XMOS-based Asynchronous USB to I2S interface

Perhaps cable adaptors are a little bit expensive compared to simple wires and 2.54 mm headers. And yes, it could be difficult for some people to solder these conns. Though, I am convinced, like you, that uBNC (or U.FL. in my case) should be taken into account thus I'll see if there's room for pinheaders besides uBNC.

Thank you qusp for feedback!
L
 
what you could do and i would certainly throw in for some of these for other projects, would be to throw a simple 4 or 7 (multiple grounds) pin 2.54 pinheader to 3-4 x bnc (with or without mclk) adapter pcb, plus perhaps a simple 1 or 2 ubnc to 1-2 fullsize board mount RCA/BNC for coax for output and input connections onto spare space on an existing job. shouldnt cost more than a few dollars each, or depending on your board manufacturer possibly almost free if spare space is utilized.

w.fl connectors can actually be soldered onto the u.fl land pattern and vice versa pretty easily, or as they are so tiny you could have both patterns in parallel, or a custom pattern could be made somewhere in between to fit both for those who have difficulty. probably both in parallel would be best.
 
I've now tried the U.FL I2S into BuffII/LegatoIII and it's without doubt better than the SPDIF. Also, I have no problem with lock using lowest DPLL bandwidth and 24/192 files. When I tried I2S/DSD from Denon 2900 using CAT6 cable, I had issues with intermittent dropout that was only solved using Acko MCU AKC12 to open up the DPLL bandwidth (this is on my 2nd Dac, BUFFII/IVYIII).

It seems Lucian's device has quite low jitter via I2S, as a lot of other users of the BuffII struggled with I2S lock and lowest DPLL bandwidth.

http://www.flickr.com/photos/56308196@N04/6028891956/http://www.flickr.com/photos/56308196@N04/6028337551/
 
Last edited:
Yes, I am thinking to let USB-bus powering aside and opt for external source though this means to make some major changes on that card to fully isolate it from USB. For now my goal is preserve this price while bring few enhancements to the nonisolated I2S outputs like realigning the signals with onboard master clocks... because I'm not fully confident in XMOS's chip :eek:... even if XS1-L1 IC makes this happen already. On the other hand, few improvements in PSU is allways a benefit.
Regards,
L

Very smart move to reclock after the xmos. As far as the galvanic isolation have you considered a manual switch to select a lower USB rate (12 Mbit/s) that would allow galvanic isolation with adum's on the USB side ( but possibly loosing hirez,) It would be a real boon for NOS DAC builders and negate resorting to the SD-players. Those things have extremely low jitter and ability to be slaved by the DAC clock, but they basically put you back to a 50 CD capacity cdplayer from the 90's in functionality.

The beauty of asych usb is the flexibility to use the computer with 3+TB storage, instant access, VST, foobar, hdcd decoding, etc. But as of yet I haven't seen an asych USB design to where the clock and outputs were isolated on the USB side such that the clock could be powered with a high quality PS with clean ground and could be shared for reclocking right at the DAC which is the ultimate low jitter solution that the SD-players have. Same high quality shunt powered noise free clock driving the DAC inputs and the xmos is the ideal situation a lot of us have been waiting on.
 
Last edited:
throw a simple 4 or 7 (multiple grounds) pin 2.54 pinheader to 3-4 x bnc (with or without mclk) adapter pcb, plus perhaps a simple 1 or 2 ubnc to 1-2 fullsize board mount RCA/BNC for coax for output and input connections onto spare space on an existing job. shouldnt cost more than a few dollars each, or depending on your board manufacturer possibly almost free if spare space is utilized.

Yes, I've also thought to these options but I still believe that it's better to minimize the number of connectors/wires in output signal's paths, especially if those signals are I2S word clock & I2S master clock. I saw on the scope what pinheaders could do to a fast clock signal... that's why I've choosed uBNC conns against pinheaders even if the latest ones are simple and common these days.
On the other hand, I could use 2.54mm headers for I2S Data and Bitclock leaving uBNCs for WordClock and MasterClock signals... but the paranoia in me says to leave the pinheaders for DC and low freq. signals only.

@ Alan: You have PM. As I've said earlier, this card have the price of 99 Euro but the shipping costs varies according to destination.

It seems Lucian's device has quite low jitter via I2S, as a lot of other users of the BuffII struggled with I2S lock and lowest DPLL bandwidth.
Thank you vanwykca for sharing your experience with WaveIO. I am confident that jitter can be further minimized and I hope future revisions of this card will bring enhancements, mostly to I2S signals. SPDIF is bound to jitter and it's a lost cause for me :Ouch:

As far as the galvanic isolation have you considered a manual switch to select a lower USB rate (12 Mbit/s) that would allow galvanic isolation with adum's on the USB side ( but possibly loosing hirez,) It would be a real boon for NOS DAC builders and negate resorting to the SD-players.

Yes I thought about that but there aren't any chances for that switch to exist on the card and that's because XMOS chip is another source of noise thus it must be isolated from the outputs too so, the USB isolation is useless in this scenario. Though, there is another solution for this and I'll implement it in future revisions of WaveIO. For now, I'll stick with I2S output reclocking and adding pinheaders to the non-isolated I2S side. Also there will be uBNC footprints compatible with U.FL and W.FL products as well, thanx to qusp!

Kind regards,
L
 
Last edited:
yes lorein, but my suggestion was for people who dont have ubnc and dont want to spend the dollars on an adapter. its just a suggestion for a cheap fix for an adapter.

for me its w.fl all the way if i can. i generally just chop a cable in half and use w.fl or u.fl on one end and hardwire the other end, but its not as neat to look at as an adapter, or even the pinheader to u.fl adapter.
 
Last edited:
micro BNC...

qusp and others, I am very interested in Lorien's board here, and would like to use the micro bnc I2S connections, as it this approach should result in the most integral communication of the I2S signals/clocks.
After a little googling, I am wondering, are folks making up their own cables, or is there someone in the US who will do custom micro bnc/coax cables? Also, are their suppliers for micro BNC jacks for the DAC side?
 
Hi Barrows, Farnell and RS stock the cable assembly and connectors. I used RS stock no 6851031 cables cut in half and one end stripped for the connection to the Buffalo. Works great and not too expensive either. You could always solder U.FL connectors to the Buffalo and use the cables as supplied. Hope this helps.
 
Thanks!

Hi Barrows, Farnell and RS stock the cable assembly and connectors. I used RS stock no 6851031 cables cut in half and one end stripped for the connection to the Buffalo. Works great and not too expensive either. You could always solder U.FL connectors to the Buffalo and use the cables as supplied. Hope this helps.

That is good info. I am thinking of soldering micro BNCs right to the Buffalo-I will be trying a different USB-I2S interface (from a friend) first, but Lorien's board appears to be really well optomized to me and I will probably try it later.
 
U.FL

Lorien,

After doing some research, I have found it is easy to purchase U.FL cables and smd jacks at Digikey here in the US. Now that I have familiarized myself with this technology, I also recommend that you keep the U.FL option on your new board version. This approach seems like a great way to insure clock/data intergrity. And, thanks for offering this approach in the first place.
 
I've scrapped a lot of notebook computers. Most with at least two of those cables in them. Some of the new notebooks with 3G have four cables in them. Of course they only have connectors on one end but if you can solder the other end they should work fine. I take it the shields have to be tied to ground on the DAC side?
 
yeah it seems you have ended up finding them at digikey, thats where i was going to say to get them. i prefer w.fl if i can, the ackodac uses these for all spdif and i2s lines, so i can connect directly. they are stupidly expensive at farnell, but quite reasonable for both u.fl and w.fl at digikey. the adapter from u.fl to sma or bnc is a bit on the pricey side, so for spdif i just chop the cable, but as my primary use is usb->i2s its not really an issue to have suboptimal spdif. for buffalo you can solder them directly for spdif, but i2s you can either solder one across 2 inputs, or i found the best way was to use a small piece of flat wire soldered from ground right along the front of that header and then its easy to solder the surface mount connectors across each input and ground.

or just scrap the buff and get an ackodac lol

you can get upgraded spiffy ones too, with spc shield, teflon cables with gold plated jacks; cant remember the brand off the top of my head. it seems you can get RF tweakers too :D
 
Last edited:
Audionote dac 2.1

Hi
I am looking at replacing the hagtech usb board in my audionote dac 2.1.
Does anybody know if the dac would work in Integer mode with this xmos based board? Does anybody have any knowledge, or thoughts, as to the difficulty of the mod, and likely benefits?

Thanks Carl