On Wed, 2 Jan 2013 19:15:05 -0800, Len Ovens wrote: > On Wed, January 2, 2013 4:05 pm, Fons Adriaensen wrote: > > On Wed, Jan 02, 2013 at 03:35:17PM -0800, Len Ovens wrote: > > Imagine that the counter and logic that define a period already start > > running on the card when ALSA sets the HW parameters (i.e. before the > > start command). And then, instead of resetting that logic and starting > > a period exactly when the start command is given, the card simply enables > > the interrupt to the PC. There will be a random delay between the start > > command and the actual start of the first period. > > > > For a single card things would just work. But it would be impossible > > to sync two of them properly. > > I would think in that case no two of them would work...Well maybe. > Assuming that the cabling is done with power off, they should be pretty > close on power up... though it is probably easy to poke holes in that too. > > The device is designed to sync with a second one, I wonder how they make > it work in windows. It must work well enough for external converters synchronised by word clock, where, it would seem to me, the problems are much worse because of all the different designs and temperature variations. I'm new to all this, but I kind of expect those devices to be synchronisable. So I don't understand why there seems to be extra difficulty doing so with ~identical cards. Does the quality of the CPU clock have any bearing at all? (I made a pretty good ntp server by running a Soekris 4501 with a clock derived from GPS some time ago. PC clocks aren't (weren't) particularly good: http://www.febo.com/pages/soekris/index.html) I noticed, while fitting the 9652s, there are two quite large crystals on board, which may say something about the clock quality. -- John. _______________________________________________ Linux-audio-user mailing list Linux-audio-user@xxxxxxxxxxxxxxxxxxxx http://lists.linuxaudio.org/listinfo/linux-audio-user
