Skip to content

Latest commit

 

History

History
87 lines (80 loc) · 4.25 KB

readme.md

File metadata and controls

87 lines (80 loc) · 4.25 KB

RTK65RX65Ndemo04: functionality: sw-toggle: LED0 250.000 us on-off, LED1 250 ms off-on

  • HOCO clock 16 MHz, PLL / 1 * 15 = 240 MHz
  • ICLK / 2 = 120 MHz, PCLKB / 4 = 60 MHz
  • TMR0 source PCLKB, CMIA0: 60 no interrupt => 1 MHz
  • TMR1 source CMIA0, CMIA1: 250 + interrupt, vector INTB149 => 2kHz
  • INTB149 acc_4kHz count + increment Micros, div 4 = Millis
  • main(): time polling with micros() and millis()

skipping smart configurator:

  • in stead of selecting smart configurator, choose Finish >

make the HardwareDebug settings:

  • Project > Properties >
  • Properties for RTK5RX65Ndemo03
    • Run/Debug Settings > RTK5RX65Ndemo03 HardwareDebug > Edit... >
  • Edit launch configuration properties
    • Debugger > Connection Settings >
      • Clock: HOCO
      • Connection with Target Board: change dropdown selection JTag to Fine
      • Power: change dropdown selection Yes to No
    • OK > Apply and Close >
  • Control-B to build the project, or: Project > Build Project >

taking over code from RTK65RX65Ndemo03:

  • copy main.c to this project folder src
  • edit some things to arrive at what is now inside the src folder
  • change inthandler.c by commenting out vector 149 to //void INT_Excep_PERIB_INTB149(void){ }.

Debug/Run:

  • ^B (build project), F11 (Debug), Confirm Perspective Switch Yes > F8 (Resume), F8
  • ^F2 will terminate the debug session

result:

  • 8 files in "generate" folder, of which 1 line was commented out,
  • 1 file in "src" folder has all code that matters for the developer
  • prog: 3004 or 3008, depending on TEST value and code in TEST 0.
  • constant: 1176
  • data: 0
  • bss: 20
  • other: 28

note: This project serves to gain some feeling about which integer sizes to use for RAM-restricted programs (I know, this one has 640 kB of it), and why you would sometimes use unsigned and signed integers of differing sizes. With a #define statements to be set for variations in functionality, you can debug and try to explain in which way and why the behaviour differs with the chosen setting.

questions:

  1. Why does TEST with calue 0 not work, and why does it work with the other values 1, 2, and 3?
    • Hints:
      1. Why is an unsigned 8-bit difference greater than 6 used for toggling the LED output and incrementing the 8-bit time value?
        • NB: a value of 256 will be understood as 0 when stored in 8 bits.
      2. Why does a signed 16-bit difference suffice to toggle the LED output and increment the 16-bit time value? The millis()-value is unsigned 32-bit, so what will happen after half a minute, or a minute, when the 16-bit time value can't store the millis()-value anymore?
        • NB: type promotion will extend the stored 16-bit time value to 32-bit before subtracting it from the 32-bit millis()-value, and then a type-cast (or bit cutt-off) to 16-bit will be done before regarding the resulting value as a signed value.
      3. Why is an unsigned 16-bit difference greater than 65286 used for toggling the LED output and incrementing the 16-bit time value?
        • NB: a value of 65536 will be understood as 0 when stored in 16 bits.
    • The above hints can be understood better when thinking of the numbers within a given size written on a disc that turns around 1 number every milli-second. Compare it to a decimal number with 1 or 2 digits, counting up and arriving back at zero after 1 (or 2 in the latter case) 9's have displayed. For a signed number at the other side of the disc the numbers must be interpreted as negative, so 5 in the decimal case would be read as -5, 6 as -4 and so on.
  2. Why does TEST with value 0 not work?
    • Try to do the calculation millis() - t_LED1 by hand, starting with t_LED1=0 and Millis=0. The first test will increase t_LED1 by 250, what is happening then when Millis is still 0, or when it is 1? Go on until Millis=7, and try to discover what the ucon sees as a result of the calculation after casting it to int8_t.
  3. Why does TEST with value 0 work when the test is expanded to if(0>dt1||6<=dt1)?
    • Hints: see the test with TEST-value 1 as an example, read the text above about signed numbers and determine when the difference will be seen as negative or non-negative.