minimal it works OK, I tried a variable radius because I liked the look but it did something weird as you moved further off axis. The only thig that changed here from the linked versions above was the edge radius.
Attachments
with a plane wave tube i can see where i was wrong about length but doesn't the diameter come to bear?
with respect to that i submit this quote from a wikipedia article:
"The duct cross section dimensions are made sufficiently small compared to the wavelength at the frequencies of interest that sound can be assumed to propagate down the duct as a plane wave with no reflections from the sides."
so does that mean that to analyze up to 14KHZ the PWT's cross section or diameter needs to be .968 inches?
with respect to that i submit this quote from a wikipedia article:
"The duct cross section dimensions are made sufficiently small compared to the wavelength at the frequencies of interest that sound can be assumed to propagate down the duct as a plane wave with no reflections from the sides."
so does that mean that to analyze up to 14KHZ the PWT's cross section or diameter needs to be .968 inches?