BUS lock mitigation?

[akkord64]

Would filling the USB ports with dielectric grease help mitigate issues with corrosion causing bus locking? Haven't thought through the marine/reef safe issues but it seems like a simple preventative option.

[akkord64]

Additionally spraying the pcb's with a protective coating (like a spray urethane/varnish) to protect from internal corrosion.

Roberto - after all the years of Bus lock issues are the common trends connections and corrosion or does it ever boil down to hardware issues (component tolerances or failures)?

[rimai]

It seems to be a resistance issue caused by corrosion.

[akkord64]

So in the scenario where a dimmer expansion is causing bus locks, looking into the schematic there is a 10k pair of pullup resistors on the bus, plus 2.2k's on the expansion hub. If the module is causing a bus lock options are 1. determine the source of corrosion adding to the bus resistance ( ie 10k + 2.2k + unknown corrosion source) and clean it. Or 2. Change the 10k's for something like 8.2k or 7.5k to provide a little extra resistance buffer? Are there any major design concerns with this approach? I see the cable lengths are another variable (capacitance on the bus) which can influence the bus stability. I am interested in your bus watchdog circuit you implemented in the Star series.

NOTALINK//electronics.stackexchange.NOTALINK//questions/1849/is-there-a-correct-resistance-value-for-i2c-pull-up-resistors
NOTALINK//www.ti.NOTALINK//lit/an/slva689/slva689.pdf

[akkord64]

Also would shielding the cables could help from spurious noise on longer cables? (poor man's solution would be wrapping tun foil around the cable as a test)

[rimai]

Yes, the cable length and the resistance from corrosion are two things to watch for.

[tngo]

Anyone come across a good solution for preventing corrosion from occurring on the usb headers and connectors?

Thanks

[conklech]

The 10k resistors are I2C pull-up resistors. 10k is shown on the datasheet for the I2C level shifter that the RA uses, but it's probably too high. Search for TI's "Application Report SLVA689: I2C Bus Pullup Resistor Calculation." (I can't post links.) Lower values give a stronger signal, at the expense of a little bit of current. The minimum is a little bit below 2k. It's the capacitance that's an issue. It takes some current to bring the entire bus back up to 5V, and a 10k resistor just doesn't pass enough. If you add some extra pull-up resistors you may see improvement. It doesn't really matter where the pull-up resistors are located.

(The actual resistance of the cable "shouldn't" be a problem. That just causes a constant voltage drop across the cable; I2C is designed to work between devices running at different voltages, so that should be OK. The I2C spec is more concerned about the bus capacitance.)