| Nerull |
Alrighty. I have an idea about synchronous reclocking my DAC. Firstly, I use an AD1896 to asynchronously reclock (24.576 MHz) and feed that to a 74HC74 flip-flop for reclocking the s/pdif line in my cd player. Is it feasible if I connect the two using a coaxial cable (to prevent interference) or is that not a sound plan? If so, how would I go about doing so (making the two synchronous)?
Thanks!
~Tom |
|
|
| A 8 |
:confused:
Perhaps you want to read that again and then think about how you can change it so that the lesser gifted of us can understand what it is you're trying to get to. |
|
|
| Garbz |
The CDplayer's S/PDIF is being reclocked via a D-Type Flipflop circuit directly from the master clock and then fed directly to the back to reduce it's jitter.
The CDplayer's internal clock is sent to a separate jack on the back of the player.
Both S/PDIF and clock are then fed into the DAC. The S/PDIF to the receiver, the clock to the AD1896 to be resampled to the systemclock required by the DAC. Thus bypassing all the jitter induced in clock recovery.
http://www.tnt-audio.com/clinica/diginterf2_e.html Normal configuration + Clock circuit. |
|
|
| igdrassil |
Yet another way to do it (and a better one, I admit) is to have a clock generator in a DAC module (somehere closer to reclock). Then you fed clock (through RS485 line or something speedy and balanced) into CD player.
In a pro audio equipment there is yet another modification of this technique. You can adapt both of yours DAC and CD transport to be a "master"-"slave" pair.
DAC is acting like a master device, providing clock for transport unit, as a CD transport (slave) has no internal generator.
Use two spdif lines - one from DAC to CD - clock one, and CD->DAC - data one.
Advantage of the "pro" technique - you can use pro- equipment with any of your transport or DAC. Yet another advantage - you may use any kind of frequency dividers/multipliers on the CD side because of good syncronous reclock on the DAC side.
I'm planning to use such way to connect external 24bit/96kHz DAC/ADC module to my Envy sound card. The only difference there will be no spdif - balanced ECL lines rulez forever.
PS: 74hc stuff sucks. Better use 74abt574 instead of 74hc74. |
|
|
| Nerull |
| Sorry! Garbz is correct. Must not have been thinking last night...:rolleyes: |
|
|
| rfbrw |
| quote: | Originally posted by igdrassil
In a pro audio equipment there is yet another modification of this technique. You can adapt both of yours DAC and CD transport to be a "master"-"slave" pair.
DAC is acting like a master device, providing clock for transport unit, as a CD transport (slave) has no internal generator.
Use two spdif lines - one from DAC to CD - clock one, and CD->DAC - data one.
Advantage of the "pro" technique - you can use pro- equipment with any of your transport or DAC. Yet another advantage - you may use any kind of frequency dividers/multipliers on the CD side because of good syncronous reclock on the DAC side.
|
That is not quite correct. In the pro arena everything is slaved to a Sync Pulse Generator like the Isochrone OCX or Tektronix SPG600 |
|
|
| Garbz |
A 8 Incase you're lost we're talking about how to interconnect digital equipment without any loss of jitter. In consumer equipment this presents a great problem. In pro equipment there is a separate and dedicated clock external to all devices which everything else is synched to.
igdrassil your technique is the same as Nerull's only that the DAC is acting as the master instead of the CD. As long as the clock line is carefully transmitted it wouldn't make much of a difference to jitter characteristics would it? Afterall as mentioned by rfbrw in pro gear the clock is external to everything.
Also can you elaborate why the 74HC series does not make for a good reclocking device? Is it propegation delay or what? I know one is CMOS the otherone BiCMOS but what does this mean to the S/PDIF line? |
|
|
| igdrassil |
External clocks are used in a studios.
I mentioned pro- (or prosumer if you like) devices as an example. They can be used as a master-slave pair. Or they both can be externally synched with clock clock generators.
Anyway, you can modify your digital source for a "slave" job adding spdif reciever and some clock dividing circuitry (to convert mck from spdif to a frequency you need in a CD player).
Also it is indeed better to place clock in a DAC module, near to reclock. In that case jitter of a cd->dac (data) and dac->cd (clock) lines are no matter to discuss.
When clock is on the CD side, jitter of the clock line became sufficient. |
|
|
| A 8 |
Nerulls talks about a dacs asrc clock at 24.576 to reclock spdif.
1. 24.576 will not work for cd or dvd
2. Why use a asrc in the first place if you want to use a synchronous clock.
3. If you want to reclock spdif whatever clock you use has to be synchronous with the source of the spdif or you'll get glitches and missed samples.
You guys then discuss the origanal post as if its doable.....thats even more confusing then the missing logic in the firts post. |
|
|
| Garbz |
No I think you misunderstood. The idea is entirely synchronous. We are trying to run 2 lines from the cdplayer. The S/PDIF, and the original clock source.
A single clock runs the CDplayer and the DAC. This is perfectly synchronous without using any form of clock recovery from the S/PDIF line. |
|
|
| rfbrw |
| quote: | Originally posted by Garbz
The idea is entirely synchronous.
|
Not with an AD1896 involved. |
|
|
| Elso Kwak |
| quote: | Originally posted by A 8
Nerulls talks about a dacs asrc clock at 24.576 to reclock spdif.
1. 24.576 will not work for cd or dvd
2. Why use a asrc in the first place if you want to use a synchronous clock.
3. If you want to reclock spdif whatever clock you use has to be synchronous with the source of the spdif or you'll get glitches and missed samples.
You guys then discuss the origanal post as if its doable.....thats even more confusing then the missing logic in the firts post. |
My Asynchronous reclocker has no glitches! You must have two flip-flops in series. Search for meta-stabilitry problem on the net.
Transporting the clock should not be a problem when using a transmission line properly; no jitter added. Jitter comes from SPDIF when recovering the clock with a PLL and data-dependent jitter.
When taking all the care of the CS8412 you can get good results though:cool: |
|
|
| A 8 |
| quote: | | No I think you misunderstood. The idea is entirely synchronous. We are trying to run 2 lines from the cdplayer. The S/PDIF, and the original clock source. |
Then its got nothing to do with the 24.576 clock.....
| quote: | | A single clock runs the CDplayer and the DAC. This is perfectly synchronous without using any form of clock recovery from the S/PDIF line. |
This is pretty basic and not new at all. As indicated before...whats the point using the AD1896 as you then break the synchrony with the source. |
|
|
| Nerull |
What I'm trying to accomplish is best said in #4 on this site, only I'm using the AD1896 instead of the DF1706.
http://www.lessloss.com/types.html
Thanks A 8 for the clock thing though (I can't use 24.576 MHz).
If I slave the cd player to the DAC as shown in that picture, and just use the AD1896 for it's oversampling alone, they should be synchronous, correct? (this is to say that my DAC has a clock frequency to match the CD players original frequency.) |
|
|
| rfbrw |
| quote: | Originally posted by Nerull
What I'm trying to accomplish is best said in #4 on this site, only I'm using the AD1896 instead of the DF1706.
http://www.lessloss.com/types.html
Thanks A 8 for the clock thing though (I can't use 24.576 MHz).
If I slave the cd player to the DAC as shown in that picture, and just use the AD1896 for it's oversampling alone, they should be synchronous, correct? (this is to say that my DAC has a clock frequency to match the CD players original frequency.) |
But AD1896 is not an oversampling filter in the same way as the DF1706. It's function is to bridge two asychronous clock domains. For this reason MCLK, in master mode, can be synchronous with the output or the input but not both. If the transport is synchronous with the input side then the output side of the AD1896 will require a separate oscillator. If the transport is synchronous with the output side, sending MCLK back becomes pointless. |
|
|
| A 8 |
| quote: | | My Asynchronous reclocker has no glitches! You must have two flip-flops in series. Search for meta-stabilitry problem on the net. |
You never fail........pushing your own stuff. I can't help wonder why you can and others can not. |
|
|
| A 8 |
| quote: | | If I slave the cd player to the DAC as shown in that picture, and just use the AD1896 for it's oversampling alone, they should be synchronous, correct? (this is to say that my DAC has a clock frequency to match the CD players original frequency.) |
A ASRC is a Asynchronous device so the answer is no.
Having said this I don't think many have though about or tested what happens if you run a ASRC on a synchronous clock....technically and provided the jitter is identical then maybe one could argue that it becomes a digital filter.
In practice this is probably not realistic and as you then have the option to do it all truly synchromous there is really no point. |
|
|
| Elso Kwak |
| quote: | Originally posted by A 8
You never fail........pushing your own stuff. I can't help wonder why you can and others can not. |
This is one of those pecularities of this forum LOL :rolleyes:
Will not do it at the place where I am mod hahaha
:clown:
Just complain with the mods here. They are on red alert 24 hour per day!
:angel: |
|
|
| Nerull |
Alrighty A 8. Thanks for clearing that up. I was thinking it through in my head today and figured how ridiculous it was.
How about if I feed the PCM1794 the master clock and send that to the CD player, if they're the same frequency. Would that be a better idea? (not to mention can I stick with the 16.9344 MHz clock?)
~Tom |
|
|
| A 8 |
| quote: | | How about if I feed the PCM1794 the master clock and send that to the CD player, if they're the same frequency. Would that be a better idea? (not to mention can I stick with the 16.9344 MHz clock?) |
Thats a very good (and classic) way to do it! |
|
|
| Garbz |
Ok I'm all clear now. Is there anything wrong with doing it the opposite way such as using the pre-existing clock in the cdplayer and feeding it through to the DAC without any sample rate converting? Other then the obvious line induced jitter ofcourse ;) .
As a matter of interest is the MCLK line in the I2S signal identical to the crystal used in cd/dvd players? |
|
|
| e.lectronick |
Is it possible to use a S/PDIF clock to drive a video capture card?Or vice versa? I found this thread because I was trying to think of a way to jerry rig a S/PDIF sound card input to sync up with the clock on a video capture card to eliminate drift between the audio and video portions of my video transfers to digital.
I'm running an LD player --->RF to S/PDIF converter box -->Soundcard + Video capture cards.
The nature of the digital audio signal (AC3) would necessitate that the digital input of the sound card (which is a bit accurate M-Audio 2496 btw = no resampling of digital bitstream) would lock to the incoming S/PDIF clock signal. What I want to do is lock the video capture card clock to the same signal. This discussion seems quite close to my own ideas. The intended result is kind of like a poor man's word clock, right?
Has anyone ever reclocked or resync'ed a PCI card to another source before? I know there was a page somewhere on the web about reclocking an AC97 sound chipset on a motherboard.
Here it is: http://www.lcaudio.com/index.php?page=66
Is this helpful to anyone? I'm definitely going to have to read up on some of the suggested subjects to get up to speed, but perhaps someone can give me a "Hell yeah" if this is not too outlandish an idea.
-Erik |
|
|
| rfbrw |
| Do both cards have a reference input ? Possibly 50Hz in the case of the video card and 48k in the case of the soundcard. |
|
|
| e.lectronick |
rfbrw
No, unfortunately, I'm using a consumer level video capture card, the Sapphire Theatrix with ATI's Theater 550 Pro chip. It's only got analog audio and video inputs. The audio card is an M-Audio Delta 2496. It has S/PDIF input. It's my understanding that the incomng digital signal carries a clock along with the data stream, and the card may then slave to that clock signal for bit-accurate recording. However, my problem is figuring out how to get the video capture card clock to march in step with that same clock signal the audio card is syncing to.
Hope this clarifies it a bit.
-Erik |
|
|
| rfbrw |
| Adding the necessary inputs would not be cost effective. |
|
|
| e.lectronick |
RFBRW, I'm sorry, but I'm not convinced of that.
No disrespect intended, of course, but you must realize the astronomical cost of professional level analog video capture cards with word clock inputs. You simply cannot buy one for less than 2 grand the last time I looked. The investment of some time and a little money to build an external clock circuit like the XO3 or something similar, and reference the two cards to it shouldn't be prohibitive by comparison, no? I have little experience in digital electronics, but I know enough to know that it isn't black magic, and for someone with a little dedication, it shouldn't be impossible.
There are plenty of folks who have reclocked their CD/DVD players and other digital devices using DIY and aftermarket circuits. It consistently produces superior results to most retail products, and certainly is more cost effective than buying retail priced pro gear implementing the same high end technology.
I appreciate that you took the time to respond, RFBRW, but I'm not giving up that easily.
I'm still open to any suggestions folks might have.
-Erik |
|
|
| rfbrw |
| Adding an XO3 or something similar to a CD/DVD is very simple. Sync'ing three devices, all in different clock domains, is somewhat harder. To further complicate matters, the LD player has a RF output. Replacing the clocks in all three devices with sync-capable versions is no small matter, particularly if you have to design said clocks from scratch. True, it will not cost $2000 but it will take a lot more than adding a flip flop or two to the data lines. |
|
|
| Nerull |
Alrighty. Do I feed the guido clock (16.9344 MHz) to the OMCK pin on the CS8416, the PCM1794's SCLK (MCLK basically, pin 7), and the cd player itself? Is that how you feed the DAC the clocks? Would I have to do anything else for this to work (will this work in the first place)?
Thanks!
~Tom |
|
|
|
