Modular multi channel DIY AD/DA Box

[rkn80]

Hello,

this is my first post on this great forum but I was already reading here for a long time.
Let me first introduce myself a little bit: My name ist Raphael and I'm a young electrical engineer from Germany.
My idea is to build a AD/DA box like the Alpha-Link from SSL, the ADA8000 from Behringer, the i/o boxes from Digidesign and similar. But my system should be more flexible. So that you can populate the box with the number of DACs and ADCs you need and later you have the option to add more channels.
So I started with designing the PCBs of the DAC and of the ADC. Both are already done and I'm going to order first prototypes of the PCB.
The DAC is based on a TI PCM4104 and you have 4 channels per PCB. Each channel offers a fully balanced floating output like the topology presented by G.D. Pontis. The DACs do run completely in hardware mode so that no additional microcontroller is needed but there is an option to control it via the SPI-interface if wanted.
The ADC is based on a TI PCM4202 and supports two channels per PCB. You can either connect a a line level signal or directly a microphone (with or without phantom power). The front end of the ADC is built like the mic design of Cohen but instead of going to the outside the outputs are going directly into the ADC. The idea is that you do not need an extra box for the micamps. Again the ADC is fully hw-controlled.
Both PCBs support I2S input/output.
The next step is to design a ADAT Lightpipe input/output card so that you can connect the box to a PC or whatever as long as it supports an ADAT interface.
The modular design will allow to add other interfaces or to upgrade the box with higher performance AD/DA later.
I would like to share the design with you and perhaps there are people how have an idea for another module or want to design a pcb that offers an additional feature and that can be integrated into the box. If so I invite you to take part in this project.
So is there interest in this project?
I'm also willing to organize a group buy of the DA or/and the AD pcb if wanted.

Raphael  

[ricothetroll]

Hi Raphael,

I'm interrested ! Those last weeks I started to take a look at how I could do that kind of project. But I'm no electronic ingeneer, so I had a lot of infos missing, and of knowledges to get to do that project properly.

I'm not so sure I can help for the theorical part, but I make homemade PCBs, if ever that's helpful...

I'd be curious to take a look at the schematics !

Pretty cool project indeed ! Congratulations !

[deuce42]

Raphael - welcome to the board:)

Your project sound awesome. I look forward to reading how this evolves.

Kindest

[jdbakker]

Interesting plan. Welcome.

Traditionally one of the more difficult things to get right is clock synchronization to external WC sources without generating excessive jitter; in a modular design clock distribution can be challenging too.

Have you seen this older thread, BTW?

JDB.

[volker]

Hi Raphael,

I've been thinking about something similar lately. Personally I would prefer to keep the ADC inputs just line level without an additional mic preamp, but keeping it modular is the great thing about it. I'm going to check out the converters you're using... Would be cool to see the schematics.


Gruß, Volker

[pedroplanet]

very interested on build one 

[guitarguy12387]

Wow. You have my attention... hah

[rkn80]

Well, the idea is to make something like an active busplane with buffers on it for the clock lines. Every module runs as a slave so that the intra-module clock lines are isolated from the system wide clocklines via buffers.
In the first step I will select one of the ADs as a master providing the system with the needed clocks. Later that can be changed with an additional module that also offers a word clock sync to the outside.
In the next days I'll publish the schematics step by step. I need to make them nice first.

Raphael

[jdbakker]

Well, the idea is to make something like an active busplane with buffers on it for the clock lines. Every module runs as a slave so that the intra-module clock lines are isolated from the system wide clocklines via buffers.

Keep in mind that every layer of buffering increases jitter, and that even low levels of power/ground noise can have significant impact on clock integrity.

Looking forward to seeing your schematics,

JDB.

[deuc224]

WOW!! got my attention as well, the DAC chips look good, is there any reason that you didnt choose the PCM 4222 for AD? Those chips are about as good as they get! i heard a shootout between the lavry gold mk3 and the PCM 4222EVM and to be totally honest, they both sounded great, extreme clarity, great depth and both true to source. One wasnt better than the other it was mostly subjective. But considering that the lavry gold is 7500 and the pcm4222 is 149...........

[playboss]

dont use 4222 but 4220... The number 4222 asks for trouble already...
The difference is in feature set , 222 is unlikely as hell .
How are you doing clock distribution? Clock on seperate daughterboard would be preferable.

[deuc224]

Could you explain? Everything on the 4222 is new and up to date.

[Whoops]

I think you have a great project!

and I don't think there's anything similar in this forum anyway.

I'm also in favour of a Line level input, instead of Mic Pre.
People here have already done their favourite Mic Pres, Neve type, API, SSL, etc, so
it would be great to keep the signal flow as simple as possible after the Mic Pre.

[playboss]

Could you explain?

No.

Since you compared an evm without psu to a Lavry converter I have nothing on earth to explain. 

[deuce42]

Could you explain?

No.

Since you compared an evm without psu to a Lavry converter I have nothing on earth to explain. 

Playboss - what gives man? Why the attitude?  If you believe somebody has some wrong information share your views rather than being heavy about it.

[playboss]

ok.
So, you guys need DSD output and 6bit quantizer output pins, thats all what 4222 gives over the 4220 . Fine, nice job.

These people who bought all these EVM's dont get the idea of these type of converters, that is , these are more like set of digital signal processing algo's than anything """analog""".
-You bought the 4222 evm, you bought a set of preset algorithms and a cheap analog front end. Now you can configure it with jumpers and leaves you with a lot of useless features,  like , DSD, flat dithered 16bit PCM out.

Comparing this to a Lavry is like , taking 1 preset of a cheap synthetiser against a vast array of presets on the expensive Lavry
- and you're free to tweak it too. It will give you a 16bit 96khz output that kills the 24/96 of the 4222 evm and its hacked power supply.

[promixe]

Let's not get all hostile here. =) While I personally think there is no comparison between the Lavry Gold series AD and a "chip converter" whatever it might be, I also think that PCM4222 is a great AD chip and it sounds awesome, so does the 4220. Also, the 4222 has -124dB SNR, which is one dB more (eventhough I think they measured it A-weighted) over 4220. But of course the argument here is - "if you REALLY know what you're doing you might get into the <-120dB spec, otherwise the noise floor of your system is gonna be above the ADC spec". =)

But I would too encourage to have the AD design based on either 4220 or 4222 and not the older 4202 series.

PS: Lavry Gold AD has an analog filter upfront (chopping everything above Nyquist beforehand) followed by custom precision laser trimmed resistor network to perform the conversion and what he calls "digital soft saturation" that sounds like nothing else out there if clipped. Most of silicon ADC chips these days have basic FIR decimation filters post the conversion that are pretty basic, and while giving near linear phase on the output (especially 422x family) it doesn't sound nearly as good as true analog filter. That's where a clear advantage of 4222 would come in with its 6-bit modulator outputs to bypass internal FIR and design your own superb decimation filter in a DSP. But I don't this particular project would require such steps. Basic 4220 will sound fine. =)

[Matthew Jacobs]

I'm hooked as well. This is exactly what I need. Can't help much on the digital side but will definitely support this project.

I vote for line inputs as well.

Nice one Raphael!

J

[detonator]

Regarding to my post  http://www.groupdiy.com/index.php?topic=35441.0 this is one I'm looking forward to. It's not always a matter of coming up with new designs, but a different approach for a concept is a way to go. So count me in.

[bitman]

I have a stuffed lightpipe board that I bought from someone on this board but never used it.
You might like it for prototyping.

$25.00

ron AT mtrecording DOT com