Timer - Oscillator per instruction

[Maverickmax]

#define osc_per_inst (12)

Hi

/* Define inital Timer 0/ Timer 1 value for ~50uS delay */

#define T_50micros (66536 - (tWord) ((OSC_FREQ / 26000) / (OSC_PER_INST)))
#define T_50micros_H (T_50micros / 256)
#define T_50micros_L (T_50micros % 256)

/* Define inital Timer 0/ Timer 1 value for ~10ms delay */
#define T_10ms (66536 - (tWord) ((OSC_FREQ * 100 ) / (OSC_PER_INST)))
#define T_10ms_H (T_10ms / 256)
#define T_10ms_L (T_10ms % 256)

/* Define inital Timer 0/ Timer 1 value for ~15mS delay */
#define T_15ms (66536 - (tWord) ((OSC_FREQ * 67 ) / (OSC_PER_INST)))
#define T_15ms_H (T_10ms / 256)
#define T_15ms_L (T_10ms % 256)

/* Define inital Timer 0/ Timer 1 value for ~20ms delay */
#define T_20ms (66536 - (tWord) ((OSC_FREQ * 20 ) / (OSC_PER_INST)))
#define T_20ms_H (T_10ms / 256)
#define T_20ms_L (T_10ms % 256)


Can somebody tell me why osc / 26000 for 50us, osc * 100 for 10ms, osc *67 for 15ms and osc *20 for 20ms?

I am using Keil software (Performance Analyser) to measure the average time for my desired time. Will I get the same result from my microcontroller chip?

Maverick Max

 

[idlebrain]

In 8051 single machine cycle consists of 12 clock cycles and timers incremented once per every machine cycle. So a running timer will be incremented:
for OSC_FREQ/OSC_PER_INST times per second.So the time for each increment is OSC_PER_INST/OSC_FREQ.

where OSC_FREQ is the clock frequency and OSC_PER_INST is 12.

In 8052 all timers are up couting timers. To generate interrupt for 20 milliseconds, the count to be
loaded is 0xFFFF - (20 milli seconds / (OSC_PER_INST/OSC_FREQ).

which equals 63536 - ((20 milli seconds * OSC_FREQ) / OSC_PER_INST)
which equals 63536 - ((20 * OSC_FREQ) / OSC_PER_INST) if OSC_FREQ is defined in KHZ

for example if we take OSC_FRQ as 12 mhz, in this case 1 millisecond needs 1000 increments.
so for 20 milliseconds it is 20000. So the macro is correct for 20 milliseconds.

I am confused with other macroes, if OSC_FREQ & OSC_PER_INST are fixed, they can not be correct up to my knowledge. Simple one is for 10 milliseconds it should be 10 not 100.


#define T_15ms (66536 - (tWord) ((OSC_FREQ * 67 ) / (OSC_PER_INST)))
will becorrect if changed to
#define T_15ms (66536 - (tWord) ((OSC_FREQ / 67 ) / (OSC_PER_INST)))

#define T_50micros (66536 - (tWord) ((OSC_FREQ / 26000) / (OSC_PER_INST)))
will be correct if changed to
#define T_50micros (66536 - (tWord) ((OSC_FREQ / 20000) / (OSC_PER_INST)))

#define T_10ms (66536 - (tWord) ((OSC_FREQ * 100 ) / (OSC_PER_INST)))
will be correct if changed to
#define T_10ms (66536 - (tWord) ((OSC_FREQ / 100 ) / (OSC_PER_INST)))

for all of the above macroes OSC_FREQ is in HZ

The logic is
15 milliseconds is 67th part of a second. 1000 / 15 = 66.66666667.
10 milliseconds is 100th part of a second. 1000 / 10 = 100
50 milliseconds is 20000th part of a second 1000000 / 50 = 20000

But this gives timing difference beacuse of rounding 66.666667 to 67


If 20 milliseconds macro also to be iwritten in same way it is
#define T_20ms (63536 - ((OSC_FREQ / 50) / OSC_PER_INST))

as 20 milliseconds is 50th part of a second, 1000/20 = 50

Please let me now If I am wrong or there is some hidden logic.

Cheers