LeafLabs Garden

A place to share, learn, and grow...

1. 

   CarlO
   Member

   Hey LeafLabs folks,

   Thanks for a great product.

   I may have found a bug in the code for the Timers.

   The following code produces a nearly perfect square wave at 999.986KHz, well within the tolerance for the crystal.

   pinMode(9,PWM);
   Timer4.setPrescaleFactor(1);
   Timer4.setOverflow(71);
   pwmWrite(9,36);

   The next piece of code, however, returns an Overflow value of 72 and produces a nearly square wave at 986.098KHz

   pinMode(9,PWM);
   int Overflow = Timer4.setPeriod(1);
   pwmWrite(9,Overflow/2);

   I have similar results using a period of 2 etc. The frequency is off by a proportion equal to the overflow value off by 1. Are you accounting for the timer count starting at zero?

   Thanks for your help,

   Carl

   Posted 5 years ago #
2. 

   CarlO
   Member

   Update:

   I modified my local copy of HardwareTimer.cpp as follows.

   Line 84 was:

   setOverflow((microseconds*72)/ps);

   Line 84 modified to:

   setOverflow(((microseconds*72)/ps)-1);

   This seems to fix the timer accuracy.

   Does anyone see any problems with fixing the code this way?

   Carl

   Posted 5 years ago #
3. 

   poslathian
   LeafLabs

   Interesting find, nice catch! Let me think about it some more. The way you have it seems to make it possible to set overflow to -1...

   I think perhaps the source of your off by 1 error was from doing an integer divide on 71 by 2, which will probably end up as a shift right operation, which is 35 - not 36.

   Posted 5 years ago #
4. 

   bnewbold
   Ex-Leaflabs

   I think you're right CarlO, I just have a sloppy off-by-one error there; I will update the library with bounds checking and test the output for accuracy.

   Posted 5 years ago #
5. 

   bnewbold
   Ex-Leaflabs

   Fixed and commited: http://github.com/leaflabs/libmaple/commit/1c6f9a9b5e9f1896a6c3d1faa895af74b6c7e3e1

   Posted 5 years ago #

RSS feed for this topic

Reply

You must log in to post.