We are using 4-Wire-SPI. The ADXL345 is connected to an STM32-nucleo board. When reading the DEVID the returned byte is always 0xCB, which is a 1 Bit-left-shifted version of 0xE5 (with new LSB with value 1).
Here is a picture with the signals (channel 1: SCLK ; channel 2: SDI; channel 3: SDO; channel 4: \CS ):

Because of the correct signals \CS, SDI and SCLK, there should be the right response on SDO wit 0xE5.
Any suggestions for the reason of this behaviour?