The only thing I can think of why not to switch to higher DCO is power
draw. The same counts for setting SMCLK to DCO instead of DCO/4. Even
though all the peripherals can divide it down, power draw is likely to
be higher. At the same time, we duty-cycle DCO so the increase might
be negligible. At the same time, the higher clock speeds and thus
lower transfer times between the mcu and off-chip peripherals could
reduce the power draw significantly (especially with the radio). I
haven't done the measurements, but my intuition is higher DCO would be
better, because the radio is such a big power hog.

The one peripheral we haven't talked about for the TelosB is the UART.
It heavily depends on the SMCLK = DCO/4 ~= 1MHz as all the constants
for its setup are derived for it.

@Eric: with Janos' software implementation of VHT, you basically get
1us synchronization for free. So the question becomes what can you do
with such high accuracy of time sync? There are certainly applications
out there, that could benefit from it. I have some work in power
metering that benefits from this high accuracy. Localization is
another one, where time sync accuracy can't be good enough.

It seems to me that we need some sort of generalized clock management
system. I see similar problems on the ARM cortex side, where every
chip has different clock configurations. How about a system module
that stores the different clock sources, and peripherals "subscribe"
to it? They can get the current clock configurations from it, and
configure the peripheral accordingly. In case clocks change, the
peripherals get informed through an event, such that they can adapt
their configurations. I attempted something like this on the Sam3u/s,
but I am far from happy with my implementation.

Cheers,

Thomas
_______________________________________________
Tinyos-help mailing list
[email protected]
https://www.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-help

Reply via email to