I wonder if am understanding your point, can you expand a bit on how you see a dome midrange driver fitting in with wide horizontal and narrow vertical response?
90% of the clarity and imaging cues we enjoy so much from music is centered within the telephone pass band, so ANY and ALL phase anomalies are a smear to those spacial cues in a stereo triangle…..it’s so easily apparent when mixing with a set of speakers who’s crossovers are involved WHEN making panning decisions of instruments in a multi track recording and the more tracks, the worse the situation becomes.
Simple….an MTM alignment……exactly what Patrick has proposed from a hybrid/line source…….adequate power handling and a close center to center spacing………but not every room benefits from narrow directivity across all frequencies…..some rooms are move lively and dynamic with a bit of floor or ceiling bounce. Again……it’s always mostly about the room when making the right choice in a speaker or design.. My tapered 6” thick suspended clouds often give the acoustic impression of an open space………..vertical directivity is NOT a desired design criteria for me.
That makes more sense, The Aries M from Follgott is quite close to this although the dome mids are used at 700 to 1.8K as a narrow vertical pattern was part of the design intent.
http://www.hannover-hardcore.de/infinity_classics/!!!/Aries M Dokumentation.pdf
I still remain confused as to Patrick's intended vertical directivity pattern, that pattern sets what can be used as a centre tweeter or central elements in a shaded array.
Hey Neil, glad to see a post from you!
Come over to the dark side and try computer based DSP under Linux. You can get 10 channels of analog balanced outputs on a Behringer UMC1820, or for better performance, a MOTU Ultralite mk5. Drop me a line if you are interested.
Thanks for sharing that…..I really like the ideas presented………interesting!
Hi,
thanks for the clarification. How much have you experimented with this? propably a lot if its your day job 🙂 do you use single speaker mono setup occasionally, does the issue exist there? See this https://www.diyaudio.com/community/threads/fixing-the-stereo-phantom-center.277519/ thread, there is some discussion and speculation whether or not and how the issue with phantom center is handled in the mix and mastering stages.
This, head and stereo setup related phasing/combfilter, is pretty nasty when speakers are good and good acoustic environment with reduced early reflections makes it worse. Hense widening response, having reflections and diffraction, what ever that gives hearing system more than exactly two sources on the ~2kHz region might actually be good.
I speculate if the issue exists with single speaker mono its speaker issue like crossover. If its problem with two speaker mono its stereo / head issue instead. Or, its speaker issue then as well, speakers are too good making perfect comb filter.
Last edited:
Yes…..an Avantone Mix Cube to isolate destructive phase interference that’s just too hard to identify in a complex mix……”you don’t know what you‘ve lost til it’s gone” lol
😀 It would be very interesting if you happen to have speakers with the issue you describe and tried to use one in mono and would comment how it compares to two in mono/stereo, if the issue persists. Its often hard to pin point issues to some specific property of loudspeaker like crossover, as it could be the listening environment, or if the issue is in the speaker but not everybody hears it as the setups/environments vary etc. Studios are special places as acoustics is usually improved and its the birth for the material everyone listens to so very much would like to understand better whats the mood there. Anyway, more data would be helpful, interesting topic 🙂 Well, carry on with the current one.
The Linux approach can be a good solution for many active speakers, but I don't see it as a good solution for the line arrays that I am most interested in. The challenge I have with the line array is that I need many channels of amplification to implement electronic curvature. I think the best way to do that is to use lots of small class D amps for the midrange and tweeter arrays, plus an extra amp for the woofer. The prototype line array I made had 20 small power amps for each side, but next I want a more modular design where I can easily scale up or down by plugging 4-channel amp modules into a motherboard. That approach is best done with custom boards, and once you're in that even darker territory, it makes sense to use a DSP module vs computer DSP.
You have always been into "rolling your own"! But you could buy hardware and be up and running in a couple of hours. Just need to add amplification to suit, and as you say small class-D amps would make sense are are cheap to buy. Make changes anytime. That would probably be more expedient and get you to the finish line faster than "DSP modules" that you DIY yourself. And if you don't like how it performs you can turn around and sell the gear.
Another option you might consider, although a bit old school, is to pick up a used Crestron CNAMPX-16x60. Great quality amp. 16 channels each with 60W@8R, 90W@4R.
I don't want to completely derail this thread, so I'll answer Charlie's post by describing a multichannel amp board that could be used for building the improved arrays mentioned here 🙄....
If you go the route of many small amps to achieve electronic curvature, you don't need much power in each amp. You can get a lot of output with lots of small amps and lots of motors. I used a mix of Analog Devices SSM2518 (2W) and SSM3302 amps (15W) for my original prototype, and they do a fine job of filling a large room. What's more important is that the amps have a digital interface, so that you can eliminate a bank of DAC's and their filters. There are no small amp modules that I am aware of with I2S or TDM interfaces, so I am making my own amps. The next prototype will use the more powerful SSM3582. But I buy the DSP modules, because there is no point in making your own when you can buy ADAU1701 modules for around $10 or ADAU1466 modules for $40. I'm not into "rolling my own", but sometimes you have to...
The DSP modules have TDM outputs for connecting directly to an array of amps, and the ADAU1466 even has SRC and SPDIF inputs, so it makes a very compact solution for digital sources. The amp/DSP board for the original prototype is only about 4" by 7", and it fits nicely inside the line array. And keeping the amp inside the speaker is essential to minimize the wiring for all those drivers. The new line array board will use 4-channel amp plug-in modules with two additional analog outputs for subwoofers. I've got some prototype amp boards on my workbench and a motherboard for debugging them. The amp boards were laid out using DesignSpark, but the motherboard was done in Kicad, and the "final" amp boards will transition to Kicad. The design files will be posted, just like the other projects I have posted on Audiodevelopers.
With SigmaStudio you can quickly implement the DSP functions using nice graphical tools--that's easy. What's hard is all the code to control the DSP in real time from a cell phone app 🙂.
If you go the route of many small amps to achieve electronic curvature, you don't need much power in each amp. You can get a lot of output with lots of small amps and lots of motors. I used a mix of Analog Devices SSM2518 (2W) and SSM3302 amps (15W) for my original prototype, and they do a fine job of filling a large room. What's more important is that the amps have a digital interface, so that you can eliminate a bank of DAC's and their filters. There are no small amp modules that I am aware of with I2S or TDM interfaces, so I am making my own amps. The next prototype will use the more powerful SSM3582. But I buy the DSP modules, because there is no point in making your own when you can buy ADAU1701 modules for around $10 or ADAU1466 modules for $40. I'm not into "rolling my own", but sometimes you have to...
The DSP modules have TDM outputs for connecting directly to an array of amps, and the ADAU1466 even has SRC and SPDIF inputs, so it makes a very compact solution for digital sources. The amp/DSP board for the original prototype is only about 4" by 7", and it fits nicely inside the line array. And keeping the amp inside the speaker is essential to minimize the wiring for all those drivers. The new line array board will use 4-channel amp plug-in modules with two additional analog outputs for subwoofers. I've got some prototype amp boards on my workbench and a motherboard for debugging them. The amp boards were laid out using DesignSpark, but the motherboard was done in Kicad, and the "final" amp boards will transition to Kicad. The design files will be posted, just like the other projects I have posted on Audiodevelopers.
With SigmaStudio you can quickly implement the DSP functions using nice graphical tools--that's easy. What's hard is all the code to control the DSP in real time from a cell phone app 🙂.
I love Follgott's stuff, he's been a big influence on me.
I'm not keen on doing a WMTMW like his latest two designs, or a WMTMW like Dave Smith's designs.
I believe it's possible to achieve something comparable using a series of passive lowpass filters on the driver elements, and then a single active high pass on all of the elements in the array.
A lot of what I'm doing is to make the array passive, to keep amplifier channels down, and avoiding bandpass filters, to keep cost and complexity down.
I'm not keen on doing a WMTMW like his latest two designs, or a WMTMW like Dave Smith's designs.
I believe it's possible to achieve something comparable using a series of passive lowpass filters on the driver elements, and then a single active high pass on all of the elements in the array.
A lot of what I'm doing is to make the array passive, to keep amplifier channels down, and avoiding bandpass filters, to keep cost and complexity down.
Today I recalled John Krutke (Zaph) had written years ago that the main reason he developed the ZA14W08 woofer is because he wanted to do an MMTMM to control vertical directivity, but all the drivers he wanted to use were just too expensive, or something to that effect.
He has the data for the MMTMM 2.5 way he developed for Madisound about halfway down this page http://zaphaudio.com/ZA5/ in case you haven't seen it.
Excerpt from the page; '
The MMTMM 2.5-way format is all about controlled vertical directivity, and it's the reason I've been wanting to do one for a long time. Directivity is important overall, as is power response and other related issues. But different issues arise when considering vertical and horizontal directivity independently. Common CD (constant directivity) designs use a waveguide high frequency element and a traditional cone driven low frequency element. While the benefits higher in frequency are obvious, lower in frequency the system is just a point source. Good or bad, a point source is the type of sound radiation that is most affected by the room. Line arrays solve the vertical low frequency directivity issue through the brute force method. However, the cost and floor to ceiling design are obviously not for everyone. A MMTMM 2.5 way design can give you a lot of benefit without as much of the cost, while smoothing the transition to a point source tweeter in a way that a line array with a single dome tweeter can't.
A MMTMM 2.5 way layout, unlike a MTM, does not have a fixed woofer center to center distance through the midrange to the low end. The effective woofer center to center distance gets wider as the frequency goes lower. This is caused by the outer woofer rolling in as needed to compensate for baffle step. As a result, provided the woofer spacing is thoughtfully laid out, vertical polar response is quite controlled between 300 and 1200 Hz, with a wider frequency spread of vertical lobing. A system like this would have far less floor and ceiling interaction in the lower midrange and midbass than an average 2-way or even a Waveguide CD design. Consider that the largest surfaces in any large room are going to be the floor and ceiling, and those surfaces are usually not an easy place to install surface treatments. The midbass and lower midrange is where the vertical directivity control is really needed. Higher frequencies are generally absorbed by the common carpet in a room.
Room response curves are normally pretty rough for any system, sometimes +/- 10 db through the bass and lower midrange. This system by nature of it's vertical layout is quite a bit smoother in that range. There is not nearly as much of the typical floor bounce dip/peak combo at 300 and 500 Hz. Remember that room responses are different for everyone, but the floor bounce is generally similar. Also in these images, you can see why I generally prefer a rolled off low end - the primary lowest frequency room node is brought down in level to minimize a one-note low end.'
He has the data for the MMTMM 2.5 way he developed for Madisound about halfway down this page http://zaphaudio.com/ZA5/ in case you haven't seen it.
Excerpt from the page; '
The MMTMM 2.5-way format is all about controlled vertical directivity, and it's the reason I've been wanting to do one for a long time. Directivity is important overall, as is power response and other related issues. But different issues arise when considering vertical and horizontal directivity independently. Common CD (constant directivity) designs use a waveguide high frequency element and a traditional cone driven low frequency element. While the benefits higher in frequency are obvious, lower in frequency the system is just a point source. Good or bad, a point source is the type of sound radiation that is most affected by the room. Line arrays solve the vertical low frequency directivity issue through the brute force method. However, the cost and floor to ceiling design are obviously not for everyone. A MMTMM 2.5 way design can give you a lot of benefit without as much of the cost, while smoothing the transition to a point source tweeter in a way that a line array with a single dome tweeter can't.
A MMTMM 2.5 way layout, unlike a MTM, does not have a fixed woofer center to center distance through the midrange to the low end. The effective woofer center to center distance gets wider as the frequency goes lower. This is caused by the outer woofer rolling in as needed to compensate for baffle step. As a result, provided the woofer spacing is thoughtfully laid out, vertical polar response is quite controlled between 300 and 1200 Hz, with a wider frequency spread of vertical lobing. A system like this would have far less floor and ceiling interaction in the lower midrange and midbass than an average 2-way or even a Waveguide CD design. Consider that the largest surfaces in any large room are going to be the floor and ceiling, and those surfaces are usually not an easy place to install surface treatments. The midbass and lower midrange is where the vertical directivity control is really needed. Higher frequencies are generally absorbed by the common carpet in a room.
Room response curves are normally pretty rough for any system, sometimes +/- 10 db through the bass and lower midrange. This system by nature of it's vertical layout is quite a bit smoother in that range. There is not nearly as much of the typical floor bounce dip/peak combo at 300 and 500 Hz. Remember that room responses are different for everyone, but the floor bounce is generally similar. Also in these images, you can see why I generally prefer a rolled off low end - the primary lowest frequency room node is brought down in level to minimize a one-note low end.'