Control Checksum doesn't seem to be operational

[simonmtl]

Hi, I am sending NMEA strings with uncalculated checksums to OpenCPN and they are accepted regardless of the "Control Checksum" setting in open CPN.

It seems to me the "control checksum" is not operational as OpenCPN should normally filter out those packets.

Does that make sense?

My setup is:
Davis anemometer connected to an arduino that generates the nmea strings, then connected to a win laptop through a USB based virtual serial port.

Here are two example strings with bad CRCs that are successfully reflected on the Open CPN dashboard:

$ARMWD,333.28,T,,,5.33,N,,,*0B
$ARMWV,333.28,R,5.33,N,A,*0B

Any comments appreciated.

[Hakan]

simonmtl...
I'm not a coder but if understand it correct sentences beginning with "A" are "known bad" and are treated as if the "control checksum" was not checked.

Here is an example how I produce the checksum in my Arduino, if you like:

//Create NMEA string like: $HCHDM,238.5,M*hh/CR/LF
String NMEA_HDM(String Mag) {
  String nmea = "$HCHDM,";
  nmea += Mag,DEC;
  nmea += ',';
  nmea += 'M';
  nmea += '*';
  nmea += String(testsum(nmea),HEX);
  nmea += '\r';
  nmea += '\n';
  return nmea;
}

// Calculates the Checksum for the NMEA string
int testsum(String strN) {
int i;
int XOR;
int c;
// Calculate testsum ignoring any $'s in the string
for (XOR = 0, i = 0; i < 80; i++) {  //strlen(strN)
  c = (unsigned char)strN[i];
  if (c == '*') break;
  if (c != '$') XOR ^= c;
  }
return XOR;
}

Thanks to checksum method author: http://timzaman.wordpress.com/code-c...cksum-xor-cpp/
Håkan

[simonmtl]

Thanks Hakan, but looks like OpenCPN doesn't care at all about the checksum for any wind related data. I tried changing the name to no avail. I have also implemented a CRC method in the arduino but I'm not using it at the moment since OpenCPN doesn't check it.

That's too bad because I occasionnally send truncated strings that corrupt the wind history graph. For instance, when wind = 300 kts is erroneously interpreted, the wind history graph becomes a flat bar with a huge spike.

Regards. Simon

[bdbcat]

simonmtl...

Hmmm. It seems that the core OCPN code checks checksums on messages he cares about. But messages are sent in the raw to the PlugIns, without checking. It then is up to the individual PlugIn to decide what to do.

This seems not great, actually. We probably should not send bad messages to the PlugIns, just on general principles.

I will correct this for the next Beta. Frankly, I expect some breakage from the installed user base, until we learn to switch off the checksum checker for badly constructed NMEA sources. We shall see.

Any comments on this proposed new policy?

Thanks
Dave

[simonmtl]

Thanks BDcat, I fully agree with you. For those who were using bad checksums, they'll be able to "fix" their issue by deactivating the "verify checksum" option.

The help files already suggest that all checksums are verified.