I2C and Prop at 33 feet

T Chap

I have been testing out some longer I2C runs, using the PCF8574 8 i/o expander with the Prop. One test uses a 33' CAT5E cable, where clk and data are on one twisted pair, and power and ground are on another pair. I didn't have any reason for choosing this wiring scheme. Another test uses two legs of cables, both 20' each off the Prop data/clk to two of the 8574 devices. These are used on the same controllers that contain motor drivers using mosfets. Soon I will buy a 250 roll of CAT5E and see what happens. The driver is the minimal I2Cdriver object.

I thought I recall reading that these type lengths were not doable with I2C, maybe I am mistaken, but in my cases everything is behaving nicely.

Anyways, just an observation.

localroger

Never put two changing signals on one twisted pair. Put ground and clk one one, and power and data on the other. Otherwise you get capacitive coupling between the two changing signal lines, which really reduces your range. (Ideally you'd twist both signals around a ground, but power works just as well and then you only need two pair.) This way the twists cause electrical interference (including from the other twisted pair) to cancel out as the twists pick it up in alternating polarity. I've run 9600 baud RS232 hundreds of feet this way when the spec says it should really only be good for 50 feet.

Mike Huselton

localroger,

You have me beat for simplicity of your explanation. Better than my college-level mish-mash-just-to-show-you-how-smart-I-am explanation

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
JMH