"you claim that if you make a RTC with software you will have higher accuracy"

Yes !

The manufacturing tolerance was higher, but the deviation over temperature was less.
So the relative accuracy was higher and after the calibration also the absolute accuracy was higher.



"disadvantage of RTC: you use the timers, not very accurate"

The timer works exact like a digital counter (e.g. 74HC93). So you get always exact pieces of the crystal clock.
Only a wrong software implementation can cause missing counts.



"i would think a software RTC as the last choise if i don't need that much accurancy"

I disagree absolute !

My kitchen clock (AT89C2051 at 11.0592MHz) deviate no more than 20 seconds over the year.



"there is a tutorial of software RTC in
http://www.8052.com/tutrtc.phtml"

I pointed it out already on many earlier postings.

The tutorial uses a wrong approach !

It consider not the variation of the interrupt responding time !

But if you consider, that the timer continues counting also after setting the overflow flag, you can always get exact pieces of the crystal clock.


Furthermore such a wrong approach must totally failed, if other interrupt sources delay the timer interrupt response by its complete execution time and not only by the different execution time of different instructions.


Almost all of my applications need already a timer to generate some delays on the program flow (e.g. flashing a LED). So there is no problem to use the same timer as RTC also.


Please look on my web page to see, how you can get a right working software RTC:

http://www.specs.de/~danni/ap...curacy.htm


Peter