Hello, i'd like to connect 3 different D/A converters (2 * stello DA100s + 1 * Violectric V800) to the lynx AES16, but i don't know if they will synchronize or if i will have to play with software delays..
thanks for your help!
Pierre
thanks for your help!
Pierre
IIUC you want to hook DACs to three SPDIF/AESEBU lines coming from a single soundcard (lynx). Here SPDIF will provide the master clock and since all three lines originate from the same clock, they should be synchronous. However each DAC model will have a different delay (latency) between digital input and analog output. IMO even a short time difference between each band will ruin the resultant sound. However, the constant delays can be offset in software.
as I said driven by the same clock synchronously, as in MCLK. spdif does not provide MCLK, it has an embedded clock, but thats a different thing, it would be a pretty bad dac that used a recovered spdif clock as the local master clock. Additionally each dac, analogue stage, receiver will all have different delays and possibly have different phase. at least one of those dacs is ansync, I dont know about the stello, but the V800 has a DSP driven async input on all inputs and probably a fifo buffer within that. so driving them synchronously isnt going to happen
Last edited:
The spdif input is controlling pace of incoming data. There is no remote feedback controlling the SPDIF transmitter. As a result, all the DACs consume the incoming data in the same pace - i.e. they run synchronously. Any asynchronous sample rate conversion will not have a long-term effect. The DACs cannot drift apart since they are all fed by realtime data stream clocked by the same clock.
Of course they will each display a different delay, IMO a relatively constant one. If the delay was not constant in the long-run, the DAC would have to vary speed of playback which would be audible, since it is fed by a constant data stream. Of course the delay will fluctuate around the mean value due to the ASRC effect.
And of course they will each have a different phase. On the other hand a 3-way x-over will mangle the phase of each band-limited channel anyway.
Anyway, I would not recommend this setup for the delay and phase reasons
Of course they will each display a different delay, IMO a relatively constant one. If the delay was not constant in the long-run, the DAC would have to vary speed of playback which would be audible, since it is fed by a constant data stream. Of course the delay will fluctuate around the mean value due to the ASRC effect.
And of course they will each have a different phase. On the other hand a 3-way x-over will mangle the phase of each band-limited channel anyway.
Anyway, I would not recommend this setup for the delay and phase reasons
haha OK if you want to split hairs ya, agreed, it will have a relatively constant level of delay and phase error for each song, but no I dont think it counts as synchronous, effectively synchronous maybe, but not truly synchronous. it transmits all at the same time of course. I think you are taking the other literal meaning of the word, I take synchronous in this technical context as meaning the Dac clocks are synchronous with the LYNX output clock; ie. they are one and the same or a multiple thereof, not just running at the same time.
for instance if there was a master wordclock, which the lynx has, or of course i2s MCLK. as with async, it may not even be using an audio FS, one may be 80MHz, one may be 22.57MHz time will still be passing, but the spdif clock has no bearing on the dac output.
however depending on the sample rate the delay and thus the offset between each dac output may change, because it will take less time for a fifo buffer/memory to become half full and start playing, so if you have a mixed sample rate playlist and one dac with a fifo one not, or both with fifo of different sizes, the difference between the dacs will vary, in short a complete mess haha.
in theory you could correct for these conditions, but...
for instance if there was a master wordclock, which the lynx has, or of course i2s MCLK. as with async, it may not even be using an audio FS, one may be 80MHz, one may be 22.57MHz time will still be passing, but the spdif clock has no bearing on the dac output.
however depending on the sample rate the delay and thus the offset between each dac output may change, because it will take less time for a fifo buffer/memory to become half full and start playing, so if you have a mixed sample rate playlist and one dac with a fifo one not, or both with fifo of different sizes, the difference between the dacs will vary, in short a complete mess haha.
in theory you could correct for these conditions, but...
Last edited:
AFAIK SPDIF receivers deploy only tiny buffers, i.e. the recovered clock by PLL is almost synchronous to the incoming SPDIF stream. ASRC by principle does not need any significant buffers either, it is just recalculating sample values based on time difference between intervals of incoming and outgoing clock ticks.
What counts here is the output analog waveform. I do not think any of the DACs will vary the delay of output analog waveform compared to time of the corresponding incoming SPDIF samples timed by decent output clock much as it would be a clear and certainly audible sign of huge jitter.
And mixing buffers with samples of different rates? The buffers are small and if not, they have to be flushed or muted before the device allows switching the sample rate, otherwise you would hear the samplerate switch as the remaining cached samples were played faster/slower. I have yet to hear that.
BTW a DAC with huge buffers is totally unusable for video playback. If they are large (AFAIK the almost half a second of HK HD970), users complain loudly and justifiably.
BTW people sometimes use multiple adaptive USB DACs (or multiple stereo USB soundcards) for multichannel playback. Adaptive USB and SPDIF reception deploys the same PLL-based clock recovery.
I would not be afraid to use multiple SPDIF/AES-EBU DACs of the same model for the x-over playback. But certainly not of different make for many reasons, not only for those related to the digital path differences.
What counts here is the output analog waveform. I do not think any of the DACs will vary the delay of output analog waveform compared to time of the corresponding incoming SPDIF samples timed by decent output clock much as it would be a clear and certainly audible sign of huge jitter.
And mixing buffers with samples of different rates? The buffers are small and if not, they have to be flushed or muted before the device allows switching the sample rate, otherwise you would hear the samplerate switch as the remaining cached samples were played faster/slower. I have yet to hear that.
BTW a DAC with huge buffers is totally unusable for video playback. If they are large (AFAIK the almost half a second of HK HD970), users complain loudly and justifiably.
BTW people sometimes use multiple adaptive USB DACs (or multiple stereo USB soundcards) for multichannel playback. Adaptive USB and SPDIF reception deploys the same PLL-based clock recovery.
I would not be afraid to use multiple SPDIF/AES-EBU DACs of the same model for the x-over playback. But certainly not of different make for many reasons, not only for those related to the digital path differences.
who said anything about playing at noticeably/audibly different speeds? it will still have a different length of time before the first edge comes along. i'm also not just talking about an spdif receiver, the V800, one of the dacs under discussion according to their material uses a DSP to buffer the data and 'lower jitter' by deconstructing it and putting it back together again. it does this for both spdif and USB
what does it matter if it has to flush the buffer? that isnt the problem, the problem is it FILLS at a different rate depending on the samplerate and buffer size, it will not start playing until it is half full. who cares if its small or large? it doesnt matter, we arent talking close enough is good enough here, if there are 2 small but different size buffers and they fill at different rates depending on the samplerate, there will be a different delay in time for each samplerate. this isnt conjecture, this is how they work and if one unit has a small one and another has a slightly larger one and the other may not have one at all, we have a constantly varying delay with a mixed samplerate playlist
i'm well aware of the problem with video with the larger i2s fifo memories, but I dont care, I dont really play video on my main audio rig and if I have to I have another dac. its worth the trouble, best digital audio i've experienced is this local memory playback connected to master clock
what does it matter if it has to flush the buffer? that isnt the problem, the problem is it FILLS at a different rate depending on the samplerate and buffer size, it will not start playing until it is half full. who cares if its small or large? it doesnt matter, we arent talking close enough is good enough here, if there are 2 small but different size buffers and they fill at different rates depending on the samplerate, there will be a different delay in time for each samplerate. this isnt conjecture, this is how they work and if one unit has a small one and another has a slightly larger one and the other may not have one at all, we have a constantly varying delay with a mixed samplerate playlist
i'm well aware of the problem with video with the larger i2s fifo memories, but I dont care, I dont really play video on my main audio rig and if I have to I have another dac. its worth the trouble, best digital audio i've experienced is this local memory playback connected to master clock
Last edited:
many thanks for the answers, that's what i suspected, the synch is not garanteed. I'll try to find another stello used (it's >1000 bucks new...)
i have made some tests in 3 ways with the 2 stellos i already own (in mono of course), 100% active filtering with convolver + Reaper, and found this setup was a big improvement compared to the 2 ways setup with passive XO for mid+ hi.
i have made some tests in 3 ways with the 2 stellos i already own (in mono of course), 100% active filtering with convolver + Reaper, and found this setup was a big improvement compared to the 2 ways setup with passive XO for mid+ hi.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.