Pins A3 and A4 wont turn on, only toggle

[scaven92]

I have the following circuit, one for each of the following pins: A0, A1, A2, A3, A4, A5, C1 (7 total LEDs).

I also have pin C2 used for to turn on the 5V supply for all of the LED's.

[PCM programmer]

[scaven92]

I tried the delay and it acts the same, the two leds connected to A3 and A4 blink for a very belief moment while all the other leds come on.

I am lost...

[PCM programmer]

To prove if it's a Read-Modify-Write problem, try setting up a shadow
register for Port A. This would be just an 'int8' variable, called
PortA_Shadow or something like that. Then do all your bit operations
on that variable. After doing a bit operation on the shadow register,
then write the entire shadow register to Port A in one operation, by
using the output_a() function.

This means that the only way you will write to Port A will be with the
output_a() function. You will never use output_bit() or output_low()
or output_high() on individual pins. Port A will only be written to
with a full port write operation.

See if that fixes your problem.

[scaven92]

Alright I will try that. If it is a read-modify-write problem, wouldn't a 500ms delay between port reads be enouph to allow that pin to settle to the correct value? The only load on that pin is the gate to the fet and the 10k pulldown, no substantial capacitive loads.

I'll let you know in a few.

Thanks for the help!

[scaven92]

It looks like it is the read-modify-write problem you have previously mentioned. I was able to turn all the LED's on.

I have a question though, even if I put a 500ms delay in between successive bit manipulations on PORTA, I still run into the problem. I don't see how there is such a load on the pin to cause the state to change and the next read of the port to not be what I set it to 500ms prior. I only have a 10k pulldown and the port in feeding directly into the gate of the fet.

For the meantime I will use the shadow variable to keep track of every bit manipulations and write the PORT as a whole.

This is still leaving me confused as to why this is happening.

Thanks,

Chris

[PCM programmer]

To clarify your circuit diagram, which side of the FET is the Source and
which side is the Drain ?

[scaven92]

I have changed my enable_led() function to accomodate this problem and everything is working smoothly.

[scaven92]

In my diagram the source of the fet is grounded and the drain connects to the cathode of the led.

[PCM programmer]

The reason is because there is a bug in your version of the compiler
(vs. 4.093). In the start-up code, it's only clearing bits 0,1,2 of the
ANSELA register. It leaves bits 3, 4 and 5 in their power-up state,
which is all 1's. This means pins A3, A4 and A5 are still set as Analog
pins, which means they read back as 0's. That's what causing the RMW
problem.

To fix this, declare ANSELA above main() with a byte statement.
Then set it to 0 at the start of main() with a line of code. Example:

[scaven92]

Wow, that has to be it. I will correct my code to handle this problem. I appreciate your help, you have saved me alot of time and made me aware of this Read-Modify-Write problem.

Should I send CCS an email with a link to this forum for this problem for future update?

Thanks again!

[PCM programmer]

They fixed it in vs. 4.099, as shown below. Both ANSELA and ANSELB
are set to 0x00.

[scaven92]

Great! I'll have to see if my 30 days of compiler updates is still valid or order an update package.

Thanks!