Hi ,
i can't believe i actually am having an issue with an 8 pin device, but here is my problem: nothing but a few glitches as an output. my thought is there must have been an enable/power race problem, and i need to toggle the enable to check it out. looking for some insight here.
here is the setup: stellaris LM3651 on a standard 3.3V side, talking fine to an i2c port - commands and responses are fine. so, SDA and SCL with 3.3V pullups on them and 3.3V is wired on the B side (higher voltage side). on the other side, i have SDA and SCL pulled up to 2.8V going in to nothing but a beagle protocol analyzer and also a logic analyzer. Vcca and OE are tied to the 2.8V rail.
here is the behavior: if the analyzers are moved to the 3.3V side, both see clock and data fine and decode command and responses fine. when they are moved to the 2.8V side, i see SDA and SCL both pulled high, and several down 8nS "glitches" and that is about it. actually, a clock glitch, then 3 clock and data glitches 42uS later, and then silence on the bus.
i have tried this on two separate devices with the same effect. even though the app note says you can tie OE with Vcca, i have a feeling the slicers inside the part might be acting to cut off the output. my next effort is to wire the OE to the 3.3V side GPIO and pull that down and then high again to see if that works.
any ideas here?