Data processing

AN3254

14/24

Doc ID 17783 Rev 1

Normally, each byte is read out as the most significant bit (MSB) first. But this can be

changed by setting the MSBF configuration bit. If this is done, each byte is read out as the

least significant bit (LSB) first.

4.1.1

Data register assembling example

The following is an example of the reading and re-arranging of STPMC1 registers.

On the left there are the eight data records as they are read, represented as hexadecimal

bytes while MSBF was cleared, on the right is the corresponding register.

1.

65 7A 7C 82

DAP = 82 7C 7A 65

2.

00 7A 0C E0

DRP = E0 0C 7A 00

3.

00 00 8C 92

DFP = 92 8C 0 00

4.

00 06 6E 22

PRD = 22 6E 06 00

5.

…

4.2

Parity check

Each bit of parity nibble is defined as odd parity of all seven corresponding bits of data

nibbles. In order to check the data record integrity, the application should execute something

similar to the following C code, given as an example:

int BadParity (unsigned char *bp)

{

register unsigned char prty = grp;

/* temporary register set to group #

(0..6)*/

prty

= *bp,

/* take the 1st byte of data */

prty ^= *(bp+1),

/* XOR it with the 2nd byte */

prty ^= *(bp+2),

/* and with the 3rd byte */

prty ^= *(bp+3),

/* and with the 4th byte */

prty ^= prty<<4, prty &= 0xF0;/* combine and remove the lower

nibble */

return (prty != 0xF0);

/* returns 1, if bad parity */

}

If the parity nibble check fails, the reading task should be repeated, but this time without the

request of latching, otherwise a new data would be latched and the previous reading would

be incorrectly lost.

In a very harsh EMI environment, it would be a good practice to read the data records twice

and then compare both readings. In this way, the probability of detecting bad readings would

be significantly improved. However, a single piece of bad data can be discarded because no

meaningful information is lost as long as the reading frequency is about 30 ms.