Quote:
I tried a bunch of high bitrate files and they worked just fine over USB. They were DVD Audio rips with 96/24 5.1 tracks with bitrates of ~6900-13824kbps. Tried WASAPI and DS and they all played like they should. I haven't had an opportunity to try optical yet, but coax does support 192/24 here, so the only input I haven't tested is the optical one.
For the USB behavior, I don't think it is so much the bit rate as it is the sample rate that is being loaded into the DAC. The only place that I have seen this happen is with 24/192 files that I got from HDtracks.com and only about half of them do it (about 7 of 15 tracks). I cannot reproduce the problem with any of the 24/96 files that I have.
The other thing is that it is sensitive to the type of digital source and USB Transfer mode being used. If I send audio data through the stock iTunes on my iMac running 10.6 and set my midi to 24/192, the problem doesn't ever show itself. However, if I bypass all of the Coreaudio processing in my iMac using something like BitPerfect, then the problem can show up. AlanHell seemed to get the problem when using a PC, but I'm not sure what kind of sourcing hardware or software he was using.
In other words, when the highest sample and bit rate possible for the DAC is used, and data is fired down the USB at the fastest transfer rate possible (probably using asynchronous mode which would add some handshaking overhead as well), it is as though some kind of processing overload in the DAC can glitch triggering the noise level to come way up to a significant level.
Note that I am not very familiar with the PC world audio support. I have heard that WASAPI can support two output modes, PUSH and EVENT DRIVEN, but this is only if the sound card in use supports both. My guess here is that one mode corresponds to USB's Adaptive (i.e., synchronous) Transfer mode and the other mode correspond's to USB's Asynchronous Transfer mode. In the former mode, the audio data is just output from the PC synchronously in real time (PUSHed?) and the DAC has to accept it and recover a clock from it. For the latter mode, when the DAC is ready for more data (based on its own internal clock), it sends a message to the PC (an EVENT?) that the PC must asynchronously respond to (DRIVEN by the event?) by sending a batch of audio data back to the DAC. The latter of these two is much more desirable in that it eliminates the source of most jitter that can occur on the data stream going to the DAC. This would normally be detected as a more open soundstage with a higher sense of clarity and focus.
Again, the hiss problem seems to be affected in part by the transfer mode being used. I'm GUESSING that the asynchronous mode is where the issue can occur, so if your sound card cannot support that mode on the PC, then you probably won't see the problem.
(By the way, If I have these PC terminology and issues are all wrong, please someone correct me! Although I think I've figured them out, I really would prefer to know what I was talking about here)
It's nice to know that the speed limitation doesn't seem to exist on the coax interface. This is consistent with the data in the instruction manual that you actually get with the HDVD800 (the S/PDIF speed limitation is only mentioned for optical). But it is also a bit confusing since I was always under the impression that most DACs taking S/PDIF input used a common S/PDIF clock recovery and data processing. The optical interface just adds two or three components to convert the optical signal into the common S/PDIF format used internal to the DAC. The Coax interface did the same for Coax (i.e., a few components to prep the electrical signal for the common S/PDIF processing in the DAC).
