Serial Communication

freeindy · 04-04-2007, 07:03 AM

Hi,

I have a bit of a problem reading a device from the usb port that sends data every second. I do receive data but i'm confused with one little thing:

The data length is usually 25 bytes (the nr between ::XX:

but SOMETIMES it gets broken into parts like this:

Code:

::25::7E 42 08 00 00 00 00 00 7D 5E 00 01 7D 5D 01 00 92 0B 00 00 00 00 47 7C 7E
::25::7E 42 08 00 00 00 00 00 7D 5E 00 01 7D 5D 01 00 92 0B 00 00 00 00 47 7C 7E
::25::7E 42 08 00 00 00 00 00 7D 5E 00 01 7D 5D 01 00 8F 0B 00 00 00 00 80 28 7E
::25::7E 42 08 00 00 00 00 00 7D 5E 00 01 7D 5D 01 00 92 0B 00 00 00 00 47 7C 7E
::25::7E 42 08 00 00 00 00 00 7D 5E 00 01 7D 5D 01 00 92 0B 00 00 00 00 47 7C 7E
::25::7E 42 08 00 00 00 00 00 7D 5E 00 01 7D 5D 01 00 93 0B 00 00 00 00 E7 39 7E
::4::7E 42 08 00
::21::00 00 00 00 7D 5E 00 01 7D 5D 01 00 92 0B 00 00 00 00 47 7C 7E
::25::7E 42 08 00 00 00 00 00 7D 5E 00 01 7D 5D 01 00 92 0B 00 00 00 00 47 7C 7E
::7::7E 42 08 00 00 00 00
::18::00 7D 5E 00 01 7D 5D 01 00 92 0B 00 00 00 00 47 7C 7E
::25::7E 42 08 00 00 00 00 00 7D 5E 00 01 7D 5D 01 00 92 0B 00 00 00 00 47 7C 7E
::25::7E 42 08 00 00 00 00 00 7D 5E 00 01 7D 5D 01 00 8F 0B 00 00 00 00 80 28 7E
::25::7E 42 08 00 00 00 00 00 7D 5E 00 01 7D 5D 01 00 92 0B 00 00 00 00 47 7C 7E

I'm not very good in serial communication. I've been reading and trying some example codes resulting to this. The settings are as following

Code:

//Get the current options for the port...
	tcgetattr(fd, &options);

	//Set the baud rates for input and output (same))
	cfsetispeed(&options, type);
	cfsetospeed(&options, type);

	//Reset control flags
	;

	//Enable the receiver and set local mode.
	options.c_cflag |= (CLOCAL | CREAD);	

	//8 bit char mode, No parity, 1 stop bit
	options.c_cflag |= (options.c_cflag & ~CSIZE) | CS8;
	options.c_cflag &= ~PARENB;
	options.c_cflag &= ~CSTOPB;
	
	//Minimum caracters to read
	options.c_cc[VMIN]  = MIN_CHAR_READ; // MIN_CHAR_READ = 0

	//Time to wait for data
	options.c_cc[VTIME] = WAIT_DATA;     //WAIT_DATA = 20 (2 s)

	//Set hardware handshake
	options.c_cflag &= ~CRTSCTS;

	//No software flow control
	options.c_iflag &= ~(IXON | IXOFF | IXANY);

	//Input character management (raw or canonical)
	options.c_lflag = IRAW_MODE;
	//options.c_lflag = ICANON_MODE;

	//Output filtering options	(raw or processed)
	options.c_oflag = ORAW_MODE;
	//options.c_oflag = OPROCESS_MODE;
	//Set the new options for the port...
	tcsetattr(fd, TCSANOW, &options);

and then I just read the buffer like this:

Code:

//Empty the recieved buffer
 	tcflush(fd, TCIFLUSH);
	
	nBytes = read(fd, &data[0], MAX_DATA_SIZE) //(MAX_DATA_SIZE = 255)

Is it because the serial buffer is full for some reason?? If so, is there anyway I can determine the the tranmitted data is only a part of it and not complete so that I can read data again till it's complete?

Thanks a million

Indy

wjevans_7d1@yahoo.co · 04-04-2007, 08:47 AM

Serial communications do not have inherent record boundaries. What constitutes a record is purely artificial.

The read() call gathers whatever bytes have been received since the prior read(). If your device sends data in spurts of 25 bytes, your read() call will usually read 25 bytes. But sometimes you can catch your device in mid-spurt, in which case you'll get whatever's available at the time.

Since you know that your data is best examined 25 bytes at a time, that's how you should look at it; just loop until you get all 25 bytes. But it can't hurt to put in a sanity check, in case a byte gets dropped or something. That may be very unlikely, but sanity checks are cheap, and if you need a sanity check and don't have one, it could take a long time to isolate the problem.

Are the first and final bytes of your 25-byte records guaranteed to be 0x7E? If so, that would be a good sanity check to use. If this device is provided by a third party, it would seem they included the 0x7E at the beginning and end of each record for precisely this purpose: so you could write a sanity check.

Hope this helps.

freeindy · 04-04-2007, 09:24 AM

Well. No sanity check yet. I was afraid of that. there is no guarantee for the byte 7E on both side. And data can vary. I guess I have to put something in the data. I wanted to avoid it but I guess I have to live with it.

Thanks anyway

Indy