16f628 A4 output

[Don P]

I'm using A4 as an output (with a pullup resistor), but something continuously keeps it low. I CAN make it high by doing an output_high followed by a delay, just as a test. But without the delay, something immediately yanks it low again.
I'm suspicious of A4 also being a comparator output, but doing setup_comparator(nc_nc_nc_nc) doesn't help.
Maybe it's the TOCKI reference, but I don't find anything helpful on that.
Any Ideas?
Regards,
Don

[PCM programmer]

What is your version of the compiler ?
The version number is given at the top of the .LST file, which
can be found in your project folder. It will be a number like
3.191, or 3.225, or 3.228, etc.

[Don P]

3.152

[PCM programmer]

setup_comparator(NC_NC_NC_NC); works correctly with your version
of the compiler. So, post a small test program that demonstrates
the problem. Show all #fuses, #use, and #include statements.

[Don P]

#include <16F628.h>
#fuses nowdt,nolvp,intrc_io,noput,nomclr,protect
// internal clock
#use delay( clock=4000000 )
#nolist
#include <string.h>
#list

/*
A4 output to channel 3 of 2003 driver chip [open collector!] relay K3
*/


/*
port usage:
A4 [open collector] relay K3 110v chime (needs pullup resistor)
*/

#rom 0x2100={1,4,2,3, // access code
0, // Access=0 Exit=1 Armed=2 Entry=3 Alarm=4 mode
10, // Exit Chime Time (1/2 secs)for exit delay
10, // Entry Chime Time (1/2 secs)for entry delay
10 } // Horn Time (1/2 secs) for intrusion alarm

#define Dial_1 pin_A1 // K6 intrusion caller

#define Intrusion pin_B0 // input

#define Access 0

int Mode = 0 ;

int Bits2 = 0x02 ;
#bit NewMode = Bits2.1 // initially = 1

void main() // M A I N
{
set_tris_a(0x20); // 0=output
set_tris_b(0xFF) ;
port_b_pullups(true) ;
setup_comparator(nc_nc_nc_nc) ; // SEEMS TO MAKE NO DIFFERENCE
Mode = read_eeprom( 0x2104 ) ; // Access=0

for(;;) // forever
{

// THIS IS STRANGE! IF THE FOLLOWING 3 LINES ARE COMMENTED-OUT, THE PROBLEM
// GOES AWAY (i.e. A4 WILL STAY ON WHEN COMMANDED.
output_bit(pin_A2,bit_test(mode,0) );
output_bit(pin_A3,bit_test(mode,1) );
output_bit(pin_A7,input(intrusion) );

switch( Mode )
{
case Access:
if( NewMode )
{


output_high(pin_A4) ; // THIS IS THE OUTPUT THAT WON'T STAY ON

delay_ms(500) ; // AND THIS DELAY LETS ME SEE THAT AT LEAST IT IS TRYING

NewMode=false ;
}
break ;
}
} // main

[Humberto]

I can�t test the code you posted because it is uncomplete.
Try a simple test like this, if it works fine, re-check your code.

[PCM programmer]

[Don P]

Sorry, been a few days getting back to this project. Humberto's suggestion doesn't illustrate the problem because I can get A4 to respond momentarily, and I have put delays after the commands to see that it actually did happen.
I also tried setting CMCON=7, but it made no difference.
Strangely, I use A0 & A4 in similar fashion. That is, they both flash HI & LO in a timer1 interrupt routine. And A0 works fine. I then tried changing the #defines for the A0 & A4 associations. The problem stayed with the A4 pin. It didn't change with the logical associations. This would indicate that it's not in the program, but somewhere in the chip's behavior with respect to A4.
If it was an RMW issue, it should move to the other pin.
So I'm not at a solution yet.
Regards,
Don

[Don P]

Oh yeah -- pullups & diodes . . .
It's a 22K pull-up on A4, which feeds an input to a 7-channel 2003 darlinton chip, that will (eventually) run the relay. The 2003 already incorporates suppression diodes. For now the 2003 output just drives an LED thru a resistor. Same for all the other "A" outputs.
Eventually they will drive relays that control chimes, horns and telephone dialers, but, for now, it's easier to watch the LEDs.
Regards,
Don

[PCM programmer]

Just do the mod with the shadow register. If it fixes the problem,
it was a RMW issue. If not, it's something else.

[Don P]

Yup, I did the shadow_Port_A thing from your example -- same.

[PCM programmer]

What's the size of the pullup resistor on pin A4 ?

Do you have a back-emf diode on the relay coil ?

[Don P]

see a couple msgs back. ( This forum doesn't seem to like cut & paste from other posts). It's 22K driving 2003, which has diode built in. And relay isn't hooked up yet; just an LED. Simple program to turn A4 on & off works the pin OK.

[PCM programmer]

To me, it still smells like RMW.

I decided to research this a little more.

Here's a thread from Piclist where someone's doing exactly what
you're doing (or what you'll eventually be doing).
http://groups-beta.google.com/group/comp.arch.embedded.piclist/browse_thread/thread/ceae357b5f2871be/07a188d57fb22a37?
They explain that the voltage on the pin will never go above 0.7v,
which is the drop across the transistor's B-E junction. They
suggest using a series resistor in the base circuit. In other words,
first comes the pic pin, then the pull-up, then the series resistor,
and then the base of the transistor.

That would be a hardware fix.
I still say that it can be fixed in software by use of a shadow
register. You just have to make sure that all writes to Port A
are done by doing byte-wide write of the shadow register.
No other writes are allowed.

I may do a test program tomorrow on a demo board to prove this.

[Ttelmah]