Is there anyone out there who knows of an economical way to connect four s/pdif (or AES/EBU) somethings to one adat? More and more stuff these days seem to come with s/pdif I/O, so I'm looking for a way to turn one of the adat lightpipes on my digiface into four extra s/pdifs (or AES/EBUs). The stuff I have found, like the Alesis AI4, is somewhat expensive for what it does; and I don't need all the sample rate conversion stuff. (I'm perfectly willing to supply all the s/pdif widgets with wordclock if I have to.)
S/Pdif to lightpipe digital patchbay?
- Thread starter BasPer
- Start date
I had a go with an STAudio digital patch bay, which if you get one that works will do the job !
http://www.musiciansfriend.com/srs7...0715126/g=home/search/d=tp?q=digital+patchbay
The AES to S/P DIF light pipe was great but.... the co-ax S/PDIF to S/PDIF light pipe was dreadful. Sent back the first two units because of this. They were also scratched and the second one was shipped with no power supply. Nice one musicians friend (?).....
Im currently waiting for an M-Audio Digipatch as a replacement 3 weeks down the line !
http://www.musiciansfriend.com/srs7/sid=030625183123012217254180715126/g=rec/search/d=tp?q=digipatch
The Digipatch should do 6 coax S/PDIFS to a S/PDIF light pipe.....I hope !
http://www.musiciansfriend.com/srs7...0715126/g=home/search/d=tp?q=digital+patchbay
The AES to S/P DIF light pipe was great but.... the co-ax S/PDIF to S/PDIF light pipe was dreadful. Sent back the first two units because of this. They were also scratched and the second one was shipped with no power supply. Nice one musicians friend (?).....
Im currently waiting for an M-Audio Digipatch as a replacement 3 weeks down the line !
http://www.musiciansfriend.com/srs7/sid=030625183123012217254180715126/g=rec/search/d=tp?q=digipatch
The Digipatch should do 6 coax S/PDIFS to a S/PDIF light pipe.....I hope !
Blue Bear Sound
New member
The problem with the Digipatch unit is that it is 16-bit only.... if you pass a 20 or 24-bit signal, it will get truncated.
(I went thru a digital patchbay search over a year ago and discvoered all these "issues"...) Ended up going with a Z*Sys Optipatch... which does 24-bit but doesn't have S/PDIF to Lightpipe.... They do have other models though - not cheap unfortunately....
(I went thru a digital patchbay search over a year ago and discvoered all these "issues"...) Ended up going with a Z*Sys Optipatch... which does 24-bit but doesn't have S/PDIF to Lightpipe.... They do have other models though - not cheap unfortunately....
Hm, I don't think it will! Because it is only a switch box and sample rate convertor, it doesn't actually aggregate several s/pdif streams into one adat stream. It only contains one Crystal CS8420 and a six-way switch. I would need something with four CS8420 and a little FPGA or someting that aggregates the four s/pdif data streams into one adat data stream. Something like the stuff from z-systems BlueBear talked about (but less exuberantly priced).Dr_simon said:
I could build something myself, but if I wanted to design electronics in my spare time I should just give up music and get a second job...
jet-rocker
New member
BasPer: I've been waiting for such a product as well. It would be perfect, relatively inexpensive, and I'm sure there is a market for it too.
I don't think the product exists yet, which is a major bummer. Let me know if you find one.
-Jtt
I don't think the product exists yet, which is a major bummer. Let me know if you find one.
-Jtt
Sorry BasPer, didnt know you wanted to mix the signal.
I have just been on to M-audio tech support re the digipatch and they sent the following:
>THE PROBLEM:
>
>Im about to buy a Digipatch and I wnat to know if it can handel 24 bit
>S/PPDI data with out dithering down to 16 bit
>
>Thanks
>Simon
Dr Simon,
The Digipatch will accept whatever signal is sent into it. It's really a
signal relay and doesn't do any processing or conversion. Standard digital
resolution is 16bit 44kHz, but optical and SPDIF connections can handle up
to 24 bit at 96kHz.
Please let us know if you have any other questions.
CK
I have just been on to M-audio tech support re the digipatch and they sent the following:
>THE PROBLEM:
>
>Im about to buy a Digipatch and I wnat to know if it can handel 24 bit
>S/PPDI data with out dithering down to 16 bit
>
>Thanks
>Simon
Dr Simon,
The Digipatch will accept whatever signal is sent into it. It's really a
signal relay and doesn't do any processing or conversion. Standard digital
resolution is 16bit 44kHz, but optical and SPDIF connections can handle up
to 24 bit at 96kHz.
Please let us know if you have any other questions.
CK
Blue Bear Sound
New member
Yes - they told me the same at the time --- unfortunately, it's not true.... after I confronted them with my bit-testing evidence it was - "oh yeah, I guess you're right, it doesn't pass more than 16-bit..." 
Use a bit tester on the digital signal - you'll see the truncation of the lower bits if you send 20 or 24-bit signals via the Lightpipe.
Use a bit tester on the digital signal - you'll see the truncation of the lower bits if you send 20 or 24-bit signals via the Lightpipe.
It's OK.Dr_simon said:
Anyway, just so there is not even more confusion after this post, I don't want to mix the signal either (if I did the M-audio SAM could do it but just at 16 bits I think); I want to aggregate it! S/pdif can carry two channels of audio, ADAT can carry eight channels. I want to take the eight channels that arrive as four s/pdif data streams and turn them into one eight-channel ADAT stream. (And possibly the other way around, but thats so simple I could probably do it without an FPGA. Or perhaps not, I havn't looked the data protocols...)Thanks for the heads up BBS, I'll check it out (extensively!!!!!) when I get it as it is the sort of thing that will wreak havoc with my sound card and I’ve had just about enough of musicians friend and there doggy digital patch bays !
Now BasPer, Gammeldansk or Underberg you say, Hummmmmmm, very intriguing !
Now BasPer, Gammeldansk or Underberg you say, Hummmmmmm, very intriguing !
Last edited:
Fletcher
New member
I just asked Samara in our office [M-A's "Digital Girl Wonder"]... she said you're shit out of luck [is that a technical term?].
She said that HOSA apparently makes a SP/DIF to 'TOSLINK' converter... but that's two channels and doesn't conform to the ADAT OPTICAL format... and it's only 16 bit.
Reckon there might be an opening here for some enterprising young company...
She said that HOSA apparently makes a SP/DIF to 'TOSLINK' converter... but that's two channels and doesn't conform to the ADAT OPTICAL format... and it's only 16 bit.
Reckon there might be an opening here for some enterprising young company...
jet-rocker
New member
You'd think so. With all the SPDIF AD/DA converters, effects boxes, etc. And all the soundcards with lightpipe ins/outs....
Seems a very obvious product. Unless there is some high cost obstacle in the way--that can't be seen from the groud.
I'll be waiting for such a product.
-Jtt
skippy
New member
The real issue isn't obvious: clock skew and drift. It's be relatively easy to do with 4 S/PDIF sources that all used the *same master clock*, because you could then put the output stream together without needing extensive buffering or sample rate conversion. Unfortunately, some schmoe is going to plug in 2 from one unit, and two from another unit (with a slightly different clock frequency), and then sue you into the ground when they get pops and clicks...
For this product to be really viable, you *have* to handle the general case of asynchronous clocks, and that'd be somewhat more expensive and challenging. You have to use phase-locked loops and multiple sections of some pretty sophisticated FIFO.
But having said that, this is already a solved problem, and there's a product out there that does it. The Z-Sys z8.8a digital detangler can do this: http://www.z-sys.com/pp_detm.html#z8 . It's a little on the pricey side (around a grand), but it has the SRC hardware that you need built in, and was designed around this problem. And though it may not seem like it, synchronizing clocks that are .000001% off (like with two units with different clock crystals) is really *exactly* the same problem as converting 44.1kHz to 48kHz, anyway...
I have the Z-Sys Optipatch (just like Bruce), because I just needed to route lightpipe at 24bits. It doesn't include the SRC hardware, so it can't do the combine/split trick. But it works like a champ- and was significantly cheaper than the z8.8a.
In general, Glenn Zelniker (the Z in Z-Sys) seems to have a pretty good handle on shipping digital data around. I've talked to him on the phone quite a bit, and his background is in industrial DSP design, similar to mine. Multirate signal processing is just one small portion of that discipline. It was nice being able to talk to him nerd-to-nerd!
Anyhoo, standard disclaimer applies: I have no connection with them other than being a happy owner of an Optipatch, which I think is the greatest thing since sliced bread. But if you really need to do this, you really need to do it _right_- so this would be worth checking out...
For this product to be really viable, you *have* to handle the general case of asynchronous clocks, and that'd be somewhat more expensive and challenging. You have to use phase-locked loops and multiple sections of some pretty sophisticated FIFO.
But having said that, this is already a solved problem, and there's a product out there that does it. The Z-Sys z8.8a digital detangler can do this: http://www.z-sys.com/pp_detm.html#z8 . It's a little on the pricey side (around a grand), but it has the SRC hardware that you need built in, and was designed around this problem. And though it may not seem like it, synchronizing clocks that are .000001% off (like with two units with different clock crystals) is really *exactly* the same problem as converting 44.1kHz to 48kHz, anyway...
I have the Z-Sys Optipatch (just like Bruce), because I just needed to route lightpipe at 24bits. It doesn't include the SRC hardware, so it can't do the combine/split trick. But it works like a champ- and was significantly cheaper than the z8.8a.
In general, Glenn Zelniker (the Z in Z-Sys) seems to have a pretty good handle on shipping digital data around. I've talked to him on the phone quite a bit, and his background is in industrial DSP design, similar to mine. Multirate signal processing is just one small portion of that discipline. It was nice being able to talk to him nerd-to-nerd!
Anyhoo, standard disclaimer applies: I have no connection with them other than being a happy owner of an Optipatch, which I think is the greatest thing since sliced bread. But if you really need to do this, you really need to do it _right_- so this would be worth checking out...
Well, it's all a question of what the widget is actually supposed to be able to do. There is quite a few variables involved. If for exemple all the spdif inputs needs to be able to act as master, there is gonna have to be a lot of sample rate conversion chips in it, and those are quite expensive. But lets say we run the entire box from wordclock (or sync from the ADAT input), and everyone else are slaves (meaning that all the spdif stuff on the inputs needs to be clocked either from an spdif output or from wordclock) and limit ourselves to one frequency at a time (either 44,1 48, 88,2 or 96) then it should be a fairly simple endeavour. So then you could run either 4 x spdif -> 1 x ADAT at 44,1 or 48; or 2 x spdif -> 1 x ADAT at 88,2 or 96. (And vice versa.)jet-rocker said:
The box would then have:
4 x spdif input
1 x ADAT lightpipe output
1 x ADAT lightpipe input
4 x spdif output
1 x wordclock input
5 x wordclock output (overkill?)
and perhaps
4 x spdif outputs
that are just copies of the first ones so you can clock the spdif input devices using an spdif output each without using a splitter box and without using up your outputs. (Or is that overkill? 'Cause connectors aren't for free.) And should the spdifs be optical or electrical, or both? I really don't think we need AES/EBU, because anyone who can afford four boxes with AES/EBU I/O can afford an AI4 too. Come to think of it I think the box should just have four spdif I/O and one ADAT I/O, and always recover clock from the ADAT input, just to keep it as simple and inexpensive as possible. (A disclaimer is in place here: I havn't looked the ADAT or spdif standards so I'm not quite on solid ground here.)
Anyway, I'd like such a box; would you? And what would you pay for it?
Yes, drift is hell. But if we only allow one master there is only skew left to deal with, and then we won't need all those PLL's. But I don't own enough boxes to make any statistcally significant guess as to how often output-only widgets like synths and ADCs have spdif and/or wordclock sync inputs.skippy said:
skippy
New member
Very few, unfortunately. And it only takes one asynchronous item to break the simple solution- which is why any commercially viable product has to do it the hard (and general!) way. You'd cut your market down by a huge factor if you said "it'll do it, but only if you have house clock: otherwise, you're on your own".
There are a lot of output-only, async devices that Joe Sixpack might want to plug into such a box. CD and Minidisc players, synths, cheap soundcards: there are more async devices than there are syncable devices, and Joe's going to demand that that product work perfectly and seamlessly with all of them.
Synchronizers are a most amusing problem in hardware design in any case. They aren't easy to get right: the problem of sampling a digital signal with a clock, where the transition times of the clock and the sampled signal are completely arbitrary, is one that vexes every microprocessor designer from cradle to grave. This is a case where synchronization failure just leads to clicks and pops, instead of a machine crashing or an airplane falling out of the sky. But people get plenty pissed about it, just the same.
If you wanted to build a 1-off unit with the caveat that you're using only house-clock-slaved devices, you could do it with a single FPGA: the only special thing it'd need would be a single onboard PLL for clock multiplication. The rest would be a couple of very simple state machines talking to each other and a handful of tiny (2-word) FIFOs. That wouldn't be too hard to do...
There are a lot of output-only, async devices that Joe Sixpack might want to plug into such a box. CD and Minidisc players, synths, cheap soundcards: there are more async devices than there are syncable devices, and Joe's going to demand that that product work perfectly and seamlessly with all of them.
Synchronizers are a most amusing problem in hardware design in any case. They aren't easy to get right: the problem of sampling a digital signal with a clock, where the transition times of the clock and the sampled signal are completely arbitrary, is one that vexes every microprocessor designer from cradle to grave. This is a case where synchronization failure just leads to clicks and pops, instead of a machine crashing or an airplane falling out of the sky. But people get plenty pissed about it, just the same.
If you wanted to build a 1-off unit with the caveat that you're using only house-clock-slaved devices, you could do it with a single FPGA: the only special thing it'd need would be a single onboard PLL for clock multiplication. The rest would be a couple of very simple state machines talking to each other and a handful of tiny (2-word) FIFOs. That wouldn't be too hard to do...