From intuition one might believe that a perfectly radiating full range source on a wide baffle would be the ideal dipole. Legions of wide range drivers mounted on barn door sized baffles seem to prove that. The contrary is true. The width of your baffle has to follow the radiated frequency range. Wide baffle for low frequencies, narrow baffle for high ones. Let me explain:
As a proof of concept let's first look at the Boxsim simulation of a 50x50 cm square baffle with a 1'' point source in the center. The picture shows the radiation pattern on axis (black), at 30? (red) and at 60? (green). You can perhaps recognize how it resembles Linkwitz' mathematical dipole radiation model in
http://www.linkwitzlab.com/models.htm#A
Boxsim will even show the same polar distribution at 1130 Hz as Linkwitz had sketched here.
http://www.linkwitzlab.com/images/graphics/2pt-src2.gif
You certainly will agree with me that this is no radiation pattern one would want to have.
For a second look let's take a baffle of 100x40 cm with the same 1'' source in the middle at 80 cm height:
It still looks usable up to 600 Hz, but not above. And the polar response still has bad patterns, for instance now at 1409 Hz:
Now you will tell me that nobody will use a 1'' full range source in real life. How about an 8'' driver on that baffle:
Ouch. An 8'' driver doesn't radiate to the side above 1 kHz. We all know that large drivers don't radiate with their full cone at treble frequencies, but with the inner part only. How about an active cone diameter of 4'', like a big wizzer cone:
This isn't really better than our 1'' source. We are stuck. We need to reduce the baffle width to get the range above 1 kHz right. We try the 4'' cone on a 4'' wide baffle:
This looks really good up to 3 kHz. From there on we obviously need a smaller driver. Let's change the 4'' cone for a 1'' dome on the same baffle as before:
And again: At 5 kHz the baffle is too wide for the 1'' dome. We need an even narrower baffle. We try 2'':
This is as good as it gets. At least if you want symmetry in your radiation pattern.
If you dare, go back to the 4'' wide baffle and lift your 1'' dome to 95 cm height and move it to 2 cm from the left baffle edge. The radiation to the left side will be better than expected (compare to the same dome on the same baffle width in picture8 ) :
But radiation to the right side now is a mess (see right diagram).
OK, if you don't bother about off axis radiation, this would be the point to leave the discussion. For all others I would like to add some comments:
When you go through the diagrams again, you will find that problems always begin ABOVE the first dipole peak of the baffle - never below. You may have noticed that some very knowledgeable dipole designers take pains to keep the working range of their drivers below the corresponding baffle peak.
As we all know, the upper usable frequency limit of a driver is determined by its cone size. As a rule of thumb the baffle width should not exceed 2.2 times the cone diameter to keep the driver below the dipole peak.
All diagrams shown and all rules mentioned only apply as long as a driver radiates fully into half space. Most tweeters fail to do that at the highest frequencies - they don't ''see'' the width of the baffle. So you can take some liberties with the baffle width at the top of the frequency range.
I have concentrated on horizontal dispersion. I know that the vertical radiation pattern is important too, but discussing both at the same time would have been disturbing somehow.
Hopefully the horizontal facts alone are disturbing enough.
Addendum: All sims were done with Boxsim
www.boxsim.de, the simulation program for Visaton drivers and an ''ideal driver'' model (symmetrical radiation to front and back, no cone breakup, linear frequency response and impedance from 0 Hz to infinity). Simulation is for anechoic conditions (''mid air''). All SPL levels are relative and have no connection to real drivers.
