I do have a timer that is overflowing every 50ms for a software RTC. I guess I could use it:

/*-------------------------------------------------------------------
    Timer 1 is implemented as a software RTC. overflows every 50 ms
    with 11.0592 MHz
    increments global soft_RTC_ticks which gives relative time in 
    seconds since the system was booted. accurate enough for now.
--------------------------------------------------------------------*/
void timer_1_interr(void) interrupt 3// using 2
{
    static unsigned char data multiplier = 0;    //this many times slower
    TR1 = 0;
    TF1 = 0; //present_keypress = 0;
    if(++multiplier == 20)
    {    multiplier = 0;
        soft_RTC_ticks++;
    }
    TH1 = SOFT_RTC_TH1_RELAOD;    //relaod timer 1
    TL1 = SOFT_RTC_TL1_RELAOD;    //decided from hit-n-trial
    TR1 = 1;
}
 

This concept of state machine in embedded system is new to me since I haven't studied this in our courses. So loops were all I could think of though I know it's bad programming practice. Anyways what I comprehend from this concept is that the states of the keyboard be updated in related variables e.g. key_down etc whenever timer ISR fires, so the program has access to the state of the keyboard which is no older than 50ms. Kindly correct me if I get this wrong. This is new to me.

And wouldn't my software RTC's accuracy be affected if I include the code to update the keyboard state in timer ISR? I had to fine tune it with hit & trial to get an acceptable error before. It was quite a pain to find the values for SOFT_RTC_TL1_RELAOD & SOFT_RTC_TH1_RELAOD that make the soft_RTC_ticks increment exactly after 1 second, considering other overheads in ISR.


-Regards,
Munish