I'm also willing to test the device and/or software if needed.
I have Windows 7x64 and Mac OSX 10.6.3 - I can compile code in both environments if needed, however I am not a skilled programmer and am not able to invest the time to learn.
Ideally it would be good to support control of a (or several) PGA2311 style chip.
I would also like to see support for controlling the output level of the processor, so that one could
- use PWM amps via I2S like miniDSP's miniAMP board.
- do the bulk of attenuation in analog domain to get the best out of the DACs, and do relatively minor volume adjustments in the DSP.
Configuring the board as a class compliant USB audio device should be all that's required?
All the processing is done on the board isn't it?
I agree with what others have said about ensuring the continued availability of the hardware. I'd also like to thank you for doing the hard work of designing the boards.
Perhaps a good option would be to release the project as a full open source package (hardware,firmware,software) once the initial boards have sold out. Maybe a license dictating not-for-profit group buy conditions only?
Alternatively, I'm curious if the firmware would be portable to Freescale's Symphony board? I'm not sure about differences in the DSPs.
🙂
I have Windows 7x64 and Mac OSX 10.6.3 - I can compile code in both environments if needed, however I am not a skilled programmer and am not able to invest the time to learn.
Ideally it would be good to support control of a (or several) PGA2311 style chip.
I would also like to see support for controlling the output level of the processor, so that one could
- use PWM amps via I2S like miniDSP's miniAMP board.
- do the bulk of attenuation in analog domain to get the best out of the DACs, and do relatively minor volume adjustments in the DSP.
Configuring the board as a class compliant USB audio device should be all that's required?
All the processing is done on the board isn't it?
I agree with what others have said about ensuring the continued availability of the hardware. I'd also like to thank you for doing the hard work of designing the boards.
Perhaps a good option would be to release the project as a full open source package (hardware,firmware,software) once the initial boards have sold out. Maybe a license dictating not-for-profit group buy conditions only?
Alternatively, I'm curious if the firmware would be portable to Freescale's Symphony board? I'm not sure about differences in the DSPs.
🙂
"Configuring the board as a class compliant USB audio device should be all that's required? All the processing is done on the board isn't it?"
Unfortunately there is a more to it than just that. The data and interface has to be formatted and presented in such a way that your favorite application can use it.
Unfortunately there is a more to it than just that. The data and interface has to be formatted and presented in such a way that your favorite application can use it.
Carefully reading the above posts I came to the conclusion that the DIY communauty is not ready yet for digital audio. That's quite amazing.
The idea of "digital volume control" is still there ... floating still the beginning of the thread and that's not normal. If you divide the audio signal by a factor, in digital domain, that's a signal degradation. If you use the output volume capability of the DACs, that's something completely different. There are DACs fitted with decent output volume control, nowadays. If you use a dedicated volume control chip like CS3310 or PGA2311, that's something different too, quite expensive nowadays, able to preserve the dynamic anywhere between 0dB to -40 dB of listening volume.
Nowadays, ADCs and DACs are single-ended or differential. Differential are more expensive and on top of this they need a separate audio-grade opamp. Differential ADCs and DACs are the best, but year after year the specs of single-ended ADCs and DACs do improve, making them a viable & practical choice. On the other hand, we now have quality 8-output CODECs like the CS42518 : a S/PDIF input with advanced PLL clock recovery, 2 x 24-bit differential ADCs and 8 x 24-bit differential DACs with Digital Output Volume control. 192 kHz compatible. 114 ADC dynamic range. 110 dB DAC dynamic range. -100 dB THD + N. The clock source can be the recovered S/PDIF clock, or can be a Quartz clock.
The idea of simplistic Xover schemes like Butteworth, Bessel, Linkwitz-Riley is also there, since the beginning of the thread, floating around, and that's not normal in a specialized, mature DIY context. The Behringer DCX2496 is an affordable piece of equipment. This is a nice 3-way digital crossover. Why bothering with a DIY-kit in such context ? Here is the DCX2496 description, coming from the Behringer BEHRINGER: DCX2496 website :
Start with its 3 analog inputs (one suitable as a digital stereo AES/EBU input) and 6 analog outputs. You get maximum flexibility in just one rack space. Superb high-end AKM® 24-bit/96 kHz A/D and D/A converters give you ultimate signal integrity and an extreme dynamic range of 113 dB. Easy connection of external digital signals with sampling rates from 32 to 96 kHz is a breeze with the integrated sample rate converter. There are precise dynamic EQs for level-dependent equalization and extremely musical parametric EQs, selectable for all inputs and outputs, and “Zero“-attack limiters on all output channels guarantee optimal signal and loudspeaker protection.
You also get four different mono and stereo output operating modes, all with individual crossover filter types (Butterworth, Bessel and Linkwitz-Riley) with selectable roll-off characteristics from 6 to 48 dB/octave. The delays for all inputs and outputs are adjustable. This allows you to manually or automatically correct for room temperature, phase and arrival time differences. An additional sum signal is easily derived from the A/B/C inputs.
Now that remote control is such a hot topic, you’ll be glad that the future-proof ULTRADRIVE PRO software enables single or multi remote control via PC through RS-232 and RS-485 interfaces. And the link option via RS-485 network interface enables cascading of several ULTRADRIVE PROs. A Windows®-based editing software is available for download free of charge. No matter what the future brings, its open architecture assures easy software updates. A PCMCIA slot allows you to store all your settings and recall them anytime you change the location—virtually taking your ULTRADRIVE PRO with you.
But don’t be fooled by the ULTRADRIVE PRO’s sleek design. Its high-power 32-bit SHARC®-DSP and ultra-high resolution CRYSTAL®/AKM® A/D & D/A converter provide outstanding audio performance—and the servo-balanced, gold-plated XLR connectors for all inputs and outputs guarantee excellent connectivity for the years to come.
Why bothering with a DIY-kit in such context ? Let me explain. You better sit down.
Everybody having played with commercially available digital Xovers (like Behringer DCX2496) knows that a DCX2496 is only there for getting your midrange and tweeter as loud as possible, without saturating them by low frequency content they cant't handle. The vast majority will thus configure the Xover as steep highpass for the midrange (say 350 Hz), and steep highpass for the tweeter (say 3.5 kHz). They will arrange all the other parameters around this core. Of course they need the corresponding lowpass at 350 Hz (woofer), and at 3.5 kHz (midrange). Smart marketing helps selling the gear saying they can get complementary outputs through the Linkwitz-Riley stuff. Of course this may be true in the digital and in the electric domain, but that can't be true in the acoustic domain because of the Bode plot of the transducer fitted in the enclosure. But they prefear ignoring this catch. Their mind are in peace once they select the Linkwitz-Riley stuff, with an order as high as possible, like a 3rd-order or a 4th-order. Then they get somebody coming with some measurement gear like audioTester, and they face the real acoustic response, always far from ideal at this naïve stage. They thus ask for a full-featured parametric equalizer at the input, say 4 corrections per channel, for correcting the global acoustic response. Using such parametric equalizer at the input, they thus may get the satisfaction of having a linear response curve, say 45 Hz to 17 kHz in a 6 dB corridor. But at this stage, they feel a little bit uncomfortable, because they remember they needed to apply sharp localized corrections around the transition band, putting the whole Linkwitz-Riley magic stuff in doubt. While experimenting with the input parametric equalizer, they may have noticed that a moderate slope Bessel Xover setting, actually demanded less intense input equalizer corrections.
At this stage they are in doubt about the whole setup, because their dream about digital complementary filtering just vanished.
An aggravating factor may be a more elaborate measurement session, still using audioTester, showing that the transducer are in opposite phase in some transition bands. Then they realize why they needed to locally boost the energy at the input, using the input parametric equalizer. Then they realize that the tranducers have their own phase behaviour, and tend to ruin the whole complementary filtering scheme. Then they dig into the specs of the digital Xover, and if they are lucky, they can find a modality where, if they don't use the max slope, they get a tiny parametric equalizer at each output. Okay, but they realize that for equalizing the transducer, they need to raise the level in a frequency band the transducer can't handle massive power. They realize that they may compensate the transducer own Bode plot, but at the expense of maximum power handling.
And this is a fatal situation.
Remember : they bought the digital Xover (and paid for one or two extra power amps) because the vendor said it was a way to throw the max power in the transducer, without damage, thus getting the loudest sound.
So at the end of the day, they know they need to chose between maximum loudness, and between truly complementary filtering. They know they can't get both.
If they are doing public address or sound reinforcement, dropping the idea of a truly complementary Xover is not an issue. They will be happy with a Behringer DCX2496. More than happy, because the Behringer DCX2496 has limiters, has a Windows interfacing software, and has digital stereo AES/EBU input.
The only reason DIY people would try another Xover than the Behringer DCX2496 would be :
- applying the digital Xover technology for domestic use and/or for audiophile use
- digital stereo AES/EBU input or digital stereo S/PDIF is mandatory
- audiophile use requires high quality differential DACs operating at 192 kHz
- a RC5 infrared remote control is needed, driving the listening volume after each DAC
- audiophile use requires a dedicated volume control chip like CS3310 or PGA2311
- digital filter design software running on Windows (and/or MAC) that takes the individual Bode plots of the transducers fitted in their enclosure, that takes a target decibel and bandwith corridor (like 6 dB corridor from 45 Hz to 17 kHz), and proposing the minimum-order IIR filtering scheme that fulfills this requirement.
- digital filter design software running on Windows (and/or MAC) that takes the individual Bode plots of the transducers fitted in their enclosure, that takes a target decibel and bandwith corridor (like 6 dB corridor from 45 Hz to 17 kHz), that takes a target phase and bandwith corridor (like 180 degree from 45 Hz to 17 kHz) and proposing the minimum-order IIR filtering scheme that fulfills this requirement.
- and the same as above, not using IIRs, but using FIRs
- and the same as above, using a mix of IIRs and FIRs
Call it a AXO24192 if you want ... (audiophile crossover 24 bits 192 kHz).
Make it so compact that it can be housed into a half width 1U rack, and you'll hit the market.
Everything else is futile ...
If your board has no S/PDIF input, no RC5 IR remote control receiver, and if you have no experience of what's described above, you better drop the project.
Your unspecified board and software are dreams that will never come true. You are going to waste a lot of time and money, designing from scratch something a company like Behringer designed ages ago. The DCX2496 is available, affordable, and has better specifications than what you may provide, recycling the DSP board you are talking about. Sorry. This is an advice, not an attack.
If you have the capability, don't try recycling those boards. Design something compliant with what got exposed above. And keep in mind that miniDSP or Behringer or any other contender may come with something new, tomorrow, that may target the DIY communauty and the audiophile communauty.
ChaparK, you and me, we are dreamer. We better select the dream that may serve the DIY communauty and the audiophile communauty. Forget about serving the public address and sound reinforcement makets : a DIY project has not the required reliability and after sales service.
ChaparK, if you have the capability designing 6-layers boards, selecting audiophile-grade components, programming 24-bit DSP with 56-bit accumulation, dealing with on-board DC/DC power supplies, understanding the Xover dilemna with an audiophile point of view, fiddling with Matlab or MathCad for selecting the algorithms (Berchin ?), porting those algorithms into a standalone Windows application, making this Windows application appealing with a nice and intuitive user interface, for sure you'll get a name, a fame ... and money maybe.
Of course you will reply to me that you are not in search of a name, a fame, and money.
In this context, a very nice way to present your DSP board is to say that it is an affordable way for doing some multichannel DSP training, DSP self teaching. Something a starting kit purchased from a DSP vendor can't handle. This would serve the DIY communauty a lot ! Reshaping the whole project, not targeting an array of applications (you'll waste your time in this), not attempting delivering a very high quality (lack of S/PDIF input, lack of multichannel listening volume control), but setting up a dedicated wiki about your board, the block diagram, schematic, the power supply to be used, the APIs, and the software development environment. Your income would be the hardware. And the more elaborate & easy to use the developing environment is, the more boards you will sell.
In this particular context, sorry to repeat myself, think about a simple API taking an ASCII file containing the routing, the IIR filters coefficients and the delays to be flashed inside the board. This will cover 80% of the needs. Please allow the users tweaking the native ASCII file format and the native DSP code, for supporting more elaborate delays, filters, routings, and later on, introducing their own modular DSP effects.
This way they will learn DSP, they will do some modular DSP programming, and they will realize how stupid it is, paying 10 USD for a miniDSP plugin that doesn't precisely fulfill their requirements. Again and again.
After 18 months or so, you may introduce a 2nd generation board embedding the CS42518 CODEC with S/PDIF input, a RC5 IR remote control receiver, 8-channel analog output, a decent built-in +8V -8V power supply, where the end-user can freely select the audio opamp (NE5532, OPA2134, ...), and where the end-user can freely select using no dedicated volume control chip, or a CS3310, or a PGA2311. During those 18 months you will have learned what the market is really asking as application software. The 2nd generation boards won't be sold as an experimental platform, but as a high quality DIY and/or audiophile digital crossover. You will charge some extra money for the Windows application software, especially if you come with a solution like described above : starting from the transducers Bode plots in their enclosures (gain and phase) for generating the digital crossovers filters, under various design constraints. The retail price for the hardware (bare board, fitted with OPA2134 opamps and no dedicated volume control chips) plus the Windows software should't exceed 249 USD. Remember your customers will need a standard +5V supply and an enclosure as extra, making the final price approx 349 eur, plus all the time and effort spent.
Regards,
Steph
The idea of "digital volume control" is still there ... floating still the beginning of the thread and that's not normal. If you divide the audio signal by a factor, in digital domain, that's a signal degradation. If you use the output volume capability of the DACs, that's something completely different. There are DACs fitted with decent output volume control, nowadays. If you use a dedicated volume control chip like CS3310 or PGA2311, that's something different too, quite expensive nowadays, able to preserve the dynamic anywhere between 0dB to -40 dB of listening volume.
Nowadays, ADCs and DACs are single-ended or differential. Differential are more expensive and on top of this they need a separate audio-grade opamp. Differential ADCs and DACs are the best, but year after year the specs of single-ended ADCs and DACs do improve, making them a viable & practical choice. On the other hand, we now have quality 8-output CODECs like the CS42518 : a S/PDIF input with advanced PLL clock recovery, 2 x 24-bit differential ADCs and 8 x 24-bit differential DACs with Digital Output Volume control. 192 kHz compatible. 114 ADC dynamic range. 110 dB DAC dynamic range. -100 dB THD + N. The clock source can be the recovered S/PDIF clock, or can be a Quartz clock.
The idea of simplistic Xover schemes like Butteworth, Bessel, Linkwitz-Riley is also there, since the beginning of the thread, floating around, and that's not normal in a specialized, mature DIY context. The Behringer DCX2496 is an affordable piece of equipment. This is a nice 3-way digital crossover. Why bothering with a DIY-kit in such context ? Here is the DCX2496 description, coming from the Behringer BEHRINGER: DCX2496 website :
Start with its 3 analog inputs (one suitable as a digital stereo AES/EBU input) and 6 analog outputs. You get maximum flexibility in just one rack space. Superb high-end AKM® 24-bit/96 kHz A/D and D/A converters give you ultimate signal integrity and an extreme dynamic range of 113 dB. Easy connection of external digital signals with sampling rates from 32 to 96 kHz is a breeze with the integrated sample rate converter. There are precise dynamic EQs for level-dependent equalization and extremely musical parametric EQs, selectable for all inputs and outputs, and “Zero“-attack limiters on all output channels guarantee optimal signal and loudspeaker protection.
You also get four different mono and stereo output operating modes, all with individual crossover filter types (Butterworth, Bessel and Linkwitz-Riley) with selectable roll-off characteristics from 6 to 48 dB/octave. The delays for all inputs and outputs are adjustable. This allows you to manually or automatically correct for room temperature, phase and arrival time differences. An additional sum signal is easily derived from the A/B/C inputs.
Now that remote control is such a hot topic, you’ll be glad that the future-proof ULTRADRIVE PRO software enables single or multi remote control via PC through RS-232 and RS-485 interfaces. And the link option via RS-485 network interface enables cascading of several ULTRADRIVE PROs. A Windows®-based editing software is available for download free of charge. No matter what the future brings, its open architecture assures easy software updates. A PCMCIA slot allows you to store all your settings and recall them anytime you change the location—virtually taking your ULTRADRIVE PRO with you.
But don’t be fooled by the ULTRADRIVE PRO’s sleek design. Its high-power 32-bit SHARC®-DSP and ultra-high resolution CRYSTAL®/AKM® A/D & D/A converter provide outstanding audio performance—and the servo-balanced, gold-plated XLR connectors for all inputs and outputs guarantee excellent connectivity for the years to come.
Why bothering with a DIY-kit in such context ? Let me explain. You better sit down.
Everybody having played with commercially available digital Xovers (like Behringer DCX2496) knows that a DCX2496 is only there for getting your midrange and tweeter as loud as possible, without saturating them by low frequency content they cant't handle. The vast majority will thus configure the Xover as steep highpass for the midrange (say 350 Hz), and steep highpass for the tweeter (say 3.5 kHz). They will arrange all the other parameters around this core. Of course they need the corresponding lowpass at 350 Hz (woofer), and at 3.5 kHz (midrange). Smart marketing helps selling the gear saying they can get complementary outputs through the Linkwitz-Riley stuff. Of course this may be true in the digital and in the electric domain, but that can't be true in the acoustic domain because of the Bode plot of the transducer fitted in the enclosure. But they prefear ignoring this catch. Their mind are in peace once they select the Linkwitz-Riley stuff, with an order as high as possible, like a 3rd-order or a 4th-order. Then they get somebody coming with some measurement gear like audioTester, and they face the real acoustic response, always far from ideal at this naïve stage. They thus ask for a full-featured parametric equalizer at the input, say 4 corrections per channel, for correcting the global acoustic response. Using such parametric equalizer at the input, they thus may get the satisfaction of having a linear response curve, say 45 Hz to 17 kHz in a 6 dB corridor. But at this stage, they feel a little bit uncomfortable, because they remember they needed to apply sharp localized corrections around the transition band, putting the whole Linkwitz-Riley magic stuff in doubt. While experimenting with the input parametric equalizer, they may have noticed that a moderate slope Bessel Xover setting, actually demanded less intense input equalizer corrections.
At this stage they are in doubt about the whole setup, because their dream about digital complementary filtering just vanished.
An aggravating factor may be a more elaborate measurement session, still using audioTester, showing that the transducer are in opposite phase in some transition bands. Then they realize why they needed to locally boost the energy at the input, using the input parametric equalizer. Then they realize that the tranducers have their own phase behaviour, and tend to ruin the whole complementary filtering scheme. Then they dig into the specs of the digital Xover, and if they are lucky, they can find a modality where, if they don't use the max slope, they get a tiny parametric equalizer at each output. Okay, but they realize that for equalizing the transducer, they need to raise the level in a frequency band the transducer can't handle massive power. They realize that they may compensate the transducer own Bode plot, but at the expense of maximum power handling.
And this is a fatal situation.
Remember : they bought the digital Xover (and paid for one or two extra power amps) because the vendor said it was a way to throw the max power in the transducer, without damage, thus getting the loudest sound.
So at the end of the day, they know they need to chose between maximum loudness, and between truly complementary filtering. They know they can't get both.
If they are doing public address or sound reinforcement, dropping the idea of a truly complementary Xover is not an issue. They will be happy with a Behringer DCX2496. More than happy, because the Behringer DCX2496 has limiters, has a Windows interfacing software, and has digital stereo AES/EBU input.
The only reason DIY people would try another Xover than the Behringer DCX2496 would be :
- applying the digital Xover technology for domestic use and/or for audiophile use
- digital stereo AES/EBU input or digital stereo S/PDIF is mandatory
- audiophile use requires high quality differential DACs operating at 192 kHz
- a RC5 infrared remote control is needed, driving the listening volume after each DAC
- audiophile use requires a dedicated volume control chip like CS3310 or PGA2311
- digital filter design software running on Windows (and/or MAC) that takes the individual Bode plots of the transducers fitted in their enclosure, that takes a target decibel and bandwith corridor (like 6 dB corridor from 45 Hz to 17 kHz), and proposing the minimum-order IIR filtering scheme that fulfills this requirement.
- digital filter design software running on Windows (and/or MAC) that takes the individual Bode plots of the transducers fitted in their enclosure, that takes a target decibel and bandwith corridor (like 6 dB corridor from 45 Hz to 17 kHz), that takes a target phase and bandwith corridor (like 180 degree from 45 Hz to 17 kHz) and proposing the minimum-order IIR filtering scheme that fulfills this requirement.
- and the same as above, not using IIRs, but using FIRs
- and the same as above, using a mix of IIRs and FIRs
Call it a AXO24192 if you want ... (audiophile crossover 24 bits 192 kHz).
Make it so compact that it can be housed into a half width 1U rack, and you'll hit the market.
Everything else is futile ...
If your board has no S/PDIF input, no RC5 IR remote control receiver, and if you have no experience of what's described above, you better drop the project.
Your unspecified board and software are dreams that will never come true. You are going to waste a lot of time and money, designing from scratch something a company like Behringer designed ages ago. The DCX2496 is available, affordable, and has better specifications than what you may provide, recycling the DSP board you are talking about. Sorry. This is an advice, not an attack.
If you have the capability, don't try recycling those boards. Design something compliant with what got exposed above. And keep in mind that miniDSP or Behringer or any other contender may come with something new, tomorrow, that may target the DIY communauty and the audiophile communauty.
ChaparK, you and me, we are dreamer. We better select the dream that may serve the DIY communauty and the audiophile communauty. Forget about serving the public address and sound reinforcement makets : a DIY project has not the required reliability and after sales service.
ChaparK, if you have the capability designing 6-layers boards, selecting audiophile-grade components, programming 24-bit DSP with 56-bit accumulation, dealing with on-board DC/DC power supplies, understanding the Xover dilemna with an audiophile point of view, fiddling with Matlab or MathCad for selecting the algorithms (Berchin ?), porting those algorithms into a standalone Windows application, making this Windows application appealing with a nice and intuitive user interface, for sure you'll get a name, a fame ... and money maybe.
Of course you will reply to me that you are not in search of a name, a fame, and money.
In this context, a very nice way to present your DSP board is to say that it is an affordable way for doing some multichannel DSP training, DSP self teaching. Something a starting kit purchased from a DSP vendor can't handle. This would serve the DIY communauty a lot ! Reshaping the whole project, not targeting an array of applications (you'll waste your time in this), not attempting delivering a very high quality (lack of S/PDIF input, lack of multichannel listening volume control), but setting up a dedicated wiki about your board, the block diagram, schematic, the power supply to be used, the APIs, and the software development environment. Your income would be the hardware. And the more elaborate & easy to use the developing environment is, the more boards you will sell.
In this particular context, sorry to repeat myself, think about a simple API taking an ASCII file containing the routing, the IIR filters coefficients and the delays to be flashed inside the board. This will cover 80% of the needs. Please allow the users tweaking the native ASCII file format and the native DSP code, for supporting more elaborate delays, filters, routings, and later on, introducing their own modular DSP effects.
This way they will learn DSP, they will do some modular DSP programming, and they will realize how stupid it is, paying 10 USD for a miniDSP plugin that doesn't precisely fulfill their requirements. Again and again.
After 18 months or so, you may introduce a 2nd generation board embedding the CS42518 CODEC with S/PDIF input, a RC5 IR remote control receiver, 8-channel analog output, a decent built-in +8V -8V power supply, where the end-user can freely select the audio opamp (NE5532, OPA2134, ...), and where the end-user can freely select using no dedicated volume control chip, or a CS3310, or a PGA2311. During those 18 months you will have learned what the market is really asking as application software. The 2nd generation boards won't be sold as an experimental platform, but as a high quality DIY and/or audiophile digital crossover. You will charge some extra money for the Windows application software, especially if you come with a solution like described above : starting from the transducers Bode plots in their enclosures (gain and phase) for generating the digital crossovers filters, under various design constraints. The retail price for the hardware (bare board, fitted with OPA2134 opamps and no dedicated volume control chips) plus the Windows software should't exceed 249 USD. Remember your customers will need a standard +5V supply and an enclosure as extra, making the final price approx 349 eur, plus all the time and effort spent.
Regards,
Steph
Last edited:
Well that was a pretty good rant, feel better now 😀
Good luck in fullfiling any dreams...If its not straightforward then you can forget about mass marketing. Linux, Mac are fine as add on but you better have a great windows version for this to actually matter.
I have followed the thread because I want what is being talked about but I think people lose sight of the trees because the forest is in the ways. I run a software company and this is a common issue with software engineers. They tend to be pedantic and very focus on building perfection (including thinking Linux and Mac needs over 50% of the resource time which is a bad way to run a business and make profits) when the client actually wants "solutions". The client here is those who want active crossover solutions. I would love for a AVR or PRE/PRO to take it to the next level and have crossover functionality that would be a perfect solution actually but we are instead left to get things done in the DIY world.
The DCX is still an incredible device for its price tag. There are many other higher priced digital crossover solutions too so its not bad at all really, only the minority thinks its bad.
Good luck in fullfiling any dreams...If its not straightforward then you can forget about mass marketing. Linux, Mac are fine as add on but you better have a great windows version for this to actually matter.
I have followed the thread because I want what is being talked about but I think people lose sight of the trees because the forest is in the ways. I run a software company and this is a common issue with software engineers. They tend to be pedantic and very focus on building perfection (including thinking Linux and Mac needs over 50% of the resource time which is a bad way to run a business and make profits) when the client actually wants "solutions". The client here is those who want active crossover solutions. I would love for a AVR or PRE/PRO to take it to the next level and have crossover functionality that would be a perfect solution actually but we are instead left to get things done in the DIY world.
The DCX is still an incredible device for its price tag. There are many other higher priced digital crossover solutions too so its not bad at all really, only the minority thinks its bad.
Last edited:
That does not make any sense. Could you clarify. The DCX is a FULL range device. Many DIYers with incredible sub setups use the DCX with great success. I have used it with great success even with a 3-way full range speaker design. The DCX also has nothing to do with how loud a driver goes, that would be the driver specs and amp power that push its SPL to the limits.
Relax, I was poking fun at Mac zealots in which think an OS that controls about 10% of the market is remotely important and to run any REAL IMPORTANT software you have to run a emulator to another OS platform, yeah real value in buing one! 😉
I fully understand Mac fan based passion. No need to argue it here...there are 100s of other sites for you to get all defensive about Macs.
Hello, I agree with you about the Linux and Mac stuff. I agree with you about the great value, useability and specs of the Behringer DCX2496.
I will now explain in detail why practically, the DCX2496 primary use is to enable as max power as possible in speaker drivers, without audiophile concern.
Take a sound reinforcment configuration with a big 2-way loudspeaker. You only need one power amp, delivering anything between 400 watt to 1600 watt. When you throw 1600 watt into your speaker, you blow your tweeter. Your tweeter blows because in the loudspeaker, if it is a cheap loudspeaker, you have a passive crossover, that can be as simple as a capacitor in series with the tweeter. This is thus a 1st order high-pass. The specification sheet of your loudspeaker says "3.5 kHz crossover frequency", and if you are not expert in analog electronics you will assume that any frequency lower than 3.5 kHz is not going to reach your tweeter. Haha ! There is the catch ! A 1st order filter has a wide transition band : the tweeter will still receive considerable energy as low as half the crossover frequency. The issue is that music contains a lot of energy between 100 Hz to 2 kHz, and less energy between 2 kHz to 20 kHz. Your tweeter will thus cumulate two different kind of stresses : a) it will need to dissipate the supplementary energy that exists in the 1st order transition band (say 850 Hz to 3.5 kHz) and b) it is unable to correctly process the lowest frequencies as the membrane (or coil) may need to travel outside of the linear displacement range. The overall impedance gets lower, the current thus increases like an electric motor taking more amps when it is blocked. You need to understand this. Okay ?
So now, you understand why a tweeter blows when overdriven because of a 1st-order passive high-pass filter.
The same applies to the midrange transducer if you are in a 3-way passive filtered loudspeaker. Say the small 8-inch medium driver(s) are passively high-pass filtered at 350 Hz, using a (big) capacitor in series. Again this is a 1st order high pass, hence with a wide transition band. Some non-neglectible energy at 85 Hz (1/4 of the crossover frequency) will be present, as supplementary energy compared to a brickwall filter. And the presence of this low frequency, lower than what the transducer got designed to handle, again asks for big membrane displacement, outside of the linear Xmax range, with as consequence a decrease in impedance, an increase of current (blocked electric motor scenario), and of course, an increase of internal temperature that leads to self-destruction.
So now, you understand why a midrange blows when overdriven because of a 1st-order passive high-pass filter.
Now, take the exact same transducers, and connect them on separate amplifiers fed by a digital crossover like the Behringer DCX2496.
On the DCX2496, select a high order filtering scheme like a 3rd order. Your transition band will be narrower : 1/3 of what we've seen above. The tweeter, high-pass filtered at 3.5 kHz, will receive amost no energy at 1.8 kHz. And you know that for such frequency, the Xmax will be respected, even if you throw 400 watt in your tweeter (if it is a respectable one). This means that you will be able to pump up the volumes, and YES, the max listening level will increase.
Same for the midrange(s). On the DCX2496, select a high order filtering scheme like a 3rd order. Filtered with a 3rd order high-pass at 350 Hz, the midrange(s) will see almost no energy at 175 Hz, and you know that for such frequency, the Xmax will be respected, even if you throw 800 watt in your midrange(s). This means that you will be able to pump up the volumes, and YES, the max listening level will increase.
This is the well known mechanism beloved by sound reinforcement people : their loudspeaker soud clearer (no fuzzy sound caused by exagerated Xmax), their loudspeakers tolerate more power at the input, making them sound louder without risking damages.
And on top of this, the Behringer DCX2496 embeds limiters, in case somebody overdrivens the mixing console in a crazy way, or in case Larsen invites himself.
Now you understand the astonishing value of the DXC2496. Sold at a small profit margin by Behringer, surely because it is an incentive to buy more amplifiers and surely also because it is the entry door for you considering buying effects processors built on the same hardware platform.
This marketing concept is now copy cat by some companies like :
- the Beglec SynQ Audio DLP6 synq-audio.com
- the Apex Audio Intelli-X48 APEX Manufacturing - Products - Intelli-Series - Intelli-X 48
Their retail price is higher than the Behringer DXC2496, but they provide maybe more flexibility and more DSP power. Do they plan allowing third-party plugins ?
I guess Behringer is watching what those competitors do (regarding signal routing, filtering and features), & also what miniDSP does (regarding making buzz and money with plugins) before actually launching the Behringer DCX24192.
The battle will be over when Behringer launches the DCX24192, don't you think ?
Must say as Behringer manager reading the DIY-audio forum, I'll need to tell my design team "take your time, no stress about the DCX24192". "The market is not yet mature for a nexgen digital crossover like steph_tsf is talking about". "Most people on diyAudio don't or don't want to understand the difference between a sound reinforcement digital crossover (like DCX2496 we designed ten years ago) and a more elaborate digital crossover taking the transducers Bode plots in consideration, enabling new tradeoffs between max power handling and phase linearity". "Let diyAudio fiddle with some miniDSP implementations (there are no at the moment) or SynthMaker implementations (there are no at the moment), this may educate the market, and then, we'll better know what kind of features will determine sales". "Remember our aim is to sell two nexgen devices taking the place of one DCX2496, creating the need for more amplifiers and more effect processors like electronically controlling the room reflexions".
Regards,
Steph
Last edited:
Thanks, I get what you are saying! I didnt know about the new DCX.
I have the BBE DS48 and the QSC DSP-30 also in testing against my DCX which runs my 3-ways (Neopro5i, PHL1120, TD12S). Im a big active fan, my new waveguides are built, measured and plugged into the DCX in no time..
I do not want to go off OT any more.
I would love a great active XO choices other then the DCX. Hypex will be coming out with something by 2011 so that should be cool. I have almost pull the trigger 4 times on their AS.200 amps/DSP.
Others are using the MiniDSP which I think has promise.
I have the BBE DS48 and the QSC DSP-30 also in testing against my DCX which runs my 3-ways (Neopro5i, PHL1120, TD12S). Im a big active fan, my new waveguides are built, measured and plugged into the DCX in no time..
I do not want to go off OT any more.
I would love a great active XO choices other then the DCX. Hypex will be coming out with something by 2011 so that should be cool. I have almost pull the trigger 4 times on their AS.200 amps/DSP.
Others are using the MiniDSP which I think has promise.
Last edited:
doug20, I think this post on SynthMaker may interest you : SynthMaker Forum • View topic - ASIO4ALL
It is about converting any x86 PC running WinXP (including x86 Netbook) into a versatile digital crossover using any multichannel ASIO card as output device like the GIGAPort AG USB. GIGAPort AG USB Sound Card with 8 Channel audio output!
SynthMaker enables you to program the DSP code of the x86 processor in pseudo assembly. SynthMaker enables you to compile a standalone VST executable if you want.
One day, SynthMaker will be ported on ARM multicore, and there will be ASIO drivers for the ARM world. Once SynthMaker/ARM will be available as an AppleStore application running on the Apple ARM platform, do you realize the buzz - and market ? Especialy if this happens with the benediction of Steinberg, becoming an Apple department, collecting royalities on every downloaded software or plugin. Think about this. It may (should ?) happen.
Steph
It is about converting any x86 PC running WinXP (including x86 Netbook) into a versatile digital crossover using any multichannel ASIO card as output device like the GIGAPort AG USB. GIGAPort AG USB Sound Card with 8 Channel audio output!
SynthMaker enables you to program the DSP code of the x86 processor in pseudo assembly. SynthMaker enables you to compile a standalone VST executable if you want.
One day, SynthMaker will be ported on ARM multicore, and there will be ASIO drivers for the ARM world. Once SynthMaker/ARM will be available as an AppleStore application running on the Apple ARM platform, do you realize the buzz - and market ? Especialy if this happens with the benediction of Steinberg, becoming an Apple department, collecting royalities on every downloaded software or plugin. Think about this. It may (should ?) happen.
Steph
I don't think so, considering the very low price tag. Anyway, the GIGAPort AG USB would be my preferred device for making experiments under SynthMaker, even if we shall expect some power supply residues, clock jitter and basic DACs without a proper listening volume adjustment. Latency is always going to be there, that's a feature of the Delta-Sigma DAC architecture. Do you know audiophile grade R2R DACs or SAR DACs, 24 bits and 192 kHz ? See you there : http://www.diyaudio.com/forums/mult...tered-neopro5i-phl1120-td12s.html#post2212800
Steph
I've been reading this thread with a lot of interest. Many thanks to all the contributors and all the great information being shared.
As a newbee, I don't have anything to contribute but I'm interest in the product as digital crossover solution BUT want to keep my Sabre32/Buffalo DAC in the chain.
So the obvious solution is DAC-> A/D ->digital crossover->D/A -> (volume control somewhere) -> amps -> drivers
BUT I also want to keep the volume control inside the DAC AND want to avoid the A/D conversion.
So another solution is SPDIF/I2S -> digital crossover -> I2S (multiple) -> 2 or more DACS -> amps -> drivers
I can control the DACs with something like Arduino and I2C protocol but more than 2 DACs could be a problem because they only have two I2C addresses
BUT, I don't want to resampling, so the digital crossover must accept native sample rate and arbitrary fs (for example, my musiland card can output I2S at 128fs) and then output processed data at the original sample rate -I don't know if this matters or not, but something like miniDSP -which I like because the GUI is good for a newbee like me- operates at 48K/64fs and thus the digital input must be through their spdif card, and everything is resampled)
PS: The Sabre32 is a multichannel DAC, but there is no DIY kit yet...
As a newbee, I don't have anything to contribute but I'm interest in the product as digital crossover solution BUT want to keep my Sabre32/Buffalo DAC in the chain.
So the obvious solution is DAC-> A/D ->digital crossover->D/A -> (volume control somewhere) -> amps -> drivers
BUT I also want to keep the volume control inside the DAC AND want to avoid the A/D conversion.
So another solution is SPDIF/I2S -> digital crossover -> I2S (multiple) -> 2 or more DACS -> amps -> drivers
I can control the DACs with something like Arduino and I2C protocol but more than 2 DACs could be a problem because they only have two I2C addresses
BUT, I don't want to resampling, so the digital crossover must accept native sample rate and arbitrary fs (for example, my musiland card can output I2S at 128fs) and then output processed data at the original sample rate -I don't know if this matters or not, but something like miniDSP -which I like because the GUI is good for a newbee like me- operates at 48K/64fs and thus the digital input must be through their spdif card, and everything is resampled)
PS: The Sabre32 is a multichannel DAC, but there is no DIY kit yet...
I am also interested to know what the I2S specs of this board are?
Number of channels in/out?
sample rate, fs?
I might like to use miniDSP's miniAMP board with this processor.
Number of channels in/out?
sample rate, fs?
I might like to use miniDSP's miniAMP board with this processor.
Hi glt and fb,
Sorry for answering late.
Hardware wise, the I2S consists of the following lines:
- 1 line in
- 3 lines out
- 2 frame sync lines (1 for ins, 1 for outs)
- 2 clock lines (1 for ins, 1 for outs)
2 channels per line is the straightforward implementation. That makes 2 inputs and 6 outputs.
Sample rate and clocks user defined - in the limit of what the dsp can do, but it's quite flexible.
Sorry for answering late.
Hardware wise, the I2S consists of the following lines:
- 1 line in
- 3 lines out
- 2 frame sync lines (1 for ins, 1 for outs)
- 2 clock lines (1 for ins, 1 for outs)
2 channels per line is the straightforward implementation. That makes 2 inputs and 6 outputs.
Sample rate and clocks user defined - in the limit of what the dsp can do, but it's quite flexible.
I think that support of various sampling frequencies is something that we should indeed include in the specs, thanks for mentionning it.
We cannot support 'arbitrary frequencies', but rather 'common frequencies'. So basically that would be 44.1/48/96, with 96 the default sampling rate (when using the onboard ADC/DAC). Don't know if 88.2 or 192 kHz are worth being supported though (cause i'm still quite happy with the good ol' CD 😀 44.1k). Any opinion?
Thanks for the reply. Yeah, didn't really mean "arbitrary", rather NOT fixed to one frequency. I think 88.2K is also important because you can purchase HD music with this sample rate.
Also if the input fs is configurable, can the output fs be independently configurable? For example, the Musiland USB card is 128fs, but the Wolfson 8741 DAC likes 256 fs
Hi glt,
Hmm late again for answering, sorry about that.
Ok for 88.2. That makes 4 sampling frequencies: 44.1/48/88.1/96k.
Regarding your USB card and DAC, honestly i don't know. I should check the datasheets of both chips, but i haven't done that so far...
When you're mentioning fs, are you referring to the master clock, or to the bit clock of I2S?
Will come back to you about that matter.
Regards,
chaparK
Hi chaparK, thanks for the reply.
I did compile a table for MCK compatibility for different dacs: SampleRate-2.jpg (image)
Also I did measure the frequency of the MCK line of the Musiland USB->SPDIF converter: H I F I D U I N O: Measuring Master Clock and fs and did get 128fs for 44.1K material.
When I connected the I2S lines from the Musiland device to the Wolfson 8741 DAC I did get music but some of the filters did not work. I concluded that it was because the Musiland MCK was 128fs and the Wolfson DAC supports 256xfs according to spec.
I did compile a table for MCK compatibility for different dacs: SampleRate-2.jpg (image)
Also I did measure the frequency of the MCK line of the Musiland USB->SPDIF converter: H I F I D U I N O: Measuring Master Clock and fs and did get 128fs for 44.1K material.
When I connected the I2S lines from the Musiland device to the Wolfson 8741 DAC I did get music but some of the filters did not work. I concluded that it was because the Musiland MCK was 128fs and the Wolfson DAC supports 256xfs according to spec.
- Status
- Not open for further replies.