For the record, I tried a slightly different approach, but not the right result, of course: it is the same method than you, just extracting the bytes differently.

1. String myCommand = "<ID01><L1><PA><FE><MA><WC><FE>Probe"; // result should be: 3E
2. byte[] bytes = myCommand.getBytes();
3. byte res = 0;
4. for (int i = 0; i < bytes.length; i++)
5. {
6.   res ^= bytes[i];
7. }
8. String checksum = hex(res);
9. println(checksum);
10. exit();

arthurG's solution works but as he said, it can be just luck, one chance out of 256... (or a brilliant intuition!)
We need more test strings/data, and/or exact reference to your source of information.

Thanks arthurG and phi.lho for your answers!!

The data sheet for the LED board which describes the XOR checksum and how messages are being composed is available here: http://www.produktinfo.conrad.com/datenblaetter/575000-599999/590998-da-01-en-LED_LAUFSCHRIFT_ROT.pdf (I reckon page 4 will me of most interest)

I found the code with the working checksum in this forums for micro controllers: http://www.mikrocontroller.net/topic/131760 (sorry, it's a German forum).

Until now I found 3 working examples of an input string (all from the forum):

1. <ID01><L1><PA><FE><MA><WC><FE>Probe3E<E> // result should be 3E
   
2. <ID01><L1><PA><FE><MA><WC><FE>test62<E> // result should be 62
   
3. <ID01><L1><PA><FE><MA><WZ><FE>  <KT>   <KD>5D<E> // result should be 5D
   

I'll get hands on the LED board within the next hour. Fingers crossed for your solutions :-)

For the records: This is the allegedly working Delphi code I tried to adopt in Processing:

1. Function SimpleCheckSum (const MyMessage : String) : byte;
   
2. Var res : byte;
3. i : integer;
4. Begin
5.       res := ord(MyMessage[1]);
6.       for i:=2 to length(MyMessage) do
7.             res := res XOR ord(MyMessage[i]);
8.       result := res;
9. End;

After having another close read through the data sheet I found proof of ArthurG's assumption that the ID doesn't belong to the checksum:

denotes the Xor Result of the data content

And the ID does explicitly not belong to the data content. Excellent! (it helps if you can read ... d'oh!)

Funny thing: I removed the exclusion of '<' and '>' but I still get the correct checksum.

Anyways. Thank you guys!!! After countless hours of trying I gave up yesterday. I would have never thought of the input string as the incorrect part of the code!

Again for the records: this is the working code right now:

1. String myCommand = "<L1><PA><FE><MA><WC><FE>test"; // result should be: 62
   
2. byte res = 0;
3. for(int i=0; i<myCommand.length(); i++){
4.   res ^= byte(myCommand.charAt(i));
5. }
6. String checksum = hex(res);
   

Come to think of it, I think the even number of symbols < and > make them cancel themselves. 0 ^ x = x; x ^ x = 0
Hence the "no difference" with the test.

Now that you mention it - makes perfectly sense :-) Thank you!!!