Re: [LAU] [OT reply to recived OT ; )] RME analog out of phase to ADAT connected Behringer analog out

On Tue, Jun 14, 2011 at 10:03 AM, Ralf Mardorf
<ralf.mardorf@email-addr-hidden-dsl.net> wrote:
> On Tue, 2011-06-14 at 08:22 -0400, Paul Davis wrote:
>> On Tue, Jun 14, 2011 at 8:11 AM, Ralf Mardorf
>> <ralf.mardorf@email-addr-hidden-dsl.net> wrote:
>>
>> > And as you could read I found an issue. At Thomann (a German dealer)
>> > they claimed there are no sync issues, what I couldn't believe. Doing
>> > this test I found out, that there are bad phases.
>>
>> You DIDN'T SYNC ANYTHING. Or at least, your description doesn't
>> mention any kind of sync.
>
> That might be correct. The optical ADAT connection can't be used for
> sync?
>
>> And if you had tried to sync, you would have
>> found that you CANNOT sync two A/D converters unless they are the
>> exact same model (so that their pre-A/D buffer is the same size).
>
> I suspect the RME using other, better converters than the Behringer
> does. This is why I expect that a completely perfect sync for the phases
> might be impossible. Thomann (the dealer) claimed I'm mistaken ;), sync
> is possible.

you're really deeply confused. let me try to illustrate (the diagram
is somewhat simplified to try to make the point more clear. the actual
reality is a bit more complex in terms of where the clock is actually
used, and so on). the section marked "buffer" refers, typically, to
some RAM into which a relatively large (say, 32 or more) samples can
be stored "in front of" the D/A converter.

digital samples-> buffer-> D/A converter digital-side>internal
buffer->D/A converter analog side
                                                   ^
                                                   |
                                                   |
                                             sample clock

the sample lock says "get another sample", so the device will fetch
another sample at precisely the same time (to within a picosecond or
so) of other devices using the same clock (if any). this is important.

but ... the way each D/A converter circuit works is independent once
it has the next sample value available. the amount of buffering inside
the D/A converter is variable from converter to converter, as is the
actual conversion process itself. So lets put 2 of them together:

digital samples-> buffer-> D/A converter digital-side>internal
buffer->D/A converter analog side
                                                   ^
                                                   |
                                                   |
                                         shared sample clock
                                                   |
                                                   |
                                                  v
digital samples-> buffer-> D/A converter digital-side>internal
buffer->D/A converter analog side

it makes no different whether the sample clock is delivered from an
internal clock crystal, the ADAT cable, word clock via BNC or any
other method. once the samples are delivered to each D/A converter,
there is no "sync" anymore.

this has nothing to with RME, and everything to do with the basics of
digital audio and digital-to-analog convertion.
_______________________________________________
Linux-audio-user mailing list
Linux-audio-user@email-addr-hidden
http://lists.linuxaudio.org/listinfo/linux-audio-user
Received on Tue Jun 14 20:15:02 2011