Em 17/1/2012 15:17, Oli Glaser escreveu:
> On 13/01/2012 17:29, Isaac Marino Bavaresco wrote:
>> Em 13/1/2012 15:08, Andre Abelian escreveu:
>>> Hi  all,
>>>
>>> I am testing pic32 and something simple I am having hard time to unders=
tand.
>>> I have 2 leds connected to G12 and G14 when I use test1 code LED1 never=
 tuns on.
>>> test2 and test3 is ok. my question is why test1 is not working. G12 and=
 G13 are digital I/O.
>>> if i use debugger step over it works.
>>>
>>> test1:
>>> LED1 =3D 1;
>>> LED2  =3D 0;
>>> delay;
>>> LED1 =3D0;
>>> LED2=3D1;
>>> delay
>>>
>>>
>>>
>>> test2:
>>> if I use&  and | logic it works
>>>
>>> LED1 | =3D 1;
>>> LED2&  =3D 0;
>>> delay;
>>> LED1&=3D0;
>>> LED2 |=3D1;
>>> delay
>>>
>>> test3:
>>> if I use LATGSET and LATGCLR it works
>>>
>>> Since G12 and G14 are not analog pins and I do not have external capaci=
tor I do not think this is Read modify write issue.
>>> my question is do you know why test1  LED1 stays off when running.
>>>
>>> here is the code.
>>>
>>> while(1)
>>> {
>>>    int short_delay;
>>>    uint8_t returnStatus1,returnStatus2;     // to test
>>>
>>>   //     led_d7_green&=3D 0;
>>>   //     led_d8_green |=3D 1;
>>>
>>>   //     led_d7_green =3D 0;
>>>   //     led_d8_green =3D 1;
>>>
>>>        LATGCLR =3D 0x5000;   // 0b0101000000000000;
>>>        LATGSET =3D 0x1000;   // 0b0001000000000000;
>>>
>>>        short_delay =3D 1000*1000;     // create short delay
>>>        while(short_delay--){};
>>>
>>>   //     led_d7_green |=3D 1;
>>>   //     led_d8_green&=3D 0;
>>>
>>>   //     led_d7_green =3D 1;
>>>   //     led_d8_green =3D 0;
>>>
>>>        LATGCLR =3D 0x5000;   // 0b0101000000000000;
>>>        LATGSET =3D 0x4000;   // 0b0100000000000000;
>>>
>>>        short_delay =3D 1000*1000;     // create short delay
>>>    while(short_delay--){};
>>>
>>>      }
>>>
>>> thanks
>>>
>>> Andre
>> Probably related to RMW (read-modify-write). When stepping-over or
>> running the code that takes more instructions, there is time for the pin
>> to settle before changing the other pin.
>>
>> When running the faster code at full speed the first pin is read with
>> its previous value and written back with the older value.
>> You are writing to register PORTG, right? Try writing to register LATG
>> instead.
>>
>>
>> Best regards,
>>
>> Isaac
>>
> Sounds like RMW to me too, given that it's the first LED (in the code)=20
> that doesn't turn on. For any fast port manipulations use the dedicated=20
> atomic registers (as you have done with LATGSET, etc in the other test)


When not in sleep, there is a synchronization circuit that makes the
signal present at the I/O pins take two SYSCLK to reach the internal bus
( Figure 12-1, page 12-3 of "PIC32 Family Reference Manual, Sect. 12 IO
Ports.pdf" ).

So, the PIC32 devices are subject to RMW issues irrespectively of
external or internal capacitance. Always wait at least 2 SYSCLK between
RMW operations.

Manipulation of the LATx registers are not subject to RMW, even when not
using the LATxSET, LATxCLR, etc.


Isaac

--=20
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist
.