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.
mainmachine said:
Anyone have any thoughts on interfacing these DACs with a PC via firewire?

Nothing that an EE with the relevant experience can't do in a man-month or two. The problem is finding one who will do it for little or no money. As much as I love IEEE1394 it's rather non-trivial, and I don't believe that enough applicable open cores are available to make this a simple matter of plug-and-play.

mainmachine said:
Here's an interesting firewire/ FPGA card that might be applicable:

http://www.orsys.de/322c6713.htm

That has only the transceivers (ie: the easy bits, which are simple enough to bolt onto any FPGA).

The idea is that it would communicate via firewire between a pc DAW and the DAC unit.

mainmachine said:
It seems like SPI would be more useful than I2S for multi channel use

...why? Apart from the fact that most ADC/DACs don't speak SPI you get more signal edges, more transitions, more noise.

mainmachine said:
Anybody else looking into this, or are the majority of people interested in ADAT lightpipe interface?

I am pretty partial to MADI myself, but MADI PC interfaces are rare and expensive. MADI would be much simpler to implement in an FPGA.

I've considered streaming low-latency audio over a (possibly dedicated) Ethernet link. The hardware is doable, of a same level of complexity as MADI, with equal or better performance. The hard bit here is that someone would need to write an ASIO (or similar) driver for it. Under Linux that would be doable, but if you want Windows compatibility (which one? XP? Vista? 7?) it gets pretty hairy pretty fast.

At the end of the day you need to get the data into/out of a DAW, and that tends to limit your options.

JDB.
[ADAT's a bit like McDonalds: nobody over the age of ten is a huge fan but you can get it everywhere]
 
JDB: Thanks for the input!

My lack of digital experience shows in regards to SPI vs. I2S. I was thinking it'd be better from a cabling standpoint.

So how can I get my PC to send audio data from my DAW software to the DAC via I2S with minimal cabling?

It would be great if I could use an available PCI card with simpler cabling than multiple ADATs, like TDIF or AES.
 
mainmachine said:
It would be great if I could use an available PCI card with simpler cabling than multiple ADATs, like TDIF or AES.

Can't get much simpler than that. TDIF is already very close to I2S, but I'm not sure if anyone still makes/sells (new) TDIF cards. Alas, cards with loads of AES links aren't exactly common or cheap either; for equivalent I/O capacity ADAT would still appear to be cheapest.

For 'one cable' it is hard to beat MADI.

JDB.
 
MADI looks great!

The least expensive TDIF solution I could find is the Sydec Mixtreme 192. Two TDIF connectors for 16 channels I/O @48kHz, 8 @ 96kHz for $595 USD direct from Sydec. Not a cost-friendly solution at all.

-Dave
 
RME makes lots of MADI hardware that is excellent, I have some PCI cards I bought for $500 used... not rare or hard to come by. It is very easy to get MADI into the computer AND you can run it over fiber for up to 3 Kilometers. My entire live recording rig runs on MADI, I would LOVE to have a way to build a large analog to MADI converter, say 64 channels (what one MADI stream holds....)
 
If the question is audio over ethernet, AES51 seems to fit the bill. IIRC it will function over ordinary switching hardware as it's a layer2 protocol and uses normal MAC addresses. There are even some provisions to pass through IP traffic so that you can have it on a casual network. Anyway, if an ASIO AES51 driver would be required, I think I could conjure that up, I've had experience with linux and windows MM device drivers, so this could be explored if there is sufficient interest.

Lacking that, I'd endorse MADI. Sure, the cards aren't cheap but the protocol is nice and the cables can be long.
 
baadc0de said:
If the question is audio over ethernet, AES51 seems to fit the bill.

Hadn't seen that one before. It looks... messy, but not too hard to put into an FPGA. Not when compared to 1394 or USB, at any rate.

baadc0de said:
Anyway, if an ASIO AES51 driver would be required, I think I could conjure that up, I've had experience with linux and windows MM device drivers, so this could be explored if there is sufficient interest.

I expect that would be very useful.

baadc0de said:
Lacking that, I'd endorse MADI. Sure, the cards aren't cheap but the protocol is nice and the cables can be long.

Yes. From a design POV I would prefer MADI, but anything that can properly run over commodity hardware such as Ethernet would have the advantage.

I'm considering a FPGA-based I2S-to-whatever board, but don't hold your breath as my to-do list is longish these days.

(mini-rant: Googled AES51 for a bit. So it appears that the AES decided to do their own implementation of ATM-over-Ethernet. Why? What was wrong with the dozen or so existing ATM-E encapsulation methods which have been developed in the past fifteen years? There's very little you can do on the IEEE802-level to make any new ATM encapsulation scheme more audio-friendly. So why not use what's already there? Would that make the committee members' products and solutions not stand out enough as Pro Audio Rated? Is it NIH? Granted, I'm rather opinionated even before reading the full standard, but experience shows that in these cases it's generally warranted if my prejudices are extreme.)

JDB.
[sure, there's nothing new here. If you look hard at MADI you'll see a bunch of FDDI-engineers shouting Here! Take our standard! It never managed to get any market acceptance, but it's technically sound! You folks do sound, right?]
 
I think your rant is spot on. We didn't need AES to invent this, but that's how it is. We can choose to ignore it and make our own if it proves too messy. In any event, I'll have some thorough look over the protocol and the ASIO lib and see what it would take to implement such a beast.
 
Wow, MADI, Audio-over-Ethernet, Firewire... if someone is willing to develop such a board I'll definitly support that.
I know that there are several OpenSource implementations of this protocols on FPGAs. What keeps me away from doing such a design is that I do not own the stuff for programming a FPGA.

Just one note about the current project status: I started to design the AES receiver/transmitter boards. Unfortunately, I had to redesign the receiver board. The problem was that a least one receiver must run as slave so that the I2S port is clocked from the master. That is needed because the PCM4104 has two data inputs but only one lrck and bck input. The DAC expects the two data inputs to be synchron. Of course that cannot be guaranteed if the receivers are sourced with AES signals form different sources and let them recover the clock signals from the incoming data stream.
So I changed to the Cirrus CS8416 receiver chip and they are always used as a master/slave pair  to source one PCM4104.
I'm now thinking if I want to keep master/slave pair design for all receiver when expanded to 24 channels (which would mean that I have 6 chips configured as master and 6 chips as slave) or if only one chip is the master and the other 11 chips are slaves.
To avoid confusion: I'm only talking about the configuration of the I2S port if the DAC I2S port is not the master. Of course it is possible to configure the DACs as master and let them clock the receivers. Besides that all the chips can be run from one system main clock.

Raphael  
 
Hi raphael, quick off topic question. Are u planning on designing a 2 channel high end ad or da? Thanks and sorry for the off topic question.
 
Well, not now. I first want to finish a multi-channel box because that is the main target. Of course it is a nice idea to have at least 2 or 4 channels with highest quality based on a PCM179x or something like that. But that would be definitly to expensive for a multi-channel box with >24 channels.

Raphael
 
Hi guys,
in the meantime I researched a little bit more in the way of diy-able interface technology.
Right now I found two interesting options that are more or less functional building blocks.

Firewire IEEE1394:  I took a look into DICE II chips from TC electronics. In the meantime it is possible to get them in single quantities e.g. at profusionplc.com.
As I understood it, they are (co-)designed by D.F.M, german digital audio specialists. They are already used in several musical devices and are widely configurable, contain a programmable dsp section etc etc.
They are even kind of affordable in small quantities.
However, it looks like the most complete building block available for that kind of task.
Implementation might still be far away from trivial. (even the package size is anything but diy-friendly)

ADAT receiver:
Interestingly, I found a completly open source FPGA core written in VHDL that claims to be tested working (status stable/fpga proven). Might be the lowest latency way possible (?) and might even be improved with integrated s/mux IF someone is capable of designing and testing VHDL in the field of DSP/audio.
http://opencores.com/project,adat_optical_feed_forward_receiver
I must confess personally I have no experience with FPGAs at all aside from some rough theoretical knowledge.

On the opencores site, there is even a MADI core available, however it is still in beta stage, means in simulation and not tested in an fpga. So maybe something to bookmark for future updates.

Kind regards,
Martin
 
Hi rkn80,
              once again thank you for doing this.I was wondering what is the status of the ADAT board?Are you
going to make it?

Best
 
syn,

I decided to delay the ADAT board for a moment because there's is already a diy receiver board for it that can be used together with the DAC. So I think it is better to design an AES input first. I'll go on with ADAT when the ADC is coming alive, starting then with an ADAT transmitter board.

Raphael
 
rkn80 said:
Wow, MADI, Audio-over-Ethernet, Firewire... if someone is willing to develop such a board I'll definitly support that.
I know that there are several OpenSource implementations of this protocols on FPGAs. What keeps me away from doing such a design is that I do not own the stuff for programming a FPGA.

How about USB then? New MacBooks don't come with FireWire support anymore (http://www.apple.com/macbook/specs.html).

This EtherSound stuff looks interesting: http://www.digigram.com/products/product_infos.php?prod_key=11100
 
It's cool in that they probably have an ASIO driver for their Ethernet/audio protocol, but the spec seems to only be available for (paying) licensees as is the FPGA code... If someone can do at least the FPGA hardware part, I'd be interested as stated in creating an AES51 ASIO driver, possibly even a CoreAudio driver. The FPGA software part would be tricky, but I could try my hand at that as well, though, it is hard without the related hardware to develop with.

For those who haven't followed through with the thread, this would mean that you could stream up to about 32 channels of audio at 96k through an inexpensive Ethernet cable that you probably use to connect to the internet; what's more, you could probably just plug the A/D D/A into your router and browse the net/email over the same cable.

For me, this would be a nice solution because I could have a remote recording rack loaded with a racked PC and converter box. After a session I could bring it home for mixing and use a simple Ethernet cable to connect both the PC and converter to my home network, where I could use that same cable to stream files off of my racked recording PC to the mixing PC and use the converters in my converter box for listening or using outboard gear, thus not needing another converter box for mixing purposes but rather reusing the one used for recording. The additional benefit of not needing to buy ADAT or MADI interface cards is there as well..
 
OK, some info on the USB interface option.
There are some nice newer USB-all-in-one adda comverter chips with spdif and direct da chip interfacing available from TI. They need minimal external components but unfortunately, they are all restricted to 16 bit (damn!)....
The dedicated usb audio receivers are a bit more delicate to implement.
ATM, usb is not a favourite for pro-audio at all. Ironically. in the meantime USB runs really stable under all known OSes....
Please be notified that almost all serious manufacturers dropped serious usb development  on pro-audio because they did not see any future for 2ch I/O. (And usb issues in the past did no good for their reputation).
(Most manufacturers i heard of refused to develop for more channels due to stability reasons, despite usb 2.0 availability)
So maybe no field for diy really...
I found ONE interface that claims to interface with ADAT from USB, but did not check it in reality yet...product is EOL, it's a B.ringer.
I guess one should get a ready-made usb card and interface to the converter via an AES/EBU or S/PDIF board.

For the ADAT option:
All I heard about the Wavefront chips is, that they spec really not that good (jitter issues etc), OTOH, they are readily available (but kind of expensive).
If one could live with that, Mikkels board is the way to go.

To be more on top of the field, an FPGA could be the solution really. extremely low latencies, and functionality could be modified/updated 'in hardware'.

I'm still investigating about the DICE chips, and think I read they can be used 'offline' (without IEEE1394 connection) like a format converter (like AES/EBU to i2s, ADAT 2 I2s, Adat to I8s and the like).
Could be a nice option for a ready-made, affordable and tested very functional fpga. needs a flash mem and sdram for the minimal system. The flash needs a minimal application to configure the functionality.

The open core adat receiver is still an option, but for myself I am a bit hesitating to get more into the field of designing and burning fpgas...but if all other options have drawbacks, why not?

Best regards,
Martin

PS: I am very interestedly lurking at the AES51 / ATMoE discussion now. I wish I had beed into fpga before...this stuff is very interesting and opens up quite a new dimension (IMHO any working solution of this kind would be mostly welcome, so I think it is not necessary to stuck to AES51 of MADI compatibility).
BTW, I would not let my router be involved into 64 channel audio streaming, hehe, but just a direct ethernet cable connection, what a great option it would be!
 

Latest posts

Back
Top