[Arm-netbook] flying squirrel weirdness on the I2C bus
Paul Sokolovsky
pmiscml at gmail.com
Wed Oct 9 17:43:41 BST 2013
Hello,
On Wed, 9 Oct 2013 14:05:04 +0100
"luke.leighton" <luke.leighton at gmail.com> wrote:
> On Wed, Oct 9, 2013 at 1:22 PM, luke.leighton
> <luke.leighton at gmail.com> wrote:
> > ahh.... something really weird going on with the flying squirrel i2c
> > addressing. i'm running i2cdetect and it's showing up 0x34 for the
> > AXP209 I2C address instead of 0x68, and so on. i _thought_ i got
> > the address for the I2C EEPROM right, from the datasheet (0xA2) -
> > but it seems that it's been shifted by one bit (0x51) as well.
> >
> > help! i have no idea what's going on here!
>
> ahh... that's interesting... i2cdetect 0 shows 0x34 for the on-board
> AXP209 and that's literally the only device on the I2C bus. actually
> it shows UU instead of 34.
>
> so the issue's not *specifically* related to the flying squirrel.
> hmmm...
I don't know why nobody replies, I guess because it's around the biggest
FAQ for I2C bus, so everyone probably thinks that everyone else knows
about it and then it must be something else. So, canonically, I2C
address is 7 bit, followed by R/W bit (Wikipedia won't let me lie on
that). Well, it's natural itch to cramp those entities together in a
byte and treat *that* as an address (with such compound address thus
being different for write (x) and read (x+1) operations). And there're
two schools of vendors who cramp stuff together and who don't. It's like
big-endian and little-endian!
Indeed, looking at AXP209 Chinese datasheet, in "Electrical
Characteristics" (sic!) table it says that "typical" "electric
characteristic" of address is 0x68. They fail to specify Min and Max
though in that table, which is 0x68 and 0x69 respectively. They mention
0x69 elsewhere though.
Randomly choosing Western vendor,
http://www.atmel.com/images/doc0336.pdf also does include R/W bit in
the address (Fig 1, p.11), so it's hard to blame someone specifically...
--
Best regards,
Paul mailto:pmiscml at gmail.com
More information about the arm-netbook
mailing list