I am trying to use the self-test feature on the ADXL345. Looking at pg 22 of the datasheet, I have a few questions. I was using application note AN-1077 as a guideline.
According to the flow chart in the application note, the test sequence should be completed in +/-16g, 13-bit (full resolution) mode. However, Table 18 in the datasheet is for 10-bit resolution. Is there a table for 13-bit or can I just use this one?
Our sensor is mounted such that the x-axis reads -1g (y and z are 0), although this may change in the future. I was using the self-test feature in the x = -1g orientation. Will this affect the results? If so, what orientation should I put the sensor in to obtain the results in Table 18 (and use the scale factors in Table 14) of the datasheet?
On page 31 of the datasheet, it states that a "Generally, part is considered to pass if the minimum magnitude of change is achieved. However, a part that changes by more than the maximum magnitude is not necessarily a failure." What does it mean if the change is more than the maximum magnitude? Also, I am assuming that for the y-axis in Table 18, this is to be interpreted as if the delta_y value is larger (more negative) than -67, it is ok (i.e., because this is the larger change)?
Are the values in Table 18 the change in bits between when self-test is on vs. off (i.e., delta = avgSelfTestOn - avgSelfTestOff)?
I am currently sampling using a data rate of 400Hz and averaging 100 samples with the self-test on and off. When the self-test is off, I get x: -248, y: -20, z: -10 (taking two's compliment of data, but not converting to g's). With the self-test on, I get: x: 33, y: -305, z: 540. These values are relatively stable give/take a few bits. I am concerned because they are far in excess of the stated maximums, and want to make sure that I am implementing the self-test feature correctly.