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.
Yes, you are totally right about how hard it *should* be to reverse the plug. Actually, there was a cheap backplane and mainboard jack combination that effectively 'allowed' a wrong insertion easily. I felt so stupid and upset that I did a bit of research in this field and noticed that I am by far not the only one that had this experience. I found several upset statements of people that destroyed high-end cameras this way (requiring hundreds of dollar-repairs). Also, I had a talk with the out-of-warranty service company for m-audio in germany, who told me that destroyed phy's are by far the most common repairs on the firewire audio interfaces, so lack of protection is an issue with firewire really.
Ironically the mainboard was equipped with the well-suited TI chip so I did not even think of how cheap some mainboards are built when it comes to the jacks nowadays. It did not require more force to plug in in reverse than it did the right way on my own computers belkin card. Surely I check orientation twice ever since...
Well, in the meantime I can laugh about this, but I guess that would be different if it had cost me a few hundred dollars more. :D

Kind regards,
Martin
P.S.: The owner told me the jack and mainboard phy still works fine with external equipment, but he threw away all firewire cables with 'soft rubber'-like plastic plugs....lucky basterd he is, hehe ;D
 
Here is the schematic of the AES receiver/transmitter board.

Raphael
 

Attachments

  • DIX4192.pdf
    141.2 KB · Views: 206
The DIX4192 is almost pin-compatible with the SRC4392 sample rate converter. It would be useful if the board supported both, as then you could feed a steady local sample clock to the converters and have the SRCs handle the timing domain conversion.

JDB.
[as Bruno Putzeys has pointed out a few times, in that case it's best to have an internal sample clock at an odd rate such as 200kHz and not something close to the external sample rates, in order to minimize SRC artefacts/distortion]
 
Indeed, a good point. I had a short look on the SRC4392 datasheet and yes it is almost pin compatible. After a first view only two pins have to be connected additional (pins that are nc on the DIX4192) and then the SRC4392 works on my pcb. I'll check that and modify the board a little bit the next days (weekend is coming ... :D).

Raphael
 
From what I know... (And I *should* know) - they are exactly p2p compatible, other than some register changes to enable the src. (The src doesn't exist on the dix part)

/R
 
Not really. Thr SRC has some functions like MUTE on pins which are not used on the DIX. So you can make the board usable for both by connecting these pins either to ground or to +3V3. More details on that when I further investigated that.

Raphael
 
Awsome work, Raphael. Great idea to have the SRC4392 beeing supported on the pcb with the option to src to an internal clock requirement, just like JDB recommended.
If I understood it right, it would be possible to adjust the internal clock speed to the best sounding sample rate of the da or ad chip. Correct?

Rochey, I guess it is just about what kind of 'compatibility definition' is used (I guess it is depending on the context in this case, e.g. upward or downward compatibility). I'm pretty sure you know your stuff....I lately saw you in a video talking about some awsome chips...

Kind regards,
Martin
 
BridgeCo has solutions for I2S/I8S <-> Ethernet:

http://www.bridgeco.com/index.php?option=com_content&task=view&id=16&Itemid=36

http://www.bridgeco.com/pdfs/consumer_audio/product_brief_DM850_v2_20080827.pdf

Using I8S I think these (if have the same core as DM1100/1500) can support upto 32 channels in and out.

Their Pro Audio section BridgeCo AG in Switzerland has been moved to ArchWave company lately, maybe you should contact them:

 http://www.archwave.net

 http://www.archwave.net/Audio-Link/Contact.aspx
 
I've now changed the AES RX/TX board supporting now the SRC4392, too.
I think next week I'll send the gerbers to the pcb factory and let make them the first prototype if there are no ne ideas in the next few days.

Regarding BridgeCo: Is their stuff available in small quantities for DIY projects like this?

Raphael
 
As far as I know, the actual bridgeco chip (870) is based on arm9 with the datasheet only available under NDA.
The predecessor 850 was used in the eary terratec noxon internet radio machine and others.
They use a linux like proprietary RTOS.
I doubt they are available in single quantities. Definitely aimed at the b2b market for time-to-market reduction for media streaming machines. Maybe it can be get in smaller quantities from some southasia chip brokers?
Nevertheless looks like very interesting chips.

PS: Maybe someone from switzerland can make a polite request to bridgeco?

Kind regards,
Martin
 
While waiting for the ordered RX/TX-board, here is the layout of the first AD-module. This has four input channels built around a PCM4204. I've decided to do to first the line input version instead for the mic input because then I can put more channels on the pcb and it makes the box cheaper.
Schematic will follow this evening.

Raphael
 

Attachments

  • ADCmodule.png
    ADCmodule.png
    107.4 KB · Views: 339
rkn80 said:
Any comments?

Looks OK, although I haven't compared it line-by-line with the data sheet.

Pity your ground plane has so many slots in it, but that's hard to avoid on a 2-layer. Personally I would have considered wire jumpers over some of the longer breaks.

JDB.
 
I know, I know.
But I wanted to avoid another source for a wiring mistake or changing to a 4-layer board. I made already a compromise with routing +/-15V and +5V to pin headers on both sides so that I don't had to route them from one border to the other border. Doing so I could route the critical (high speed) lines without crossing a split in the ground plane.

Raphael
 
Hi Raphael,

any specific reasons to not use the internal high pass filter? Since you are dc coupling the outputs of the receiver stages...

Did you think about protecting the inputs of the ADC? With +-15V in the previous stages, the inputs can be easily driven way beyond the maximum ratings.

How do you want to set the low pass filter? Obviously you want the corner frequency to be as high as possible. On the other hand you have different sampling frequencies to choose dynamically. Hmmm. How is that done in commercial units, do they really switch the whole filter?

Concerning the layout: not that I have too much experience in mixed signal design, but one of the first things that comes to my mind is to use separate ground planes for analog and digital. Any thoughts about that?


Please forgive all my questioning, I'm just trying to learn something along the way ;D.


Volker
 
volker said:
any specific reasons to not use the internal high pass filter? Since you are dc coupling the outputs of the receiver stages...

No.

volker said:
Did you think about protecting the inputs of the ADC? With +-15V in the previous stages, the inputs can be easily driven way beyond the maximum ratings.

The evaluation boards of the PCM4204 do not have a protection around the inputs but are driven with +/-15V stages. So I decided that it is not needed. If one of the inputs is overdriven the clip pin will flag that.

volker said:
How do you want to set the low pass filter? Obviously you want the corner frequency to be as high as possible. On the other hand you have different sampling frequencies to choose dynamically. Hmmm. How is that done in commercial units, do they really switch the whole filter?
I think commercial units have only one filter for all sampling frequencies because you don't see switching relais etc. normally. The question is do these units always sample with the selected sampling frequency or do they use a fixed sampling rate and generate the selected one via a sample rate converter.

volker said:
Concerning the layout: not that I have too much experience in mixed signal design, but one of the first things that comes to my mind is to use separate ground planes for analog and digital. Any thoughts about that?

Well that is wrong. You should not use separated ground planes for AD/DA conversion. There are many applications notes and even the datasheets state that AGND and DGND should use one ground plane and not two separated planes.

Raphael
 
[quote author=rkn80]Well that is wrong. You should not use separated ground planes for AD/DA conversion. There are many applications notes and even the datasheets state that AGND and DGND should use one ground plane and not two separated planes.[/quote]

Yes, I just saw the remark in the datasheet related to the ground plane. Will have to read up on that topic.

[quote author=rkn80]The evaluation boards of the PCM4204 do not have a protection around the inputs but are driven with +/-15V stages. So I decided that it is not needed. If one of the inputs is overdriven the clip pin will flag that.[/quote]

Sure, the clip flag will show too much input voltage, but it will not protect the ADC. The datasheet states 300mV below/beyond ground/VCC as the absolute maximum, otherwise the ADC is likely to be damaged. Also depends on the input attenuator of course. But a little two-diode-package seems to be a small price in comparison to desoldering a TQFP-64. It's also mentioned on page 15, but then the diodes don't appear in the example circuit.


Thanks for answering!

Volker
 
Short before Christmas the ADC prototype board arrived. As you can see I placed three modules on one pcb so 12 channels in total. The AES RX/TX board did not arrive yet so I'll first stuff and test the ADC board. Hopefully after Christmas I can report some good news.

Raphael
 

Attachments

  • IMG_0034.JPG
    IMG_0034.JPG
    45.1 KB · Views: 291
Nice PCB layout. Analog and digital grounds use same symbol at schematic, but they separated at pcb, right? :)
Just a stupid question... Matter of curiosity, which clock are you going to use for this beast,
configuration, PLL corner freq., oscillator type, etc.....
 
Back
Top