Re: [Gumstix-users] I2C Communication Problems

[Dave H. <dhy...@gm...>]

Hi Joss,

On Thu, Sep 15, 2011 at 6:45 AM, Joss <Log...@gm...> wrote:
>
> Hello Dave, the community,
>
> As the I2C communication problems are still occuring I attached an image
> which shows the hardware connections currently used. The code used still
> remains the same as can be read throughout this topic. I use the following
> hardware:
> Gumstix with an Overo Tobi extension board, a Phillips dual bi-directional
> bus buffer P82B96 as level-shifter and a Honeywell HMC6343 3-Axis Compass.
>
> http://old.nabble.com/file/p32471727/Bildschirmfoto.png

I think you're missing the pullups on the 1.8v side?

These level shifters only drive to ground, and require pullups to
pullup to Vcc. You need pullups on either side.

> The adress byte and an ACK-bit is being sent, as can be seen in the image of
> the initial post in this topic. But no data byte with the command, what the
> HMC should do, is being sent. Only the first byte with the adress (0x32 HMC
> related). The logic analyzer I use is connected to both sides of the level
> shifter and the outcome in terms of analyzing the SDA and SCL lines is the
> same for both of 'em...

So the trace looks good, and the host should be sending out the
command as soon as SCL goes high (it looks like somebody is stretching
it for some reason).

The fact that ACK is being driven low means that the device is
responding. The evidence suggests that issue is host side.

The fact that you're seeing good signal on both sides suggests that
the pullups might already be present on the 1.8v side.

-- 
Dave Hylands
Shuswap, BC, Canada
http://www.davehylands.com