You ramble on a lot about nothing really. You do not confuse me if you actually realized I run a little software company, i have pretty good success for almost 20 years in that business 
I find it very humorous that you post saying this is all easy and free with then truth about it is that NOTHING is free. Proper sound cards, Silent PCs are not exactly free. This discussion reminds me of all my meetings with IBMers, unix guys and then big linux fans. First was 15 years ago the Unix guys hammered away at the reliability **, while I told the client he could save 500K with a reliable non unix solution (I got to pocket the 100K savings...thank you ). The second was 6 years ago we hammered away between linux and windows, again we had some like yourself boosting the simplicity of linux, they throw out obscenely geek topics try to make themselves sound smart....Again, they had no idea how to talk to real people, management types that write the cheques. I remember the owner telling me to never let that linux guy talk in a meeting again.
Your type never ceases to amaze me with your disconnect to the real world. The best and brightest individuals are the ones that actually never TRY confuse people in discussions. They understand the REAL goals and they know the audience, they also never try to impress anyone with "high scrabble value" words....you know the silly little idea like posting transducers instead of just saying drivers
Now Im all for your last point!!
To answer your questions about the DCX and what I do... I measure my drivers, I pick some slopes that work within the limits of the drivers, I dial things in on the DCX and I listen. I then swap different slopes out and listen some more. I also run ARTA pink Noise to see my on and off axis in room response. Yes, I use the DCX not only for XOs but I would correct driver and in room issues with it. I do not need to know crap about the raw transfer functions to do anything here. I do not need to be aware of SynthMaker or use it. I can measure with ARTA and HOLM with great success there isnt anything else needed. If the results please myself and any of my family and friends then the goals are might. Im not in DIY to please someone like you with anything I build, I do not even build speakers thinking I would present them at a DIY event, Im not looking for speaker awards, etc. I do not belong to audiophile clubs, I wouldnt waste my time build and trying to impress audiophiles since I have been to some DIY events and its not the world I would enjoy very much since the subjectivity goes beyond anything real.
There are simply many, many different goals that exist. None of us should assume any goal is better or worse. If you think Im stupid for how I use the DCX so be it. Last I checked I do not report to you or anyone else. My friends and family do not look to your approval and I surely make ZERO dollars from your opinion on how the DCX or similar should be used so how remotely possible could your opinion be the only opinion that matters?
What is your role in this thread again?
I find it very humorous that you post saying this is all easy and free with then truth about it is that NOTHING is free. Proper sound cards, Silent PCs are not exactly free. This discussion reminds me of all my meetings with IBMers, unix guys and then big linux fans. First was 15 years ago the Unix guys hammered away at the reliability **, while I told the client he could save 500K with a reliable non unix solution (I got to pocket the 100K savings...thank you
). The second was 6 years ago we hammered away between linux and windows, again we had some like yourself boosting the simplicity of linux, they throw out obscenely geek topics try to make themselves sound smart....Again, they had no idea how to talk to real people, management types that write the cheques. I remember the owner telling me to never let that linux guy talk in a meeting again.Your type never ceases to amaze me with your disconnect to the real world. The best and brightest individuals are the ones that actually never TRY confuse people in discussions. They understand the REAL goals and they know the audience, they also never try to impress anyone with "high scrabble value" words....you know the silly little idea like posting transducers instead of just saying drivers
Now Im all for your last point!!
To answer your questions about the DCX and what I do... I measure my drivers, I pick some slopes that work within the limits of the drivers, I dial things in on the DCX and I listen. I then swap different slopes out and listen some more. I also run ARTA pink Noise to see my on and off axis in room response. Yes, I use the DCX not only for XOs but I would correct driver and in room issues with it. I do not need to know crap about the raw transfer functions to do anything here. I do not need to be aware of SynthMaker or use it. I can measure with ARTA and HOLM with great success there isnt anything else needed. If the results please myself and any of my family and friends then the goals are might. Im not in DIY to please someone like you with anything I build, I do not even build speakers thinking I would present them at a DIY event, Im not looking for speaker awards, etc. I do not belong to audiophile clubs, I wouldnt waste my time build and trying to impress audiophiles since I have been to some DIY events and its not the world I would enjoy very much since the subjectivity goes beyond anything real.
There are simply many, many different goals that exist. None of us should assume any goal is better or worse. If you think Im stupid for how I use the DCX so be it. Last I checked I do not report to you or anyone else. My friends and family do not look to your approval and I surely make ZERO dollars from your opinion on how the DCX or similar should be used so how remotely possible could your opinion be the only opinion that matters?
What is your role in this thread again?
Last edited:
hello Doug, nice synthesis, thanks for taking the time. Currently and here we are issuing ideas, suggestions and comments on a "DSP Xover project" thread on diyAudio. There are three words there : DSP (audio DSP namely), Xover and project. Post #1 of this thread is :
P.S. 1
As you may have noticed, the Freescale K60 at 150 MHz is maybe not powerful enough for executing the Philips DSS930 IIR + FIR compound. There we may see a justification for using a Freescale DSP56321 at 275 MHz.
P.S. 2
As you may have noticed, the whole software can be tested and prototyped using SynthMaker, including the real-time optimization of the polar pattern using an ASIO multichannel USB audio subsystem having 4 or 8 mike inputs.
On one side we have the imminent arrival of ARM Cortex-M4 low cost DSP like the Freescale Kinetis microcontrollers. On the other side we have miniDSP, with the newest plugins enabling you to enter your own IIR coefficients. And like you pointed out, we also have the DCX2496, eventually fitted with audiophile mods. How to be different now ? How to earn and occupy a solid market niche with a DSP56321 clocked at 275 MHz ? What I'm suggesting is to dig into the AES library and exhumate some publications made ages ago, when audio DSP was a futuristic possibility. There you will (re)discover the Lipshitz-Vanderkooy linear phase delay-compensated crossover. There you may also (re)discover the Erik Baekgaard "missing link" 2nd order 3-way crossover, also providing a linear phase. There you may find numerous approaches for modelling, compensating and linearizing the Bode plot of the unfiltered drivers. You may also read the AES publication about the Philips ESS930 system (a clever IIR + FIR compound), in 1993, when audio DSP went more practical. Those are my requirements about a Xover project in 2010. Of course such Xover needs to embed a Bode plot measuring modality. Doug, I'm afraid it is your turn now to summarize your requirements. Shortly you will agree that most commercial Xover schemes appear dumb or stupid, in front of what a freshly designed DSP Xover can do for you. Which doesn't mean that people loving oldschool DSP Xovers, tuning oldschool DSP Xovers and listening to oldschool DSP Xovers are dumb and stupid. The miniDSP marketing positionning is wide, now with the possibility of entering the IIR coefficients manually instead of relying on "textbook filtering". This is a nice beginning, a long term asset for the Analog Devices SigmaDSP in the audiophile communauty. You need to understand that once there will be a Lipshitz-Vanderkooy or a Eric Baekgaard implementation on the Freescale DSP56321 at 275 MHz platform, companies like Behringer and/or miniDSP will copy them. One way to preserve the Freescale DSP56321 at 275 MHz platform markets would be to put the emphasis on a built-in audio analyzer able to draw the Bode plots of each unfiltered driver, automatically calculating the IIR coefficients in the context of a Lipshitz-Vanderkooy delay compensated crossover, or in the context of a Baekgaard "missing link" crossover, or in the context of a compound IIR + FIR approach like the Philips DSS930. If you undertstand this, you'll realize that a survival condition for the Freescale DSP56321 at 275 MHz platform is to provide a measuring mike input, possibly 4 or 8 measuring mike inputs if you want to deal with simultaneous polar pattern measurements, with a realtime optimization during the setup. In a nutshell, the DSP56321 at 275 MHz platform needs to be marketed as a multiway speaker lab, well above the concept of "textbook filtering", well above the concept of a "programmable IIR bank". What's worrying me is that within a few weeks, Freescale Kinetis application engineers may issue a design note showing how one can configure the built-in I2S and/or SPI for emulating a 8-channel TDM interface. If this happens, I let you imagine the BOM cost decrease when opting for a Freescale K60 (Kinetis) clocked at 150 MHz, instead of a DSP56321 at 275 MHz. What's worrying me is that all what's listed above may be possible using a two-chip design like a Freescale K60 and a Cirrus CS4207 (also having a S/PDIF input). Okay, this is not audiophile quality, but this would represent a giant leap forward and a mass market, possibly. Attached you will find a little survey (.xls in a .zip) about DACs that may be used, replacing the CS4207 by a few audiophile DACs hooked on a common TDM bus. Cheers.
P.S. 1
As you may have noticed, the Freescale K60 at 150 MHz is maybe not powerful enough for executing the Philips DSS930 IIR + FIR compound. There we may see a justification for using a Freescale DSP56321 at 275 MHz.
P.S. 2
As you may have noticed, the whole software can be tested and prototyped using SynthMaker, including the real-time optimization of the polar pattern using an ASIO multichannel USB audio subsystem having 4 or 8 mike inputs.
Last edited:
That sounds absurd !! What's stopping them from doing it now ?? Why do they have to wait for someone on diy forum to do it first ?
It's easy to throw DSP chip part numbers around like telephone numbers but actually implementing the hardware and software is a totally different ball game !! Maybe try it yourself some day
Well I have just sat and read this post with great interest!
I am a complete noob to DSP (and diy HiFi in general) but am an engineer and thus take such things as a challenge/hobby and take great delight getting things to work.
Recently I have decided to make a three way active speaker system and came across the miniDSP boards. They looked perfect for my application until I started to look further into it. I have basically convinced myself to build a pre-amo that has dsp crossovers built in. By the time you add the required two boards, two minidigi's and a 24/96 USB streamer into it it is no longer a cheap sollution. Add to that some input switching, analogue volume control for 8 channels etc etc it starts to get expensive!
What interests me about your board is that it would teach me about DSP, that it can accept and transmit i2S digital and of course the reasonable price. I would be very interested in getting my hands on one!
My goal would be:
Digital input section (various toslink, spdif, i2S inputs) > i2S
Analogue input selection > DSP board adc
Switch between A/D inputs on board with simple jumper/button
DSP crossover
i2S digital out (3*2 channel) > off-board DAC * 3 with volume control
Dac > 3 * power amp via balanced connections.
Is this feasible?
Keep up the good work, and thanks for any replys to my noob questions in advance!
Jai
I am a complete noob to DSP (and diy HiFi in general) but am an engineer and thus take such things as a challenge/hobby and take great delight getting things to work.
Recently I have decided to make a three way active speaker system and came across the miniDSP boards. They looked perfect for my application until I started to look further into it. I have basically convinced myself to build a pre-amo that has dsp crossovers built in. By the time you add the required two boards, two minidigi's and a 24/96 USB streamer into it it is no longer a cheap sollution. Add to that some input switching, analogue volume control for 8 channels etc etc it starts to get expensive!
What interests me about your board is that it would teach me about DSP, that it can accept and transmit i2S digital and of course the reasonable price. I would be very interested in getting my hands on one!
My goal would be:
Digital input section (various toslink, spdif, i2S inputs) > i2S
Analogue input selection > DSP board adc
Switch between A/D inputs on board with simple jumper/button
DSP crossover
i2S digital out (3*2 channel) > off-board DAC * 3 with volume control
Dac > 3 * power amp via balanced connections.
Is this feasible?
Keep up the good work, and thanks for any replys to my noob questions in advance!
Jai
Hi Jay,
The hardware is described in great detail i think on page 1 of this thread. Basically there are 2 audio sections: one path which is hardwired to the onboard codec, an other path with I2S open-ended. Each path has 2 inputs and 6 outputs.
In order to use the I2S interface you need to bring your own hardware (spdif, USB receiver etc) and wire it. Note that you can route in your signal through the onboard D/A converter and retrieve outputs on the I2S lines or vice-versa, and you can switch anytime.
The I2S interface is not plug-and-play. You need to add your connector and set the devices on both sides for having the communication on.
Hope this helps!
chaparK
I'm pretty sure I'm there in terms of understanding, but need to draw some block/flow diagrams to describe what I mean.
I have basically got round to realising I am going to need to develop some new skills to get exactly where I want to be: Such as a micro-controller to take care of the front end (user I/O).
I shall get back later today (I really shouldn't be using work time to construct block diagrams of hifi.. ha!)
Thanks for the reply, this project looks exciting - just possibly a tiny bit out of my skill-set but I'm sure I can learn!
A basic question for you: I like the look of the minidsp boards very much, but when I sketch up my exact requirements it starts to add up in price. I am also concerned that in these days of HD downloads I would be missing some of the picture due to the internal sample rate of 48k on them. Will this board internally be using a higher sample rate? ie 96 or 192?
Also: if I were to use the board with i2S input (via a digital resampler/source selector thingy such as the minidigi) could I first use the on-board DAC's and at a later date add three (2 channel) DAC's that accept an I2S input from this DSP? (edit: thinking about it, I mean is it possible for the DSP to accept the I2S in (slave clock mode) AND output I2S (master clock mode); and also will the unit output an pass through I2S signal so I can expand it from 6 channel to 10 channel with a minidsp being used for the sub output channels). 10 channels - excessive of course lol .
Many thanks..
Jai
I have basically got round to realising I am going to need to develop some new skills to get exactly where I want to be: Such as a micro-controller to take care of the front end (user I/O).
I shall get back later today (I really shouldn't be using work time to construct block diagrams of hifi.. ha!)
Thanks for the reply, this project looks exciting - just possibly a tiny bit out of my skill-set but I'm sure I can learn!
A basic question for you: I like the look of the minidsp boards very much, but when I sketch up my exact requirements it starts to add up in price. I am also concerned that in these days of HD downloads I would be missing some of the picture due to the internal sample rate of 48k on them. Will this board internally be using a higher sample rate? ie 96 or 192?
Also: if I were to use the board with i2S input (via a digital resampler/source selector thingy such as the minidigi) could I first use the on-board DAC's and at a later date add three (2 channel) DAC's that accept an I2S input from this DSP? (edit: thinking about it, I mean is it possible for the DSP to accept the I2S in (slave clock mode) AND output I2S (master clock mode); and also will the unit output an pass through I2S signal so I can expand it from 6 channel to 10 channel with a minidsp being used for the sub output channels). 10 channels - excessive of course lol .
Many thanks..
Jai
Last edited:
Interesting stuff here on the thread. I agree steph_tsf that the ARM Cortex M4 might be a game-changer in terms of getting DSP more open and available to the average DIYer. I'd be more inclined towards the NXP offering which is due out any day now :
NXP Semiconductors - Microcontrollers [Products - ARM Cortex-M4 Core Microcontrollers (MCUs) and Microcontroller (MCU) Functions]
Scroll down to the bottom and watch the two vids - most informative, particularly about how the M4 improves on the M3 in relation to DSP tasks. Bottom line is - if you want FIR then M4 only takes you so far, but for more general purpose DSP (IIR biquads perhaps) its really attractive. And from NXP, its bound to be affordable - I have an NXP Cortex M3 based dev board which cost just $16.
NXP Semiconductors - Microcontrollers [Products - ARM Cortex-M4 Core Microcontrollers (MCUs) and Microcontroller (MCU) Functions]
Scroll down to the bottom and watch the two vids - most informative, particularly about how the M4 improves on the M3 in relation to DSP tasks. Bottom line is - if you want FIR then M4 only takes you so far, but for more general purpose DSP (IIR biquads perhaps) its really attractive. And from NXP, its bound to be affordable - I have an NXP Cortex M3 based dev board which cost just $16.
Jai,
It's planned to support various sampling frequencies. 96k is a sure bet.
Regarding your DACs, it should be doable with some work. For example, there's only 1 clock line for all 3 stereo output lines, so you'll have to find a way to deliver the clock to each DAC. Same applies to FS.
Best,
chaparK
Ha..
I'm in no hurry as I have yet to buy any power amps or get a USB to SPDIF/i2S streamer; BUT am quite excited about the prospect of this DSP.
Any news?
I've not got around to it yet but will attach a few pictures/flow diagrams of what I would intend to do with the board to see if it is feasible.
Hope the development is going well and things are prgressing guys.
Also: Can anyone point me in the right direction for I2S implementation as stated in the above post? ie how would I get spidif nito the card. And also the best method of getting a clean, good sounding output that has volume control (for use in my own pre-amp/dsp).
Kind regards,
Jai
I'm in no hurry as I have yet to buy any power amps or get a USB to SPDIF/i2S streamer; BUT am quite excited about the prospect of this DSP.
Any news?
I've not got around to it yet but will attach a few pictures/flow diagrams of what I would intend to do with the board to see if it is feasible.
Hope the development is going well and things are prgressing guys.
Also: Can anyone point me in the right direction for I2S implementation as stated in the above post? ie how would I get spidif nito the card. And also the best method of getting a clean, good sounding output that has volume control (for use in my own pre-amp/dsp).
Kind regards,
Jai
Last edited:
MiniDSP - miniDIGI
is one option for getting SPDIF in. There are a few other choices around too.
As for volume control, I assume there will be digital volume control. If that's not suitable, your best bet is a PGA chip solution, ie.
IRVC2 remote volume control with LCD
http://www.diyaudio.com/forums/digital-line-level/175167-vc23-my-new-pga2310-project.html
is one option for getting SPDIF in. There are a few other choices around too.
As for volume control, I assume there will be digital volume control. If that's not suitable, your best bet is a PGA chip solution, ie.
IRVC2 remote volume control with LCD
http://www.diyaudio.com/forums/digital-line-level/175167-vc23-my-new-pga2310-project.html
I was looking at this unit. Very flexible for the price.
I quite like the look of the twisted pear 4:1 mux unit but it is a little pricey.
I am particularly interested in devices that can be controlled via I2C. The arduino and lcduino (up-coming) units look like a fantastic and easy (ish!) way to make my ideal pre-amp/crossover unit all in one. Are there any I2C serial control ports on the DSP units? Ie will anything be switchable from a unit such as the arduino or will it be jumper based (not too much of a problem, I imagine the arduino will be able to control relays to open/close jumpers on such a board).
Regards,
Jai
I quite like the look of the twisted pear 4:1 mux unit but it is a little pricey.
I am particularly interested in devices that can be controlled via I2C. The arduino and lcduino (up-coming) units look like a fantastic and easy (ish!) way to make my ideal pre-amp/crossover unit all in one. Are there any I2C serial control ports on the DSP units? Ie will anything be switchable from a unit such as the arduino or will it be jumper based (not too much of a problem, I imagine the arduino will be able to control relays to open/close jumpers on such a board).
Regards,
Jai
ChaparK, this is exactly I was hoping to see on DIYAudio. I just came across this thread and I would be very excited to see this came alive. The fact I like a lot is that your project will give a lots of flexibility in the terms of I/O and that people could choose their own input boards and more importantly DAC boards. One thing that someone in the beginning of this thread mentioned, but I have not seen any discussion developed: Bohdan of Bodzio fame, maker of exceptional speaker design software Sound Easy has developed and is further developing, really great software - the Ultimate Eq. Just to mention he is also from Australia. He was actively looking for DSP developer / programer to partner with. So far the Ultimate EQ is the best software out there, because it combines exceptional measuring features found in Sound Easy with DSP capabilities for Xover making and filtering. With that, by executing measurement, a negative mirror of speaker response is applied which in return creates a ruler flat response not achievable by any other means. If you are not aware of Bohdan's quest, I will be very happy to provide you with the link.
I think that Steph_tsf was trying to make us aware to that approach and that Doug20 was missing that. Just having DSP based crossover with various slopes and having available EQ features is something, as Steph_tsf pointed out, that was done 10 years ago by Behringer and is readily available for only $ 250. The revolutionary capability of modern DSP would be best used if there are available combined features of measurement or at least import function, and mirror correction. This is used for speaker response correction as well as for room response. I think since ChaparK is at the point of designing completely new software / hardware than it would be very smart going this route since nothing like this exist, at least for the DIY community.
I will be very interested in ordering this board, in a moment you announce it to be available. ChaparK, good luck with this project and thank you for making this available to us!
I think that Steph_tsf was trying to make us aware to that approach and that Doug20 was missing that. Just having DSP based crossover with various slopes and having available EQ features is something, as Steph_tsf pointed out, that was done 10 years ago by Behringer and is readily available for only $ 250. The revolutionary capability of modern DSP would be best used if there are available combined features of measurement or at least import function, and mirror correction. This is used for speaker response correction as well as for room response. I think since ChaparK is at the point of designing completely new software / hardware than it would be very smart going this route since nothing like this exist, at least for the DIY community.
I will be very interested in ordering this board, in a moment you announce it to be available. ChaparK, good luck with this project and thank you for making this available to us!
Last edited:
Is it posssible for it to run a convolution program like BruteFIR? As crossover, delay, PEQ etc can all be done within the filter, just having a convolution engine running various can solve a whole lot of problem
This is my current implementation. Using PC running BruteFIR
The Building of a Pure Digital System
This is my current implementation. Using PC running BruteFIR
The Building of a Pure Digital System
Thanks you so much, AR2 ! One or two analog inputs for measurement mikes should be added on the list.
Must say that I love the ackcheng approach : bruteFIR.
Today it takes a whole PC running Linux.
Very soon it will only take a BeagleBoard running Linux.
BeagleBoard.org
On the BeagleBoard, try attaching a WM8580 CODEC using the TDM mode (kind of multichannel I2S). You get a full digital (S/PDIF input) stereo 3-way crossover. With two analog inputs as extra. Mouser has the WM8580 in stock, no minimum order.
WM8580 codec
Later on, a custom board hosting a NXP LPC43xx (dual core with a Cortex-M4 core as MACC unit) could be developed, running a simplistic OS instead of Linux, with the Cortex-M4 running the DSP code of bruteFIR.
NXP LPC43xx Series Device Highlight
Today you can try the DSP code on the Cortex-M4 using the Freescale Kinetis K60.
http://be.mouser.com/ProductDetail/...GAEpiMZZMvu0Nwh4cA1wT%2bVLtNmflj2Dk0zJTRM3Dw=
The synchronous serial interface of the Freescale Kinetis K60 seems specifically designed for interfacing multichannel audio CODECS using TDM. How lucky we are !
Must say that I love the ackcheng approach : bruteFIR.
Today it takes a whole PC running Linux.
Very soon it will only take a BeagleBoard running Linux.
BeagleBoard.org
On the BeagleBoard, try attaching a WM8580 CODEC using the TDM mode (kind of multichannel I2S). You get a full digital (S/PDIF input) stereo 3-way crossover. With two analog inputs as extra. Mouser has the WM8580 in stock, no minimum order.
WM8580 codec
Later on, a custom board hosting a NXP LPC43xx (dual core with a Cortex-M4 core as MACC unit) could be developed, running a simplistic OS instead of Linux, with the Cortex-M4 running the DSP code of bruteFIR.
NXP LPC43xx Series Device Highlight
Today you can try the DSP code on the Cortex-M4 using the Freescale Kinetis K60.
http://be.mouser.com/ProductDetail/...GAEpiMZZMvu0Nwh4cA1wT%2bVLtNmflj2Dk0zJTRM3Dw=
The synchronous serial interface of the Freescale Kinetis K60 seems specifically designed for interfacing multichannel audio CODECS using TDM. How lucky we are !
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.