I'd be far more interested in a USB in/I2S in/I2S out crossover board than a 8 channel USB 'sound card'. There are already plenty of multichannel sound cards available. Of course if it can do both that's the best of both worlds.
The world needs more DIY aimed DSP crossover solutions.....
If this board has more 'grunt' than the miniDSP boards it would be very popular.
The world needs more DIY aimed DSP crossover solutions.....
If this board has more 'grunt' than the miniDSP boards it would be very popular.
I would have thought that the multichannel support would be handled entirely with the USB2 standard audio driver.
You might need a micro to configure the Xmos chip so that it is flagged as 8 channel capable or something though. I'd have thought, however, that besides software, the only added hardware complexity, in going from 2 channel to 8 channel, would be the addition of 3 more I2S data lines on the header. Ditto for a data input. I'd want 8 channels for going surround sound eventually.
RE DSP, the sigmaDSP chip that the minidsp is built around is fully capable of handling 24bit/192 khz. What happens though is that the number of computations it can process per time frame decreases proportionally as the sample frequency increases. So if you're pushing the limits with a 48khz system, you'd need something 4x more powerful to handle 192.
The minidsp chip uses one of the lower power sigma chips, the advantage is that it comes with A/D D/A conversion built in, so makes for a very affordable 2 way stereo DSP. Although in absolute terms the A/D D/A converters are very low quality as far as top notch converters go. But then the basic minidps board only costs $99, so you can't expect ESS sabre or PCM1792/4 quality converters.
I've been experimenting with the daddy of the chips, the ADAU1442, in analogue devices purpose built software allowing you to program the DSP without any coding knowledge. Currently I am using about 1/3 its processing power running 24/192 over 4 stereo channels. As you can see in the attached image this uses quite a lot of xover building blocks too so there's plenty of power left available for other things.
I am currently designing a PCB for the ADAU1442 and intend on seeing just what it's like. If the TPA USB interface ends up being what I want I'll probably be buying one to feed into the DSP board.
It might be worth TPA considering such a board themselves. I think there'd probably be a decent amount of interest in a similar ADAU1442 DSP board, with I2S inputs and outputs. Just another board to add to the mix
In the interest of quality though, the ADAU1442 has an extremely flexible way of handling the various clocks, so I'd imagine very low jitter should still be on the cards, if it's configured right. That however I will try and find out for myself.
fb, I sincerely doubt that the Xmos chip is more capable then the minidsp.
You might need a micro to configure the Xmos chip so that it is flagged as 8 channel capable or something though. I'd have thought, however, that besides software, the only added hardware complexity, in going from 2 channel to 8 channel, would be the addition of 3 more I2S data lines on the header. Ditto for a data input. I'd want 8 channels for going surround sound eventually.
RE DSP, the sigmaDSP chip that the minidsp is built around is fully capable of handling 24bit/192 khz. What happens though is that the number of computations it can process per time frame decreases proportionally as the sample frequency increases. So if you're pushing the limits with a 48khz system, you'd need something 4x more powerful to handle 192.
The minidsp chip uses one of the lower power sigma chips, the advantage is that it comes with A/D D/A conversion built in, so makes for a very affordable 2 way stereo DSP. Although in absolute terms the A/D D/A converters are very low quality as far as top notch converters go. But then the basic minidps board only costs $99, so you can't expect ESS sabre or PCM1792/4 quality converters.
I've been experimenting with the daddy of the chips, the ADAU1442, in analogue devices purpose built software allowing you to program the DSP without any coding knowledge. Currently I am using about 1/3 its processing power running 24/192 over 4 stereo channels. As you can see in the attached image this uses quite a lot of xover building blocks too so there's plenty of power left available for other things.
I am currently designing a PCB for the ADAU1442 and intend on seeing just what it's like. If the TPA USB interface ends up being what I want I'll probably be buying one to feed into the DSP board.
It might be worth TPA considering such a board themselves. I think there'd probably be a decent amount of interest in a similar ADAU1442 DSP board, with I2S inputs and outputs. Just another board to add to the mix
In the interest of quality though, the ADAU1442 has an extremely flexible way of handling the various clocks, so I'd imagine very low jitter should still be on the cards, if it's configured right. That however I will try and find out for myself.
fb, I sincerely doubt that the Xmos chip is more capable then the minidsp.
Attachments
Sigmastudio certainly looks great. Is it possible to use it with 'retail' boards, or just the eval boards?
Having recently ordered 3x Opus, I'm in the market for a I2S based DSP.
miniDSP isn't powerful enough for the price (I'd need 2 boards for stereo 3 way), and this project is progressing slowly due to the firmware development process, something the SigmaDSP platform seems to make obsolete.
Having recently ordered 3x Opus, I'm in the market for a I2S based DSP.
miniDSP isn't powerful enough for the price (I'd need 2 boards for stereo 3 way), and this project is progressing slowly due to the firmware development process, something the SigmaDSP platform seems to make obsolete.
Yep, I am investigating supporting 8 channel right off the bat. But the problem is that with the L1 chip there are (surprisingly) very few open 1bit IOs.
I am seeing what I can do though.
I would think hardware a better approach. The model I had in mind, and I've only recently started researching this, is to USB a stereo stream from the PC (or SPDIF from a transport) and then use a DSP to run the crossover algorithms (a la miniDSP). This would output separate I2S or SPDIF streams to individual high quality dacs, creating a fully separate stereo pair for above and below the cross over point.
Of course this gets expensive for a 3 way speaker system but isn't too bad for a two way speaker (such as mine
).I would have thought that the multichannel support would be handled entirely with the USB2 standard audio driver.
You might need a micro to configure the Xmos chip so that it is flagged as 8 channel capable or something though. I'd have thought, however, that besides software, the only added hardware complexity, in going from 2 channel to 8 channel, would be the addition of 3 more I2S data lines on the header. Ditto for a data input. I'd want 8 channels for going surround sound eventually.
RE DSP, the sigmaDSP chip that the minidsp is built around is fully capable of handling 24bit/192 khz. What happens though is that the number of computations it can process per time frame decreases proportionally as the sample frequency increases. So if you're pushing the limits with a 48khz system, you'd need something 4x more powerful to handle 192.
The minidsp chip uses one of the lower power sigma chips, the advantage is that it comes with A/D D/A conversion built in, so makes for a very affordable 2 way stereo DSP. Although in absolute terms the A/D D/A converters are very low quality as far as top notch converters go. But then the basic minidps board only costs $99, so you can't expect ESS sabre or PCM1792/4 quality converters.
I've been experimenting with the daddy of the chips, the ADAU1442, in analogue devices purpose built software allowing you to program the DSP without any coding knowledge. Currently I am using about 1/3 its processing power running 24/192 over 4 stereo channels. As you can see in the attached image this uses quite a lot of xover building blocks too so there's plenty of power left available for other things.
I am currently designing a PCB for the ADAU1442 and intend on seeing just what it's like. If the TPA USB interface ends up being what I want I'll probably be buying one to feed into the DSP board.
It might be worth TPA considering such a board themselves. I think there'd probably be a decent amount of interest in a similar ADAU1442 DSP board, with I2S inputs and outputs. Just another board to add to the mix
In the interest of quality though, the ADAU1442 has an extremely flexible way of handling the various clocks, so I'd imagine very low jitter should still be on the cards, if it's configured right. That however I will try and find out for myself.
fb, I sincerely doubt that the Xmos chip is more capable then the minidsp.
5th E
An excellent and informative post - thanks. It echoes my thoughts regarding the current miniDSP offering. I am not interested in doing digital analogue conversion at the DSP (unless it's a Sabre32 dac!) just want to get the crossover implemented in the digital domain and use downstream dacs to handle the conversion.
Mark
I agree, that would be such a great thing to be able to do. I've been thinking along these lines for sometime but haven't seen a solution that really cuts the mustard yet. Just think of what you could do with a really high quality USB Transport - USB DSP crossover that outputs I2S - DACs - Poweramps. Not sure where would be optimum point to incorporate volume control.
Dual/triple-ganged Volumite pots at the DAC or incorporate into the DSP - miniDSP allows that.
I want to keep a balanced signal out from the DACs as my Hypex amps sound noticeably better fed a balanced diet and balanced preamps are more complex and costly than they're worth most of the time.
Mark
It depends on how handy you are with the DIY and how the retail implementation has been designed.
AD manufacture a board specifically for interfacing the DSP evaluation modules with sigmastudio. The board is basically a USB to I2C/SPI converter. If you could modify the retail unit so that you could get access to the control pins, then I am sure you could directly control the chip via ADs USB-I2C board.
SigmaStudio is free on request from AD, but they didn't want to release the firmware programmed into the USB to I2C converter. Naturally this means I'd have to buy theirs, if I want to guarantee success. It isn't outrageously expensive @ $99. Note that AD don't allow a company to provide SigmaStudio along with their product, the company would have to write their own software.
Sigmastudio will output the code that's actually used to program the DSP chip, in both hex and binary. So I am sure it wouldn't be too much trouble for someone competent at micro controller programming to write their own program that could upload the hex file to the DSP chip. Currently I know zero about programming, but I have a project coming up where I intend to learn!
The brilliant/interesting thing about the sigmadsp chips is that they aren't hugely expensive. I think the ADAU1442 will cost me something like £15 each? And to make a board with I2S inputs/outputs would require nothing more then the DSP chip, a few caps and resistors, a crystal, a voltage regulator, a E2PROM and the headers. So it's cheap and small seems perfect for TPA board
If you're intending on buying a retail product, to use with ADs USB-I2C module I will say a couple of things.
The 1442 can operate either with a micro controller sending it the code, or on boot it can grab the code from some E2PROM memory. The E2PROM memory isn't built into the chip and has to be provided separately, but once in place it will allow the chip to self boot, without the need for a micro controller. (Instead of the micro sending the 1442 the code on boot up, the 1442 grabs the same code from the memory instead. You have to select what boot mode you want to use by altering the state of one of the 1442s pins.)
You would have to make sure that the retail product is configured in the correct way, or that you could modify it to be, otherwise it'd be a pain to use. Do you know of any retail products that work around the sigma chips besides the minidsp?
That's what I've been looking for a long time.I would think hardware a better approach. The model I had in mind, and I've only recently started researching this, is to USB a stereo stream from the PC (or SPDIF from a transport) and then use a DSP to run the crossover algorithms (a la miniDSP). This would output separate I2S or SPDIF streams to individual high quality dacs, creating a fully separate stereo pair for above and below the cross over point.
Of course this gets expensive for a 3 way speaker system but isn't too bad for a two way speaker (such as mine
Exactly, my aim is to find a surround decoder that outputs NOT 6 analog outputs, but 6 spdif outputs (or even I2S). That way I can use 3 external high quality dacs.
Nowbody looks for this solution, but it doesn't seem an strange thing. Are any of you using a Home cinema sorround at home? Are you using the crappy integrated dacs of the receiver?
Regards,
First decide if you want to implement it in the digital or the analog domain. I would go digital, otherwise would be a pain to control 8 channels. You need 8 mono pots or 4 dual pots or any other similar solution. LDRs could be a solution, but matching 8 and 8 can be a bigger pain.
Going balance only doubles this problem
A solution would be to use a software volume before the transport, that is, the media player itself volume control.
It depends on how handy you are with the DIY and how the retail implementation has been designed.
AD manufacture a board specifically for interfacing the DSP evaluation modules with sigmastudio. The board is basically a USB to I2C/SPI converter. If you could modify the retail unit so that you could get access to the control pins, then I am sure you could directly control the chip via ADs USB-I2C board.
SigmaStudio is free on request from AD, but they didn't want to release the firmware programmed into the USB to I2C converter. Naturally this means I'd have to buy theirs, if I want to guarantee success. It isn't outrageously expensive @ $99. Note that AD don't allow a company to provide SigmaStudio along with their product, the company would have to write their own software.
Sigmastudio will output the code that's actually used to program the DSP chip, in both hex and binary. So I am sure it wouldn't be too much trouble for someone competent at micro controller programming to write their own program that could upload the hex file to the DSP chip. Currently I know zero about programming, but I have a project coming up where I intend to learn!
The brilliant/interesting thing about the sigmadsp chips is that they aren't hugely expensive. I think the ADAU1442 will cost me something like £15 each? And to make a board with I2S inputs/outputs would require nothing more then the DSP chip, a few caps and resistors, a crystal, a voltage regulator, a E2PROM and the headers. So it's cheap and small seems perfect for TPA board
If you're intending on buying a retail product, to use with ADs USB-I2C module I will say a couple of things.
The 1442 can operate either with a micro controller sending it the code, or on boot it can grab the code from some E2PROM memory. The E2PROM memory isn't built into the chip and has to be provided separately, but once in place it will allow the chip to self boot, without the need for a micro controller. (Instead of the micro sending the 1442 the code on boot up, the 1442 grabs the same code from the memory instead. You have to select what boot mode you want to use by altering the state of one of the 1442s pins.)
You would have to make sure that the retail product is configured in the correct way, or that you could modify it to be, otherwise it'd be a pain to use. Do you know of any retail products that work around the sigma chips besides the minidsp?
I don't know any other retail products. I was thinking more along the lines of whether or not low-level DSP programming was required for a finished (retail or DIY) product.
Having just spent hours and hours laying out my first set of PCBs for a Midibox project, I'd be having a shot at my own ADAU1442 board if I wasn't rather scared of soldering a chip with 0.5mm spacing between its pins.
If the SMD soldering were magically done it's not hard to imagine a cheap and easy digital only DSP board booting from E2PROM, and the firmware done in Sigmastudio via AD's USB dongle.
Hey guys, this is all very interesting, but not really in the scope for this module.
It is not intended to be a DSP. You can interface with a DSP if you like.
It is quite simply a USB Audio Class 2 Device with PCM and SPDIF output.
Russ - you're right. Sorry, dude.
A high quality USB to SPDIF module would be well received.
Interestingly the XMOS guys are based a couple of miles from me...
Mark
If you
use a Mac, Pure Music can handle all of your digital cross over needs with very high precision, then output the signals to multiple TPA DACs. Doing the crossover in the computer seems to be a much simpler approach to me, with less hardware.
Although USB2 should handle multiple channels quite effectively, in the real world some people seem to be having problems getting USB2 to operate glitch free with just two channels of 24/192? I wonder how easy it will be to get 6 channels (for a three way xover) of 24/192 operating glitch free through a USB2/I2S interface, in the real world.
use a Mac, Pure Music can handle all of your digital cross over needs with very high precision, then output the signals to multiple TPA DACs. Doing the crossover in the computer seems to be a much simpler approach to me, with less hardware.
Although USB2 should handle multiple channels quite effectively, in the real world some people seem to be having problems getting USB2 to operate glitch free with just two channels of 24/192? I wonder how easy it will be to get 6 channels (for a three way xover) of 24/192 operating glitch free through a USB2/I2S interface, in the real world.
Greetings Barrows,
I'm running crossover/delay software for 8 channels as well, but on a PC. If I wanted to switch that interface hardware over to my Mac Pro as a source, I would have more and different driver, routing and software issues. I inquired here about running filters in XMOS hoping for a greater measure of 'plug-and-play'. I didn't realize that regardless of the number of cores, the XMOS chips seem to have only 16 1-bit ports (if I'm reading the table correctly).
From reading Russ's posts for a number of years now, I believe that he shares enthusiasm for an external crossover solution, right? It's more a matter of how many problems to tackle at once. Eight channel, plug-and-play USB -> every imaginable format would surely be sweet, but the XMOS is going to enforce limited choices. I personally prefer I2S I/O.
Let us know over in the other thread how that tweaked legato sounds after you fire it up, OK?
Cheers,
Frank
Another solution, if you can set up the micro controller, is to use the PGA series of chips from TI. These should have incredibly good tracking.
This is what I'm aiming for. I've managed to make and solder my own PCBs with lead spacing the same as the Sigma chips before, although this was with a 10 pin CS2000.
I've been experimenting with solder paste, as used in wave/reflow soldering and putting the PCBs in the oven. This was for soldering high power LEDs which do NOT like high temperatures at all. The oven method lets you heat everything up slowly and stops you exceeding the bust temperature of the LEDs.
The paste contains a lot of flux so the volume of the paste shrinks considerably when it melts. Now the solder likes to adhere to metal things rather then the FR4 epoxy, preventing legs from being soldered together. This worked wonderfully when soldering Cree XP-G LEDs, so I'm hoping I can use a similar thing with the AD chips. If not it's out with the 0.2mm soldering iron cartridge and solder wick
Using an outboard DSP chip wins hands down over the PC, simply because this places no constraints on the sound configuration of the PC. It keeps everything separate, allowing you total freedom with whatever you do on the PC.
My PC isn't a music only machine, I need it to be able to play games/stream videos via web browsers, play movies and record (why I want a 2 channel I2S input with the Xmos
)/playback. All software based DSP stuff that I've seen, appear to come with severe limitations when it comes to using all sorts of other programs.
I strongly agree with this... I have a dac kit from china with onboard usb and it was such a pain to cut an acceptable hole on the chassis.
Russ I hope you are doing the board this way!
Pietro
Sorry guys...
My 2 cents: I strongly disagree. I would much prefer a board mounted USB connector. Having to DIY a very high speed (USB2) cable harness to a separately mounted jack just seems like a great way to have signal degradation. With a board mounted jack, Russ can play close attention to impedance of the traces, and get everything right. Considering how sensitive even async USB is to cable selection, I do not want to risk introducing problems via flying wiring from board to connector. Remember, USB2 data transmission is a very speed signal.
My 2 cents: I strongly disagree. I would much prefer a board mounted USB connector. Having to DIY a very high speed (USB2) cable harness to a separately mounted jack just seems like a great way to have signal degradation. With a board mounted jack, Russ can play close attention to impedance of the traces, and get everything right. Considering how sensitive even async USB is to cable selection, I do not want to risk introducing problems via flying wiring from board to connector. Remember, USB2 data transmission is a very speed signal.
- Status
- Not open for further replies.