Outboard S/PDIF out
Posted by: jtwang on 09 October 2008
For about a year I've been using my Mac Mini's optical digital output into my SuperNAIT. This is a setup that's been advocated by many on this forum. However I couldn't shake the feeling that something was missing compared to Ken Christianson's excellent setup where he uses a TC Electronic Konnekt 8 for its S/PDIF instead of the onboard S/PDIF output on his laptop.
In theory, digital should be a "perfect" link. I was trained as an electrical engineer and blindly believed this. Unfortunately, hearing is believing...
On a whim, I tried playing the same CD using the coax S/PDIF output from my cheap $100 DVD player. The difference was night and day. It sounded like a shroud had been lifted from the music - I heard much better dynamic range and much more of Naim's trademark tightness.
So I've now ordered a Konnekt 8 as well as an Echo Audio Audiofire 2 to test, purely for their S/PDIF out function and not their DACs. Who knows why there was such a huge difference. Perhaps there is something wrong with Apple's drivers or their particular implementation of S/PDIF. I will keep you all updated when the new kit gets in...
(Before you ask, yes, my Mac's Audio MIDI Utility was set up to have the same sample rate as the original file, 24-bit output. FLAC files on the Mac were played with Cog, which sounded substantially better than iTunes/Quicktime. Volume on the Mac was set to 100%)
In theory, digital should be a "perfect" link. I was trained as an electrical engineer and blindly believed this. Unfortunately, hearing is believing...
On a whim, I tried playing the same CD using the coax S/PDIF output from my cheap $100 DVD player. The difference was night and day. It sounded like a shroud had been lifted from the music - I heard much better dynamic range and much more of Naim's trademark tightness.
So I've now ordered a Konnekt 8 as well as an Echo Audio Audiofire 2 to test, purely for their S/PDIF out function and not their DACs. Who knows why there was such a huge difference. Perhaps there is something wrong with Apple's drivers or their particular implementation of S/PDIF. I will keep you all updated when the new kit gets in...
(Before you ask, yes, my Mac's Audio MIDI Utility was set up to have the same sample rate as the original file, 24-bit output. FLAC files on the Mac were played with Cog, which sounded substantially better than iTunes/Quicktime. Volume on the Mac was set to 100%)
Posted on: 09 October 2008 by pcstockton
that all sounds completely plausible.
I have consistently said in this area of the forum, that the act of extracting the data from the PC/MAC is of utmost importance.
There are many ways to do it, and I personally feel some are better than others.
I think this aspect of "distributed audio" is where jitter becomes an issue and is most likely the prevalent reason for sound differences.
Using some kind of word/super clocked convertor to place between the PC/Streamer/Server etc, and the DAC is crucial.
Whether using a USB, PCI card, mini-digi toslink, or spdif output(s), and/or ASIO, driverless, MAC output etc, this NEEDS to be addressed.
-2cents, P
I have consistently said in this area of the forum, that the act of extracting the data from the PC/MAC is of utmost importance.
There are many ways to do it, and I personally feel some are better than others.
I think this aspect of "distributed audio" is where jitter becomes an issue and is most likely the prevalent reason for sound differences.
Using some kind of word/super clocked convertor to place between the PC/Streamer/Server etc, and the DAC is crucial.
Whether using a USB, PCI card, mini-digi toslink, or spdif output(s), and/or ASIO, driverless, MAC output etc, this NEEDS to be addressed.
-2cents, P
Posted on: 09 October 2008 by james n
I'll be interested to hear your findings with the card. Perhaps the Supernait is more sensitive to its source than other DACs. I know my n-Vi is more particular of its source than my Lavry.
Posted on: 09 October 2008 by winkyincanada
PC, you say that jitter is likely the issue. My (perhaps incorrect) understanding is that the SN DAC re-clocks the S/PDIF signal (much like a Lavry?). Does this not eliminate the jitter error associated with the Mac-to-DAC digital transfer in this regard?
I'm tempted to try the Konnect8 myself, but want to understand the theoretical benefits of a different digital protocol first.
I'm tempted to try the Konnect8 myself, but want to understand the theoretical benefits of a different digital protocol first.
Posted on: 09 October 2008 by Eric Barry
Winky, the best explanations I have seen indicate that "reclocking" takes correct amplitudes with inaccurate timing and converts them into incorrect amplitudes with correct times. They measure as having no jitter but they are not bit transparent.
If reclocking worked as advertised, it should not matter what you feed it. But that does not seem to be the case.
The one exception would be the slaved clock offered by Empirical Audio, or the memory buffer once offered by Genesis (the Digital Lens) and a few others and soon to be revived by PS Audio. With those you take the bitstream, put it in a memory buffer, then spit out a new bitstream according to a new clock. Such a method is bit transparent, but does eliminate the jitter in the input signal, or at least reduce it to the jitter of the new clock. I haven't heard it but I see no fly in the ointment.
By the way, the Lavry originally promised to do this, but apparently had trouble locking onto signals because of the slight variation between its clock and that of the signals it received, so the feature was removed. Because of the difference in clocks, you need either slave the clock (the Empirical Audio solution) or have a buffer and read ahead (the Digital Lens).
If reclocking worked as advertised, it should not matter what you feed it. But that does not seem to be the case.
The one exception would be the slaved clock offered by Empirical Audio, or the memory buffer once offered by Genesis (the Digital Lens) and a few others and soon to be revived by PS Audio. With those you take the bitstream, put it in a memory buffer, then spit out a new bitstream according to a new clock. Such a method is bit transparent, but does eliminate the jitter in the input signal, or at least reduce it to the jitter of the new clock. I haven't heard it but I see no fly in the ointment.
By the way, the Lavry originally promised to do this, but apparently had trouble locking onto signals because of the slight variation between its clock and that of the signals it received, so the feature was removed. Because of the difference in clocks, you need either slave the clock (the Empirical Audio solution) or have a buffer and read ahead (the Digital Lens).
Posted on: 09 October 2008 by winkyincanada
Thanks, Eric.
Seems that maybe we can't get something for nothing! Once we convert a bitstream to S/PDIF there is an error associated with timing that we can't fully recover. Having said that, the fact that computers operate many, many times "faster" than is required for audio frequency reproduction leads me to wonder whether the issue of jitter is isn't all all bit hypothetical in the first place - at least as it relates to audible degradation.
Seems that maybe we can't get something for nothing! Once we convert a bitstream to S/PDIF there is an error associated with timing that we can't fully recover. Having said that, the fact that computers operate many, many times "faster" than is required for audio frequency reproduction leads me to wonder whether the issue of jitter is isn't all all bit hypothetical in the first place - at least as it relates to audible degradation.
Posted on: 19 October 2008 by jtwang
Well, the fact that the computer's CPU operates at a faster clock than the audio sampling rate is totally irrelevant since that has no bearing on what the audio system in the computer is sending out. The CPU simply sends a bunch of bits to the audio subsystem, and it's up to that subsystem to output those bits in a way that an amp or a DAC can understand. This is why Firewire audio cards can sound better than built-in outputs - you are replacing the part that converts "CPU bits" into "audio bits" so to speak.
I heard almost no difference between the konnekt 8 and the Audiofire2, except that the konnekt 8's drivers kept cutting audio out and losing signal entirely. Judging from tc electronic's support forums, this is a fairly widespread problem. So the konnekt 8 is going back to the store.
The Audiofire2, however, is staying. The difference between it and the Mac's built-in output is truly dramatic. The best I can describe it is a veil being lifted off the music. In particular, highs are much clearer and crisper. The sound is as a whole less muddy - it sounds at least as good as it did coming out of my DVD player.
Incidentally, I happen to like my CD rips with the Audiofire2 clocked at 88.2kHz more than at 44.1kHz. Clearly it's not adding any information by clocking at twice the sample rate, but somehow it sounds (slightly) better.
I heard almost no difference between the konnekt 8 and the Audiofire2, except that the konnekt 8's drivers kept cutting audio out and losing signal entirely. Judging from tc electronic's support forums, this is a fairly widespread problem. So the konnekt 8 is going back to the store.
The Audiofire2, however, is staying. The difference between it and the Mac's built-in output is truly dramatic. The best I can describe it is a veil being lifted off the music. In particular, highs are much clearer and crisper. The sound is as a whole less muddy - it sounds at least as good as it did coming out of my DVD player.
Incidentally, I happen to like my CD rips with the Audiofire2 clocked at 88.2kHz more than at 44.1kHz. Clearly it's not adding any information by clocking at twice the sample rate, but somehow it sounds (slightly) better.
Posted on: 19 October 2008 by js
TC should work and I like being able to use a standard dig cable instead of a break out cable.
Posted on: 21 October 2008 by goldfinch
quote:
By the way, the Lavry originally promised to do this, but apparently had trouble locking onto signals because of the slight variation between its clock and that of the signals it received, so the feature was removed. Because of the difference in clocks, you need either slave the clock (the Empirical Audio solution) or have a buffer and read ahead (the Digital Lens).
Does it mean Lavry canīt reclock properly?
Would it be possible to use an external for the input signal before feeding the lavry?
I use a Lynx AES16 + Lavry DA10 and I wonder if it is worth adding an external clock (this soundcad has this facility)
Posted on: 22 November 2008 by js
Thanks for bringing the echo by today. stock < echo < TC and all rather evident.quote:Originally posted by jtwang:
Well, the fact that the computer's CPU operates at a faster clock than the audio sampling rate is totally irrelevant since that has no bearing on what the audio system in the computer is sending out. The CPU simply sends a bunch of bits to the audio subsystem, and it's up to that subsystem to output those bits in a way that an amp or a DAC can understand. This is why Firewire audio cards can sound better than built-in outputs - you are replacing the part that converts "CPU bits" into "audio bits" so to speak.
I heard almost no difference between the konnekt 8 and the Audiofire2, except that the konnekt 8's drivers kept cutting audio out and losing signal entirely. Judging from tc electronic's support forums, this is a fairly widespread problem. So the konnekt 8 is going back to the store.
The Audiofire2, however, is staying. The difference between it and the Mac's built-in output is truly dramatic. The best I can describe it is a veil being lifted off the music. In particular, highs are much clearer and crisper. The sound is as a whole less muddy - it sounds at least as good as it did coming out of my DVD player.
Incidentally, I happen to like my CD rips with the Audiofire2 clocked at 88.2kHz more than at 44.1kHz. Clearly it's not adding any information by clocking at twice the sample rate, but somehow it sounds (slightly) better.
Posted on: 24 November 2008 by Obsessed1
JT, I hope your experience is better than mine with the Konnect 8. I had trouble getting the settings right, with zero support from TC.
Even when the settings finally were correct, I found the sound from MAC to TC to DAC inferior to MAC to DAC every time, using AKG 701s to hear the differences. In addition, the TC was prone to making random spurious electronic noises.
Maybe I just got a duff unit, but make sure that if you buy the TC, you can take it back for a full refund.
Even when the settings finally were correct, I found the sound from MAC to TC to DAC inferior to MAC to DAC every time, using AKG 701s to hear the differences. In addition, the TC was prone to making random spurious electronic noises.
Maybe I just got a duff unit, but make sure that if you buy the TC, you can take it back for a full refund.
Posted on: 25 November 2008 by js
My guess, if it wasn't just driver incompatibilty that happens at times with any computer, is that you didn't get the buffer sizes right which could cause the breaks and could lead to inferior sound. The TC can show up a fly in the ointment. For instance, those that have experience with an active speaker setup understand that keeping everything in the chain at it's optimum has greater importance. The result for instance will sound worse if a properly functioning TC output is reclocked which indicates to me that it does a better job than setups where that isn't the case. Maybe it's driver isn't great with leapord and I've heard of duff firewire drivers in some versions of MACs so who really knows. I wasn't there so don't take this as more than the speculation it is.quote:Originally posted by Obsessed1:
JT, I hope your experience is better than mine with the Konnect 8. I had trouble getting the settings right, with zero support from TC.
Even when the settings finally were correct, I found the sound from MAC to TC to DAC inferior to MAC to DAC every time, using AKG 701s to hear the differences. In addition, the TC was prone to making random spurious electronic noises.
Maybe I just got a duff unit, but make sure that if you buy the TC, you can take it back for a full refund.
It's made for the Pro market and the settings are very common and standard fare to that community. It has a few so it can accomplish some required tasks like mixing. I'm not involved with recording and didn't find them difficult to use on a PC but if there's an incompatability of some sort with your particular MAC than there's probably nothing reasonable to do to fix it. You could always try a more expensive Lynx. Always a good idea to know that you can return something in case it doesn't work out.
Posted on: 28 November 2008 by js
JT, if you have that MAC available, bring it in and maybe we can try to get the TC installed and come up with a how to. DD's has a lot of experience at this, I'm sure he'd help.