CRC16 calculation not quite right

Question

I am working from http://www.ross.net/crc/download/crc_v3.txt, and using the 16-bit polynomial 0x8005.

My message is 0xAE.

This site http://www.lammertbies.nl/comm/info/crc-calculation.html generates a correct calculation from other data I have.

This is the output from my code, describing each step.

Poly: 1010000000000001
Initial message: 01110101
Message: 011101010000000000000000 24
crcreg: 0000000000000000
crcreg: 0000000000000001
crcreg: 0000000000000011
crcreg: 0000000000000111
crcreg: 0000000000001110
crcreg: 0000000000011101
crcreg: 0000000000111010
crcreg: 0000000001110101
crcreg: 0000000011101010
crcreg: 0000000111010100
crcreg: 0000001110101000
crcreg: 0000011101010000
crcreg: 0000111010100000
crcreg: 0001110101000000
crcreg: 0011101010000000
crcreg: 0111010100000000
crcreg: 1110101000000000
crcreg: 1101010000000000  //Here we had a 1 pop off the shift reg, so we XOR in the poly.
^poly:  1010000000000001
=crcreg:0111010000000001

crcreg: 1110100000000010
crcreg: 1101000000000100
^poly:  1010000000000001
=crcreg:0111000000000101

crcreg: 1110000000001010
crcreg: 1100000000010100
^poly:  1010000000000001
=crcreg:0110000000010101

crcreg: 1100000000101010
crcreg: 1000000001010100
^poly:  1010000000000001
=crcreg:0010000001010101

CRC:    0010000001010101

4 aa
R-CRC:  1010101000000100   //Reversed, just in case MSB/LSB display got hosed.

55 20

The expected CRC16 is 0xBC81

Answer 1

Your polynomial is reversed: You need the coefficients for higher powers of x to be represented by the leftmost bits. Try reversing the bits of your polynomial (representing it as 1000000000000101) and I think you'll get the correct result.

You may also need to reverse your input and output, depending on how the particular implementation of CRC you're using is defined (LSB-first or MSB-first).