https://www.diyaudio.com/community/...-array-28-tc9fd18.349356/page-13#post-6590599
Job done... limited vertical directivity and wide horizontal directivity.
A WWmmmmTmmmmWW, but created with active DSP
Loosely based on the work we did in nc535's thread...
That is very good looking response, it is very hard problem to get wide horizontal and that much more narrower vertical for wide bandwidth.
DSP enables lot of systems that wouldn't work passive and I see no point restricting oneself to passive crossovers in year 2022. Or why not if the concept is sufficiently simple, just don't expect miracles with passives, DSP is the miracle that enables any crossover and quite extreme loudspeaker concepts. I like to think crossovers can be assumed trivial these days with VituixCAD and allowing oneself to use DSP in the first place.
One would have to check at all important listening distances if it all holds up, but I think it has a lot of potential,
as does bbutterfield's fractal array concept... I'm already convinced of the advantages of the limited vertical dispersion.
As long as standing output still resembles the balance of the design (seated) listening axis. And these concepts including
my revised array do that even better than most multi-way speakers..
I'd expect more of that to come in future speaker concepts...
as does bbutterfield's fractal array concept... I'm already convinced of the advantages of the limited vertical dispersion.
As long as standing output still resembles the balance of the design (seated) listening axis. And these concepts including
my revised array do that even better than most multi-way speakers..
I'd expect more of that to come in future speaker concepts...
^I've got question regarding this for you who have had line arrays and limited vertical directivity concepts alike: does the narrow vertical pattern matter mucho when DI and power response are smooth as butter, "predicted inroom response" is similar as on axis response? Even if tweeter went bit wider response at top octave than array on the midrange, how much would it really matter if one listens standing up? I guess this is highly dependent on the room and all but I suppose the sound is nice anywhere in the room and just comes more into focus at listening spot? Or is this the spot where the nice graphs break out from reality, even with nice smooth predicted in room response in simulator the response is not nice in reality?
I guess this would be typical for MTM configurations with waveguide tweeter. Midrange has wider directivity on horizontal plane than the waveguide, but narrower in vertical plane giving somewhat flat DI even though directivity is not constant on either plane. But how does this kind of setup sound? This is just a simulation.
I guess this would be typical for MTM configurations with waveguide tweeter. Midrange has wider directivity on horizontal plane than the waveguide, but narrower in vertical plane giving somewhat flat DI even though directivity is not constant on either plane. But how does this kind of setup sound? This is just a simulation.
Attachments
Last edited:
Let me disagree: PA line arrays wouldn't be used if it was so difficult to implement. But how i see things is from where freq range it does matter to have control?
For horizontal it seems 1khz vicinity ( lower=better) is the place where control should happen for an average ( one octave lower or even more is probably even 'better').
For vertical i wonder. There is things to be gained if floor/ceiling bounce could be mitigated so maybe from circa 200hz and up.
For dsp/passive ( i think we should rather say FIR/IIR) i don't think this is such black and white. Both approach have pro and cons in my view.
I don't know the answer to that question but I am now convinced that it is the high vertical directivity that makes the line array work well over a wide variety of recordings and genres where some other concepts are intolerable with some music.
I think these posts from kimmosto sum it up well in terms of options to aim for and where some problems can be
https://www.audiosciencereview.com/forum/index.php?threads/some-help-with-lobing.22661/post-755427
https://www.audiosciencereview.com/forum/index.php?threads/some-help-with-lobing.22661/post-753404
^^ Good point with PA line arrays 😀 I guess I was thinking only with the dome tweeter and not so bulky home concept, totally missed this one 😀
.
And yes passives can work if the application is so, but some systems just need the added flexibility and I see no point limiting oneself to passives and hope they'd work and when they don't just use DSP, not too expensive today. Besides IIR or FIR filters there is also delay what comes in handy, can't do easily with passives whereas with DSP no problem, for example in this particular topic delays would allow flat baffle.
.
And yes passives can work if the application is so, but some systems just need the added flexibility and I see no point limiting oneself to passives and hope they'd work and when they don't just use DSP, not too expensive today. Besides IIR or FIR filters there is also delay what comes in handy, can't do easily with passives whereas with DSP no problem, for example in this particular topic delays would allow flat baffle.
Look at predicted in-room response with floor/ceiling and wall reflections. One by one to give you an idea...
Smooth DI should bring well balanced in-room sound. If it deviates or if it's different from tweeter to midrange than it becomes much harder to EQ (or find) the pleasing tonal balance.
Smooth but sloped DI is still easier to listen to than varying directivity in certain frequency area's. While EQ or simply the balance used might work well, I wouldn't be surprised
if it's success varied among different musical genre's. Where a smoother DI would probably do every thing quite well.
But in general, we listen to a lot of indirect sound from the room that messes up that perfect frequency response we've created. The more that room response follows
the direct response, the more balanced it will sound in more than one spot in the room.
However, I am not a fan of wide dispersion to create the illusion of a wide stage like some Toole followers advertise. As that to me is a gimmick that gets old over time...
Same sauce over everything. That's why I absorb early reflections and add ambience later. For my favorite genre's, it just works better (pop/rock/studio work + life recordings)
If I were a fan of classical music it might have been different.
Disclaimer: if I had a bigger listening space, I could have done things passively instead of using extra ambience speakers. Like well placed diffusion etc.
You are not far away from the 'quasi transient perfect filters' school of thinking that was active in France 10/15y ago. Jmlc started the dance followed by many others.
He had a spreadsheet where you could play with filter type/order/ freq/ delay or physical offset. I will try to find it.
I find interesting you come to same things from another perspective.
We didn't comment your sim, but i like the wider one better ( last one). I think the smoothest profile is important, and part of why the XA have great review.
To bad @SpeakerDave is not active here this much since he left loudspeaker design for another professional path.
I would really like to listen to his pov.
We mostly agree but i get why you don't want the ergonomic issues which can come with multiamped system.
Most can be solved: you don't need this much power when using a lot of membrane area and a 2u rack could handle a bunch of classd module, a racked 1U soundcard and a laptop. Add an Spl vol8 if you want a real volume knob/mute. Let's say 0,5m2.
Less space and weight than my pair of SL-1200mk2.
But it have a cost, there will be cables,... after living with a multiamped system for more than 10years i find simplicity appealing.
For instance, here's a picture of the Snell Acoustics XA Reference, designed by @SpeakerDave. The crossover point is 250Hz for the woofers. The slope is something like 18dB per octave. This will tend to introduce a delay of about 3ms - or about forty inches!!! In the real world, the actual delay will depend a lot on the passive crossover configuration. Theoretically, a third order crossover introduces 270 degrees of delay (90 degrees for each order.) The net effect is that the woofer is delayed by about 3ms. (Sound travels 13.5 inches in one second, 250Hz is fifty four inches long, 75% of fifty four inches is 40.5 inches.)
Hi Patrick,
hey, crossovers simply don't work like that.....an 18dB/oct xover at 250Hz won't begin to introduce 3ms delay.
The electrical high-pass side won't measure any delay from a straight linear signal, if the high pass side is measured separately.
And the electrical low pass side will measure about 1.25ms delayed relative to straight signal.
But even that 1.25ms delay is misleading, because no delay between the two sides is what is needed, to have the two sides sum back to straight unfiltered linear response.
After tinkering for a while in Vituixcad I began to realize that the baffle needs to CONVERGE in order to produce a flat wavefront. This probably seems counter-intuitive; the Snell baffle is flat. Why would it generate a curved wavefront? And the answer is in the passive crossover. A third order crossover at 250Hz (like in the Snell XA Reference) produces about 40.5 inches in delay! Due to the fact that the tweeter in the center of the array isn't delayed remotely as much by the crossover, the wavefront of the array is significantly curved. The wavefront isn't curved physically; it's curved by the crossover.
With that knowledge in hand, I began to tinker with WMTMW arrays, and manipulating the curvature of the loudspeaker baffle, to achieve beamwidths which wouldn't be otherwise possible.
Also must say I don't think the idea of expanding, straight, or converging arrays will effect ultimate vertical pattern control coverage angle.
I think those effect the apparent vertical acoustic center that's perceived at different listening heights, but not the overall vertical pattern so much.
Or at least that is what my beam steering experiments have led me to think......experiments with line arrays using 24 TC9's, and 8 amp channels/DSP, running 3 drivers a piece for creating concave, convex, and straight line delay timings
I'm only converging the array to get a desired wavefront.
In the Keele CBTs, the wavefront shape is a fraction of the curvature. For instance, a 45 degree convex CBT will get you a wavefront that's about 30 degrees.
The reason that the Keele CBTs produce a curved wavefront is because the baffle is curved and there's no inductors or caps in the crossover.
Once you start adding inductors and caps, they delay your signal.
So one option is to go active and use multiple DSP channels.
But the thought I had, is that you could manipulate the DEPTH of the drivers to compensate for the delay introduced by the xover. And when you do this, you wind up with a concave baffle.
So it LOOKS like it's focused on one spot in the room, but it can generate any wavefront you want, within reason. I'll probably aim for a flat wavefront, but you could also generate a concave or convex wavefront if you wanted.
I did the math in my head and I was off by something like 300% - 400%
My original idea looked like this:
But as I began to work out the details in VituixCad, I found the delays were fairly subtle:
But there seems to be merit to the idea; this concave array has beamwidth that's more consistent than a flat baffle.
If I was willing to live with a passive crossover that had fifty components per side I could simply do a WMTMW like the Snell eXpanding Array. The approach that I'm going for uses a series of cascaded low-pass filters. The upside of going that route is that it reduces the parts count of the crossover by about 40%, because there's no high pass on the midbasses. This also increases the power handling by a LOT, because all the midbasses are working in tandem below about 250Hz. So it's almost like having the displacement of a 10" subwoofer, but distributed vertically (which should reduce room modes.) The downside is that this MUST be a hybrid loudspeaker, it requires an electronic filter to reduce the hump in the low frequencies. But I'm cool with that, a $100 miniDSP is cheaper than six passive high pass filters.
🤔
Looks like the preferred speaker was the Salon 2 and has a pretty wide response vertically as well as a very wide horizontal response. (..not to surprising given the small 4" mid. (down to 450 Hz) and steep crossover filters.)
Makes me wonder if something closer to OMNI wouldn't have the highest preference (..and preferably point-source at that)..
I understand what you want to do, but from what I know of acoustics sound won't conform to your drawing. I think you want the effect of a phase plug by creating equal path lengths (through delay or filters) to create a flat wavefront, which could occur at a defined point but it will not always continue to be flat beyond that or before that point.
Ideal beamwidth is dependent on application. In a different room the panel would have likely come to a different conclusion.