I recently purchased a number of Pololu LSM303D sensors for use in an iceberg tracking beacon project. Low-power operation is one of the major priorities for the project, and I’m experiencing some issues with the LSM303D’s quiescent draw.
The LSM303D datasheet states a current consumption in power-down mode of 1 uA. However, I am measuring a minimum current draw value of ~47 uA. I am connecting a 3.3V source directly to the VDD pin in order to bypass the onboard regulator and am also writing to the appropriate accelerometer (CTRL1) and magnetometer (CTRL7) registers to enable power-down mode. I can observe the current draw dropping as it switches from taking measurements (~500 uA) to power-down mode mode (~47 uA).
I’ve taken a look at the schematic for the board, and am having trouble identifying where this draw could be coming from. While I assume the regulator is completely off, could it be causing the problem? I’d prefer to avoid having to physically remove it if possible.
I scoped the onboard regulator and you are correct, it is on even though power is supplied only to the VDD pin. Looking at the schematic, I assumed VDD shouldn’t have been able to reach the VIN pin of the regulator. I’m still not sure how or why it does.
It looks like my options are to either remove the regulator, cut the trace to the ON/OFF pin, or pin-power the sensor.
Yes, LDOs like the one used on that sensor can draw small currents like what you are noticing when voltage is supplied across their output. Cutting the trace to the ON/OFF pin might help your situation, but, unfortunately, the trace connecting IN to ON/OFF on that board runs under the LDO, making removing that connection infeasible. It also looks like there is not a way to cut the VDD trace going to OUT on the LDO without removing the VDD connection to the rest of the circuit. Given those things, it does seem like removing the LDO is the most straightforward solution for eliminating that extra draw in current.
Thanks for your reply. I removed the LDO and things have seemed to improve.
Initially, I had tried powering the LSM303D using a GPIO pin, which seemed to work well at first, but as soon as I had another I2C device on the bus, it turned itself back on, sourcing 0.5-1.0 mA across the SCL/SDA, even when in deep sleep. I assume the SCL/SDA lines were feeding VIN. It didn’t appear that writing to the registers to power-down the LSM303D had any effect in this state. I’m glad I’ve been able to troubleshoot all of these issues, but I can’t say it was an overly enjoyable experience.
