8 channel AD converter idea/discussion

[Rochey]

ADAT's SMUX protocol only support 4x96KHz channels.

I'm told this is more of a bandwidth limitation on the plastic fiber used to connect devices optically. Although maybe one of the other guys here can give some more details.

Adding AES (or S/PDIF) to the unit isn't difficult - as the ADAT devices expect either left justified or right justified data, hanging a DIT4096 off the stream (with a buffer) should be relatively easy.

The hardest thing (in my opinion) with anything like this is the clocking and synchronisation. Thats one thing the ADAT devices are useful for

Tobwen: if you need any assistance with you're efforts, please give me a shout.

cheers

R

 

[Kev]

[quote author="Rochey"]... The hardest thing (in my opinion) with anything like this is the clocking and synchronisation. Thats one thing the ADAT devices are useful for .. [/quote]
and that's where my opinion of all this sits also

the ADAT optical send return gives the quickest way to sync a simple 8 channel add on
Ai3 and ADA8000 type of thing.

AND
the only reason people want this is because the DAW interface of choice has one ... perhaps two of these fitted.

IF a high class ADDA is rquired the clock and word distribution becomes the main issue and this something where the APPLICATION NOTES is just not enough.
Much design and layout .. trial and error ... seems to be what seperates the various products.

By the time you have a good working proto a new chip arrives on the scene.
... there is another thread here that mentions one of the ADK chips, the RME and a change of chip post proto or early model.

so NOT easy for a single person to stay at the cutting edge.
AND
at the bottom of the heap the mass manufacturers are in the sit position to make things very cheaply.

Hard to find the advantages in DIY unless it is to make a product that's just not out there in the market place.

BUT
DIY is always fun so don't let me talk you out of anything

good luck people and happy DIY
:thumb:

 

[TobWen]

yeah, jitter is always a big problem ...

that's why the clock is important, like KEV and Rochey quoted.
but wait until it's finished. then we can start a discussion

 

[TobWen]

I requested to remove it ... you gave me the idea to release some prototypes before posting here

so I first will create some prototypes and THEN post information here

 

[jdbakker]

[quote author="Rochey"]The hardest thing (in my opinion) with anything like this is the clocking and synchronisation.[/quote]

Indeed.

I'm working on yet another similar design; one of the things to keep in mind is that in single-speed mode (ie 44k1 / 48k) the PCM420x ADCs really like to have a 128x bit clock, whereas the AL1401A expects a 64x bit clock. This is solveable, but you'd have to use the 'C'-word (CPLD) :wink:

[quote author="Rochey"]Thats one thing the ADAT devices are useful for [/quote]

I wouldn't use those; they have next to zero info on the output jitter (~1.5ns on the word clock is all they mention). It should be easy enough to use a TI PLL17xx (if you don't mind a bit of jitter), or have a voltage-controlled crystal oscillator (VCXO) which can be locked to the incoming word clock with a phase locked loop (PLL). The word clock is slow enough to implement said PLL in an AVR (or any other microcontroller with an on-chip timer/counter).

Does anyone know for sure whether word clock as supplied on 75Ohm BNCs is guaranteed to be TTL-compatible ? Maybe I should try and get that AES-doc you mentioned earlier.

JDB.

 

[hejsan]

[quote author="Rochey"]...Adding AES (or S/PDIF) to the unit isn't difficult - as the ADAT devices expect either left justified or right justified data, hanging a DIT4096 off the stream (with a buffer) should be relatively easy...[/quote]

Yeeeaaah! :roll: Just hang an o'l DIT4096 off the stream... that ought'a do it.. :?: :?:
Just kidding, I hardly understand a word youre saying, but don't be discouraged!! This is THE most awesome project on the block. Can't wait for results.

And think, An average card has 8 x analog inputs + 2 x spdif that no one uses + 2 x AES EBU that nobody knows what is + 4 x 96kHz ADAT that the occational nut uses.

If you figure out this "analog to spdif" converter, and then do what Rochey said about ADAT and then do the whole AES thing, then I'll have 16 useable 96kHz channels on my motu!! HAHAHAHAHAHAHA!!!!! I can feel WORLD DOMINATION one step closer!!!!.... :twisted:

 

[Rochey]

[quote author="hejsan"][quote author="Rochey"]...Adding AES (or S/PDIF) to the unit isn't difficult - as the ADAT devices expect either left justified or right justified data, hanging a DIT4096 off the stream (with a buffer) should be relatively easy...[/quote]

Yeeeaaah! :roll: Just hang an o'l DIT4096 off the stream... that ought'a do it.. :?: :?:
Just kidding, I hardly understand a word youre saying, but don't be discouraged!! This is THE most awesome project on the block. Can't wait for results.

And think, An average card has 8 x analog inputs + 2 x spdif that no one uses + 2 x AES EBU that nobody knows what is + 4 x 96kHz ADAT that the occational nut uses.

If you figure out this "analog to spdif" converter, and then do what Rochey said about ADAT and then do the whole AES thing, then I'll have 16 useable 96kHz channels on my motu!! HAHAHAHAHAHAHA!!!!! I can feel WORLD DOMINATION one step closer!!!!.... :twisted:[/quote]


A valuable contribution that's just put a grin on my face... thanks hejsan. :green:

Things are hectic at the moment, but once I have some time, I'll contribute more.

cheers

R

 

[mhelin]

Have you seen this one:

http://www.dfm-audio.com/download/dfm_diceii-ikit_infosheet_1.001_en.pdf

We could DIY a nice firewire ADAT I/O "soundcard" using these DFM modules. Wonder how much they ask for these.

E-mail: mailto:[email protected]
Internet: http://www.dfm-audio.com

Seems they have everything we've ever thought including DSP speaker crossover modules and A/D's with mic pres.

 

[Rochey]

I already know the guys at DFM. The module is in no-way aimed at the DIY'er.

To put it into perspective, small quantities of the board would cost more than buying a small firewire soundcard. (once you include the licensing cost of the software).

>> The dedicated business model for DICEII iKit is a onetime fee of 4500 Euro
>> which includes the customisation of the GUI as well as all the stuff
>> listed in the COS+ Software package section including 10 hours support.
>> Additionally there has to be a commitment of min. 10 boards. Prices are
>> arating at 450 EUR @ 10 units, 290 EUR up to units and 195 EUR up to 250
>> units. For 1000´s we should talk seperately.


I think the way forward is to use something like one of the behringer control boxes with the ADAT interface on it, and then create a 8in 8out interface box.

 

[mhelin]

Ok, too expensive. Someone know someone at PMI audio? Maybe they could reserve an extra batch of Firewire cards for DIY'ers? They use BridgeGo ( http://www.bridgeco.net/index.php ) like M-Audio as an add-on for the new Toft mixer and a separate breakout box.

 

[Guest]

Bump
Rochey?

 

[Rochey]

Too busy to even think guys

New job - run off my feet.

Things are still on the wannadoo list

 

[Rochey]

Gents,

I just got a PM as a bump for this project. I'm still up to my eyeballs in a new job, but I'm hoping to get some time this weekend to put a basic (read: on paper) 8in8out design together.

However - and this is the crunch. My workload is far too high at the moment to complete the project myself. I really wouldn't mind putting something together with one or two of the regulars here - thus splitting the workload down into easy chunks that can be done in parallel.

Any takers on the idea? - if so, PM with your basic skillset and maybe we can put a small team (3 people or so) to attack various sides of the project.

Let me know.



R

 

[Rochey]

Ahhh... after quite a delay, the boy is back in town

For those that have followed this project, the delay has been blamed on my recent move to Texas with work. Let's just say that the difference between the UK and Texas is bigger than I expected

The new job is very relevant to the parts that are being dicussed in the projects on this page and on GroupDiy... in fact, I do wonder how many of my customers actually read groupdiy

Anyway - I digress...

I've had to relearn my own Schematic. (I always hate that...).
I've also updated the blog a little.

As the initial intention - This is a modular design based around various functions. More on the Modular system in posts to come.

Cheers

R

 

[Rochey]

Another day, another post.

This project is slowly building, with interest shown from a few others on this board into making this into a range of cards, each compatable with the other through a "motherboard"

I'm still chewing on this one - here's what I have so far. This pic is on my blog, which doesn't like hotlinks i think. Right clock on the picture and clck "display picture"

The idea is that things like Power are bussed on the board (+15V, -15V, 5V). Clocks are also bussed around. More importantly, using a bus structure allows the clocks to be generated by any of the cards and carried to others. However, I suspect that something like 3-state buffers may be needed in that case.

I2C runs on a bus type basis, whilst SPI (running on different pins on the same connector) will have to be clocked in and clocked back out of each card.

The "Data" card could then have something like an ADAT connector. In other words, a card that recieves multiple channels of left justified digital audio.

At this point, left justified was chosen because the ADAT chips by Wavefront don't accept I2S.

Feedback comments etc are appreciated

Many thanks to those of you who emailed me after my last post. This looks like it could develop into something pretty good.

Cheers

R

 

[Svart]

:sad:

I get a broken image..

If you alt click and "view image" you can see it.

very professional flow chart!

 

[Rochey]

i wonder if a cardedge connector would work here? something like a PCI connector...

or maybe stick to ribbon cables... any thoughts?

 

[jdbakker]

Card edge needs plating (preferably gold), which isn't very DIY-proof. Besides, a card edge backplane enforces a maximum card spacing which might come back to haunt you down the road (when, say, you have a rev3 of your board where you want to use really big caps, or when you find out that EMI-wise it might be good to have some more clearance between PCBs).

On my work-in-progress system I use ribbon cable with IDC connectors, with every other conductor grounded, and separate impedance-matched cables for clock. Works pretty well on the bench, so far.

IME, connectors, switches and other mechanicals account for well over half intermittent or permanent failures, so be careful what you pick.

JDB.
[the 'separate cables for clock' may be a bit overkill, but I do like to keep my ADC clocks as jitter-free as reasonably possible]

 

[Rochey]

Hia JD,

in that case, it might be easier to put each section on a 2x5 IDC - for ease.

Power (+15V AGND -15V +5V DGND)
Clocks (Wordclock, 64fS, 256fS, X,Y)
Control (4 pins of SPI, 2 pins of I2C)
Left Just I/O = Data In, bitclock, wordclock, Data Out.

it also means I could be a tight bastard and put the motherboard on single layer PCB

which means... more :sam: for me.

 

[walter]

I am interested in building an eight channel pre to ADAT interface, and i like the title to this thread, so.... I dig it up. Alot of discussion on clocking in another thread, the clock seems to be a major factor. You could just use an external clock, but that would be no fun. I recently purchased the M-audio profire lightbridge=four ADAT in/out to Firewire, my incentive to persue this project. Using the B'ringer ADA8000 as a model(because I have one to mangle, not because it's that great, but because it functions) The power supply works, but could be beefed up a bit, the clock works, but there may be a better clock(I'll have to check the jitter), the DtoA works. Change to the pre-amp section is what I feel will make the greatest impact. Perhaps four tube pre-amps/four solid state. There is no software involved for this DA/AD, It should be an easy build, just configuring hardware, Layout. Anyone?