Audiophile version
Hi,
I am looking into creating an audiophile digital crossover. As co-developer of the APOX volume control, I think there may be merit in a unit designed for best audio performance.
Instead of just allowing Nth order slopes, I am hoping to allow arbitrary mag/phase plots to be downloaded and the appropriate filter coefficients calculated. Of course, standard butterworth and other types would be supported.
I am not sure that a DSP is really needed for the calculations. I would rather use a couple of high speed microcontrollers or even a low end X86 type chip (built in FPU).
Could people list a few high level specs that they would like to be included...
In general, I was anticipating:
SPDIF or I2S inputs
Analog Inputs: 24/96 ADC
six or eight channels 24/96 DAC (3 or 4 way)
Digital output?
Volume control via PGA43XX or PGA23XX in analog domain
Etc...
Dale
Hi,
I am looking into creating an audiophile digital crossover. As co-developer of the APOX volume control, I think there may be merit in a unit designed for best audio performance.
Instead of just allowing Nth order slopes, I am hoping to allow arbitrary mag/phase plots to be downloaded and the appropriate filter coefficients calculated. Of course, standard butterworth and other types would be supported.
I am not sure that a DSP is really needed for the calculations. I would rather use a couple of high speed microcontrollers or even a low end X86 type chip (built in FPU).
Could people list a few high level specs that they would like to be included...
In general, I was anticipating:
SPDIF or I2S inputs
Analog Inputs: 24/96 ADC
six or eight channels 24/96 DAC (3 or 4 way)
Digital output?
Volume control via PGA43XX or PGA23XX in analog domain
Etc...
Dale
Member
Joined 2002
You'll be a lot happier with a SHARC than with some x86 CPU. The performance per unit heat produced will be much much higher, and the pin count will be lower.
I've been playing with digital x-overs in software, and I'd like to see phase, slope, 3dB frequency, and passband attenuation all controlled in software.
I guess it really depends on how far toward the high end you want to go. Each channel could have digital outputs, single-ended and balanced analog outputs, etc. The main concern quickly becomes having enough room for all the connectors 🙂
I've been playing with digital x-overs in software, and I'd like to see phase, slope, 3dB frequency, and passband attenuation all controlled in software.
I guess it really depends on how far toward the high end you want to go. Each channel could have digital outputs, single-ended and balanced analog outputs, etc. The main concern quickly becomes having enough room for all the connectors 🙂
Great idea Dale
Ok, my suggestions-
Good to have
2 ins, and 3+ sub, or 4 +sub outs.
Option of balanced or unbalanced inputs and outputs
Programable/controlable via Usb and a Unix program, as users of both Macs and Pcs can get Linux/Unix running on their machines.
Adjustable delay for time alignment of drivers.
If possible
Linkwitz transform or other EQ on the sub or bass outputs
Maybe two or three bands of adjustable eq for room correction
Volume controls optional for those that don't need them.
Ok, my suggestions-
Good to have
2 ins, and 3+ sub, or 4 +sub outs.
Option of balanced or unbalanced inputs and outputs
Programable/controlable via Usb and a Unix program, as users of both Macs and Pcs can get Linux/Unix running on their machines.
Adjustable delay for time alignment of drivers.
If possible
Linkwitz transform or other EQ on the sub or bass outputs
Maybe two or three bands of adjustable eq for room correction
Volume controls optional for those that don't need them.
Hi Dale,
Now you’re talkin’! I’ll think about this some more, but here are my first thoughts.
Digital time delay per channel
An extra channel per channel for notch filters, or baffle step comp., or FR tailoring.
Channel by-pass
Level per channel
System configuration memory and/or programmable pre-sets
Rodd Yamashita
Now you’re talkin’! I’ll think about this some more, but here are my first thoughts.
Digital time delay per channel
An extra channel per channel for notch filters, or baffle step comp., or FR tailoring.
Channel by-pass
Level per channel
System configuration memory and/or programmable pre-sets
Rodd Yamashita
Rodd
Channel mutes perhaps, channel bypass could lead to blown tweeters
I think level per channel might be easier to handle in the analogue domain rather than digital, to save processing power, but hey, what do I know😉
At least two presets is definitely a good idea, then you could have one for music and one for HT for instance.
More general thoughts-
A pink noise generator for system setup would be handy...
Channel mutes perhaps, channel bypass could lead to blown tweeters
I think level per channel might be easier to handle in the analogue domain rather than digital, to save processing power, but hey, what do I know😉
At least two presets is definitely a good idea, then you could have one for music and one for HT for instance.
More general thoughts-
A pink noise generator for system setup would be handy...
Hi Al,pinkmouse said:Rodd
Channel mutes perhaps, channel bypass could lead to blown tweeters
Yeah, I guess I could wire straight through channels externally. I thought about that because my Heathkit xover has a by-pass, but it is for the bass range only, so in general, you're probably right.
Rodd Yamashita
Member
Joined 2002
Is this going to be balanced or non-balanced.
Will it be adjustable.?
will there be one main in and 4 or so out's..
My idea.
Main in
Sub out
hi out
mid out
low out
All adjustable..
Desplay ? yes no.?
Will it be adjustable.?
will there be one main in and 4 or so out's..
My idea.
Main in
Sub out
hi out
mid out
low out
All adjustable..
Desplay ? yes no.?
Another quick thought, a fixed high Q high pass filter at say 10Hz, for those of us that still use vinyl would be nice🙂
you asked...
I want:
2x 1394 input with HDCP (audio and video)
1x DVI input with HDCP
1x HDMI input
4x 1394 output
1x DVI output
and enough MACs and memory to do 1000 taps on each output channel at 96k and a delay on the video.
I know getting the product approved for HDCP on the 1394 costs $10000. Who knows how much the HDMI certification will cost. I'm afraid the RIAA and MPAA have ruined digital for us DIYers.
The good news is, though, that you should be able to do all this with a Windows Media Player Plugin on Windows Longhorn when it comes out in 2005. If Microsoft makes proper use of hyperthreading it should be close to hard real time too. I'm using Windows now to do my DSP with a program called Sounds Logical and an M-Audio Delta 1010. I've given up on using speciallized DSP chips. It's just so much more convenient to do your DSP on a PC. It gives you a lot more time to play around with different filters, which in my book is what DIY DSP is all about.
If you're living in the here and now and just want i2s or AES/EBU ins and outs and don't like PCs, get yourself one of the multitudes of cheap EVM modules from TI, Motorola, etc. You will never be able to get close to those prices with a DIY product. You could, however, add some value by making a kit to reroute the i2s outputs going to the DACS into AES/EBU outputs and use outboard DACs. You could apply the AES/EBU kits to CD and DVD-Audio players as well--foil those damn RIAA bastards.
Good luck on the project, John
I want:
2x 1394 input with HDCP (audio and video)
1x DVI input with HDCP
1x HDMI input
4x 1394 output
1x DVI output
and enough MACs and memory to do 1000 taps on each output channel at 96k and a delay on the video.
I know getting the product approved for HDCP on the 1394 costs $10000. Who knows how much the HDMI certification will cost. I'm afraid the RIAA and MPAA have ruined digital for us DIYers.
The good news is, though, that you should be able to do all this with a Windows Media Player Plugin on Windows Longhorn when it comes out in 2005. If Microsoft makes proper use of hyperthreading it should be close to hard real time too. I'm using Windows now to do my DSP with a program called Sounds Logical and an M-Audio Delta 1010. I've given up on using speciallized DSP chips. It's just so much more convenient to do your DSP on a PC. It gives you a lot more time to play around with different filters, which in my book is what DIY DSP is all about.
If you're living in the here and now and just want i2s or AES/EBU ins and outs and don't like PCs, get yourself one of the multitudes of cheap EVM modules from TI, Motorola, etc. You will never be able to get close to those prices with a DIY product. You could, however, add some value by making a kit to reroute the i2s outputs going to the DACS into AES/EBU outputs and use outboard DACs. You could apply the AES/EBU kits to CD and DVD-Audio players as well--foil those damn RIAA bastards.
Good luck on the project, John
New idea😉
Very basic cut down circuit with no external controls and just one analogue/ digital in and 3/4 analogue outputs, programmable by remote that could be built into an active speaker.
Very basic cut down circuit with no external controls and just one analogue/ digital in and 3/4 analogue outputs, programmable by remote that could be built into an active speaker.
Like Pinkmouse, I'd hope the modular route was possible; I'd like to see the basic core be as low cost as possible, especially as I'd probably be bypassing a lot of the frills anyway.
I'd like to see some low-cost core that had digital I/O only. ADC/DACs would certainly add a lot to the cost, and many of us might want to do both on our own.
I did like the idea of L+R sub-out; programmable bass management would be a godsend.
I'd like to see some low-cost core that had digital I/O only. ADC/DACs would certainly add a lot to the cost, and many of us might want to do both on our own.
I did like the idea of L+R sub-out; programmable bass management would be a godsend.
The simplest solution, and a good place to start out, might be to use a pair of TI TAS3103 processors since these use the i2c interface and could therefore plug directly into the APOX-bus. However they have only biquads and so will not allow you to do long FIR's.
If this is too little, then a DSP is the only solution. A microcontroller or x86 will be too slow and/or too difficult to interface to i2s signals. You could choose between TI C67xx, SHARC 21161 or Motorola 563xx. None of these are dip-chip though 🙂
I would leave the ADC, DAC and PGA off the board and let people use their own designs here. Too many different opinions on DAC design to satisfy everyone. SPDIF input would be nice though.
ojg
If this is too little, then a DSP is the only solution. A microcontroller or x86 will be too slow and/or too difficult to interface to i2s signals. You could choose between TI C67xx, SHARC 21161 or Motorola 563xx. None of these are dip-chip though 🙂
I would leave the ADC, DAC and PGA off the board and let people use their own designs here. Too many different opinions on DAC design to satisfy everyone. SPDIF input would be nice though.
ojg
Hi,
Some great ideas. Thanks!
Right now, I am chossing the DSP based on development costs. The Sharc devices have GNU support.
The Motorola CodeWarrior is now on sale at $499, which is not too bad.
The TI devices seem to have the most expensive support costs.
I was planning on a daughter card arrangement with the analog circuits on it.
Has anyone used some of these tool sets?
The SHARC ADSP-21065L is my first choice, with the Motorola
DSP56367 as second...
Dale
Some great ideas. Thanks!
Right now, I am chossing the DSP based on development costs. The Sharc devices have GNU support.
The Motorola CodeWarrior is now on sale at $499, which is not too bad.
The TI devices seem to have the most expensive support costs.
I was planning on a daughter card arrangement with the analog circuits on it.
Has anyone used some of these tool sets?
The SHARC ADSP-21065L is my first choice, with the Motorola
DSP56367 as second...
Dale
Member
Joined 2002
Hi Jason,
$499 is the cost for the CodeWarrior Compiler/linker. I would also need to buy a development kit (board), so that I can play.
The SHARC ADSP-21065 is about $40.00, but there would be a fair bit of glue chips to support it. If you want SPDIF, I would probably add a CS8515 or CS8516.
Dale
$499 is the cost for the CodeWarrior Compiler/linker. I would also need to buy a development kit (board), so that I can play.
The SHARC ADSP-21065 is about $40.00, but there would be a fair bit of glue chips to support it. If you want SPDIF, I would probably add a CS8515 or CS8516.
Dale
Hi Dale
I'm not sure if it's a recommendation or not, but most of the pro audio kit I have looked inside recently seems to use the Sharc chipset😉
Obviously this is a little premature, but another thing comes to mind. Would it be possible, as the board will be using mostly SMT, to get the boards partially prefabbed, as I, and I suspect many others, am a bit wary of that aspect of the DIY.
I'm not sure if it's a recommendation or not, but most of the pro audio kit I have looked inside recently seems to use the Sharc chipset😉
Obviously this is a little premature, but another thing comes to mind. Would it be possible, as the board will be using mostly SMT, to get the boards partially prefabbed, as I, and I suspect many others, am a bit wary of that aspect of the DIY.
You are correct. I would not attempt to hand solder a 200 pin PQFP package. We would have the SMT components assembled and tested...
Yes, the SHARC seems to be a good choice, except for the lack of built-in flash memory.
I would probably have a second microprocessor to handle the user interface and external communications.
Would people be opposed to using a PC for configuration. I know many use MACs, but ....
There are too many parameters to do a simple UI. The PC could also do the coefficient generation. That would make the actual DSP code much simpler...
Dale
Yes, the SHARC seems to be a good choice, except for the lack of built-in flash memory.
I would probably have a second microprocessor to handle the user interface and external communications.
Would people be opposed to using a PC for configuration. I know many use MACs, but ....
There are too many parameters to do a simple UI. The PC could also do the coefficient generation. That would make the actual DSP code much simpler...
Dale
harvardian said:
Pah, you hate us, you really hate us!😀😀
Seriously, I'm sure we could all cope if we had to, but I still think USB would be a nice touch, and probably faster as well if the setup is going to invove a lot of data shifting. Hmm, but then that probably involves writing drivers and such like...Oh well...😉
- Status
- Not open for further replies.