CCS C send string data with using interrupt

cengav4r · Joined: 31 May 2012 Posts: 10

Hi,
I am getting some data over serial and I want to send (random) Gps data when coming data equals 'A'.Unfortunately, I can not send more than 15 character always.If my string size lower as char data="123465789012"; it works properly, does not freeze pic and others.If my string size getting bigger as data="123465789012345";, pic send this data just one time then system is freezing and not respond.

I want to show to you what I have done and what I got;

asmallri · Joined: 12 Aug 2004 Posts: 1635 Location: Perth, Australia

This is a classic example of how NOT to write an interrupt handler.

For this class of application you would typically implement a ring buffer in the receive interrupt handler. The code is short and sweet. Get the received character, put it in the ring buffer, return from interrupt.

Then in the mainline code test if the ring buffer has anything in it, if so, go process it.

There are lots of examples, supplied with the compiler, in the forum, available on the internet.
_________________
Regards, Andrew

http://www.brushelectronics.com/software
Home of Ethernet, SD card and Encrypted Serial Bootloaders for PICs!!

Ttelmah · Joined: 11 Mar 2010 Posts: 19535

You need to understand time....

The PIC has just two characters of hardware buffering on output. If you arrive at the interrupt, and this is empty, two characters can be send almost immediately, but everything beyond this, has to wait for a character to be sent, before proceeding. Now, the receive also has a couple of characters of buffering, and the serial from the GPS, is probably not quite 'non stop', but after a certain number of characters (in your case 15), you will reach the point, where all the buffers are full, and if another character arrives at this time, the UART will then become hung....

So you always need ideally to get out of the ISR, before another character arrives.

There are two ways of doing this:
1) When data is to be sent, you set a flag, and then send this data in the main code.
2) You put the data to send into a buffer, and use an interrupt driven transmit.
3) Just store the character received, and do the entire job in the outside code.

CCS supplies an example of the second, in ex_stisr.c, and the core of the third in ex_sisr.c.

However there is then the question of whether you send anything else in the 'outside' code.
If you do, you can't use (directly) the buffered putc (from the example), inside the interrupt. There are ways of doing this, but it involves very careful control of the enable for the transmit interrupt, and possibly having two copies of the buffer write code, one for inside interrupts, and one for outside. Needs good understanding of what you are doing.
Easier probably to use method 1, or method 3 at this point.

On question 2, if you are using interrupt driven RX, you wouldn't use either approach shown.
Instead you would look for the start of message character, and then write the received data as it arrives, character at a time to a storage buffer, and when you see the end of line character, set a flag to say 'message received'.
Then look at the message in your main code when this flag is set (remember to clear the flag).