Leak through the discharge resistor
0.94 mA with the sensor off
Field marks. With the load switch off, the cell supplies 0.94 mA: 0.47 mA through each pull-up. The sensor’s rail settles at 0.14 V, and SDA and SCL at 0.78 V.
Look-alike. A sensor that is off. A meter on its rail reads 0.14 V, which passes for nothing. The bus shows it. An idle I2C line should sit at the supply, and these sit at 0.78 V. That is under the 0.9 V an input reads as low, so every chip on the bus sees both lines held low.
Range. At 3.3 V the leak is 1.06 mA. With 2.2 kΩ pull-ups at 3.0 V it is 1.87 mA, and with 10 kΩ it is 0.46 mA. The diode hardly matters: changing its saturation current over four decades moves the 3.3 V figure from 1.02 to 1.11 mA.
Habitat. Battery boards that cut a sensor’s supply with a load switch that has a discharge resistor, and leave the I2C pull-ups on a rail that stays up.
597