[time-nuts] Clock Correction Algorithm

Miguel Gonçalves mail at miguelgoncalves.com
Thu Nov 24 11:32:04 EST 2011

Hi Atilla!

On 24/11/2011, at 14:42, Attila Kinali <attila at kinali.ch> wrote:

> On Thu, 24 Nov 2011 14:21:37 +0000
> Miguel Gonçalves <mail at miguelgoncalves.com> wrote:
>> I am using an Arduino Uno that presumably is running at 16 MHz. 
>> I am using a 16 bit timer with a 256 pre-scaler:
>> 16000000 / 256 = 62500 Hz
>> 1 / 62500 = 16 us
>> 65536 - 62500 = 3036
>> I am setting the counter to 3036 and let it overflow after 65535.
> I'm not quite sure i understood you correctly, and i dont know anything
> about the arduino and the avr32. But usually, you set a timer to do a
> certain repetition rate. Ie raises an interrupt ever x clock cycles.
> Or to put it differently, you let the timer run freely, but let it call
> you when it's time to do something.

It's really easy: the timer counter is increased 62500 times per second so on each increase 16 us have passed. I set the start of the counter to 3036 to let it run to 65535. From 3036 to 65535 one second has passed at the 62500 Hz rate  When it overflows a routine is called that simply updates a global boolean flag.

In my main loop I just have something like

while (true) {
  while (!flag) {}
  /* a second has passed... this code is executed between seconds */
  flag = false;

> How do you check for the overflow? If you check a flag in your main loop,
> this would explain the "huge" difference, as this polling will add a
> (not so) constant delay each time the timer overflows, until you start
> your timer again.

I have to check this flag. I do all the processing between seconds. I believe the interrupt service routine should be kept as short as possible. 

Also I checked with a simple counter and the above loop (the inner one) is executed 480000 times per second. I also know that if I don't measure this it will run much more. Right?

Thanks for your help!


More information about the time-nuts mailing list