Hello,

I have master and one slave hanging on the bus,

The point of the problem is I receive in most cases distorted characters.

I know exactly that answer must comprise 4 characters. It is so, however they are all distorted.

Moreover, sometimes I receive those 4 characters in correct form.

Allegedly it happens after long shutdown. I didn't still figure out a reason of this phenomenon.

Msorich

Assuming correct bitrate, parity, start & stopbits first thing I would check is termination of the bus. A lot has been written but in short make sure both ends of the bus are terminated with correct termination. If still a problem you could check the biasing (see doc).

Another issue could be timing. rs485 should switch fast (enough) beteween send & recv. Depending on the protocol used you can lower bitrate and/or inter frame silence, pause time (or whatever it is called)

@enof,

Sorry, I didn't write, all options I've done:

• cable: twisted pair copper multistrand wire, shilded;
• half-duplex, ground wire -ok;
• ridiculous cable length, less than 1 meter; Anyway I've tried termination and biasing;
• identical comm settings on both participants (Baud rate, parity and so on);

Update...

The timing might be a good idea...

I used so called toggle command for SI module. That command allows toggling such commands like: transmit and receive as quckly as possible. It gave nothing. The problem still persists.

Msorich

Kaulquappe

Hello Msorich,
which error number is reported?
How do you recognize the end of the frame (fixed length or character delay time exeeded or defined character)?
I would call the receive block unconditionally.

Regards,
Kaulquappe

That's the point there's no errors at all. Just I receive a proper number of characters. True, they are all distorted. Sometimes (I still didn't catch conditions of that phenomenone) I receive a proper number of characters and they all in correct form, so to speak readable by ASC|| table.

Kaulquappe

How do you recognize the end of the frame (fixed length or character delay time exeeded or defined character)?
I would call the receive block unconditionally.

Regards,
Kaulquappe

Do you mean a character at the end of commands?

If I got you correctly. It's  so-called Carriage Return or its  hex variant 0D.

Answers always go with CRLF or their hex variant 0D0A at the end. Of course, in case when there's no distortion there.

Msorich

Update: a pic is added.

Kaulquappe

due to the picture you attached, you did not parameterize CR LF as end of the frame. You obviously use "character delay time expired". 

 No, I did. There's an alternative for CR chatacter, it's ';'. in hex 3B. At the moment I'm using the latter one. I tried both, it doesn't matter which one is used.

Kaulquappe

 And it looks like you work without paritiy. Am I right?

As the error is kind of sporadic, I still think of cable / termination / electrical noise.

Regards,
Kaulquappe

 Both, the master  and its slave are set to: 9600 Baud, 8 bits, even parity, 1 stop bit.

Kaulquappe

As the error is kind of sporadic, I still think of cable / termination / electrical noise.

Regards,
Kaulquappe

• I excluded cable version already;
• termination & biasing... maybe, but having so ridiculous cable length;
• electrical noise, maybe...

One more reason to consider purchasing of digital oscilloscope with record option to see difference.

Msorich

There's description of so called Error pattern in manual.

• The error pattern contains the character that is written into the FIFO, if a character was received with an error. Reasons are e.g. parity errors,
  exceeded value range, superimposed noise.

So I have C6 24 24 C1

Here (in table) I don't quite understand for what stands lower line for. See attachment.

Msorich