How make a checksum?

cvargcal · Joined: 17 Feb 2015 Posts: 134

Hi, I have this data and I should simulate a device so I need make the checksum:

PCM programmer · Joined: 06 Sep 2003 Posts: 21708

Tell us the manufacturer and model number of the device that produces this
message and checksum. Always do this.

Ttelmah · Joined: 11 Mar 2010 Posts: 19219

Thing to understand, is that there are thousands of different checksums.
Simple XOR, then polynomial ones. These than have individual parameters
on the terms used and the seed values. Actually probably _billions_ of
possible sums. This is why we need to know what device is involved....
With a single example, there are still an almost infinite number of ways
of getting this value.

temtronic · Posted: Wed Mar 24, 2021 4:21 am

I haven't needed to make CRC generators in decades so..
gee, the first thing I saw was that the <ID> appears to be a 64 bit number!
I don't know if the current compiler can do 64 bit 'math' and as others have said we need to know the device mfr. They will(should) supply an application note on 'CRC generation' for this device, whtever it is.

Ttelmah · Joined: 11 Mar 2010 Posts: 19219

He actually talks as if the checksum could be a simple XOR of every
byte after the second $, up to where the checksum is output, and then
is simply a 2 byte hex value of this XOR. If so probably starts from zero.

cvargcal · Joined: 17 Feb 2015 Posts: 134

Thanks for answers...

I want use the fifo protocol, i need "simulate" this device in one server that work with this device, so I need make a real checksum.
http://www.fifotrack.com/download/protocol/FIFOTRACK%20A300%20and%20A500%20GPRS%20Protocol%20and%20Command%20List%20V1.3.pdf

So the protocol is A01, in page 6 show every space... I don't undertand how yo get checksum of 2 bytes with a sum of more size.

PCM programmer · Joined: 06 Sep 2003 Posts: 21708

Ttelmah · Joined: 11 Mar 2010 Posts: 19219

It's wrong really to think of it in terms of a 'sum'. It isn't. It's a bitwise parity.

If you have (say) 'abcd', these have ascii values of 0x61, 0x62, 0x63 & 0x64.
Now in C the XOR operator is ^.

0x61^0x62^0x63^0x64 gives 4.

The result will always be an 8bit value if all the source bytes are 8bit.

This would be output as hex, in the format required with:

printf("*%02X", val);

Which would print *04.

Like PCM, I read it as XORing every byte after the $$, including the comma
before the checkbyte.
It is a very poor check value. It can detect any single bit error, but not
ones with two bits, 4 bit, 6 bits, 8bits etc..