[ntp:questions] Re: WLAN synchronization @ 10-100us

Hal Murray hmurray at suespammers.org
Mon Dec 8 08:55:33 UTC 2003


>I'm looking for a way to synchronize the clock of several nodes over a
>WLAN network. The network will be controlled in such a way that no
>other traffic will be sent during synchronization. I can't use
>standard NTP protocol over UDP/IP due to the extremely high
>requirments in synchronization. The clocks in the nodes must be
>synchronized to the same clock within less than 100us.

Do you need absolute time, or just to have all your machines
agree so time differences will be correct?

There are two parts to your problem and/or error budget.

First is to get the time correct.  Second is to keep it
correct for as long as you need to.  That requires getting
the frequency right, or rather figuring out what the frequency
of your local crystal actually is so you can correct for it.

Ballpark crystal "drift" is 50 ppm.  You didn't specify how long
you need the clocks to stay in sync.  50 ppm is 50 microseconds
per second, or 50 ms per 1000 seconds (15 minutes).

The first thing that I would suggest, would be to run NTP on
all your machines and turn on the statistics logging and see
how far off they think they are on your normal network setup
and learn the drift of each crystal.

Drift depends upon temperature - ballpark of 1 ppm per 10 C.

-- 
The suespammers.org mail server is located in California.  So are all my
other mailboxes.  Please do not send unsolicited bulk e-mail or unsolicited
commercial e-mail to my suespammers.org address or any of my other addresses.
These are my opinions, not necessarily my employer's.  I hate spam.




More information about the questions mailing list