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.
Would this backplane board include an actual masterclock, or just the routing? If the clock is included, I might be interested. I don't know how you would plan to set up the power rails on this board, but I hope it doesn't involve regulators. I already have my rails completely set up.

This kind of board would also be an ideal place to buffer the clipping flags to actual leds.
 
Not sure yet.
For some people it it convenient to have the master clock on it but for other people it is not a good idea because the have their own high quality clock or want to use a wordclock input whatever.
The same thoughts about the power rails.
So whatever you do it is right and wrong at the same time. ;)
Clipping and error LEDs are already on the ToDo list.
 
humble observations:
-it may be wise to have each backplane board able to accept at least 4 modules (since that's when it seems a board is critical).
-if clocking is included, then maybe include a clock jumper to select between master/clock through, which could feed from the extension connector you were discussing?  the clocking pinout at this connector could then be used as a word in from either the diy master or an external source to feed the other backplane boards.

this would cut down on wiring between individual backplane boards, which is the task at hand, and provide everyone the option for a master clock.  it would also net 16 channels for each backplane board, which would make for a nice molex out to DB25  ;D
as far as power distribution goes, i'd love to see local regs... but that debate is for braver, smarter men than i!
 
click on the pictures in my previous post to see larger versions.

EDIT:
grantlack said:
humble observations:
-it may be wise to have each backplane board able to accept at least 4 modules (since that's when it seems a board is critical).
-if clocking is included, then maybe include a clock jumper to select between master/clock through, which could feed from the extension connector you were discussing?  the clocking pinout at this connector could then be used as a word in from either the diy master or an external source to feed the other backplane boards.

this would cut down on wiring between individual backplane boards, which is the task at hand, and provide everyone the option for a master clock.  it would also net 16 channels for each backplane board, which would make for a nice molex out to DB25  ;D
as far as power distribution goes, i'd love to see local regs... but that debate is for braver, smarter men than i!
Also, I agree with everything Grant said.


Cheers,
B.
 
Very nice work thumbs up! Got some patience.... But on the picture of the board alone i could see some solder bridges on the PCM4104, maybe because of picture on my iphone small screen, don't mean to offend maybe just the ground tabs and so it's tottaly normal  ???, just an observation,i don't think you would miss that tough   ::)  
 
Indeed. The bridges occur due to a connection of these pins on the PCB when you handsolder them. But you have to make sure that they are only there where they should be and not somewhere else. Better you look twice... ;)
Well done, Kingston. Are they already working?

Raphael
 
educated myself a bit more on power regulation this morning... promoted myself from a know-nothing to a brash, fumbling novice  ;).  now for more undereducated thoughts on the backplane... i'm sure it'll be mostly trivial, but i've got a real itch for discussion on this project!
if regulation is linear, either on or off the backplane, an iteration of the jung circuit would be superb.  performance would obviously be great, so it would be more a question of cost/size/efficiency.
as far as switching options go, the ADP2114 or TI's TPS54286 offerings might be viable as well.  obviously an on-board only solution in these packages.  the power handling characteristics sit nicely for 4 converter modules, corresponding AES, and fudge factor for clock.  switching frequencies are well outside the audio band, so rather conventional filtering should take care of that.  still trying to learn up on how crucial synchronization would be for daisy chained boards, as that would steer selection toward the analog devices chip.  they happen to be cheaper and can sync to higher switch freqs, but Vin is pretty limited and they only come in 32lfcsp   :'(

EDIT: that analog devices chip wouldn't provide an elegant solution at all.  didn't take the 15V into account.  who wants to deal with solder bridges on that thing anyway...
 
The project is not dead it just smells like it... ;)

During the last few weeks I designed the carrier board. Today I finished the design. Here you can see a preview of it. Tomorrow I'm going to etch that board myself.
Some features:
- takes 6 modules
- connection to the outside via SUB-D25
- connectors for daughter boards if XLRs or BNS or whatever is needed instead of SUB-D.
- all voltages for the boards are generated from the +/-15V supply via standard linear regulators (but they can be left out and replaced by a Jung-regulator or something like that)
- the board offers expansion ports to connect it to a second board, maybe you want 24 channel DAC with AES input then two of these carrier boards are connected together
- it offeres a galvanic isolation: noise in the ground system of the digital interface part cannot enter into the ground system of the DAC or the ADC.

I'll keep you informed about the progress. Perhaps some of you are interested in that carrier board.

Raphael

PS: You cannot see the slots for the modules with the size like the ADC module. These slots are on the other side of the board.



 

Attachments

  • mainboard.jpg
    mainboard.jpg
    258.6 KB · Views: 134
Here is now my DIY-etched board. Let's see wether I can get it running or not...

Raphael

P.S.: I forgot to tell, the board will get LEDs for the Clip/Error signals.

 

Attachments

  • DSC_0003.jpg
    DSC_0003.jpg
    297 KB · Views: 148
rkn80 said:
P.S.: I forgot to tell, the board will get LEDs for the Clip/Error signals.

Spotted those in the layout, well done all around! Don't have my schematics handy to follow traces from pinouts, did you doing anything with the clock i/o routing?
 
Hi,

for long time now there have been no replies on this topic.
First the bad news: The mainboard and the group buy of the DAC modules is still delayed due to TI.
Therefore, I swiched back to the clock board in the meantime. I reworked the generation of the clocks in stand alone mode as you can see in the attached schematic and picture of the layout.
Currently I'm adding a word clock in option. In the previous version we discussed a solution built around a microcontroller. I decided to discarded that for two reasons:
1) I guess for many members it will be a big contra if you need to flash a microcontroller first.
2) I found a device that enables me to do it wihtout a microcontroller but still with good results.
I'lll keep you informed about that as soon as I've drawn the schematic.
Anyway here is the board only with the parts that are needed to generate and distribute the needed I2S clocks.
Comments are welcome.

Raphael

 

Attachments

  • cdist.pdf
    93.9 KB · Views: 54
Looks interesting. Nice to see you using a sync divider chain.

I'd recommend against using shared buffers for MCLK/BCLK/LRCLK. Ground bounce in the common supply lines will induce periodic jitter (albeit at Fs and multiples, so it depends on the modulator what the impact is). Consider single gate logic (in SOT-23-5) with sufficient decoupling, not much harder to solder than the parts you're already using. Similarly, your selectors will have some edge feedthrough, albeit much less than the aforementioned ground bounce issue.

MCLK is the most sensitive line jitter-wise.

What are the 100R input resistors expected to achieve? As placed they don't offer any characteristic impedance termination; as far as I can tell all they do now is increasing jitter somewhat (both through Johnson noise and by slowing down the edge).

Do you have any parts in mind for the clock oscillator?

JDB.
 
For the clock oscillator I used FXO-HC73 from FOX. But that is not the only solution as long as it fits onto the HC73 footprint.

Which 100R do you mean? Some of them are only there to have a 0805 footprint at the place to have the option to solder a resistor there during debugging/testing the PCB. I think I'll replace most of them with 0R and kick them out in the final design.

Raphael
 
Back
Top