Post Go back to editing

ADXL355 - How many folks seeing unexplained failures?

Category: Hardware
Product Number: ADXL355

The ADXL355 is a powerful sensor with great specs.  However, we've seen a number of failures where the sensor "latches" or gets stuck (e.g. -20 G) and experiences "permanent damage".  The common explanation is the sensor was subject to an acceleration/shock condition beyond range.  Or, repeated shock/vibration causing wear of the anti-stiction material and then the sensing element getting stuck at one extreme.  From our experience, it is hard to replicate the problem in a lab environment.  Simply subjecting the sensor to repeated shock/vibration may not lead to this failure condition.  Yet, we've had sensors in the field experience this damage.  And, it is hard to know what the actual shock and vibration conditions were for that particular sensor.

It looks like we're not alone.  This forum has a number of other folks who experiences this.  For example, see:

  1. Permanent damage on ADXL355
  2. ADXL355 shock problem. (sensor is not working)
  3. Q:Adxl355 data is abnormal
  4. ADXL355 Latch-up condition

My question is:

  1. How many other people have experienced this failure using the ADXL355?
  2. Is this truly explained by subjecting the sensor to large shock/vibration?  If so, how to replicate this? 
  3. Is there a design flaw or limitation where the sensor could still fail to do other factors?  If so, what conditions should be avoided to prevent damage?

Ultimately we don't want surprises where the sensor fails and we don't know why.  Any information from other ADXL 355 users or AD engineers would be appreciated.

  • For context, the Analog Devices reliability handbook:

    has a section on MEMS reliability (starts on page 71).  It goes through the different ways

    MEMS devices can fail, in a bottom-up way, and what can trigger the failure. 

  • Hi  , 

    To achieve high sensitivity and low noise there is a compromise to make on the MEMS sensor design: the springs must be softened. This is one of the main reasons why the high amplitude-high frequency vibrations could lead to stiction/latch-up, especially if the vibrations are in the vicinity of the sensor resonant frequency (~2.4kHz). 

    You can find in the ADXL355 datasheet absolute maximum rating section the maximum sine vibration acceleration the sensor can handle. 

    Test condition A of the military spec highlighted is defined as 20g peak acceleration for a harmonic waveform in the range of 20Hz to 2kHz. But I would limit the input vibration frequency range to ~1.5kHz, to be safe. 

    It is very important to understand the vibration profile the sensor is being expose to in the field, if you wish to replicate the issue.

    In general, we recommend mitigating the high frequency vibrations using soft materials like rubber or foam. 

    I hope this helps, 


  • Thanks   for explaining the trade-off and the suggested limits.  Two follow-up questions:

    1. The fact that other people experienced problems (see links in the original post) was helpful but also concerning.  For any given MEMS inertial sensor, is typical for this many folks to have unexplained/latch-up issues and post about it?  Or, is the high-sensitivity design of the ADXL, and the poor understanding or attention to installation environment, causing this many people to have issues?

    2. What is a good sensor or procedure for understanding the vibration profile of the environment?  I understand there are accelerators designed specifically for vibration sensing - and it appears ADXL355 is not one of them.  Actually, MEMS probably are the wrong sensor type.  

      - It may be a misconception, but folks often look at the same accelerometer output (e.g. ADXL355) to understand the vibration environment they are operating in.  And use that to justify using the sensor (again, the ADXL355).  Comments on that?

  • Hi  , 

    1) I think in general customers expect the sensor to work in any environment conditions and we surely aim to makes sensors like it, but the ADXL355 is not intended to be use in harsh vibration environments. That is why the note of the vibration excitation limit in the datasheet. I think we (ADI) need to make a better job at communicating this in the datasheet. I will work on it. 

    2)  To answer this question, I would like to ask if you can share details about the end application? 



  • Yes, we tested several ADXL355, in different condition, and we realized the unstable behaviour.
    Shock can produce suddenly the lock of one or more channel (x,y,z), or can start to come out with unpredictable measurement values. Only the turn-off turn-on can replace some normality. The ADXL355, despite its very good performance (sensitivity and noise) at low value of vibration, seems not really reliable.