Andy, to give you a good answer we need to know a bit more about
what kind of data rates you're using and what you define as 'accurate
time stamp'. All accuracy is relative. For Radio Astronomy, sub
nanosecond
is required if you're running a wide baseline interferometer. For river
levels, minutes is sufficient.
GPS 1PPS will get you to +/- 30 nSec or so short term. 1E-10 to -12 with
a long term average and a disciplined clock. A tweaked and compensated
Dallas clock, a few seconds a month. A fine tuned PIC about the same.
Reference time is also available on GOES satellites (485Mhz or so).
What do you actually need? Absolute time, or relative? Battery backed
up, or can you survive a power outage and several minute recovery?
If you lock your sampling to an external clock you don't really need to
time stamp your samples. You just need to store the time stamp of the
first (or last) sample, and then compute the time based on sample
number.
Saves one heck of a lot of memory.
Ramana wrote:
{Quote hidden}>
> Hi,
> I am using the PIC as replacement for a dallas clock chip because I can
> integrate some other functionality to it. Mine ia also a data logging
> type. I am using the PIC also to monitor the changes in any Analog
> signal to wake up the main processor which is sort of power hungry when
> not in sleep mode.
>
> So the point is why don't you use the timer in PIC to maintain a real
> time clock. Is there anything I am mising
> rgds
> ramana
>
> {Original Message removed}