RS232 reception on PIC18F (SOLVED)

strombom · Joined: 04 May 2007 Posts: 4

Hi,

I have trouble receiving data using the USART on a PIC18F1320.

Main routine:

PCM programmer · Joined: 06 Sep 2003 Posts: 21708

Try a more simple test program. See if it works. If it doesn't work,
post your compiler version. Also post a description of your hardware.

strombom · Joined: 04 May 2007 Posts: 4

Hi,

Unfortunately that code does not make any difference. Still there are often this type of errors:

The first byte is always correct but some of the last ones usually have errors.

When I change to software uart however (FORCE_SW), it works perfectly.

The hardware consists of a PIC18F1320 running from a 20MHz crystal, I have stripped it down to a minimum, I measure directly on TX/RX pins. The TX/RX go through a MAX232 level converter and then the PC serial port. The 5V power supply is very clean.

PCH version 4.093

/Johan

FvM · Joined: 27 Aug 2008 Posts: 2337 Location: Germany

You don't perform any additional, interrupt base action while echoing the UART data? You can clearly see, that the receiver looses part of the UART bitstream but reproduces it correctly from the erroneously assumed start bit, it's the case in both tests.

If no interrupts are disturbing the reception, it may be the case, that the receiver assumes an overrun error and restarts the reception. You may want to check (if you already didn't) the Rx signal directly at the processor pin, also if it has a correct logic level. The logic analyzer must not necessarily use the same threshold as the PIC.

Ttelmah · Guest

Interrupt actions would not cause a mid character break in the data on the hardware UART.

I must admit, I'd be most suspicious of the second proposed idea, the signal level.

RB4, uses TTL signal levels, _when used as a conventional input_, so would use these levels when you program a software UART,explaining why this works. It uses Schmitt input levels, when you use the hardware UART. try the simple expedient, of adding a 2K2 resistor, from this pin, to +5v. If it starts working, you have found the problem. Then, look at your MAX232, and check the version. The original MAX232, only warranted TTL outputs. The latter MAX232A, has CMOS outputs, which provided they are only driving a single input, should easily meet the Schmitt input thresholds.
If your analyser supports variable input thresholds try at 0.8*Vcc, and see if it now duplicates the problem.
Obviously then investigate the wiring to this pin. One 'problem' with logic analysers, is that if a signal does not cross their threshold, you will see lovely 'square' edges and tops, even with a signal that is perhaps not making good contact....

Best Wishes

FvM · Joined: 27 Aug 2008 Posts: 2337 Location: Germany

strombom · Joined: 04 May 2007 Posts: 4

Ttelmah: You were right about the level problem.

It seems the hardware USART configures the RX pin as Schmitt Trigger and requires >4.0V. The output from the RS232 level converter barely gives out 4.0V for a logic one for some reason. I have no more time to look into this, but using software rs232 solved the problem so I stuck with that.