Modular multi channel DIY AD/DA Box

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
jdbakker said:
Have any specific questions? I generally use the Xilinx stuff at work, and several of the Digilent eval boards are cheap and powerful enough.

-- snip --

I'd suggest sticking to 100TX, as that is nearly universally compatible and 'light' enough to consider implementation on a fast (ARM) microcontroller instead of (/in addition to) an FPGA. Three dozen channels at 24/96 should be enough for a start, no?

JDB.

I'm looking for a FPGA core that is cheap, easy to obtain in the US and EU, is DIY solder-able (no funky BGA packaging), available as an eval/devel board with 100Mbit Ethernet in an arrangement that would be easy to copy to our interface board and has a high enough clock to push 32 bidirectional 96k/24-bit channels to Ethernet (though the reality of 100tx being able to handle this data transfer rate uncompressed is questionable). EDIT: and the eval board should have enough pins available to actually test bi-directional 24-ch i2s operation.

Possibly this one?


We had a discussion with rkn80 about how many channels it would be cool to push.. there was an idea of supporting up to almost 200.. that's hard even for gigabit to handle reliably as a developer from a next-door firm tells me they have a hard time getting 500mbits out of a gigabit chip. I'm happy to keep it 100tx, though that might limit people to 24ch in/out until the driver would get multihead.
 
Spartan3 looks like a reasonable choice for price and availability (can't say too much about it's capabilities yet :D), and it's available in QFP packages.
Just recently I was also looking into getting a development board, maybe you want to take a look here as well.
There are also several with on board ethernet, though the ones with gigabit are a bit more pricey. Good discounts for students ;).
Still an ARM or something similar is probably simpler to implement for an actual working unit, as JDB pointed out. They also come with hardware integrated ethernet.


For me something like 32-48 channels out would be totally enough, in maybe something like 24 max. ASIO only would be fine for me as well.


I'm really excited about your progress! Be sure to keep us posted.
 
baadc0de said:
I'm looking for a FPGA core that is cheap, easy to obtain in the US and EU, is DIY solder-able (no funky BGA packaging), available as an eval/devel board with 100Mbit Ethernet in an arrangement that would be easy to copy to our interface board and has a high enough clock to push 32 bidirectional 96k/24-bit channels to Ethernet (though the reality of 100tx being able to handle this data transfer rate uncompressed is questionable).

While several smaller FPGAs are available in QFP, I wouldn't consider those DIY solderable, at least not for the average DIYer. There's also the question how cost/time-effective it is to spin a multi-layer PCB just for this project. I would suggest looking at an existing FPGA board, preferably with Ethernet connectivity, and using that as the core for the digital transmitter.

baadc0de said:

Looks workable, but it might be cheaper to order them directly from Digilent.

As for channel count: it's a bit hackish, but you could always have each FPGA board handle up to 32 channels to/from 100B-TX and connecting these to ports on a 100/1000-switch (or multiple PCIe boards) if you need more channels.

JDB.
 
Just an idea:

What about using a Blackfin DSP for it? Some of the evaluation boards are cheap and come with a 10/100 Ethernet. Most of the basic software like TCP/IP stack etc. is available, I2S is already there, you can program it in C and you get some DSP power.
I've seen DIY boards with Blackfins on it already in the OpenHardware group.

Raphael
 
An update: I've just finished the design of an simpler AES input board that can be used together with the DAC modules and does not need a microcontroller to be programmed.
This module offers 2 stereo / 4 mono AES inputs and the same I2S connector layout like on the other boards. Next week I'll order a prototype of it and as soon as the module is working I'll start the group buy for the DAcs and this module.

Raphael
 
volker said:
Spartan3 looks like a reasonable choice for price and availability (can't say too much about it's capabilities yet :D), and it's available in QFP packages.

i'd avoid the original Spartan 3 series. Actually, I'd avoid all but the Spartan 3AN parts and here's why:
a) onboard config EEPROM (which looks like an SPI part and is big enough to be used for more than just config data),
b) requires only two supplies, +1.2V for core and +3.3V for the I/O (unless you need a +2.5V or other rail).

Having said that, S3E has more package options. S3AN50 is the only flavor in a non-BGA package.

The S3A/AN starter kit has an Ethernet PHY and connector. Do the MAC in the FPGA, which is the 700 part.

Oh, do you need 5V I/O capability? That limits options to the older families.

-a
 
Hi there,

I am new to the board and new to DIY Audio Stuff. I really think this Project rocks, but i guess i will wait a little until i will give it a try. i think i will start with the SSL Compressor clone or something. Is there a SVN (Sourceforge or whatever) and maybe a WIKI for the drivers yet? Or is baadc0de doing all the work by himself?  If help is needed, I would be happy to help. If a svn server or a wiki is needed, i could set them up easily on a free webspace somewhere. If there is work with the actual programming of the drivers, i think i could help as well. i know C. C++, C# and Assembler but i dont have much experience with ic programming yet, but i am looking into it.

Best Regards,

Klaus
 
Minotaurus said:
Hi there,

I am new to the board and new to DIY Audio Stuff. I really think this Project rocks, but i guess i will wait a little until i will give it a try. i think i will start with the SSL Compressor clone or something. Is there a SVN (Sourceforge or whatever) and maybe a WIKI for the drivers yet? Or is baadc0de doing all the work by himself?  If help is needed, I would be happy to help. If a svn server or a wiki is needed, i could set them up easily on a free webspace somewhere. If there is work with the actual programming of the drivers, i think i could help as well. i know C. C++, C# and Assembler but i dont have much experience with ic programming yet, but i am looking into it.

Best Regards,

Klaus

Hi. The drivers are being written in C(++), by me alone, currently targeting windows only. I am currently wrestling with WDF to get best of both worlds (faster access to ethernet and still remain a userspace driver). I'm not ready to release source code or commit to a multi-developer setup yet, however, once the hardware to support has been selected/built and tested [AD boards, DA boards, IF board], I will open the drivers and have other people, if not anything else, check the driver for errors and spot possible improvements.

Andy Peters said:
volker said:
Spartan3 looks like a reasonable choice for price and availability (can't say too much about it's capabilities yet :D), and it's available in QFP packages.

i'd avoid the original Spartan 3 series. Actually, I'd avoid all but the Spartan 3AN parts and here's why:
a) onboard config EEPROM (which looks like an SPI part and is big enough to be used for more than just config data),
b) requires only two supplies, +1.2V for core and +3.3V for the I/O (unless you need a +2.5V or other rail).

Having said that, S3E has more package options. S3AN50 is the only flavor in a non-BGA package.

The S3A/AN starter kit has an Ethernet PHY and connector. Do the MAC in the FPGA, which is the 700 part.

Oh, do you need 5V I/O capability? That limits options to the older families.

-a

I/O is required to be a multitude of TTL pins. And ofcourse, some kind of ethernet, the faster the better :) So you would suggest the S3AN50?
 
baadc0de said:
I/O is required to be a multitude of TTL pins. And ofcourse, some kind of ethernet, the faster the better :) So you would suggest the S3AN50?

As long as the inputs don't have to deal with 5V TTL (3.3V LVTTL works well) then all's well with the S3AN parts.

-a
 
@baadc0de:
It's  been a long while since I wrote about my experiments to develop a WDM driver to send audio to ethernet.
I looked into the code again and re-built the driver i had from the DDK.
As of to my knowledge, the best template for a driver we would need are the msvad samples.
I managed to get the WINDDK 7600.16385.0 sample driver vadpcmex.sys working without crashing at all now and after a few tweaks it shows up it's interface correctly to Sonar with Kernel Streaming WDM, to DirectX, whatever... with up to 8 channels (for now) up to 96khz/24bit. It is only capable to stream the audio to a wave file on disk as long as i reduce to 2 channels (harddriver performance reasons...), so I am not sure about the realtime capability for ethernet device, but pretty optimistic.
As far as I know there are no 'known working' examples of such a kind of driver for Win7 right now, but to my knowledge msvad is working up to Vista at least, and there you will have to deal with WDM anyway....Why wouldn't be a kernel mode driver be preferable?  ???

Kind regards,
Martin
 
Audio over ethernet is something like what this guys do ?

http://www.networksound.com/Digsnake.html
 
My first SMD!

SMD.JPG


It sure ain't neat, but it works!

To all the people who aren't sure if they're capable of this (I included myself in this category until I tried), just dive in and give it a go.  It'll take some practise to get it looking as neat as my through hole work, but I've checked out the work in the photo with a multimeter, and it measures as it should.

The brown marks you can see on the photo are where I have added flux before soldering.  This does seem to make the components 'take' to the board more easily, but it doesn't look very nice.  Is there any problem except cosmetics with leaving it there?

 
You can always clean the flux residue with isopropilene alcohol or special solvent.
Good flux, don't afraid to put more, that's all. 0204 smd's are also solderable:)
 
It has been a bit quiet on this front lately. What has everybody been doing? I'm currently one AES-EBU cable (to RME HDSP) short of starting testing the input cards I have. I built a PSU based on some super regulators, even for the 3.3V line.

rkn80, What's happening with the output part of this project? Are there going to be PCB available?

And how should clocking be set up with the between input and output using your AES-EBU PCB's?
 
Perhaps everybody is soldering?  ::)

Seriously I've been away for some days and currently I'm processing and shipping the last orders for AES outputs. I'm planning to start the DAC PCBs order phase next week but I'm not sure if that is a good idea due to holiday time.  ???

Regarding clocking I'll write something in the build help thread.

Raphael
 
Curious - how much would the total cost be?

Say if it's a unit similar to the SSL Alpha Link with 24ins / 24outs.

Been following this post for awhile now...  excited/impatient/skeptical/curious/optimistic ... :D
 
Well,... i kept putting it off in favor of other things, until now job & family & studio mainly dominate my time. I don't feel like I can pursue this much further, so I'm selling my A/D kits (at a discount, ofcourse)..

There is much potential in this project (especially with AES in/out boards and similar) and I hope somebody picks up the idea of an ethernet controller.. When pressed for inputs for larger sessions, i simply had to buy an alphalink system :-[ and the incentive is pretty much gone..
 
Back
Top