need help with em4102.c manchester decoding

[epv]

Hi. I'm playing with an EM4095-based RFID basestation. (for reference, it's the Parallax/Grand Design one) As a sort of challenge, I wiped the Parallax-supplied code and have been trying to get the em4095.c/em4102.c drivers from CCS to work on it.

After a little work making it fit on the onboard 16f627A (ie, read-only, MOD tied to ground, pic has no CCP2, etc..) I feel like I have it close to working, but I can't get the decoding to work out.

The pulse widths measure almost perfect 32/64 clocks, header_search_4102() finds headers, etc, but decode_data_4102(data) fails to decode the data because the row-parity check fails; here is the 50 bit stream read from each tag (plus what i get if i read another 10 bits further) :

tag1: 00000111100000000110000000010110001101111110101111+0100011111
tag2: 00000010010001101010101001111011101111101100011101+0110011111

Both tags produce these respective bitstreams absolutely reliably, and with no tag present I get dribs and drabs of random noise with the occasional 9-ones header found, but nothing repeatable.

Are extra bits getting stuck in somewhere? I'm not really sure where to proceed from here. If I pretend the initial 5 0's are spurious, it still doesn't pass parity check.

If nothing else, it's certainly been educational...

eric

PCM compiler version 4.049;8/6/2007 (linux)

[kolio]

According EMMarin 4102 datasheet page 4 table and your sequence:
00000111100000000110000000010110001101111110101111
it will look like:
0000 0
1111 0
0000 0
0011 0
0000 0
0010 1
1000 1
1011 1
1110 1
0111 1
These are data nibbles, followed by even row parity bits and they are obviously OK.
But looking in em4102.c , we'll find a bit comparision expressions, in which I already stopped to trust!
As far as I remember there were already compliants about wrong bit comparision, weren't there?
So, try to use "^" instead of "==" or "!=" operators and write your own routine.
wish you luck

[epv]