Writing to two different I/O at the same time

towpew · Joined: 25 Sep 2015 Posts: 24 Location: sweden

Hello
I have a C code question.
I have seen in the manual that it is possible to write to different bits at the same time using a struct like this.

Ttelmah · Joined: 11 Mar 2010 Posts: 19513

Nothing really to do with CCS. The _processor_ cannot write to two memory addresses at the same time.

towpew · Joined: 25 Sep 2015 Posts: 24 Location: sweden

Yea Ok i understand that.

But I was curious how the actual processor uses LCD segments on different ports, but it could may be done in the hardware.

Ttelmah · Joined: 11 Mar 2010 Posts: 19513

Totally hardware.

When you use a peripheral, it disables the normal port I/O functions. Things like the LCD, are done from larger latches, connected directly to the pins (except for the switches to disconnect them when normal port I/O is involved). Effectively the pin is no longer portx.y, but just a pin connected to the LCD controller.

PCM programmer · Joined: 06 Sep 2003 Posts: 21708

asmallri · Joined: 12 Aug 2004 Posts: 1634 Location: Perth, Australia

temtronic · Posted: Sat Oct 10, 2015 5:32 am

There is a 'thread' somewhere here about writing to two ports so try using the 'search' feature. Also the 'flex_LCD' driver does this as well, it's located in the 'code library'.
Be aware that NO method will allow the PIC to write 'simultaneously' to 2 or more I/O ports. Can't be done due to the PIC having 8 bit 'wide' ports.Now depending on processor speed and what the data controls, your program controls it probably will be fine.
In the PC World it's called 'multitasking' where doing 2 or more things very fast makes it 'appear' that both actions occour at the same time( AKA simultaneously).

Jay

PCM programmer · Joined: 06 Sep 2003 Posts: 21708

This is the thread.
http://www.ccsinfo.com/forum/viewtopic.php?t=18949
That's a very old thread. Nowadays, you would use getenv() to get
the port register addresses, instead of entering them as numbers.
Example:

Ttelmah · Joined: 11 Mar 2010 Posts: 19513

One thing that has been ignored through this thread, is the simple question of 'why'?.

Now the systems being pointed to, allow you to 'hide' the separate nature of the pins, to some extent. Great, if this is just for simplicity. However if this is for a 'real' electronic reason (needing signals to switch together), then a deep breath has to be taken, and a big step back made. In such cases, hiding the separate nature can be very dangerous. If (for instance), these are pins controlling devices being switched on/off, then the sequencing may well need to be different, when going from one state to another. So if (perhaps) one low of pins turns on a particular driver, and others turn on another driver the other way, the sequences have to become 'turn off the drivers that are on, and only then turn on the other drivers'. If the PIN identities are hidden in a wrapper, the sequencing becomes that of the wrapper function, implying that the order may well be wrong in one direction, if not the other...
In such systems it becomes vital to control the timings yourself, and work out what order things have to happen to be safe, and ensure the code handles the changes in this sequence in both directions. Get the sequencing wrong when dealing with anything involving power, and nasty things can start to happen.

So big question. Why?.