Re: [linux-audio-dev] MIDI sync issues; mmc, mtc, ...

New Message Reply About this list Date view Thread view Subject view Author view Other groups

Subject: Re: [linux-audio-dev] MIDI sync issues; mmc, mtc, ...
From: Paul Barton-Davis (pbd_AT_Op.Net)
Date: Tue Dec 19 2000 - 23:00:00 EET


>> > I recorded multiple tracks this way, and to my suprise,
>> > they stayed nicely in sync.
>> >
>> > Now is this the expected behaviour, or is it more common that the speed
>> > of various hw devices (and of course, soundcards) vary from time to time,
>> > sooner or later resulting in lost sync? Uhm, maybe a little dumb question,
>> > but this is something that's hard to verify yourself. Just because my two
>> > devices x and y stay in sync doesn't say much really... :(
>>
>> Kai, I think you already know the answer: you may be lucky now and then
>> but in general two devices will not stay in sync. In a five minute track,
>> if the x-tals are accurate to 10 ppm, the drift will be 3 ms, i.e. too
>> large. I have a soundcard with two independent chips and even they won't
>> stay in sync even though they see the same temperature and voltage drift.
>
>Eh? I think MMC re-syncs once per second (or is it beat?), which means
>that your maximum drift (at 10ppm) would be about 10 microseconds. Not that
>I beleive 10ppm is feasible over MIDI, but in general MMC is fine as
>long as you maintain tempo maps between the two devices.

MMC doesn't have anything to do with sync. You're thinking about MTC,
which is a MIDI representation of SMPTE time. Typically, you'd expect
to see one MTC message every 2msecs or so, and you'd expect the
receiving device to use a phase-locked loop to remain as in sync with
the incoming MTC as it can.

There is no particular reason to expect things synced with either MTC
or SMPTE to drift by more than a couple of SMPTE frames, and in
general, no more than 1. This means basically about 1/30sec, or about
33msec.

While things are running, once the PLL's in the receivers have
established the rate, unless you change the sender's rate, I wouldn't
expect them to drift to any detectable degree.

--p


New Message Reply About this list Date view Thread view Subject view Author view Other groups

This archive was generated by hypermail 2b28 : Wed Dec 20 2000 - 00:09:58 EET