Revision 23:881d93b8749a, committed 2016-09-08

Comitter:

DavidEGrayson

Date:

Thu Sep 08 23:53:42 2016 +0000

Parent:

22:5368af3ff07d

Child:

24:5c01a6fa1556

Commit message:

Added a comment.;

Changed in this revision

--- a/PololuLedStrip.cpp	Thu Sep 08 23:14:02 2016 +0000
+++ b/PololuLedStrip.cpp	Thu Sep 08 23:53:42 2016 +0000
@@ -20,11 +20,15 @@
         // Try to generally compute what the delays should be for a wide range of clock frequencies.
         
         // The fudge factors below were experimentally chosen so that we would have
-        // ~100/840 ns pulses and a ~1430 ns period on the mbed NXP LPC1768 (96 MHz Cortex-M3).
+        // ~100 ns and ~840 ns pulses and a ~1430 ns period on the mbed NXP LPC1768 (96 MHz Cortex-M3).
         // There seem to be some ~100 ns inconsistencies in the timing depending on which example program is
         // running; the most likely explanation is some kind of flash caching that affects the timing.
         // If you ever change these numbers, it is important to check the the subtractions below
         // will not overflow in the worst case (smallest possible f_mhz).
+        //
+        // On an STM32F303K8 (72 MHz Cortex-M4), these delays give us ~170 ns and ~840 ns pulses
+        // and a ~1595 ns period, and there were no timing differences between the two
+        // example programs.
         led_strip_write_delays[0] = 750*f_mhz/1000 - 33;
         led_strip_write_delays[1] = 550*f_mhz/1000 - 20;    
     }