Steve, I'm with speaker dave on this one...but then I would rather put my speakers in the dead end, so make of it what you will 😉
Would you mind elaborating on your last comment vis-a-vis the boom below 50Hz?
Would you mind elaborating on your last comment vis-a-vis the boom below 50Hz?
I'd have to agree with that. Popular wisdom is that narrow baffles image better but that is over simplistic and doesn't really look into the reasons WHY that might be the case and therefore why it is only a generalisation not a hard and fast rule.
I've heard narrow baffle speakers that have terrible imaging, and also heard wide baffle speakers that have excellent pinpoint imaging.
At its heart the difference between narrow and wide baffles (ignoring incidental differences in panel vibrations/bracing, internal reflections and standing waves etc) is directivity and diffraction.
A wide baffle will maintain 2pi directivity to a lower frequency, typically an octave or so lower and change the power response into the room at low midrange frequencies for the same on axis response.
If you compare a baffle with a baffle step frequency of 250Hz vs one with a 600Hz baffle step frequency and both are equalised flat on axis, the wide baffle will have about 3dB (not 6dB) less power response in that 250-600Hz range than the narrow baffle.
This is low enough in frequency for our hearing integration time to be long enough to include room reflections (thus power response) into our perception of tonal balance, so we WILL hear a difference in tonal balance in a room between the two speakers even if their anechoic on axis response is identical.
The larger baffle will also have a 3dB better direct to reflected ratio in room over that frequency range which means in a practical room the amplitude of peaks and dips caused by room boundaries (especially the wall behind the speaker) will be significantly reduced leading to a slightly flatter response at the listening position, and I think its this which contributes to a sense of better imaging at lower frequencies with a wider baffle.
That lower midrange region is particularly difficult to deal with, its too high in frequency to use helper subs to perform modal smoothing, yet too low in frequency for our hearing mechanism to start to separate the first arrival and later reflections - they all get mushed in together and give a perception of non flat frequency response, comb filtering etc so anything we can do to reduce the summed amplitude variations at the listening position in this frequency range is beneficial IMO. In this low frequency range the wide baffle is the clear winner.
That's only half the story though. The other half of the story is diffraction.
A wide baffle will have an increased delay time between the direct arrival and the diffraction from the cabinet edge. If that delay time is kept small (well under a millisecond) then it will tend to "fuse" together into a single auditory event in our hearing system, so we hear the frequency response imbalance caused but we don't perceive it as a separate event. It sounds like one sound source to us.
In theory we could EQ the response error too, but only on one axis. Off axis any EQ would tend to make things worse, particularly at high frequencies.
On a large baffle the amplitude of the diffraction events is reduced a bit because the wave has travelled further to get there, but the time delay can now be long enough (over a millisecond or so) that it starts to be perceived as a separate event, and this is damaging for imaging beyond just the frequency response error.
As well as that the physical size of the effective radiating surface (cone plus baffle outline where diffraction radiates from) is much larger as well, and its my opinion that when the baffle gets wide enough and there is a lot of diffraction that this starts to localize the source of the sound because our brain can "triangulate" the depth of the speakers better than it can with a true point source due to having three horizontally offset and identifiable discrete sound sources. (Left baffle edge, driver, right baffle edge) We can discern angular location of sound sources in the horizontal plane down to I believe under 1 degree, so this is enough resolution for us to be able to sense that its not just one point source in space. Three points can be triangulated to a distance, one point cant. The ambiguous distance of a true point source is part of what allows the image within the recording to be perceived at distances beyond the physical speaker rather than localised at the same depth as the speaker.
So yes, a large baffle which has a lot of high frequency diffraction from the baffle edge can be located more easily than a narrow baffle which more closely approximates a point source, and imaging is not as good. With similarly poor attention to diffraction in the design a narrow baffle will tend to image better because its physically smaller and has diffraction time delays below the fusing threshold.
However if you eliminate diffraction (as much as possible) then this advantage goes away. It doesn't seem to be necessary to eliminate diffraction entirely (almost impossible anyway) but if you eliminate it from high midrange frequencies up, (say 2Khz upward) it seems to be sufficient, probably because that is the frequency range where most of the precise image location and triangulation is coming from.
One way of doing it is controlled directivity drivers - large mids and waveguide tweeters for example. If the driver itself is not radiating significant energy along the baffle surface then there won't be baffle edge diffraction even if the baffle edge is sharp. Of course in practice its best to attack the problem from as many angles as possible - driver directivity, baffle edge rounding, maybe some absorption etc.
If the end result is that the large baffle has little or no baffle diffraction coming from the edges at "high" frequencies then it will image well regardless of its larger size - the driver forms a point source and there is no secondary radiation from the baffle either before or after the fusing threshold. No baffle edge diffraction will also tend to lead to a smoother off axis polar response variation if the drivers themselves (waveguide etc) are well designed, this is beneficial to imaging stability too.
If you can make a wide baffle that has no significant diffraction at high frequencies then the wide baffle can actually image better overall due to the improvement in direct/reflected ratio in the low midrange frequencies.
Last edited:
Simon, can you help me understand why a wide baffle would have an issue with diffraction at all? Most people claim that wide baffles are most objectionable with tweeters and small mids but this doesn't make any sense. Let's say we decide to use a very wide baffle to mount the tweeter, in this case a 24 inch square which is still practical to use inside a room.
A 24" wavelength represents a frequency of roughly 550Hz and means that the baffle step really starts to occur at around 275Hz. Regardless of exactly where the baffle step is in frequency, the tweeter simply has no output below 2000Hz. So at all frequencies where the tweeter operates, the baffles acts as a perfect reflected and all frequencies produced by tweeter are equally boosted by 6db.
So this means there is no diffraction at all and that the wide baffle is actually a solution for diffraction without the need to round corners or apply wool.
What did you mean by this: "So yes, a large baffle which has a lot of high frequency diffraction from the baffle edge can be located more easily than a narrow baffle which more closely approximates a point source, and imaging is not as good. With similarly poor attention to diffraction in the design a narrow baffle will tend to image better because its physically smaller and has diffraction time delays below the fusing threshold."
I don't see how a large baffle would have any diffraction as you describe. Can you please elaborate?
A 24" wavelength represents a frequency of roughly 550Hz and means that the baffle step really starts to occur at around 275Hz. Regardless of exactly where the baffle step is in frequency, the tweeter simply has no output below 2000Hz. So at all frequencies where the tweeter operates, the baffles acts as a perfect reflected and all frequencies produced by tweeter are equally boosted by 6db.
So this means there is no diffraction at all and that the wide baffle is actually a solution for diffraction without the need to round corners or apply wool.
What did you mean by this: "So yes, a large baffle which has a lot of high frequency diffraction from the baffle edge can be located more easily than a narrow baffle which more closely approximates a point source, and imaging is not as good. With similarly poor attention to diffraction in the design a narrow baffle will tend to image better because its physically smaller and has diffraction time delays below the fusing threshold."
I don't see how a large baffle would have any diffraction as you describe. Can you please elaborate?
Sound waves aren't a "single event" but a series of events. Along the baffles surface, wave lengths smaller than the baffles width, act as a "segmented string". Edge diffraction causes the "string" to fold back on itself disrupting each of the following "segments". That "smears" the original signal to varying degrees (depending on the frequency). It might look like a "whirlpool" of sound waves at and near the baffle edges as a result. I'm sure each frequency reacts a little differently when it encounters an edge. Each frequency may have a different angle of deflection.
Last edited:
Puppet, again this makes no sense to me. You are talking about edge diffraction. However, in the example I gave of a tweeter on a wide baffle, there is simply no diffraction at all. So regardless of how many events you have, the wavelength at high frequency is not large enough to cause diffraction and all high frequencies will have a 6db gain and reflected directly into the room with perfect 2pi radiation. So what diffraction and smearing are you talking about?
Sorry for my feeble attempt at an explanation. I believe that our understanding of this parts ways when you state that there "is no diffraction at all" for a tweeter mounted as you have described. (we're not speaking of a baffle of infinite size)
Sound waves will still follow the baffles surface (regardless of wavelength) until it encounters an edge ... whether that be the drivers mounting flange and/or the end/edge of the baffle. A narrower baffle will "limit", due to size, the amount of wave content (string length) along the baffles surface. The diffraction ill effects go down proportionally. This is my understanding.
Sound waves will still follow the baffles surface (regardless of wavelength) until it encounters an edge ... whether that be the drivers mounting flange and/or the end/edge of the baffle. A narrower baffle will "limit", due to size, the amount of wave content (string length) along the baffles surface. The diffraction ill effects go down proportionally. This is my understanding.
[A]udioari, consider your example. The tweeter starts to move forward through the first quarter of its cycle. (24 x (the speed of sound in inches)) later, a compression wave reaches the edge of the baffle and starts to diffract.
You're speaking as though the soundwave never reaches the edge of the baffle, but if that were true, it would never reach your ears, either. Since the medium is a gas, sound propagates in a sphere unless constrained. The baffle constrains the wave to a hemisphere until it reaches the edge. Then all hell breaks loose. 😉
You're speaking as though the soundwave never reaches the edge of the baffle, but if that were true, it would never reach your ears, either. Since the medium is a gas, sound propagates in a sphere unless constrained. The baffle constrains the wave to a hemisphere until it reaches the edge. Then all hell breaks loose. 😉
Last edited:
Right ... and the "equator" of the hemisphere is supported by the baffle .. until the baffles surface is gone. This point is where the sound waves hemisphere edge encounters the edge of the baffle.
There is too wide and too narrow, I have found the max baffle width
(flat baffle) is 16 inchs best is 8-14 . With a contoured baffle you can play around with these numbers a bit , you can increase baffle width when contouring, but the flat baffle is best below 16 inchs.
(flat baffle) is 16 inchs best is 8-14 . With a contoured baffle you can play around with these numbers a bit , you can increase baffle width when contouring, but the flat baffle is best below 16 inchs.
The narrow baffle has 2 advantages.
First, if it's small relative to the piston size, the edge is illuminated less because of the inherent directionality of the driver. Second, the diffraction artifacts rely upon pathlength differential- the delayed diffracted energy has to have a long enough delay based upon the distance to the baffle edge to have meaningful phase differential. The narrower the baffle, the higher a frequency at which artifacts can begin, and thus there's less total diffraction.
A wider baffle avoids diffraction until the edge, and so in that respect, the long pathlength means that a lower SPL illuminates the baffle edge, to diffract and interfere with the direct wave. In some ways it's easier to damp diffraction via roundovers and baffle treatment on a wide baffle as well. Another advantage is that because the bulk of the energy remains hemispherical, you tend to be able to better match directivity, allowing for designs like Audio Note/Snell with 8" 1" combos to not be completely off the reservation in that respect.
Both have their advantages, and in either case attention to detail is key to ensure best performance. Knowing what I know now, my inclination is that larger baffles with thorough edge termination are the better choice for typical cone 'n dome speakers, both for directivity control and reduced baffle step loss.
To have this discussion without the addition of speaker placement in a room takes only a small portion of the direct to reflective additions and subtractions into effect. Yes a live end dead ended room would be nice also with all devices flush with the walls, we also then need to look at the room dimensions and parallel walls versus angular walls that are used in a studio to counter room affects and also the use of selective reflective and absorptive surfaces. The width of an enclosure must also take into affect the depth of the enclosure and the angular diffraction affects of the enclosure. Having developed many years ago absorptive baffle faces for JBL there is also that to think about. What if the surface of the front baffle is not a perfectly reflective surface but is a tuned absorptive surface? This is a very involved subject that just can't be simply stated. There are many factors affecting the localization of a loudspeaker both on axis response and reflective affects. Every baffle consideration has its own condition, whether it be 2pi, 4pi or open baffle with the comb filter effects that come with those designs. Simplification of just looking at the baffle width does not consider enough variables to be by itself realistic except for in an open space.
Wide curved baffle can do miracles KS-483 DIY archive of Kimmo Saunisto
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.
- Status
- Not open for further replies.