DAC’s and multichannel

Time to lock initially may change as the clock circuits synch up, but the data should be aligned very closely. Moving a speaker slightly would probably cause more shift. And, the phase alignment of the Trinnov has more impact (and could correct any delay).

Points to ponder:

• 1/44 kHz = 22 us DAC clock period
• Speed of sound is about 1130 ft/s = 13,560 in/s; about 1.13 ft/ms
• One sample at 44 kS/s is thus about 0.31 inches; changing the distance to the speaker by this amount (moving yourself or the speaker) changes the delay by one DAC clock cycle (at 44 kS/s).
• The wavelength of the sound at 1 kHz is 13,560/1000 = 13.56 inches; at 100 Hz, 135.6 inches (11.3 feet)

The delays relative to the frequencies and distances are such that they are likely buried in the mud, especially at subwoofer frequencies. There is probably more phase shift in the crossovers. My guess is that time alignment is not likely an issue in this scenario. Unless I am misunderstanding the problem, possible...

HTH - Don

I am with Don. Word clock synchronization is used to produce the content we listen to. If it is wrong, heaven help us as we are starting with the wrong content .

Ivory said:

Thank you Don and Amir for your replies.

I’m not mainly worried about the PLL locking to the Word Clock or a one clock delay between channels, my concern is about the processing time.
Before getting the DAD, I spoke with Daniel Weiss about connecting several DAC1 MK2 for the multichannel and two ways cross-over and he didn’t recommended it with the DAC1. What I understood was that the clock lock and processing time of the DAC1 may change at every power on causing possible time misalignments.

The installation manual of my DAD AX24 says: Processing delay < 1ms. What about connecting another DAC with a processing delay of 2 ms or 0 ms?
I’m probably missing some important point.
Thank you for your help.

Luigi

Click to expand...

Correct me if I'm wrong, but misalignment in the range of say 0-3 ms are will bw outside the audible range. As long as you run all channels through the Trinnov, the Trinnov time alignment will take care of business, and you can throw any combination of DACs into the mix.

My setup is a little different, but my experience may offer some insight; I have a 5.0 system and a 4 channel Trinnov (strictly for economic reasons). I use an extrenal DAC for mains (using Trinnov digi out), the Trinnov DAC for center channel, and another DAC for the surrounds. The surrounds run through the Trinnov as well, but there is not processing, only systems dely added (Trinnov allows you to as "delay lines" beyond the number of processes channels). This systems delay is in the order of 40ms.

When I used the DAC in an old Meridian 568 for my surrounds, the surrounds were audibly misaligned with mains/center. I presume because this old Meridian DAC takes a lot of time to do D/A conversion (the sound from the surround channels appeared to arrive late). When I changed my meridian to a Bel Canto DAC3 for processing of surround channels, this audible misalignment went away.

From the 5.1 guide of NARAS

"Delays in cinema are not coherent. The front speakers are typically mounted in a
fairly flat screenwall, with the L and R speakers toed in at an angle of 21° or less.
As a result, for most listeners, the center channel arrives approximately 1ms
before the left and right channels. The surround channels have an additional
delay of 10ms, plus the natural delay from the front channels to the listening
position. For this reason, phantom image panning is not dependable from
theater to theater, or from listening position to listening position. Signals shared
between channels (sometimes called “shouldering”) will combine only up to
600Hz or so. Mixes will translate best when decorrelated between channels
(best described as taking a “multichannel mono” approach)."

You can read all about timing in 5.1 systems HERE

Bruce B said:

From the 5.1 guide of NARAS

"Delays in cinema are not coherent. The front speakers are typically mounted in a
fairly flat screenwall, with the L and R speakers toed in at an angle of 21° or less.
As a result, for most listeners, the center channel arrives approximately 1ms
before the left and right channels. The surround channels have an additional
delay of 10ms, plus the natural delay from the front channels to the listening
position. For this reason, phantom image panning is not dependable from
theater to theater, or from listening position to listening position. Signals shared
between channels (sometimes called “shouldering”) will combine only up to
600Hz or so. Mixes will translate best when decorrelated between channels
(best described as taking a “multichannel mono” approach)."

You can read all about timing in 5.1 systems HERE

Click to expand...

But that would be a function of delay choices made in the creation of the 5.1 mix. A 5.1 system running through a Trinnov would have delay set so that all channels arrive at the listening position at the same time, if all channels have no delay at the source. Correct?

edorr said:

But that would be a function of delay choices made in the creation of the 5.1 mix. A 5.1 system running through a Trinnov would have delay set so that all channels arrive at the listening position at the same time, if all channels have no delay at the source. Correct?

Click to expand...

Correct....

Ivory said:

The Trinnov can compensate different DAC’s latency and comply with the recommended timing for a 5+1 system after the calibration, but my concern is if the processing delays on different DAC's normally remain constant after every power on.

Luigi

Click to expand...

I think this is strictly a theoretical concern. No one will be able to hear a shift of 1ms in delay between channels.

Ivory said:

Thank you Edorr for your reply.
I’m not certain that 1 ms is inaudible, it is like moving on a loudspeaker the mid/high speakers 34 cm away from the woofer. I will investigate further.

Click to expand...

You are correct. My math was off. 34cm (1ft) is probably audible. However, if DonH50 math is correct, 1ms would correspond wo about 100 samples. Why would the locking related misalignment be so high? So I think 1ms is off and the difference will be far lower, and hence inaudible.

Several things going on, here are some random thoughts:

1. I am not sure what "lock time" means. The time for a PLL to lock to carrier does vary and may take 100 ns or 1 s, depends upon the design and other factors. Once locked, it should stay locked. Lock-in happens when you first send a signal across, and all standards I know include start-up and idle patterns to ensure there is no "lock time" seen in acquiring the data.

2. I suspect the lag is processing-related. A DSP takes time to do its magic, and earlier DSPs took longer (slower and less processing power). Seems to me processing lag could be a ms. Certainly many cycles of the DSP clock, but that does not correspond to sample clocks (or very loosely). In an AVR it is accounted for and all channels are delayed equally. Add a stand-alone box, and the lag may no longer match, obviously. Most room correction programs would compensate for that, however, assuming the lag is within its range.

3. There is also some lag/phase shift through the analog components at the output of the system. This again will delay the output and potentially cause time errors. And again Trinnov or whatever should correct for them.

4. I do not recall our sensitivity to delays. I suspect 1 ms is audible, but in what way I am not sure. If you move a speaker 1 foot closer to you and adjust the balance so volume is identical, you can still tell one is closer. I believe there is a thread someplace where we went through the number and found our ability to localize allows us to distinguish down into the us range. There is also the "echo" effect, wherein a delayed version of a signal creates an echo or reverberation-like sound. I think that is more in the ms range but am not sure.

5. Feel free to check my math as I do not always check it closely enough when posting on the fly. I am often doing something else between writing and reading (like practicing).

usual disclaimers apply - Don