Hi Pierre,
I've followed your progress avidly for the last year or so & have been drooling with anticipation for it's realisation as I think it will be the real next step for those interested in high end audio.
It addresses all of the known & emerging issues with PC audio (PS, jitter, grounding, clock) and marries them to a FPGA for flexibility (digital crossover, digital filters, etc) along with the ability to handle some new concepts (LDR vol control) & probably future audio developments.
It's great to now see a group project emerging. I would love to pitch in but my programming skills are long in the past. However, I lecture in a college of technology & I'm sure I could find some young guns who would be delighted to take on the challenge of writing drivers for JACK/Linux and/or Windoze - only problem is college is not back until Sept.
I have built & continue to build & experiment with DACs, I/Vs & have just started with LDR vol control. So, if I can offer my services in this area I would be delighted to participate.
John
I've followed your progress avidly for the last year or so & have been drooling with anticipation for it's realisation as I think it will be the real next step for those interested in high end audio.
It addresses all of the known & emerging issues with PC audio (PS, jitter, grounding, clock) and marries them to a FPGA for flexibility (digital crossover, digital filters, etc) along with the ability to handle some new concepts (LDR vol control) & probably future audio developments.
It's great to now see a group project emerging. I would love to pitch in but my programming skills are long in the past. However, I lecture in a college of technology & I'm sure I could find some young guns who would be delighted to take on the challenge of writing drivers for JACK/Linux and/or Windoze - only problem is college is not back until Sept.
I have built & continue to build & experiment with DACs, I/Vs & have just started with LDR vol control. So, if I can offer my services in this area I would be delighted to participate.
John
Re: Re: Hackers wanted for Ethernet DAC team project.
I want to switch to BGA for the FPGA, so I'm now looking for a PCB fab that can build the PCB prototype : 4 layers (maybe 6), 6.6 mil track/gap, and especially 0.2mm drill ! Any hnts ?
Purpose is to connect it to a DAC using I2S, so yeah, you probably could. The Master clock is not on the FPGA board, of course, it has to be on the DAC board with the reclocking circuits.
For now, just a few. Later, maybe we can produce more... but it would be unmanageable to handle newbies on a R&D project
Very cool we'll see about that when the FPGA hardware prototype works. Free student exploitation is always welcome (heh heh)...
Loot at http://audio.peufeu.com, the version 1 is described. This version will be a little different (faster RAM and another MAC chip), but this doesn't change the system architecture.
It seems the software part of the team is already crowded
I want to switch to BGA for the FPGA, so I'm now looking for a PCB fab that can build the PCB prototype : 4 layers (maybe 6), 6.6 mil track/gap, and especially 0.2mm drill ! Any hnts ?
lykkedk said:Peufeu, i was just wondering if i could connect it to my dac, that's all
Could you tell me, where you bought your'e fpga ?, and howto connect it to the i2s bus then ?
Purpose is to connect it to a DAC using I2S, so yeah, you probably could. The Master clock is not on the FPGA board, of course, it has to be on the DAC board with the reclocking circuits.
Tazzz said:
For now, just a few. Later, maybe we can produce more... but it would be unmanageable to handle newbies on a R&D project
jkeny said:
Very cool
we'll see about that when the FPGA hardware prototype works. Free student exploitation is always welcome (heh heh)...JonHarrison said:
Loot at http://audio.peufeu.com, the version 1 is described. This version will be a little different (faster RAM and another MAC chip), but this doesn't change the system architecture.
It seems the software part of the team is already crowded
Hi Peufeu,
I’ll also be watching your progress with eager anticipation!
You shouldn’t have problems getting your board fabricated.
I’m currently working on the initial stages of a PCB layout for a 484 ball BGA part. I think we’ll go straight to an 8 layer board to get the all the balls wired out and to ensure signal integrity. Don't forget that you'll need at least 2 layers just for power and 0V. Possibly three if you decide on a local plane for the core voltage.
I’ll be able to offer specific advice about design rules once we get a bit further into the design and once you’ve chosen your FPGA package. I’m also happy to review your schematics and layout.
Cheers, Ralph
I’ll also be watching your progress with eager anticipation!
You shouldn’t have problems getting your board fabricated.
I’m currently working on the initial stages of a PCB layout for a 484 ball BGA part. I think we’ll go straight to an 8 layer board to get the all the balls wired out and to ensure signal integrity. Don't forget that you'll need at least 2 layers just for power and 0V. Possibly three if you decide on a local plane for the core voltage.
I’ll be able to offer specific advice about design rules once we get a bit further into the design and once you’ve chosen your FPGA package. I’m also happy to review your schematics and layout.
Cheers, Ralph
Thanks for the offer.
I will not be needing 484 pins, though !
The PQ208 package has enough I/0 pins for the design ; however it has big drawbacks :
- its EMC performance is bad.
- the fattest available FPGA is Spartan-3E 500. I would rather start the project with a larger one even if, in the end, only 50% is used, rather than hitting the wall. I have made an estimation of the needed slices and Spartan-3E 500 is more than enough for the base system, but things like oversampling / FIR filtering are another question.
- there are enough IO pins left for a DAC, but not really enough for future-proofing (ie. consider all the interesting features, like volume control,
So, a BGA256 should be enough IO-wise. However, I might use a 320 ball BGA (without routing all the pins) simply because this one has a larger range of Spartan FPGAs available, which would allow using a more powerful one.
Right now, the main problem is finding a PCB prototyping house that is :
- European (no import taxes)
- Handles 1mm BGAs : the condition is [ track + 2x gap + via diameter ] < 1 mm
- Isn't horrendously expensive
So far, hard to find !
I will not be needing 484 pins, though !
The PQ208 package has enough I/0 pins for the design ; however it has big drawbacks :
- its EMC performance is bad.
- the fattest available FPGA is Spartan-3E 500. I would rather start the project with a larger one even if, in the end, only 50% is used, rather than hitting the wall. I have made an estimation of the needed slices and Spartan-3E 500 is more than enough for the base system, but things like oversampling / FIR filtering are another question.
- there are enough IO pins left for a DAC, but not really enough for future-proofing (ie. consider all the interesting features, like volume control,
So, a BGA256 should be enough IO-wise. However, I might use a 320 ball BGA (without routing all the pins) simply because this one has a larger range of Spartan FPGAs available, which would allow using a more powerful one.
Right now, the main problem is finding a PCB prototyping house that is :
- European (no import taxes)
- Handles 1mm BGAs : the condition is [ track + 2x gap + via diameter ] < 1 mm
- Isn't horrendously expensive
So far, hard to find !
Hi Peufeu,
Have you tried this company ? http://www.realtime.ie/services.html
Edit: Another one http://www.litho-circuits.com/pcb_assembly.html
Might be worth firing them a RFQ?
John
Have you tried this company ? http://www.realtime.ie/services.html
Edit: Another one http://www.litho-circuits.com/pcb_assembly.html
Might be worth firing them a RFQ?
John
Have you tried this for schematic S/W with BGA layout?
http://www.bme.ie/bae/baedoc/uh_en.htm#BAEUH_XX
Edit: I hope I'm not bombarding you with irrelevant links but here's another PCB prototyping house http://www.mirustechnologies.com/manufacturing.html
John
http://www.bme.ie/bae/baedoc/uh_en.htm#BAEUH_XX
Edit: I hope I'm not bombarding you with irrelevant links but here's another PCB prototyping house http://www.mirustechnologies.com/manufacturing.html
John
I haven't managed to make much progress on starting on the Jack driver investigations, but I do have a question:
Is there anything in the plan to allow input from external I2S sources using a different master clock? I'm thinking specifically of the possibility of tapping into the I2S lines of a dvd player (i.e. http://www.diyaudio.com/forums/showthread.php?s=&threadid=95972&perpage=25&pagenumber=1) to get multi-channel dvd-a/sacd input capability. The idea would be to stream this back to the PC for recording, processing etc, which would then get played back 'normally' over the multi-channel output.
Seems like a cool capability, but it might complicate clock management. This is particularly true if you have to detect the incoming clock configuration to determine sample rate etc.
Is there anything in the plan to allow input from external I2S sources using a different master clock? I'm thinking specifically of the possibility of tapping into the I2S lines of a dvd player (i.e. http://www.diyaudio.com/forums/showthread.php?s=&threadid=95972&perpage=25&pagenumber=1) to get multi-channel dvd-a/sacd input capability. The idea would be to stream this back to the PC for recording, processing etc, which would then get played back 'normally' over the multi-channel output.
Seems like a cool capability, but it might complicate clock management. This is particularly true if you have to detect the incoming clock configuration to determine sample rate etc.
Hello Peufeu,
great to see developments with your project. I've followed your adventures with the atmel board and with the diy fpga mk.II. At one moment I was ready to buy the atmel board you previously tryied but then I've seen you've runned in troubles with it.
For now I have two ak4396 and two ad1955 chips waiting for something like this. This means 8 channels, crossover-ready.
I think it's a good thing to use a bigger FPGA just to be sure you have the space for future development. However I somehow doubt that crossover and drc or other signal procesing should be done in fpga. Like you've mentioned including these would result in big source of heat and emi in the middle of the dac... something to avoid.
What do you consider for upsampling? Again, doing it in the pc it seems a good choise as there is a better control on the algoritm used. (I've seen some tests for software upsamplers, the quality seems to vary a lot). Another advantage is the offline upsampling thus not made in real time. Some other options could include hardware upsampling like ad1896 or the src4192.
Good luck
great to see developments with your project. I've followed your adventures with the atmel board and with the diy fpga mk.II. At one moment I was ready to buy the atmel board you previously tryied but then I've seen you've runned in troubles with it.
For now I have two ak4396 and two ad1955 chips waiting for something like this. This means 8 channels, crossover-ready
. I think it's a good thing to use a bigger FPGA just to be sure you have the space for future development. However I somehow doubt that crossover and drc or other signal procesing should be done in fpga. Like you've mentioned including these would result in big source of heat and emi in the middle of the dac... something to avoid.
What do you consider for upsampling? Again, doing it in the pc it seems a good choise as there is a better control on the algoritm used. (I've seen some tests for software upsamplers, the quality seems to vary a lot). Another advantage is the offline upsampling thus not made in real time. Some other options could include hardware upsampling like ad1896 or the src4192.
Good luck
Well, I'm not dead (yet), but I had a very hectic time since the last email. Tax forms, preparing to sue some of my tenents over something like 4000 euro of unpaid rents (one of them completely destroyed the apartment), even made holes in the floor, etc, well that's life. So, I didn't have too much time to learn Altium and do this BGA layout, as you can guess.
Tomorrow I leave for a much needed 3 week vacation in Asia, and when I'm back, I will have to finish destroying said appartment and put new floor tiles, new kitchen, fix the leaky bathroom, change the broken windows, remove the cat "droppings" from everywhere, plug the holes in the walls, and rent it to someone new. Phew.
So, I be back when I be back, but the project isn't dead.
See you.
Tomorrow I leave for a much needed 3 week vacation in Asia, and when I'm back, I will have to finish destroying said appartment and put new floor tiles, new kitchen, fix the leaky bathroom, change the broken windows, remove the cat "droppings" from everywhere, plug the holes in the walls, and rent it to someone new. Phew.
So, I be back when I be back, but the project isn't dead.
See you.
Well, I'll be OK and the guy is actually quite nice, but well, he's the type who attracts the bad luck from hell since he has two left hands and tried to install a home cinema with projector (and he couldn't handle a power drill so, huge holes everywhere) ; he got fired (hence, no money and rent) and went in holiday and lent the flat to a friend of his who is apparently into chainsaw sculpting or something, I don't know, but how can you do that to a floor, I mean it looks like someone dropped spike mines on it., you can see through the boards. All this in a 12 square meters parisian student room of course (including the home cinema, but all that remains now are the drill holes for the screen, 5.1, projector, and it looks like he tried a few different places for each speaker. He also had a movie production studio with 5 PC's in there. I don't know where he slept.). I'm so glad I don't live in paris anymore.
Then again, it could be worse. I know someone who bought a flat where the previous tenent had ripped the floorboards and burned them in the chimney for heating.
Then again, it could be worse. I know someone who bought a flat where the previous tenent had ripped the floorboards and burned them in the chimney for heating.
Maybe a stupid remark, but (besides the obvious fun of doing your own FPGA), I would like to add some thoughts:
- what about a TOSLINK input? My old PD-S703 has one of these optical outputs, which I suspect is accurate (I have read your tests about comparing CD output to your original CD-R, by the way, but I don't have any special sound equipment on my PC).
- I would prefer a standalone box, that doesn't need an extra computer for working. The Ethernet idea is very nifty, I might add.
- I see that the recent wave of singe-board computers are sporting fast ethernet and rather fast CPUs (e.g. 600+ MHz ARM chips). Wouldn't these systems be able to do all the processing in software, with some assembler libraries thrown in?
Surely, this project looks lots of fun, though!
Cheers,
N.Fotis
- what about a TOSLINK input? My old PD-S703 has one of these optical outputs, which I suspect is accurate (I have read your tests about comparing CD output to your original CD-R, by the way, but I don't have any special sound equipment on my PC).
- I would prefer a standalone box, that doesn't need an extra computer for working. The Ethernet idea is very nifty, I might add.
- I see that the recent wave of singe-board computers are sporting fast ethernet and rather fast CPUs (e.g. 600+ MHz ARM chips). Wouldn't these systems be able to do all the processing in software, with some assembler libraries thrown in?
Surely, this project looks lots of fun, though!
Cheers,
N.Fotis
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.