I2C bus corrosion

[Flappy]

Hello,
I have been replacing a lot of SAA1064 ICs on boards that have been exposed to rainwater and I have noticed that the corrosion is almost completely isolated to the clock and data pins.  Are conductors that carry AC with a DC offset more susceptible to electrolysis?  Does frequency play a part?  Is there such thing as corrosion resistant data protocol? 

[AndyC_772]

Yes, they are. That's precisely why telephone networks are powered from -48V with respect to earth, rather than +48V. You could argue that the negative bias constitutes a "corrosion resistant" protocol at the hardware layer.

Better to keep the water out in the first place, though.

[Rerouter]

For limited DC bias, your looking for Manchester encoding, it makes sure there is an even amount of data sent as a positive and a negative, and a clock is generally safe

The issue with I2C is that it idles high, which means it has a voltage gradient to the ground surrounding those pins, allowing for corrosion,

In buried wiring they will run a sacrificial anode which sits at a positive voltage in relation to the wiring to be protected, so that it gets eaten away first. e.g your I2C bus in this case is playing the sacrificial wires to the ground plan.

[jpanhalt]

Just remember Michael Faraday's definition of "anode" as the electrode at which oxidation (corrosion) occurs.   Ford used to use a positive ground until the mid-1950's.   There is debate as to its effects: https://mechanics.stackexchange.com/questions/1934/what-are-the-tradeoffs-for-positive-vs-negative-ground

John