XMOS-based Asynchronous USB to I2S interface

I finally got the WaveIO up and working with the Buffalo IIISE this weekend. I'm impressed. Very good sound. Thank you Lorien for your great work.

It has been interesting comparing the unisolated and isolated i2s. It seems that the latter gives more energy and less compression in the midrange. Can't explain why, but I'm quite sure of my findings.
 
is 10 cm too long?

Be sure your I2S wires are well away from any power wires. Shorter I2S lines are always better.

Hi Brian,

AC is all 6 inches away (and well twisted and trans. are in shielded inner housing) and DC is about 3 inches away (width of wave board) also well twisted. Since the mac recognized the Luckit board at 384, 24 bit, I left it at that level, with integer output from Audirvana. I have not changed any of the dip switch settings, which worked great with toslink at 24/96—should I change any of these? Would (not home, not sure if this is possible) my setting the sound out to the Luckit at a lower frequency (96 or 192) be a way of seeing if the length of the uFL cables was limiting bandwidth?
 
I am confused. When I look at the connection to your BNC connector in the photos you linked to, it looks like the center connector (hot lead) is connected to the circular pad; and the shield (return) is connected to the square pad on the pair of isolated SPDIF outputs. Yet, in post 1874, Lucian said the "squared pad is hot" and the "round one is for returning currents" via the cable's shield. If your unit works wired that way, is it somehow because the isolation on that output makes the polarity moot?
 
For the non-isolated out are you using uFL cables? I just tried this last night and am having problems. Did you end up seting any of the dip switches to non-default settings?

I did use uFL cables for the non-isolated, but cut the wire in one end and pushed the screening a little bit back (I don't have uFL-connectors on the Buffalo. Used a separate ground wire from Buffalo i2s-ground to WaveIO ground. No changes in the dip-switches. Just remember that i2s doesn't work if you do not to short the IP_S on the Buffalo IIISE card!
 
[QUOTEI did use uFL cables for the non-isolated, but cut the wire in one end and pushed the screening a little bit back (I don't have uFL-connectors on the Buffalo. Used a separate ground wire from Buffalo i2s-ground to WaveIO ground. No changes in the dip-switches. Just remember that i2s doesn't work if you do not to short the IP_S on the Buffalo IIISE card![/QUOTE]

Out of curiosity, how long are the cables in the one end cut off form? I'm getting sound, but there is "hash" surrounding it, more noticeable on quiet, simple music; with a full jazz combo going it's still there but harder to notice. Also, do you know the frequency that you're sending to the BIIIse?
 
Out of curiosity, how long are the cables in the one end cut off form? I'm getting sound, but there is "hash" surrounding it, more noticeable on quiet, simple music; with a full jazz combo going it's still there but harder to notice. Also, do you know the frequency that you're sending to the BIIIse?

I guess that the cable length that I used for the non-isolated i2s connection was around 8 cm. Maybe you have some ground issues. WaveIO has "cabinet" ground on one of its legs. I use it. Are you using a separate power supply for the WaveIO? I made my own power supply and cut the 5+ wire in the Usb-cable to be on the safe side.
 
Other than the board power supply "ground" and the grounds coming from the BIIIse, I haven't done anything to ground the Wave board. My DAC has 3 power supplies, and the Wave and BIIIse each use side off of the same CT transformer. The housing is non conductive, but I have a shielded interal box around the AC components.

My plan is to eventually have a switch on the USB cable ground, so that once a connection has been made, I can see if I get better results with no ground connection between the computer and Wave card; but that's gotta wait until everything is working correctly. I don't have what I am used to thinking of as a ground issue (old, analog mindset—ground issues would be heard as hum or other noise)—instead what I get is complete silence when no signal is present, but distortion when there is a signal. This may well be a ground issue with digital...
 
Have my Wave I/O completed (SPDIF isolated output wired with BNC center conductor going to square pad of output pair and shield to circular pad; never got any feedback as to whether this is correct or wrong - see my post #1965 above ). Problem: no output. I get a "host active" yellow LED lit, and a "streaming" green LED lit, when attempting to stream an 88.2 khz FLAC file from Media Monkey on my Lenovo ThinkPad (running Windows 7), but no music. Also, despite the resolution of this file, my 48 khz red LED lights up, as though what is coming into the Wave is lower-rez. I downloaded and installed the latest 2.15 driver. Any ideas on what could be wrong would be greatly appreciated. Saw some discussion about issues with drivers but I thought that pertained to Windows 8 (?).
 
Have my Wave I/O completed (SPDIF isolated output wired with BNC center conductor going to square pad of output pair and shield to circular pad; never got any feedback as to whether this is correct or wrong .......

Square pad is gnd. This is working:
 

Attachments

  • PC240002_2.JPG
    PC240002_2.JPG
    490.5 KB · Views: 507
Okay. Thank you, Ian! I will try reversing the connections, so that they are configured as your photo shows. (On the un-isolated pair, I got a clear continuity check between the round pad and ground; but on the isolated pair, of course, neither pad showed any continuity to ground, but I assumed the same pattern would hold.)

I should have put more stock in your photo to begin with. But, it's easy enough to reverse my wires, so...no worries!

Thanks again.
 
Yeah - I guess I should have said "square pad is for shield". :)

Anyway my brother simply soldered a right-angled plastic bodied BNC to his PCB and its shield pin was at the left (when viewed from behind). This seemed pretty standard on the ones I saw at Element14 as well. And that's why I soldered my wires that way around. His works too.
 
The fault is entirely mine because of lack of documentation or even some decent explicative text on silkscreen layers (even though the latter will be addressed very soon). I apologies for that! I attached a picture with a bare board in which you can clearly see the polarity of Murata 1:1 trafo. The dot is on the upper-left corner and, according to the respective datasheet, the output dot should be mirrored hence the square pad is "hot" signal while the round is for returning currents. Anyway, if any of you wants to try Newava alternatives then the output dot will be switched so all infos above will become useless.
@ Ian: in your case (along with other few WaveIOs) things got even harder to connect your SPDIF output because there's no white pant of any of the Murata trafos. That's because I had the "pleasant" opportunity to get a big batch of trafos on which the white descriptions on top of it were simply dissolved by the alcohol I'm normally using to clean up boards - this got me by surprise as long as from 2010 I didn't saw that happen. I hope it will not be any problems with that!
Kind regards,
L
 

Attachments

  • Trafo.jpg
    Trafo.jpg
    157.7 KB · Views: 489
Last edited:
I found some S/PDIF AC waveforms on the 'net and they all seem to look generally symmetrical, so I guess the polarity is unimportant. After all it's not like an analogue audio signal where the coax shield is tied to circuit ground. Mine is working "apparently backwards" after all. I'll reverse it (for "correctness") and report back. Sorry Zapped if I caused you any confusion.
 
Lucian, are you saying that I have connected my BNC incorrectly? It is working after all.

What would be the symptom of an inversion?
I'm not saying anything rather than offering infos gathered by a perfectionist which is trying only to be as clear as possible in this manner (BTW, I'm feeling guilty for this lack of documentation :eek:). On another hand I didn't tried to 'reverse' the SPDIF signals on purpose and see what's happen ... OR, I did, but I wasn't aware about that :) ... The bottom line is that you can leave the wires as are right now or try the other alternative and see if there are any improvements. I wil look forward to your feedback!
 
Well, my unit is working after I reversed the wires to agree with Ian's photo, and it didn't have any output before, so I am hesitant to switch back.

I still have an issue with all files (low rez or hi rez) showing ( on my array of LEDs) a 48 kHz sampling rate, but I'm wondering if that is due to an incorrect setting in the Media Monkey player I'm using. I was thinking of switching to JRiver, which appears to offer greater specificity in terms of file specifications, playback settings, and what not. For the moment, I am enjoying the sounding coming out of my system. My wife says: "that sounds great!" and her ears are better than mine. So...... Thanks to both Ian and Lucian for their help!