digidesign 442 I/O as standalone DAC for S/PDIF transport

Hi,

I've been given a Digidesign 442 audio interface box - the Nubus Mac and controller card is long gone.

Rather than put this into landfill as toxic waste, I was wondering whether anyone has any experience of getting these units to 'free-run' as DAC boxes - taking S/PDIF from the RCA input jack and putting the analog audio out of the XLR connectors.

The S/PDIF decoder is the Crystal Semi CS8412

The DAC/filter pair is AD1860/NPC SM5813

I've found datasheets for the CS8412 and AD186x - no luck at NPC.co.jp for the digital filter, sadly.

I've tried feeding 2-channel 44kHz PCM from my CD player into the S/PDIF input to see if the box has any autonomous capability - but it stays stuck on AES/EBU 48kHz :-(

Opinions about the practicality of this project welcome !
 
rfbrw said:
It's been a long time since I last saw one, but doesn't it have a sync input? If it expects a 48K wordclock signal or freeruns at 48k without one, you may have to change the sync internally or by some other means. Or does it lock to video?


It has 2 BNCs on the back - Slave Clock IN and OUT

There is no word clock input, BUT...

It came with another box - the Video Slave driver - that I assume is the gizmo that locks audio sampling to video frame rates - this appears also to take a master word clock input and produce a slave clock signal - presumably for the 442 IO when it is required to sample video and audio together.

I know little about S/PDIF, but doesn't it have its own embedded clock in the data stream ?

So you think that a 44kHz square wave at TTL or EIA-485 line levels will persuade the box to accept S/PDIF 44kHz ?

thanks for the URL for the dig filter, btw !
 

rfbrw

Member
2001-10-26 11:51 pm
.
When an external reference is required the system can be configured to ignore the SPDIF derived clock. Wordclock at Fs rates and TTL levels is what you need but be aware some models also used Superclock at 256Fs. Simply put you need to build or buy a sync pulse generator ideally one that covers both options.
 
rfbrw said:
When an external reference is required the system can be configured to ignore the SPDIF derived clock.

I don't want it to ignore the SPDIF clock, though ! And presumably the box is configured to do this by ProTools sending data from the Nubus card (gone) to the SCSI connector on the back of the box ? I had decided that if the complexity of the project escalated to the point where it required some form of embedded controller hanging off the back of the box SCSI connector in order to boot it into a '44kHz pass-though' mode, it would take forever, and I would never finish it. The address mapping for the 442's devices is certainly Digidesign's IP.


Wordclock at Fs rates and TTL levels is what you need but be aware some models also used Superclock at 256Fs. Simply put you need to build or buy a sync pulse generator ideally one that covers both options.

11.2896 MHz ? !!!! - at least that frequency is available as a crystal module.

Reading the datasheets, I can now see a bigger problem, though...

The CD data from the transport will be arriving as 16-bit words, but the output AD1860 DAC is 18-bit and presumably that's what ProTools originally sent from the host Mac - isn't this a source of incompatibility in getting the box to operate as a standalone DAC - the interface would never have been designed to incorporate the logic to internally bit-extend 16-bit words to 18-bits for the DAC ?
 

rfbrw

Member
2001-10-26 11:51 pm
.
If all you want to do is run it as a standalone dac with no external sync, i.e. CS8412 - SM5813 - AD1860, then thats a doddle. All you need is a little pcb surgery.The wordlength problem insofar as it is a problem will be addressed by the digital filter.
 
rfbrw said:
If all you want to do is run it as a standalone dac with no external sync, i.e. CS8412 - SM5813 - AD1860, then thats a doddle. All you need is a little pcb surgery.The wordlength problem insofar as it is a problem will be addressed by the digital filter.


have you done this, or has anyone to your knowledge done this here was essentially my first question :)

It's exactly the 'pcb surgery' required that I was trying to get info about e.g. which control pins to hack on each chip and what values to set them to - as the CS8412 datasheet mentions the 8 data formats it can be configured to decode, but nowhere relates these to real transport formats like CD-audio. Is it msb first, &c. &c....
 

rfbrw

Member
2001-10-26 11:51 pm
.
transport.nut said:



have you done this, or has anyone to your knowledge done this here was essentially my first question :)

It's exactly the 'pcb surgery' required that I was trying to get info about e.g. which control pins to hack on each chip and what values to set them to - as the CS8412 datasheet mentions the 8 data formats it can be configured to decode, but nowhere relates these to real transport formats like CD-audio. Is it msb first, &c. &c....


Putting these things together is a bit like Lego or Meccano. With enough knowledge and an ability to decipher datasheets, you can interconnect anything.
The SM5813 expects 16bit LSB justified data valid on the rising edge of bitclock (SCLK on the CS8412, Pin 12 and BCKI on the SM5813, Pin2). In order to use the embedded SPDIF clock, bitclock and L/R clock (FSYNC on the CS8412, Pin11 and LRCI on the SM5813, Pin28) have to be configured as outputs. This equates to mode 5 of the CS8412. To configure this mode the mode pins have to set as follows,
Pin 17 M3 = 0
Pin 18 M2 = 1
Pin 24 M1 = 0
Pin 23 M0 = 1
The SM5813 also requires a master or system clock on Pin 6 XTI.
With an external reference, system clock at 192/256/384/512Fs, bitclock at 64Fs and L/R clock at Fs would be generated. In this case all the clocks will be generated by the CS8412. The CS8412 only does 256Fs (MCLK Pin19), so this should be connected to XTI on the SM5813. To configure the SM5813 to accept system clock at 256Fs, /CKDV, Pin 4, has to be connected to V+ and /CKSL, Pin3 has to be connected to GND. You may also have to check /RST Pin 14. This may be connected to the computer port to allow an external reset. It can be connected to ground.
On the output side of the SM5813, the wordlength has to set to 18bits to match the AD1860. Connect /OW20, Pin 16 to V+ and /OW18, Pin17 to ground but they should already be set as should most of the output side settings. The connections to the dacs should need no alteration.
 

Attachments

  • dac.gif
    dac.gif
    10.8 KB · Views: 1,400
Hi,

also got such thing today and wanna try.
On the support site it says that the later 882 can be used stand alone but no idea if this is valid also for the 442.

When I switch on it searches ( first slave LED is on and 44,1k blinks ) but does not find digital input.
In the end it shows Master sync, analog input, 44,1k, AES.

No idea if the unit is ok but 1xFS, 256xFS, word clock out test points are ok.

Do your units behave the same ?

If I can't get it working I will remove all those chips...
 

newmedia

Member
2005-08-07 6:36 pm
Hello:

I just landed 8x 442 I/O units (as well as 12x VSD's) -- seems these have become pretty useless to modern Protoolers.

Since the value is minimal but the engineering is well done (by 1990 standards, which is when the board was first laid out), I'd like to re-purpose these -- just like many others, I suspect.

Did Digidesign ever release any schematics or internal details? Are there multiple versions? Will they work with anything other than Digidesign Mac/PC cards?

There are two (unused at the moment) jumpers I can spot -- marked JP3 and JP4 -- what are they used for?

I've read the thread but can't tell if anyone ever got these working as standalone Analog-to-SPDIF converters -- how did it turn out?

Newmedia
New York City
 
This may be of interest:

http://news.palmos.com/read/messages?id=1101

a post from 5 years ago by David Clementson - the guy at Digidesign who designed the 442 circuit board. There's also an email address in his signature:

E-mail is dclement
domain is digidesign.com

Now, would someone who knows the right questions to ask and is more likely than me to understand the answers like to try emailing this man and then telling us all if he has any useful information on either making the 442 do D>A or (preferably) A>D?
 

redrabbit

Member
2004-10-28 4:25 am
I have a friend , who has a friend, that has altered this box to use it as a simple "line in" device.....but that's all I know. I'll press him for details/info.

These used 442's are so cheap these days, that I am using one to power a Green Mic Pre. They have a +15, -15, 0, switching psu inside.......gut the rest, and I've got a 1 unit rack space box with power supply, and several usefull holes in the back. All I need to do is make a new faceplate.

=RR=
 

redrabbit

Member
2004-10-28 4:25 am
I talked to this gentleman today, and while I did not understand everything he said, here is the info I DID understand:

He does NOT make use of any of the digital side of the 442. He simply turns them into Line IN / Line OUT buffering devices that produce about +18dB of gain......adjustable via the trimpots already on the board + the meters still work.
He has meticulously signal traced the board, and figured out where to install jumpers and/or cut copper traces. He warned me that this is a 3-layer circuit board, so you don't know everything by just following a trace with your eyes.
To avoid any digital noise from entering the analog lines, he pops out all the digital IC's, most importantly the two metal crystals.
I asked him if it is possible to acsess the digital side of the board, and he said that he never felt it was worth his time, as the unit has only 16bit resolution, that he has no need for it.

He also reminded me about using the box with the powersupply for other projects. You can also cut (dremel tool) the board at the place where the XLR's are mounted, and use those too, because they fit the holes in the back. You can easily see where to cut, basicly sawing through the small ferrites, as those were only there to keep the digital side noise away.

He mentioned he has sold these mods, I did not ask how much.

Sorry, if that was not the answer you wanted, but he convinced me, to look elsewhere for digital conversion.

=RR=