Regarding after adding several g from the 0 g state

Hello S1640  thanks for posting.

Here you have some raw measurements from ADXL357B at 10g FSR (first screenshot) and 40g FSR (second one) alongside the raw output values in LSB.

The platform used for this was: EVAL-ADXL35x-SDP Evaluation Board | Analog Devices

What I did was placing the sensor on a start position with gravity facing against the Z axis (1), then rotate 90º towards X axis (2), and finally come back to the initial position. (1)

Hope this helps solving your inquiry.

best regards,

Mario SM


Offset_test_10g.txt


Offset_test_40g.txt

 Thank you for your support. I would like to confirm you some issue.  
I would like to know if, for example, if an impact is added in a short period of time at each full scale, for example, 0g to 10g / 10g to 0g, will the value return to 0g-offset value? I'm wondering if the ±40g full scale will be slower to return to the 0g offset value after a sudden impact than the ±10g. Please give your opinion.

In sheet [Offset test 10g.txt], From the graph, it appears that the X-axis is moving away from 0 acceleration but the Y-axis is not from around 5700 samples. What is this?

Does "Number of Samples" refer to the time axis? Please also tell us the definition of "Number of Samples."

In the graph, I thought that the X axis is the direction in which acceleration is added and the LSB increases, and the Z axis is the direction in which acceleration is decreased and the LSB decreases, but what does it mean that the LSB decreases in the same direction on both the Z axis and the X axis?

Best regards,

Hello S1640 

S1640 said:

Thank you for your support. I would like to confirm you some issue.  
I would like to know if, for example, if an impact is added in a short period of time at each full scale, for example, 0g to 10g / 10g to 0g, will the value return to 0g-offset value? I'm wondering if the ±40g full scale will be slower to return to the 0g offset value after a sudden impact than the ±10g. Please give your opinion.

As far as you do not go greatly over the full scale range there should be no delay.

(Only if you match the sensor's stoppers gee limit you could face some short of delay, but that limit is over the full scale range for sensor's safety purposes and not featured in the DS)

S1640 said:

n sheet [Offset test 10g.txt], From the graph, it appears that the X-axis is moving away from 0 acceleration but the Y-axis is not from around 5700 samples. What is this?

There is no acceleration on the Y-axis during the whole test, thus, it remains close to 0g. (only some collateral acc. from not making a perfect perpendicular movement, since I flipped the sensor with my hands)

S1640 said:

Does "Number of Samples" refer to the time axis? Please also tell us the definition of "Number of Samples."

The number of samples is the number of acceleration samples (X-Y-Z axis measurements per sample in this case) taken. The ODR for the test was 500 Hz in order to match the DS config, you can use the period to estimate the time passed by.

S1640 said:

In the graph, I thought that the X axis is the direction in which acceleration is added and the LSB increases, and the Z axis is the direction in which acceleration is decreased and the LSB decreases, but what does it mean that the LSB decreases in the same direction on both the Z axis and the X axis?

As you can observe in the shared reference scheme, the Z-axis was facing gravity on its positive direction (LSB increases), while after the flip the X-axis faces against gravity, negative direction. (LSB decreases) 

Z axis comes from 1 gee to 0, and then again to 1 gee, while X axis goes from 0 to -1 gee to end at 0g again.

This can be counter intuitive but comes from the working principle of an accelerometer, you can find further info in the following links:

What is the meaning of output of an accelerometer - Electrical Engineering Stack Exchange

Accelerometer Specifications - Quick Definitions | Analog Devices

Best Engineering practices when using ADXL accelerometers [Analog Devices Wiki]

hope this solves your questions, have a nice day

Mario SM

Thank you for your quick response.
You say that "as far as you do not go greatly over the full scale range there should be no delay."
Am I correct in understanding that even if a sudden, instantaneous shock is added, for example, 0g → 10g → 0g, it will follow accurately and eventually return to the 0g offset value?

In sheet [Offset test 10g.txt], From the graph, it appears that the X-axis is moving away from 0 acceleration but the Y-axis is not from around 5700 samples. What is this?→Sorry, I am sure that there is no acceleration on the Y-axis during the whole test. What I want to confirm is that at around 5700 at number of samples, the X-axis value is shifted from the value before the impact was applied.

Do you have a 10g test instead of a 1g test?

Also, at for example 3215 number of samples, The values ​​on all axes are significantly different. Can I just ignore this as noise? This value is significantly different from the 0 offset value.

Best regards,

Thank you for your quick response.
You say that "as far as you do not go greatly over the full scale range there should be no delay."
Am I correct in understanding that even if a sudden, instantaneous shock is added, for example, 0g → 10g → 0g, it will follow accurately and eventually return to the 0g offset value?

In sheet [Offset test 10g.txt], From the graph, it appears that the X-axis is moving away from 0 acceleration but the Y-axis is not from around 5700 samples. What is this?→Sorry, I am sure that there is no acceleration on the Y-axis during the whole test. What I want to confirm is that at around 5700 at number of samples, the X-axis value is shifted from the value before the impact was applied.

Do you have a 10g test instead of a 1g test?

Also, at for example 3215 number of samples, The values ​​on all axes are significantly different. Can I just ignore this as noise? This value is significantly different from the 0 offset value.

Best regards,

Hello S1640 

In relation to the first three questions: Yes, I would not expect any mayor offset, that value around 5700 samp comes from the manipulation by hand. I will try to run a new test with more gees, in a fixed position so there is no undesired accelerations. I may be able to run it late next week.

Regarding the samples around the sample 3215, until I kept the sensor fixed you can expect the sensor to track any small acceleration from my fingers, simple test and really sensible sensor. I hope this example helped after all, but I will try to run a new one that suits your needs better.

have a nice weekend,

Mario SM

Thank you for your quick support.
It's very helpful.

For now, I plan to reply to the customer that there will be no deviation from the 0g offset value.