ADIS 16475 shows modulation with 100 sec period when excitated with harmonic oscillation of larger than 1.05 g

Hi,

we use ADIS16475-2 and have strong problems under vibration.

Therefore we tested it on a shaker at 0.95 g, 1.05g and 3 g at 10 Hz, 100 Hz and 250 Hz. Up to 0.95 g the accelerometers work as expected.

For 1.05 g or higher, there appears a modulation on the accelerometers and gyros with 100 s period. This is independant on the excitation amplitude or frequency. It seems that the sensor adds this modulation as soon as 1 g excitation is exceeded.

Also filtering will not help - the amplitude becomes smaller, but same behavior. So it is not an aliasing issure, especially because it also happens at 10 Hz excitation.


We tested the same setup with ADIS 16465 without any problem. The behavior is repeatable on all (5) tested units. Unfortunately.

Attached is a filgure with ADIS 16475. 

I guess that the sensor detects excitation larger than 1 g and performrs internally something terrible... (?)

Need yur help because this effect might be otherwise a show stopper for a larger project in automotive testing.

 
BR
Edgar

  • Hi!

    additional information: This effect happens when exitation is applied on z axis. If the vibration is applied in parallel to the PCB, no effect is observed. We only measure the effect on ADIS 17475 and ADIS 16477 (ballgrnd mounted devices) - using the same setup (electronics etc.) with other types of ADIS (e.g. ADIS 16465) we see no impact! 

    So my question: what happens inside the ADIS 16475 plastic package ???

    WE appreciate all comments / hints which can help to solve the problem.

  • Hello  

    Thank you for sharing this with us.  I am sorry that you initial message had no responses. This seems like a never-ending story, but I am in the process of re-organizing how our team supports this forum, but we are still a bit short on resources, at the moment.  For now, I will do my best to help you.  Just making sure that I understand this correctly, you are seeing a resonance at 0.01Hz (frequency of 100s period) on the ADIS16475 and ADIS16477, but never on the ADIS16465 or ADIS16467 devices.  Am I understanding this correctly? 

    I will need to give this some thought, as it is tough to imagine a physical mechanism, inside of a device with these types of dimensions, which would have that low of a resonant frequency.  

    Best, NevadaMark

  • Could it be possible to get pictures of the setup, which illustrate the mechanical support for these parts, during the z-axis vibration exposure? Thank you! 

  • Hi Mark,

    thank you very much for your reply! Initially we thought that the ADIS itself has a problem at 100 Hz harmonic excitation. But when we isolated the ADIS from the PCB and mounted it hard on the vibration table, the impact disappeared. Initially we thought that a certain bending of the pcb would cause the problem, but locally more stiffness of the PCB did not help.

    After many tests and physical resonance analysis on our vibration table we found out that there is an eigenfrequency of the PCB at 550 Hz (amplitude magnification of about 10). This "kills" the ADIS sensor performance (it produces a significant bias). Coning/sculling compensation at 1 kHz helped  by 30 % (sampling at 1 kHz or 2 kHz) but nonlinearity of the sensor has still an impact.

    So, early this week we decided to redesign the PCB to damp the Eigenfrequencies. A quick test yesterday showed significant improvement within our application.  

    The 100 second signal (our first observation) has its root cause by sampling a 10 Hz or 100 Hz or 250 Hz harmonic vibration signal with 1000 Hz ADIS sampling frequency! Both oscillators (those of the vibration table and those of our electronics to sample the ADIS) are not synchronised but quite equal (but with a difference of about 0.01 Hz. So finally the 100 sec period was an observation but not the root cause of the problem - the root cause (see above) was/s the Eigenfrequency of the PCB.

    So, the problem is now solved!

    My suggestion to Analog Devices: Impement a simple coning / sculling algorithm on the ADIS and make the user choose the output frequency. Example: User provides 2 kHz sampling frequency and selects a devider by 4. Then the internal coning/sculling algorithm reduces from 2 kHz to 500 Hz and provides corrected data at 500 Hz. And on this way you might be able to offer even 4 kHz internal sampling, which makes your ADIS much more resistant against such impacts whicht the need to increase SPI data volume.  

    One other question: During our tests we also filled the total ADIS with glueing t assure that no resonances happen inside the ADIS. This worked well and opening the device we identified als the three axes accelerometer and the two gyro packages. We are very interested to buy the gyro package as stand-alone device for another project, but we have not found the component in your catalogue. Can you give us more information?

    Please do not hesitate to contact me for further questions.

    Best regards,

    Edgar.

  • Hello Edgar,

    Thank you, very much, for sharing your learning experience with us.  When I first read this, I was wondering if there was opportunity for resonance, which was folding back onto DC, but didn't have the time for fully mature and share my thoughts, when I last posted.  This is some solid engineering work! 

    Mark