rs232 problem - SOLVED

[gjs_rsdi]

I am sending from the PIC 19 bytes that should be:
55h,AAh,16*00h,FFh
I get:
19*00h
The message is sent once a second. (enable_interrupts(INT_TBE);)
Tested with two different terminals, both same.
The program was working fine and suddenly stopped working. I made this test program to try to find out what is the problem.
Testing in MPLAB simulator, no problem.
I am using basic FTDI USB to TTL converter
CCS PCM C Compiler, Version 3.249, 37061
Help much appreciated.

[gjs_rsdi]

I have rewritten the software, same problem

[Mike Walne]

I have not got a snowball in Hades chance of solving your problem.

If you want help we need short, complete, compilable code to copy. paste and test.

Your code is none of the above.

a) It's too long,
b) We don't have your Scom1.h etc.
c) No main()
.............................

Explains why you have had no response to your original post.

Mike

[gjs_rsdi]

Thanks for the attention Mike.
I shortened the program and put all the headers on the same page as the .c
I am sending now just 5 bytes.
It is compiling, working in the MPLAB simulator, working with the board, but I get: 00,00,00,00,00
instead of 55,AA,00,00,FF

[alan]

You disable INT_TBE and enable INT_RDA, but you have the interrupt pointing to INT_TBE

Regards

[gjs_rsdi]

Hi Alan

I am enabling INT_TBE ones in 2 seconds using timer0 and disable it in case 4 of the switch.
My mistake. I disable the INT_RDA but it is the same.
I am getting every 2 seconds 00,00,00,00,00

Joe

[Ttelmah]

I also have to point out that on the original code, cases 2, to 17 were all the same code. As such separate code is not needed for these. When I originally saw this, I stopped looking....

On the new example, the enable_interrupts(GLOBAL), does not want to be in the loop. setup_oscillator(FALSE)!. What is this going to do?. See below.
The approach of sending a message using INT_TBE, is perfectly sensible and acceptable, but the actual writing of it is really bad. Do you really have an external clock oscillator module?. If not, the code is not going to run properly. Old comment about is you are using the hardware UART, you _must_ have ERRORS in the #use RS232. Your use of the WDT, is pointless/appalling. The key thing with a watchdog, is it only does some good, if you write the code so that the watchdog is only reset, if the chip is running correctly. Scattering 'restart' instructions round the code, means that it'll get restarted, even if the code is doing unexpected things. In which case you might as well not have the watchdog at all. Waste of code space and time.
The old key thing with debugging, is KISS. Step back, and get rid of all the interrupts, and printf a message. Do you get anything?. Anything other than '0'?. You may well have a simple hardware problem that started the code not working.
If you are going to use the watchdog, specify a divisor for it (assuming that things like this will 'default' to what you want is always dangerous). Then write the code so the watchdog _only_ gets reset if the code _is_ running properly. Get rid of all but one restart_wdt, and make sure that the code sequence _will_ reach this if the code is working, and can't if it isn't.

Your setup_oscillator(FALSE); line, may well be the problem. FALSE codes as '0', and setup_oscillator(0) will switch to about 40Khz operation....

If you say you are getting something every 2 seconds, then either his is the result of a watchdog triggering, or your oscillator is not 8MHz. Timer1, with an /8 prescaler will divide the master oscillator by 4*8*65536 = 2097152. With a counter running 0 to 60 (61 counts), you would get the interrupt being enabled every 16 seconds if your master clock was 4Mhz....
Your chip is not running at the speed you think.

[gjs_rsdi]

Ttelmah, thanks for the input.
I changed the program according to your advice (I hope I didn't missed something) except the timer0.
I have an 8Mhz external clock oscillator to free A6 for I/O.
I checked in the simulator, I got timer0 interrupt every 32.768ms, with the counter I am enabling the TBE approximately every 2 seconds.
While compiling, I am getting worning: Variable never used: rs232_errors.
The corected program:

[gjs_rsdi]

Finishes testing with an other board that have a 8Mhz crystal, changed the fuse to HS but I am getting the same result.

Joe

[Mike Walne]

Please confirm the following:-

1) Simple LED flasher works at the correct speed.
2) Simple "Hello World" code also operates OK.

Mike

[gjs_rsdi]

I reprogrammed the software that was working before with this board.
The hardware includes an AC motor that works at 5 different speeds controlled by PWM, keyboard with 4 buttons and 5 LED's that are blinking in certain situations.
All buttons functions are correct.
LED's blinking or not, correct for the specific situations.
Motor turns in the 5 correct speeds.
Serial communication does not work. It was working before with the same software.

Joe

[gjs_rsdi]

I added the LED per Mike advice to flash every second (500ms on/500ms off) and it works correct.
I didn't change the serial communication as I can read just binary on my terminals. I also don't know how to write "Hello World"
Still read 00,00,00,00,00 instead of 55,AA,00,00,FF
every 2 led flahings.
Still get: Warning 202:Variable never used: rs232_errors
Joe

[Ttelmah]

You still have not done the basic 'hello world' test without interrupts. As I said:
"Step back, and get rid of all the interrupts, and printf a message."

You will get the rs232_errors warning. This was a stupidity by CCS, where when you add 'ERRORS' as part of the error handling code they add this variable, and if you don't use it they give this warning. Understand it is a _warning_ not an error. You can either just ignore it (what 90% of people do), or add single line to access the variable (which stops the warning), or even a dummy function to access the variable, that is never called (which stops the warning, and since it is never called, costs nothing in code...), or just tell the compiler to ignore this warning.

The simulator tells you nothing. It'll believe what you tell it about the clock, which can be wrong. You should not get an interrupt at 32768mSec, with an 8MHz clock, and T1_INTERNAL|T1_DIV_BY_8. The T1 prescaler is fed off Fosc/4. T1 counts to 65536. With an /8 prescaler the interrupt should be at 1/4 the rate you are seeing in the simulator.
Look at the timer1 block diagram in the data sheet. Note Fosc/4 when the internal source is selected. If you are seeing an interrupt at the rate you say, then it tells us something is wrong (oscillator is 32Mhz, not 8MHz, etc..).

[gjs_rsdi]

Ttelmah

I am not using timer1. Using timer0.
setup_timer_0(RTCC_INTERNAL|RTCC_DIV_256); with 8Mhz is giving interrupt at 32.768ms.
The LED is on for 500ms and of for 500ms with the last program.
Tested with the watch for 60 seconds.
I don't know how to printf "hello world".
I will try to learn.
Joe

[Mike Walne]

Main() code fragment something like this should send "Hello World" to a PC at ~1 second intervals.