Unstable serial comm when using internal osc?

dave4 · Guest

I'm using an internal osc and the serial comm is unstable. Sometime the data is good and sometime it is garbled. Can anyone relate this or I am doing something wrong?

Thanks

RossJ · Joined: 25 Aug 2004 Posts: 66

Which PIC are you using?
Are you working at room temperature (i.e. 25C)?
Are you autonegotiating the baud rate or is this fixed (at both ends)?

Most PIC18s have a factory calibrated INTOSC which is typically accurate to +/-1% at 25C. The INRC oscillator is very inaccurate and can be +/-10% or more and is NOT appropriate for baud rate generation.

USART comms should be reliable with baud rate errors of a couple of percent. Serial comms using the USART is timing critical since there is no separate clock (as is the case with say I2C). So clock error/drift could well be responsible for data errors.

If you suspect clock/baud rate accuracy problems, test the clock's accuracy. Either use an oscilloscop or some other suitable test instrument, or write a routine to output a fixed length pulse on an IO (say 60 seconds) and time it with a stop watch.

SET · Joined: 15 Nov 2005 Posts: 161 Location: Glasgow, UK

The INTOSC is pretty stable and good enough for UART clocking - we had one project where we were running at a non-standard Baud rate (was around 230kBaud), and the fixed error between both sides was around 3% - and with this we never lost data. So with standard Baud rates if you get within 2% it should be fine.

Ken Johnson · Joined: 23 Mar 2006 Posts: 197 Location: Lewisburg, WV

Using the internal osc on a 485 network caused me problems. The baud rate was ok, but, using a master-query where each slave would respond in a time slot, based on an id, sometimes failed. My time slots were too close together, and by the time id #25 reported, it might crash over top of #24 or #26. Doesn't sound like this is your problem, but thought I'd throw it in just in case Smile

Ken

Douglas Kennedy · Joined: 07 Sep 2003 Posts: 755 Location: Florida

Its a long shot and it depends on the chip you are using ...but could it be you overwrote the internal rom location that stores the calibration constant
Ex for 12F629
#rom 0x3FF = {0x3490}
//// sets the internal oscillator to mid freq (0x3480) (0x34B0) near 4mhz
//// 0x3FF is top of rom and must contain a retlw instruction
//// with the value for the internal osc calibration
/// retlw is 0x34kk so kk Ex 80 is the value placed in OSCAL
/// osccal is xxxx xx.. max 1111 11.. mid 1000 00.. min 0000 00..
/// 0xFC 0x80 0x00
/// if the location 0x3FF is erased in error the osc won't run
/// freq affects baud rate one 629 needs 3470 the other 34B0

dave4 · Guest

i'm using PIC12F683

dave4 · Guest

Here is my code

Guest · Posted: Tue Aug 28, 2007 3:41 pm

Guest · Posted: Tue Aug 28, 2007 5:34 pm

SherpaDoug · Joined: 07 Sep 2003 Posts: 1640 Location: Cape Cod Mass USA

If you look at the 12F683 datasheet figures 16-38,16-39,16-40 you will see that the HFINTOSC can easily drift over 3% with changes of voltage and temperature. I have seen desktop PCs that were off by almost 2% sitting on my desk. Using a software UART sometimes if 9600 baud doesn't work I try 9408 baud or 9792 baud (9600 +/- 2%) to see if it works better.
_________________
The search for better is endless. Instead simply find very good and get the job done.