TOSLINK is terrible for jitter. I have to think the only reason anyone uses it is cost. BNC (not RCA!) coax is definitely the way to go, even if a balanced AES/EBU connection is available. While the data is ultimately digital (yes, bits is bits), the signal is analog RF and you have to deal with things like reflections from poor implementation (hello, RCA). If swapping BNC coax cables makes the sound change, the implementation in either the DAC or the source is broken.
As for clocking, source the clock from the DAC and slave the source to it. Unless you have a lot more participants in the clock domain, there is no need for a multi-output clock appliance like the Big Ben.
As for clocking, source the clock from the DAC and slave the source to it. Unless you have a lot more participants in the clock domain, there is no need for a multi-output clock appliance like the Big Ben.
There's simply no mechanism for the data to get corrupted. Digital data at audio bitrates is trivial to transmit over these kinds of distances (a few metres) and bit errors just don't happen. The only thing that can go wrong is the timing of the bits, which is what jitter is a measure of.
Traditionally, the device sending the data in S/PDIF format will act as the clock master, and the receiver will attempt to derive the clock timing by looking at the data it gets. This is difficult to do well, even if the data coming in has perfect timing, and due to imperfect oscillators in the transmitting equipment and imperfections in the cable, the timing of the data arriving at the receiver won't be perfect.
The receiver is thus unable to convert the data values (which it does receive correctly) into analogue values at the correct times. This is problem that jitter causes.
Now, if you use the DAC's oscillator as the clock source, the data source will be sending bits exactly when the DAC is expecting them to arrive. The timing of this data will get corrupted a bit during transmission - either by cable problems or the source's bad design - but it doesn't matter because the DAC doesn't have to derive any timing information from it. So long as the source's jitter is less than the length of a bit (which it always will be at these data rates, even from a terrible cheap digital source) then the DAC receives the correct data, and converts the values to analogue when its own clock oscillator tells it to.
Provided it can transmit the correct data bits, the quality of the source thus plays no part at all in what the DAC is actually producing.
It's true, but has nothing to do with this, jitter or anything else. Even if a DAC was provided with perfect 24-bit data (easy) and a theoretically perfect, jitter-free clock (impossible), it still wouldn't produce analogue signals with 24 bits of information in them. This is because of thermal noise (I think somebody on here quoted Planck's Constant in this context quite recently) and can't be overcome. It's not a "quality" problem, it's a fundamental physical limitation.
You're right, though - the limit seems to be around 20 bits of resolution, or ~120dB.
Wingfeather said:
You are not accounting for the typically poorly implemented coax interfaces giving rise to signal reflections (yes, even at these data rates). While you might want to lump them in with jitter, they really aren't.
steaxauce said:
Anagram's G5 can get to full 24 bir resolution:
http://www.anagramtech.com/products/ip/q5src/
I was thinking to pair it with an AKM "pro" DAC chip.
Telstar, in answer to your original question, using a computer instead of a transport is definitely the way to go.
I've posted a thread on my experience of doing this for the last 3 years or so:
http://www.diyaudio.com/forums/showthread.php?threadid=116683
As I mention in my post, my computer setup (I use a dead quiet Sony laptop) sounds better than my Esoteric P70 transport, which itself is a superb transport.
Incidentally, I have a FF800 sound card also. This sounds better than the MOTU 896HD if an external DAC is not used. With my Esoteric D70 DAC in place, both sound cards sound identical when slaved to it.
Telstar, I'd be interested in knowing which sound card / DAC combo you've decided to go for.
Mani.
I've posted a thread on my experience of doing this for the last 3 years or so:
http://www.diyaudio.com/forums/showthread.php?threadid=116683
As I mention in my post, my computer setup (I use a dead quiet Sony laptop) sounds better than my Esoteric P70 transport, which itself is a superb transport.
Incidentally, I have a FF800 sound card also. This sounds better than the MOTU 896HD if an external DAC is not used. With my Esoteric D70 DAC in place, both sound cards sound identical when slaved to it.
Telstar, I'd be interested in knowing which sound card / DAC combo you've decided to go for.
Mani.
manisandher said:
Thanks for the input, I will read your thread.
It seems that both you, I, and Peter (the maker of XXHE) have a VRDS transport, which is recognized as the best. We have different models but the difference is very little: i have a vrds 10se and Peter has a P700. Your P70 should be slightly better. But I have put a XO3 reclock on mine and changed the AC cable 🙂
I'm eager to do A/B tests with 16/44.1k audio files and my TwinDAC. Then, once i'm sure of the superiority of the computer playback, I'll invest in high-resolution hardware and software.
On xxhe forums there are one thread or two about soundcards and DACs. I am thinking of getting a cheaper firewire soundcard to use as digital transport to the DAC. There should be a couple of options available, around 200€, but I cannot find anybody that used one only for this task for comparison. if the situation remains the same i'll get a FF400 🙂
I would gladly save the money for the Fireface to invest in a better way (like biamping). The fact that Peter found a difference just changing the firewire controller on his computer is making me wonder.
Really? My understanding is the following (please correct me if I'm wrong):
CD audio has a data rate of 1.411 Mbits/sec. If the high-frequency analogue signal has a bandwidth of five times this (I'm just plucking that number out of the air, but I would guess it's conservative) then the half-wavelength of the signal (at 66% of the velocity of light) is 14.17m.
So, unless your S/PDIF cable is at the very least this long, it's not a transmission line and the rules of characteristic impedances and correct termination don't apply. And, even if it is this long, only the higher harmonics of the signal will be reflected - it's going to be a long cable before these reflections are large enough to cause bit errors.
It's common knowledge that RCA plugs are made for cheap consumer audio and don't work to S/PDIF specification, but I don't think it's fair to say they never work in practical situations.
I'm sure you're right and there is little measurable difference between good and bad coax implementations.
I think this is what Wingfeather and I have been saying...manisandher said:
Telstar said:
It should make you wonder. It should make you wonder whether the difference is real or not. If these were the results of blind tests done with multiple individuals in controlled conditions or even the opinions of several respected hi-fi reviewers then I could understand if not entirely sympathise with your desire to have the very best at all cost. Over-engineered is, unfortunately, not a compliment.
Why not spend some of your obvious energy and superfluous cash on improving your skill with a musical instrument, or acquiring some further electronic design skills, either of which will result in greater enhancement of your self esteem than spending ever-increasing quantities of money in the, forgive me, somewhat hysterical and misplaced if not superstitious pursuit of ever-decreasing increments in quality of what is, after all, in most instances, only a re-creation of the real thing?
wakibaki said:
If i didnt trust Peter opinion, I wouldnt bother.
He has a high-efficiency system with horn speakers and those reveal the tiniest details, so I believe that the difference is true, but that on a less sensitive system, it wouldnt be noticeable.
I do think that the cause is the firewire controller - onboard vs a pci card. It has been said that TI makes the best firewire controllers, probably the one on his mainboard was a via or something of much lesser quality.
Manisandher instead is using the same firewire controller on which he tried two different quality soundcards, using only their digital out. The fact that they sound the same it is true as should be.
Now, if somebody can help me find a cheaper firewire soundcard with digital output, I would be grateful to save that money 🙂
Telstar,
Before I got my RME FF800, I used their Multiface II unit. It's cheaper than the FF800, but you'll need to get one of their PCI, PCI Express or Cardbus interfaces (none of which are cheap).
Unfortunately, I never managed to compare the SQ of the Multiface II (with Cardbus) with either my MOTU 896HD or FF800, as I didn't have either of these at the time.
Mani.
Before I got my RME FF800, I used their Multiface II unit. It's cheaper than the FF800, but you'll need to get one of their PCI, PCI Express or Cardbus interfaces (none of which are cheap).
Unfortunately, I never managed to compare the SQ of the Multiface II (with Cardbus) with either my MOTU 896HD or FF800, as I didn't have either of these at the time.
Mani.
manisandher said:
The first thing I want for now (that is until i find the ultimate true 24bit DAC) is a cheap FIREWIRE interface that can allow me to:
1) Have SQ comparable to the Fireface when used as digital (spdif out) transport only. I need this to compare the SQ of my modded Teac with the Computer transport, using 16/44.1 sources and the NOS TwinDAC+.
2) Use exclusive mode with xxhighend.
3) Costs less than 300€ and have decent SQ on its own for being left to the studio pc (vista workstation with Tannoy Reveal 5s acrtive speakers) when I'll find the ultimate high-res dac.
I was considering:
-ESI Duafire (not sure of the SQ)
-Something from m-audio (but i dont trust much their quality)
-A used FF400, but they go to 400-500€ at least.
-Something else?
-Apogee duet does NOT work under Vista, thanks to those i*iots that think only about mac
Consider the Apogee Mini-DAC Firewire? It's true, you can't get it for less than a little under $900 USD, but it's an excellent firewire DAC in its own right, and it has digital thru mode (an s/pdif digital output). It also operates in 24/192. You may just end up deciding to keep the Apogee.
steaxauce said:
No offence, but for that price I'll get a fireface 400 on ebay. Which I also will keep for the studio pc.
b-square said:
I have to disagree with that statement.
Jitter from a TOSLINK transmitter can be easily filtered out.
Jitter from the source clock cannot be easily filtered out.
Whats the difference? Let us say we have a source clock for the SPDIF transmitter that has bad jitter at 200hz. This ugly clock would generate bits that were a few nanoseconds too short for 2.5 milliseconds, then generate bits that were a few nanoseconds too long for 2.5 milliseconds. The SPDIF receiver chip would receive words timed too close together at first and then the words will be too far apart.
This will sound bad no matter what transmission media you use. TOSLINK, Coax, BNC or whatever. Unless your receiver chip has a very good PLL.
IF,,,
If we use a source clock for the SPDIF transmitter that has no jitter, then the bits sent to the TOSLINK transmitter will be the right length and arrive at the right time. The optical transmitter and receiver do vary in their sensitivities, meaning that the received pulse train duty cycle may not be exactly 50% ones and 50% zeros. The rise and fall edges of each particular bit may vary a few nanoseconds, but THE CENTER OF THE BIT WILL STILL BE ON TIME.
Any SPDIF receiver chip should be able to reject the edge jitter from TOSLINK !!!!
Unfortunately, some do not.
Check out the newer receiver chip by AKM and Wolfson. Do a search and find out how bad the old crappy receiver chips by Cirrus and the others are.
Telstar said:
Okay, I just thought I would suggest it because it seemed like it would be better suited to your needs, since you only need two channels and the Apogee has a higher quality DAC.
steaxauce said:
Ah, this could be useful for the studio pc. How are the drivers?
BTW i think to have found a cheap solution, for the time being:
http://www.terratecproducer.it/prodotti/phase/phase24_fw.shtml
It costs less than 200 euros. What do you think?
About the ultimate high-res 24 bit ladder dac, last night i found the MSB Platinum DAC 3.
I need to wait for some listening impressions online (and I would like a lower price, but that's the current audio market). Not in a hurry anyway 🙂
rossl said:
rossl said:
Ok!
I agree with you on the need for a good clock, though. Except it should be at the DAC end, not the source end. Since that is typically not how such things are done, best to not introduce further jitter in the link between the two. Just my $0.02!
- Status
- Not open for further replies.