[vdr] Re: Seperation of sat clock code

[Sergei Haller <Sergei.Haller@math.uni-giessen.de>]

On Mon, 29 Apr 2002, Carsten Koch (CK) wrote:

CK> Sergei Haller wrote:
CK> ...
CK> > because the times delivered by the NTP protocol have to be exact up to
CK> > milliseconds (or even microseconds, I can't recall that right now)
CK> > I don't know in which format the time is broadcasted in DVB streams, but
CK> > assume as simple as "YYYY MM.DD hh:mm:ss" or something like that.
CK> >
CK> > even more, the ntp protocol has to measure the (network) transfer time
CK> > from the time server to the client. So in our case this would be the
CK> > transfer time from the broadcasting server via the satellite in the sky
CK> > and to the DVB card. I doubt it's possible to measure that. (again: to
CK> > measure in milliseconds)
CK>
CK> Both of that is true.
CK>
CK> But: Even if we only get the seconds, I would assume that we either get
CK> them more than once per second or we get them at the moment when the
CK> milliseconds are zero (plus network delays).
CK> And: Isn't the network delay well-known? After all, the speed of light
CK> is constant and the distance to the satllite is (roughly) constant as
CK> well.
CK>
CK> So maybe the idea to create a "satelite clock device" is not that
CK> bad after all?

Maybe. I'm just very careful about that:
 - how exact was the time string at the broadcast time.
 - as you said, the distance to the satellite is _roughly_ constant
 - the distance depends on the geographical position of the DVB card
 - how do we know, to which satellite the LNB is pointed?
 - how long is the signal delay at the satellite? at the in-house cables?
   inside the computer/driver?
 - what about DVB-C and DVB-T cards?

and again: all that values have to be exact up to milli- (or even micro-)
seconds.

Onother thing: the NTP protocol is _very_ complicated. Who is going to
implement it for the DVB cards?

At the same time the TP is _very_ simple and doesn't claim to be _that_
exact or to care about network delays.

just my 2 cents

c ya
        Sergei
--
--------------------------------------------------------------------
         eMail:       Sergei.Haller@math.uni-giessen.de
--------------------------------------------------------------------
Be careful of reading health books, you might die of a misprint.
                -- Mark Twain