Clock x ports doubts

[fagundesbr]

Posted on November 19, 2014 at 12:13

Hello

i�m new in STM8s105 microcontroller and cosmic c compiler.

i wrote a small program:

i set the clock with internall clock with prescaler 8:

CLK_CKDIVR= 0b00011000;

it means that my uc will operate at 2 MHz, right?

i set the port D3 as output pin, push pull fast:

PD_DDR=0B00001000;

PD_CR1=0B00001000;

PD_CR2=0B00001000;

PD_ODR=0B00000000;

in the main, i wrote this code:

main()

{

PD_ODR=0B00000000;

PD_ODR=0B00001000;

PD_ODR=0B00000000;

PD_ODR=0B00001000;

PD_ODR=0B00000000;

PD_ODR=0B00001000;

PD_ODR=0B00000000;

PD_ODR=0B00001000;

for (;;) {}

}

in my mind, if i set the port, it will in high level for 500ns, turn off for 500ms and so on... but in the scope i see times of 1us.

what i do wrong?

thanks a o lot!

#stm8s105-ports-clock

[zzdz2]

Posted on November 19, 2014 at 17:27

Hi,

Yes, it looks like the CPU clock is set to 2MHz.

Can you see the resulting assembly code from your compiler, maybe it would explain something here.

I did some similar tests but with different compiler, if I remember correctly it was compiled to a single cycle MOV instructions and it really took single cycle to set the port.

But if your code takes two cycles it's still not that bad.

[fagundesbr]

Posted on November 19, 2014 at 18:19

hello

i checked the assembly code, but it´s like ok:

0x810a <main+2>             0x725F500F      CLR   0x500f              CLR   _PD_ODR 

main.c:43 PD_ODR=0B00001000; 

0x810e <main+6>             0x3508500F      MOV   0x500f,#0x08        MOV   _PD_ODR,#0x08 

main.c:59 PD_ODR=0B00000000; 

0x8112 <main+10>            0x725F500F      CLR   0x500f              CLR   _PD_ODR 

main.c:60 PD_ODR=0B00001000; 

0x8116 <main+14>            0x3508500F      MOV   0x500f,#0x08        MOV   _PD_ODR,#0x08 

main.c:62 PD_ODR=0B00000000; 

0x811a <main+18>            0x725F500F      CLR   0x500f              CLR   _PD_ODR 

main.c:63 PD_ODR=0B00001000; 

0x811e <main+22>            0x3508500F      MOV   0x500f,#0x08        MOV   _PD_ODR,#0x08 

main.c:65 PD_ODR=0B00000000; 

0x8122 <main+26>            0x725F500F      CLR   0x500f 

i see in the assembly paper - PM0044 that MOV and CLR takes one machine cycle.... but in my tests i don´t see this..

[zzdz2]

Posted on November 20, 2014 at 09:56

I don't know, it should be single cycle.

Maybe it's about flash read time.

The code in your assembly listing is not aligned to 32-bit/4-byte address so it would need two reads per instruction.

It's quite possible it would run twice faster when properly aligned.

[fagundesbr]

Posted on November 21, 2014 at 21:12

ok, how i make this alignment??

[Ian Legg]

Posted on November 24, 2014 at 09:36

Hello,

Given your source is a C file and not assembler then I would have expected the compiler to handle the alignment correctly.

It should be possible to place the pin toggling code in a function and then align that function to a 4 byte alignment but somehow that feels a bit of a workaround!?

Thanks,

Ian

[fagundesbr]

Posted on November 26, 2014 at 13:05

Hi

but how i check the correct clock and change the status orf the port´s?

how i set my clock to 2MHz and see a toggle of port of 500ns?

it is very confuse to me.. in the pic of microchip, i can see the correct operation of this.

[Ian Legg]

Posted on November 26, 2014 at 16:59

Hi,

You can check the speed of a running clock on the CLK_CCO pin via the CLK_CCOR register.

You can change the frequency of the running clock via the SYSCLK register.

Does that help?

Thanks,

Ian

[fagundesbr]

Posted on November 26, 2014 at 19:10

humm no....

how i toggle a pin on/off with 500ns, in my example?

if the machine cycle for the MOV command is 1, if i on/off a pin, the time will be 500ns, right?

[Ian Legg]

Posted on November 27, 2014 at 09:50

Ah ok... well the code you showed previously looks to use one cycle to write a 0 to the pin and one cycle to write 1 one to the pin, so in theory with a 2MHz clock you'll see 500ns off and 500ns on. However, you appear to be seeing 1us off and 1us on, which implies a 1MHz clock.

That's why I said verify that you are actually using the HSI clock at the rate you think you are via the CLK_CCO pin.

Did you correct the alignment of the code as mentioned by another forum user as what he said is correct?

Thanks,

Ian

[zzdz2]

Posted on November 27, 2014 at 19:06

I think the most feasible way to do the alignment is to put two NOP asm instructions before the gpio output block. It may work.

I guess I was just lucky when I did the test and I got aligned instructions by chance.