software serial port baudrate vs oscillator frequency

[Leef_me]

I am using a 16F630 with internal RC oscillator.

I am trying to use the PIC with a software uart by the USE RS232
statement. I tried this, and it appears to have the right bit timing, but the resulting data coming from the PIC into my PC seems to garbled.

Does anyone have information on the usable bit-banged baudrate that is possible?

The serial port of my PC is a TTL level to USB serial port emulator.

Thanks for any info.

[Ttelmah]

Do you know whether this USB emulator, gives positive or negative serial logic?. If you don't know for sure, post a part number.
What version is your compiler?. There have been timing 'issues', with some compiler versions, and software serial.
Now there are three seperate 'issues'. The first is the basic clock accuracy of the chip itself, the second is how accurately, the software serial can be, and the final one, is the serial timing accuracy of the USB adapter. The first is given in the data sheet (look very carefully, at the supply voltage limits). The second is dependant on the selected baud rate, and how closely this can be generated in instruction times. If (for instance) you choose 115200bps, with a 4MHz clock, then the possible emulation 'rates', are with 9 instructions, 111111.11bps, and with 8 instructions, 125000bps. The former is in error by 3.7%, and the latter by 8.5%. The maximum error allowed will depend slightly on the word length etc., but a total error of 4%, is a good guideline. Hence you can see that even if everything else is 'exact', this sort of rate, is borderline. You should also be aware, that the serial port on a PC, is not guaranteed to give accurate timings. Often 1% or more error, is common, and in some cases much worse (I have seen a number of 'brand name' laptops, return timings in error by over 3%..). How accurate the USB adapter is, will depend on the unit, but quite a few USB peripherals use ceramic resonators, that go significantly out of spec, if in cool enviroments...
Now the internal oscillator, if your supply range is 'in spec', is warranted to 1%. I you tried 115200bps, and had 3.7% error in the timings, then the oscillator happened to be out, in the same direction, and further the USB adapter itself had another 1% timing error,you could have a total error of just under 6%. Unlikely to work. Conversely, at 38400 bps, the bit timing error, would drop to just 0.1%, and everything ought to work.
As a general comment, I'd suggest that any rate, which has a bit time below perhaps 25 instruction cycles, is likely to give trouble. So for the 4MHz clock, perhaps a limit at around 40000bps.

Best Wishes

[Leef_me]

Hi Ttelmah, thanks for the reply.

Pololu USB-to-Serial Adapter
http://www.pololu.com/products/pololu/0391/

Seems to be normal positive logic.

Compiler is PCWH version 3.180

I wanted just a standard low rate debugging port at 9600 or less

My supply voltage is probably 4.7V, 6 inch leads from a 7805 regulator.
Perhaps that seems a bit low, but the specs seem to allow lower than 3.5V
for 4mhz at 0*C to 85*C

I'll be checking the other details you suggested, possible tomorrow.

[ckielstra]

From the Pololu website

[Ttelmah]

[Leef_me]

Hi Gang!

I finally got back to checking out the problem.
What I failed to state in various replys was that I am using Microchip's ICD2 with the AC162052 ICD header to debug pic16F630 code.



When I write a simple loop
[code]

#use delay(clock=4000000)
#use rs232(baud=9600,xmit=pin_C3,rcv=pin_A2,stream=COM)


do {
delay_ms(100);
fputc('U',COM); // alternate 1 and 0 for debugging

} while (TRUE);

[/code]

and program it in a pic16F630 it works just fine.
When I use the ICD2 and the header, the output is gibberish.

I scoped the output for the pic16F630 and get about 100us per bit time.

I scoped the output for the ICD2 setup and get about 80us per bit time.

If I change the ICD2 code to #use delay(clock=4500000) then the bit times are closer to 100us and the Pololu USB-serial adapter can receive the data at 9600 bps. YMMV.

Thanks for your assistance.

Leef_me

[ckielstra]

What are your #fuses settings and compiler version?

[imad84]

Hello, u must don t use the RC oscillator

it is not precise, the communication with the RS232 need a BER less than 3%, try to use a quartz which is more precise

Rgardless