I've built a lot of horns and waveguides. What you're describing, it sounds like a multicell horn. And multicell horns have quite a few problems, which is why you never see them any more.
There are a couple of ways to control directivity:
1) put a radiator on a waveguide
or
2) Change the shape of the radiator to control the directivity.
For instance, the reason that many planar is curved is because the curvature widens the horizontal directivity. If you wanted to control the vertical and horizontal directivity, you could make the planar convex. (That would be difficult to build)
Another option is to make the planar in rings, which is what Quad does. Basically each ring plays a different frequency band, and the crossover slope and shape can dictated the wavefront shape. This is because crossovers introduce delay, so you can curve the wavefront by having the high frequencies "lead" the midrange and low frequencies. (via the delay introduced in the xover.)
There's a thread here somewhere about a build in the shape of a dynamic driver cone. I just don't get how curving the radiating surface 'away' from the listening position is an over all solution. With this thinking it would be better to flip dynamic drivers around on the baffle. I have not heard a curved ESL panel but it seems that although a flat paneled one narrows dispersion, it also images better. I think I will conduct an experiment and turn my Monitor 3 around so the outer panels face inward just to see how that differs
l
The shape of the wavefront is defined by the shape of the loudspeaker. If you want a diverging (convex) wavefront, then you want a convex speaker. If you want a converging (concave) wavefront, then you want a concave speaker. If you want a flat wavefront, then you want a flat speaker.
Keep in mind that the curvature can be accomplished by physically curving the speaker, or by curving the speaker electronically. The latter solution is typically what most designers do; in a two-way or three-way speaker it's fairly simply to curve the wavefront using delay.
Depends on the wavelength. For instance, 5000Hz is 6.8cm long. So if you're listening to a 2.5cm tweeter, the wavelengths are so long, the tweeter dome imposes no directivity. (Because the wavelengths are longer than the tweeter itself.)
But when we're talking about planars, where the planar is bigger than the wavefront it's playing, the shape of the diaphragm plays a big role.
Okay, I'm not quite following you here. If curvature has to do with dispersion, how does delay accomplish this? I understand about time alignment as such which in the case of speakers like B&W 801, or Kef 105, it's accomplished by physically time-aligning the drivers. But what does dispersion have to do with that? I'm referring to your statement about curving the speaker electronically.
The shape of the wavefront is defined by the shape of the loudspeaker. If you want a diverging (convex) wavefront, then you want a convex speaker. If you want a converging (concave) wavefront, then you want a concave speaker. If you want a flat wavefront, then you want a flat speaker.
Keep in mind that the curvature can be accomplished by physically curving the speaker, or by curving the speaker electronically. The latter solution is typically what most designers do; in a two-way or three-way speaker it's fairly simply to curve the wavefront using delay.
Depends on the wavelength. For instance, 5000Hz is 6.8cm long. So if you're listening to a 2.5cm tweeter, the wavelengths are so long, the tweeter dome imposes no directivity. (Because the wavelengths are longer than the tweeter itself.)
But when we're talking about planars, where the planar is bigger than the wavefront it's playing, the shape of the diaphragm plays a big role.
You can physically curve a loudspeaker to get a specific wavefront, or you can electronically curve it.
For instance, this Martin Logan is curved horizontally so that the beamwidth is somewhere around 45 degrees. If the panel was flat, the beamwidth would be zero degrees, and then it would have an exceptionally narrow sweet spot.
With a loudspeaker array, you can curve the wavefront electronically using delay. This requires a LOT of DSP and amplification channels, but it's routinely done for prosound apps. As amplification and DSP gets cheaper you'll start to see it done at home. The Beolab 90 has something like 30 channels of DSP and amplification and it can vary it's beamwidth electronically.
As for the actuall practive of beamforming, it's exactly what you think it is; you change the wavefront with delay. For instance, an array that is physically flat can be electronically curved by progressively delaying the elements on the top and the bottom. You can also curve an array asymmetrically, for instance you can make the wavefront point UP or DOWN.
Hi
The electronic constructions and techniques used for controlling dispersion are collectively known as phased arrays (see wikipedia).
Yes, the most sophisticated versions do require a lot of DSP, however there are two cases where very useful improvements in the dispersion characteristics of ESLs can be achieved with passive networks.
The Quad ESL 63 and subsequent models uses an array of circular/annular segments in a flat ESL panel. Each segment behaves electrically like a capacitor (C), and is coupled to its neighbour with an inductor (L) to form an LC transmission line. This has the effect of introducing a small delay between the acoustic radiations from each segment. This is used to construct an acoustic radiation pattern that looks as though it comes from a point source - expanding spherical wavefronts. In theory, with an infinite number of such segments, it would yield a perfect 360 degree dipole dispersion - i.e. a perfect figure 8 polar response ay all frequencies. In practice, the inductors have to be quite large so Quad use only 7 segments, so the polar response is not perfect - but is a substantial improvement. Have a look at the polar plots in the Quad ESL datasheets.
The second possibility is the segmented line source in which a tall (ideally floor-to-ceiling) ESL panel is segmented as a couple of dozen vertical strips all coupled to neighbouring strips by resistors. When driven at the centre strip, this yields a cylindrical wavefront and a dispersion pattern that is near perfect up to frequencies of a couple of kilohertz, and very smooth roll-off at frequencies above that giving a perfect listening zone about 20 degrees wide (instead of 1 degree wide). Quite a number of DIYers have built them here - it is very simple compared to the quad solution and makes a very nice loudspeaker. See
Another segmented ESL . There is also an Excel spreadsheet around here to help.
Hopefully others here who collect the links can tell you were to find other links on this design.
Curved ESLs, like the martin logins, are a partial solution the the dispersion problem - they make a useful improvement to the dispersion but at the expense of audible distortion due to the asymmetry between forward and backward movements of the membrane. Still sound better than conventional speaker though
The electronic constructions and techniques used for controlling dispersion are collectively known as phased arrays (see wikipedia).
Yes, the most sophisticated versions do require a lot of DSP, however there are two cases where very useful improvements in the dispersion characteristics of ESLs can be achieved with passive networks.
The Quad ESL 63 and subsequent models uses an array of circular/annular segments in a flat ESL panel. Each segment behaves electrically like a capacitor (C), and is coupled to its neighbour with an inductor (L) to form an LC transmission line. This has the effect of introducing a small delay between the acoustic radiations from each segment. This is used to construct an acoustic radiation pattern that looks as though it comes from a point source - expanding spherical wavefronts. In theory, with an infinite number of such segments, it would yield a perfect 360 degree dipole dispersion - i.e. a perfect figure 8 polar response ay all frequencies. In practice, the inductors have to be quite large so Quad use only 7 segments, so the polar response is not perfect - but is a substantial improvement. Have a look at the polar plots in the Quad ESL datasheets.
The second possibility is the segmented line source in which a tall (ideally floor-to-ceiling) ESL panel is segmented as a couple of dozen vertical strips all coupled to neighbouring strips by resistors. When driven at the centre strip, this yields a cylindrical wavefront and a dispersion pattern that is near perfect up to frequencies of a couple of kilohertz, and very smooth roll-off at frequencies above that giving a perfect listening zone about 20 degrees wide (instead of 1 degree wide). Quite a number of DIYers have built them here - it is very simple compared to the quad solution and makes a very nice loudspeaker. See
Another segmented ESL . There is also an Excel spreadsheet around here to help.
Hopefully others here who collect the links can tell you were to find other links on this design.
Curved ESLs, like the martin logins, are a partial solution the the dispersion problem - they make a useful improvement to the dispersion but at the expense of audible distortion due to the asymmetry between forward and backward movements of the membrane. Still sound better than conventional speaker though
As I've already mentioned, dispersion with my Monitor 3 Acoustats is not a problem...at the listening position. I would not want to compromise what they do well. I sit 12' away, they are 8' apart, 4' to the front wall, 4' to the back, toed in to my ears....exactly to the millimeter. Placement geometry is perfect. I found that precise placement yields a sweet spot comparable to any other speaker I've heard. There's no head in a vice issue. But when I get up and walk away, that fabulous linearity is gone.
This is a power response problem; total acoustic energy radiated into the room has significantly different spectral balance than acoustic response at listening position. The designs with more uniform dispersion characteristics that golfnut mentioned would solve your problem. But, it sounds like you are happy with your current speakers and just looking for a way to improve things outside the listening area.
Is it more the “getting up” or the “walking away” that kills the spectral balance?
The Monitor3s are short speakers with essentially zero dispersion of highs in the vertical direction. It is possible that simply tipping the speakers back an additional 10 degrees or so might improve things to your liking when you are not in the sweet spot.
Martin Logan and JansZen both addressed this issue on their designs with shorter panels by adding a dome tweeter
that can be adjusted as needed for HF ambience. ML calls this feature “NAC” (Natural Ambience Compensation)...JansZen settled on the slightly less descriptive name “Airlayer”. In either case, the goal is to add back some of the mids and highs that are missing from the room response for non-critical listening (aka party mode) or listening from another room (aka kitchen mode). You may even find, as several reviewers did, that it improves your experience in the listening area as well.This is a power response problem; total acoustic energy radiated into the room has significantly different spectral balance than acoustic response at listening position. The designs with more uniform dispersion characteristics that golfnut mentioned would solve your problem. But, it sounds like you are happy with your current speakers and just looking for a way to improve things outside the listening area.
Is it more the “getting up” or the “walking away” that kills the spectral balance?
The Monitor3s are short speakers with essentially zero dispersion of highs in the vertical direction. It is possible that simply tipping the speakers back an additional 10 degrees or so might improve things to your liking when you are not in the sweet spot.
Martin Logan and JansZen both addressed this issue on their designs with shorter panels by adding a dome tweeter
I have them leaning back 3deg. right now which is perpendicular to my ears at 12 ft away. They sound perfect like that. However I do have the HF adjustment up a notch on the OTLs, I'm 61
If I stand up, there's not much difference. It's when I'm no longer directly in front of them that things fall apart. Depth perspective of HF is superb which I take it is attributable to the ESL's resolution. I would definitely not want to change that. I just want to hear the same thing when I'm over in the kitchen. I guess I can't have everything, eh? Mind you I'm not complaining. If I turn them up LOUD, there's no lack of anything anywhere.
Hmmmm....the first comment makes me think it is not the spectral balance that you miss in the kitchen, it is more that you miss listening in the near field where the direct sound of the loudspeakers dominants the sound coming from the room(modes and reflections). If that is that case, not sure there is much that can be done short of moving the speakers into the kitch. But your second comment makes it sound like all you have to do is turn up the volume and then you are happy in the kitchen, so am still uncertain as to what you are describing.
When I said LOUD, I meant full tilt. You're right, it's that perfect sweet spot I want everywhere. That spot is not as pronounced with dynamic drivers so more of a change when you move out of it.
Discopete
Your posts ...make it sound like your Shocket....really?....
This thing we all call {stereo image} we all hear ...when we set in front of 2 speakers...happens inside our head...not in the room!
You must know.....for 40 years..I have been in Audio bizz an this hobby...all other also would like to have the great sound an good image....every where ....or when thay stand up..or other rooms.....that comes from just 2 speakers...
good luck with that
Your posts ...make it sound like your Shocket....really?....
This thing we all call {stereo image} we all hear ...when we set in front of 2 speakers...happens inside our head...not in the room!
You must know.....for 40 years..I have been in Audio bizz an this hobby...all other also would like to have the great sound an good image....every where ....or when thay stand up..or other rooms.....that comes from just 2 speakers...
good luck with that
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.