The RFID is at 13.56MHz.
What we have done so far is to test the accelerometer on the PCB with and without the RFID active. When we saw that the chip showed undesired behavior with the RFID active, we tried relocating it. This was done by removing the device from the board and glue a new device on top of the microcontroller (which is slightly further away from the RFID antenna). This new device has been connected to the PCB by wires.
What we had seen earlier was that this did not resolve the issue. Interrupts still occurred even though the accelerometer was completely stationary on the desk. It did also respond correctly to actual movement as we expect. Today I also hooked up a logic analyzer to the SDA, SCL and interrupt lines to see what the exact behavior was on the signal wires that connect to the accelerometer in the stationary situation.
In the attachments I have included 2 pictures, the first is the overall behavior of these three signals observed (both analog as digital). The second is a zoomed in view on one of the interrupts.
These figures show that the signals on the wires are well defined in the sense that they do not show any spikes. Every interrupt that occurs, is a well-defined transient from 3V3 down to 0V. This leads us to the conclusion that there is no coupling of signals into the wires. So any misbehavior of the sensor can only be caused by external effects that occur due to coupling of signals directly into the sensor itself.
In the attached pictures, the RFID field is only activated after 17.5 seconds. So it seems that the relocation of the sensor made the problem worse in the sense that now even the microcontroller alone influences the interrupt-behavior of the sensor. And even on this new location, the RFID field (13.56MHz) still seems to have an impact, as after 17.5 seconds, the amount of interrupts increases significantly.
The tests above seem to indicate that the sensor is extremely susceptible to external electronic field influences. In the documentation it is clearly stated that below the chip (between the pins) there should be no PCB tracks. In the PCB we have a ground plane below the chip. Please advise on what the preferred solution is for the PCB design / neighborhood with or distance from other signals.
We are currently also evaluating moving the device to a location next to the Bluetooth module. This module is completely shielded by a grounded cage and should therefore negate most of the effects of the RFID field. Although then the Bluetooth module may interfere with the sensor, which is what we still have to evaluate.
These are the only tests we have performed so far. The RFID antenna is completely surrounding the entire board, it is not really possible to test this without significant board changes.
We have received a few sample modules (STEVAL-MKI151V1 - Evaluation Board, LIS2DH12 ) to do some additional measurements independent from the current PCBA design.
