Posted on February 15, 2008 at 09:48There is a bug in tim.c in the function
TIM_PWMOModeConfig()
The pulse width value is increased if the requested value is less than 5
/* Set the Duty Cycle value */
if (XDutyCycle < 5)
{
XDutyCycle = 4;
}
This shouldn't be there as it means that the minimum value of PWM can never be set.
The next line is:
TIMx->OCAR = XDutyCycle - 4;
This is also wrong, as 5 needs to be subtracted not 4.
I have tested this with timer 3 where the supplied values to the function are 1 for PWM and 0xC0 for frequency giving 125KHz with an on time of about 200nS for an expected on time of 30nS.
The library function puts 0 into OCAR and 0xBF into OCBR.
Correcting the code as outlined puts 0xFFFC into OCAR and 0xbf into OCBR.
This gives 125KHz with 30nS on time as expected.
Charles Beal
