> check what I wrote carefully, the secondary detection might be like > the winbond, but the primary ($58) is $31 and that is not winbond but > Asus in this case, that is why I have the 1st and 2nd detection. The > fact the other 2 ID's are like winbond simply shows you that Asus > bought them from winbond:) Oh sorry, I forgot to mention one thing. In our detection script, we ignore the LSB of register $58. Thus the AS99127F rev2 ID ($31) and the W83782D ($30) look the same, that's why I got confused. I don't know why we ignore this bit, I couldn't read anything about that in the datasheets. I will take a look at our CVS logs tommorow and see what made us add this in the past. > a diffrence we have -i think- is that MBM does not use the $48 > register, MBM detects a sensor chip.... Asus on $2D.... then the 2 > LM75's.... I just name/label these 2 LM75's Asus 2 and 3 but IF the > chip on $2D is a Asus, if it's a winbond I name then winbond 2 and > 3.... internaly they are LM75's so I really don't need the $48 to show > me where the LM75's are, if there are more then 2 LM75's on a board > (not found one so far) the extra LM75's are LM75's while those 2 on > the default adres are Asus 2 and 3 The register $48 is supposed to contain the I2C address of the main chip itself (that is the address we are "talking" to at the moment). I don't know what the idea was, but at least this adds some reliability to the detection. Looks like Winbond only does this, the Asus chipsets seen so far did not. The subaddresses are computer using register $4A (but that's a different thing). Thanks again for your help. -- Jean Delvare http://www.ensicaen.ismra.fr/~delvare/
