I remember reading an AES paper a while back. The calculations somewhat suggests that perfect CD may not be the best solution. Bear in mind that we want IID to change in proportion with ITD, however, it seems that delay change is proportional to distance, but not intensity is somewhat proportional to the square of distance. So the pattern depends on how you desire to toe them in.
Hello,
Constant directivity is a strange thing. Not that it is something extraordinary. Instead it is very easy to achieve over a wide bandwidth.
Here's a typical 2-way bookshelf speaker with 6.5" bass-midrange with 1" dome tweeter. Response curves are 0 deg to 90 deg horisontal.
From 100Hz to 10kHz this simple speaker is having more constant directivity than some of the complex speakers posted in this thread
- Elias
Constant directivity is a strange thing. Not that it is something extraordinary. Instead it is very easy to achieve over a wide bandwidth.
Here's a typical 2-way bookshelf speaker with 6.5" bass-midrange with 1" dome tweeter. Response curves are 0 deg to 90 deg horisontal.
An externally hosted image should be here but it was not working when we last tested it.
From 100Hz to 10kHz this simple speaker is having more constant directivity than some of the complex speakers posted in this thread
- Elias
LOL
we possibly had *the whole discussion* not only once !
So – I'm pretty confident that we will arrive at roughly the same conclusions as before *if* we go through all that again (and again, and again, and ...).
This whole "CD *must* have falling slope" thing is just another mantra, some like to sell as physic's law – like some other such mantras we already came across around here...
Michael
May I suggest - we'd better might look at the topic the other way around !
So
– who's to explain where *exactly* those CD behaviour can be found in the (new) SUMMA polar plot below ?
Summa
An externally hosted image should be here but it was not working when we last tested it.
Just overlay a high lighting frame on the CD area in that plot - and then re-post again - *please*
Michael
So - no one to solve such a simple task ???
Michael
The data you show is obviously highly smoothed. It does not reflect how typical bookshelf 2-way boxes behave. Here's a more realistic plot (although one-third octave smoothed!):
Last edited:
You are obviously wrong. My data is 1/3 octave wavelet smoothed. Your plot shows only a poorly executed design of a 2-way box.
- Elias
- Elias
A 70 dB scale,lots of smoothing and a minimum number of curves will make any loudspeaker look good.
But its not just CD that is desired it highly directional CD.
Venture an opinion on how any of this would perform? Just wildly curious. Inspired by the ER18RNX, the MCM H-65 waveguide curves, and perhaps a bic pen. I'm having fun with Sketchup lately and am getting carried away.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
Michael,
I would call the red area above "CD". The 90° response is only 5 dB off from the 15° response (to which the curves are normalized IMHO) at the right end of the area. I believe that's not shabby at all.
The light red area would be "CD" for me too - up to about 50°.
I know that you would ask about that notorious 3 kHz on-axis dip next. Let me answer with the response of two "CD" devices which are NOT waveguides, but dipoles:
First my own two Dayton ND20FA tweeters mounted back-to-back. Response is taken at 15° intervals up to 90° (red) and normalized to the power response (green line):
Second "cuibono" with his baffleless B&G Neo3 and TB midrange below:
In both cases you can see CD up to the on-axis-dip and beaming above that dip. The frequency of the dip is a function of the dipole length. I can show you horizontal measurements of a non-symmetrical setup for the Daytons with NO sign of that dip at all - but the vertical measurements are really bad.
With dipoles you can do a lot more experiments with CD behaviour than with WGs - its just another piece of baffle (or no baffle at all) against building a whole new waveguide. And my guess after those experiments, measurements and simulations is:
That dip is not a fault of an individual driver or baffle contour - it must be an integral part of a CD device IMHO. I know how to move that dip above 15 kHz in a dipole, but that would call for something like a real point source - useless because of the total lack of power. I know how to make that dip disappear, but it will always result in a severe loss of CD in the vicinity.
Since the dip vanishes with going off-axis, it is obviously the result of symmetry. I wonder, if Earl could give an easy-to-understand explanation (no mathematical formula please
) where it derives from.Rudolf
Attachments
"CD" does not have a standardized definition, which is what allows people to define it at will. No loudspeaker could ever be completely CD across its entire bandwidth unless it was a pure omni and even that would be a design challenge.
"CD" is a target, some speakers get closer than others. And I need to point out once again that when talking about CD we need to also define the design angle intent.
Not all the figures come through on my browser, so I am not sure what you are refering to. I would suggest that if you want me to look at data, that it be only a 50 dB scale and no more. Its simply too hard to tell anything on these huge scales that are often shown.
As too someone else's rudness being my problem, wow that's arrogant! People like that do not deserve a response.
"CD" is a target, some speakers get closer than others. And I need to point out once again that when talking about CD we need to also define the design angle intent.
Not all the figures come through on my browser, so I am not sure what you are refering to. I would suggest that if you want me to look at data, that it be only a 50 dB scale and no more. Its simply too hard to tell anything on these huge scales that are often shown.
As too someone else's rudness being my problem, wow that's arrogant! People like that do not deserve a response.
Then the dip that you are refering to is the axial dip on my waveguides? I've explained this many times, but I'll do it again.
No matter how gradual the trnsition from waveguide to baffle there will be a diffraction at this edge. In the Abbey two things happen at the same frequency. The first is the diffraction and the second is a standing wave across the mouth caused by this diffraction. These two things will always occur. Now at some frequency the diffraction from the edge will cause a cancellation directly on axis since the distances from the edge to the axial point are exactly equal only precisely on-axis. This is why it goes away off axis and why you don't see it in non-circular mouths either. In the Abbey the standing wave across the mouth also happens to have an effect directly on-axis and in the Abbey this coccurs at the same frequency as the diffraction hole. This is why the Abbey is a worst case for this axial hole. In all the other waveguides the two effects occur at different frequencies.
This axial hole will also not be seen in any waveguide that has HOM. Thats because the HOM will fill in this hole since they are at different phases than the primary mode. Only a highly coherent single mode wavefront will exhibit this hole. Its like the center spot in the shdow zone of an optically illuminated circular disk. Unless the light is coherent this spot will not be evident.
No matter how gradual the trnsition from waveguide to baffle there will be a diffraction at this edge. In the Abbey two things happen at the same frequency. The first is the diffraction and the second is a standing wave across the mouth caused by this diffraction. These two things will always occur. Now at some frequency the diffraction from the edge will cause a cancellation directly on axis since the distances from the edge to the axial point are exactly equal only precisely on-axis. This is why it goes away off axis and why you don't see it in non-circular mouths either. In the Abbey the standing wave across the mouth also happens to have an effect directly on-axis and in the Abbey this coccurs at the same frequency as the diffraction hole. This is why the Abbey is a worst case for this axial hole. In all the other waveguides the two effects occur at different frequencies.
This axial hole will also not be seen in any waveguide that has HOM. Thats because the HOM will fill in this hole since they are at different phases than the primary mode. Only a highly coherent single mode wavefront will exhibit this hole. Its like the center spot in the shdow zone of an optically illuminated circular disk. Unless the light is coherent this spot will not be evident.
Well, then most 2-way speakers are poorly executed designs. What make and model is the speaker you showed?
That's plain wrong - as I've shown....
##################
Well spoken *in the end*, Earl
Your hindsight - assuming you feel yourself included in that bunch of people that define CD applying to their design "at will" - being actually the end of any discussion in that department IMO....
Michael
Rudolf, I've seen your beautiful work on back to back tweeters .
The point on your highlighted areas is that there is - and must be - some tolerance to what we call CD.
Nothing gets "more true or the *one* truth" on earth by simply preaching / repeating any of these silly mantras. ("IF the axial response does not fall then the device cannot be CD, its that simple." - that is)
Michael
The point on your highlighted areas is that there is - and must be - some tolerance to what we call CD.
Nothing gets "more true or the *one* truth" on earth by simply preaching / repeating any of these silly mantras. ("IF the axial response does not fall then the device cannot be CD, its that simple." - that is)
Michael