Lpc2000

At 09:27 AM 5/11/06 +0100, Andy wrote:
>Robert,
>I am running a shared clock network over ethernet, so to reduce task jitter
>on the slave nodes the timer0 interrupt needs to be as close as possible.

Do you need to have your message sent as close to 1mS as possible?  If so 
I'm surprised that ethernet uncertainties don't swamp any jitter from the 
interrupt.  Never mind the additional uncertainties introduced by whatever 
higher level protocol you are running on ethernet (probably TCP/IP consider 
you are using WIZnet?)

Or do you just need to know the time that you sent the packet as accurately 
as possible.  If that's the case (and I think it should be with a good 
clock synchronization algorithm) then there are other approaches that will 
give you greater accuracy without the interrupt overhead. With the timers 
on the LPC series it is quite easy to setup a polled timer that will give 
you time with an precision of a few tenth's of a uS that will not exhibit 
any significant wrap around oddities as long as you poll it every few 
minutes.  Accuracy is actually limited in most cases by the crystal. Chalk 
one up for 32bit timers, one of the features I really like on this micro.

See

http://www.aeolusdevelopment.com/AppNotes/LPC210X/an-timerperformance.pdf

It's an app note I did on taking some performance measurements on such a clock.


Robert

" 'Freedom' has no meaning of itself.  There are always restrictions,   be 
they legal, genetic, or physical.  If you don't believe me, try to chew a 
radio signal. "  -- Kelvin Throop, III
http://www.aeolusdevelopment.com/