Hi,
Graham C said:
I haven't read the article, but I think open baffles are one of the most misrepresented speaker types about
Maybe, maybe not. I find the majority of speakers to be misrepresented, simply because the engineering fails to account for application.
But to cover the behaviour of an open baffle in room briefly....
For starters we assume the open baffle is in an entierly open space (say 15m Up in the air.
First, if the wavelength of the radiated sound is small, compared to the baffle dimnsions all sound is radited out perpendicular to the baffle, nothing cancels. There is of course rear radiation, however the rear of conventional cone drivers have a natural acoustic lowpass filter formed by the magent and basket and the air trapped between basket and cone. The less skelletal and the more "oldfashioned" the driver frame is the better the attenuation of higher frequencies towards the rear.
Secondly, as the wavelength of the sound approaches the dimensions of the baffle some of the sound from the rear will fold forward and will be delayed "just so" which introduces a response peak at that frequency. If we have the driver entered in a circular baffle this peak is strongest at about 6db and fairly narrow. With a suitable design of rectangular baffle this can be kept to around 2db and over a wide bandwidth and can be used to offset some driver frequency respone aberation if present.
Thirdly, as we further increase the wavelength we observe that the rear wave progressively cancels the frontwave. This leads to the much vaunted 6db/8ve rolloff. However this rolloff, as you rightly observe covers only the baffles own action.
All of this is covered in great detail on Sigfried Linkwitz website using first order approximations, Mr. Backman accounts to fifth order effects, making his model more accurate, including even higher order effects will make the model even more accurate, but computationally more intensive.
The Drivers response is superimposed on what the baffle does. If we select a drive unit (or have one designed) whose response is more or less the inverse of that caused by baffle (this would imply a depression somewhere in the midrange and a boost towards higher and lower frequencies) we can achieve a pretty flat measured response, on axis, anechonic.
Fourth, as we now proceed to place our open baffle into a room we find that we have a number of further effects. These ARE actually covered by Siegfried Linkwith on his website, but not in the section on dipole theory, but in the section on crossovers, where they are treated as undesirable deviations from the "ideal".
I personally consider them instead as usefull allies in attaining a fairly wide bandwidth from a fairly compact dipole.
First, if our baffle is sitting on the floor, we have the floor image, which ultimatly adds theoretically 6db SPL at low frequencies, the exact amount with frequency being determined by the geometry between baffle, driver and floor.
We can use that effect conciously to extend the LF response of our speaker, in this case any anechonic measurement should be taken with the speaker placed in a halve-space envoironment (in other words on a floor that is sound reflective), easier approximated outdoors.
Second, at low frequencies our open baffle is very close to an ideal dipole, meaning it radiates as much sound towards the rear as it does to the front. In that it is actually identical to any other speaker, the main differences are that no sound is radiated at a 90 degree angle compared to the front axis and that the rear radiation is 180 degree out of phase with the front radiation.
Regardless what particlular phase or polarity the rear radiation has, when reflected forward by the wall behind the speaker it will lead to a comb filter effect with the direct sound, if we have a speaker that radiates sound omnidirectional at low frequencies we will get additional comb filtering from reflections from the side wall.
So, regardless what speaker we use, this comb filter pattern must develop, the one produced by a dipole is simply less complex and different to that of a conventional speaker. This combfilter pattern should also make clear why often moving a speaker as little as a few mm hcan have a profound effect on percieved sound.
Finally, dipoles interact differnet with room modes than monopoles (or indeed unipoles). The simple reason is that dipoles work by modulating air velocity and monopoles by modulating air pressure. As a result a monopole (traditional speaker) will excite room modes maximally when placed at room boundaries, which are pressure maxima zones, while a dipole will excite room modes maximally when placed at velocity maxima, that is right smack bang in the rooms center.
The reverse observe also holds. Thus a dipole will have the least room mode excitation when placed in a corner and a conventional speaker will have the least room mode excitation when placed in the room center.
Each of the above issues, together with actual driver behaviour determine what happens with Dipoles, making them in the end as simple and predictable as traditional speakers, however the theory and mathematical tools to predict them are still thin on the ground and where practically non existent when the actual design work on what became a commercial product in the transparence was designed.
Past that I feel that using dipole (or unipole) radiators throughout the modal range of the room has great merit and is among the possible solutions the best possible compromise (shedsize horns that retain teir directivity to below 50Hz are not practical in almost all situations).
Ciao T
