I have a project involving 2 Leos (genuino, not knock-offs). Both are HID enabled. One is an i2c bus master. The other is a slave. A third i2c slave is a wiichuck.
So far so good. With this configuration I am able to read the wiichuck using the master, then use EasyTransfer (great lib!) to share data with the slave Leo. Everything works great. The whole setup is very stable.
Then I tried to add a fourth i2c device to my bus. I should mention that I am using one of the power buses on a breadboard as my i2c bus, SCL on one row of holes and SDA on the other. This has worked very well so far... until I added that fourth i2c device.
The fourth device is an Adafruit 7segment 4digit LED readout with i2c backpack. It responds to the example sketch correctly so I think it's working properly. I set its address to hex 71 which does not conflict with any other devices on the bus. I added code to Leo 1 (the master) to init the device and write a value to the display. And things got weird.
My reads of the wiichuck started to be unreliable; with zero stick input, the joy Y position should read as a steady value of 123. But now all of a sudden the value is fluctuating every 10 or so loops, to 119. It looks fairly periodic, though I have not timed it accurately. This gives me a relative movement of -4 wii units, and I'm using this to control screen mouse -- so my screen mouse is slowly but surely crawling vertically.
If I remove the 7 segment display from the bus, the problem disappears. If I reconnect it, the problem reappears. If it is connected but idle (I don't init a matrix device using the adafruit library, I don't write to it at all) it still seems to corrupt bus traffic. [Interestingly if I leave the LED display connected to the i2c bus but unplug its dc power, the bus master Leo seems to hang; apparently the i2c master does not respond well to slaves that fail to respond?]
At any rate it appears that the 7segment display for some reason interferes with my reads of another i2c slave. Do I have a weirdly defective 7seg display or is this a common i2c problem? is there a known workaround? do I have to noise-filter this bus somehow? I wouldn't know where to begin, so I am hoping that there is a tried&true fix for this problem.
It's frustrating, as my experience with i2c until now has been terrific -- very easy and trouble-free. I was planning to use i2c throughout my project as the backbone for the architecture. Hope I don't have to abandon it!