PIC processor pin contention

If the control bits are reversed in a single instruction, I think the outputs switch with only propagation differences, from the same clock edge (probably on the rising edge of Q4, the last edge of the 4 cycle instruction clock). At worst, I suspect there will be a few nanoseconds of higher than normal load across the supply terminals. Good bypass practice should handle it.

You could set up an experiment and make some measurements, with the outputs connected through a series pair of resistors. Watch the mid point with a scope. Set one output state to 1 and the other to 0. If there is any overlap, there will be a period of time when the divider will produce a voltage half way between the rails, instead of just swinging cleanly between logic states.

Then switch the 2 output states and check again.

Surely the simple solution is to preceed the instruction that changes the pin direction by one that programs both as an input. That way no contention will occur because, for the duration of a couple of instructions, both pins will be floating (i.e. input). A pull-up or pull-down resistor could be used to ensure a definite logic level on the pins if this is a concern.

TRISB=0b11; // both input ** TRISB=0b10; // Bit 0 output : : TRISB=0b11; // both input ** TRISB=0b01; // Bit 1 output

Hi,

I do a lot of work with PIC's, inculing some unconventional designs. I would like to know why you need to short two pins together. Something I could use?

Luhan

Exactly the arrangement I would use.

Personally, I'd not connect two bidirectional PIC pins together regardless of the firmware. Treat your PIC to a $0.001 resistor and be sure in all possible states (including the billions of b*gg*r*d-up ones).

Best regards, Spehro Pefhany

There are output buffer chips like MAX4427 which are rated for this transition and even specify limits on it.

John,

I connected a couple 1.5K resistors between the pins and the signal looks clean on a low bandwidth scope at the resistors junction.

But, just to make sure there are no problems, I think I'll use the sugestion of setting both pins to input, and then changing one to an output with the next instruction. One more line will only waste one more microsecond.

Thanks,

-Bill

Yes, good idea.

-Bill

I'm using one I/O line to drive a common line to several toggle switches. The other side of the switches are connected to other I/O lines with pullup resistors. This way I can read the state of several switches without any dedicated I/O lines. The I/O pins can be used to monitor the switches and also operate as I/O, but not at the same time.

-Bill

I'll file that factoid away for future reference.

Thanks for reporting the trial.

Bill snipped-for-privacy@att.net wrote: [Luhan wrote]

I think you didn't say which PIC you're using, but it seems likely to me that you can get rid of the pullup resistors. For example, the 16F873A has built-in weak pullups on ports B and C lines (but not on ports A, D, E). With software control of whether weak pullups are on or off, you might be able to share a few more lines between switch reading and other I/O, as well as lowering the time and cost of board building.

Also, I agree with Spehro Pefhany's earlier remark about putting some resistors in series with your output lines, say around 1K. (On an AVR board where I use some lines for both i/o and switch reading, by looking at min/max specs for weak pullup resistances, high and low thresholds of LCD i/o and mcu i/o, etc, I calculated I could have up to about 3K in series with those lines. I used

-jiw

I believe that only port B has programmable pull-up resistors on the

Then what you need to do is drive the common line low when reading the switches, and float (input) the common line when using the other lines as outputs. There will then be no contention for outputs tied together at any time.

Luhan

Andrew wrote: [James Waldby wrote]

You are right - I thought the port C structures on p. 46 of DS39582A (16F873A.pdf) were like the port B ones on p. 44, but now see that the P device is an output driver in one case and a pullup in the other.

-jiw