Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.

Modeling vibration transmission from motor to PCB

Please login with a confirmed email address before reporting spam

Hello all,

I am developping a prototype of PCB that will hold a motor running at constant speed along with a couple of accelerometers. In order to estimate the effect of vibration induced by the motor on the other components I am building a simplifided model using only the toolbox Solid Mechanics.

Attached is my model which is inspired by the blasting_rock one. It looks like that: https://i.ibb.co/4szh23D/model.png

The top material is Aluminium 3003-H18. I am doing a material sweep for the remaining two bottom blocks, material 1 is the same Aluminium 3003-H18, material 2 is FR4 (Circuit Board).

On the top face I am applying 4 periods of a time dependent sinus stress along Z, with max value at 1MN/m^2. One of the remaining side of this whole thing is set as symetric, the other side is low-reflecting boundary. The bottom side is free.

I am doing a time dependent study and I am comparing the displacement on the input point and output point as pointed out below: https://i.ibb.co/4P4Gc3p/model-input-Output.png

For the first case since both materials are Aluminium I want to observe that the displacement wave is the same between input and output (propagation attenuation rather negligeable at this scale). For the FR4, since it has a lower impedance, I want to visualise the displacement wave is less important, which would reflect the non-perfect match of impedance at the interface between materials.

Below are the results: https://i.ibb.co/M6SHBz9/model-output.png Each color correspond to a material switch, input is the dashed line, solid is the output. Although a little bit of propagation time is visible as phase on this graph, I don't see any difference in the wave magnitude. No impedance match effect is visualized which is a problem.

Am I missing something in my model? Is there a parameter that let the wave be attenuated, even more on interfaces?

Side problem, I had to set a very high stress otherwise the resolution on the displacement would be extremely low. I think that would be linked to my mesh that is too big. Any tip on that would be much appreciated. The concrete case for the motor is a MEMS motor that will spin at around 20 kHz and will produce a super small vibration, hence very far from my MN/m^2.

Any help or tip on how to visualise impedance match effect would be much appreciated.

Thanks in advance



4 Replies Last Post 07.01.2019, 10:31 GMT-5
Edgar J. Kaiser Certified Consultant

Please login with a confirmed email address before reporting spam

Posted: 6 years ago 07.01.2019, 06:34 GMT-5

Thomas,

your model dimensions are much smaller than the wavelength of the vibration at 2 kHz in those materials. So it is essentially quasi-static.

Cheers Edgar

-------------------
Edgar J. Kaiser
emPhys Physical Technology
www.emphys.com
Thomas, your model dimensions are much smaller than the wavelength of the vibration at 2 kHz in those materials. So it is essentially quasi-static. Cheers Edgar

Please login with a confirmed email address before reporting spam

Posted: 6 years ago 07.01.2019, 09:22 GMT-5

Thomas,

your model dimensions are much smaller than the wavelength of the vibration at 2 kHz in those materials. So it is essentially quasi-static.

Cheers Edgar

That is correct. I changed the frequency to 100 kHz and removed the boundary load to prescibed displacement, it works much better. Here are the results: https://i.ibb.co/yqW6XjB/displacement-graph.png

From that I think I could derive a complex impedance that includes the interface and the transmission line.

>Thomas, > >your model dimensions are much smaller than the wavelength of the vibration at 2 kHz in those materials. So it is essentially quasi-static. > >Cheers >Edgar That is correct. I changed the frequency to 100 kHz and removed the boundary load to prescibed displacement, it works much better. Here are the results: ![https://i.ibb.co/yqW6XjB/displacement-graph.png](https://i.ibb.co/yqW6XjB/displacement-graph.png) From that I think I could derive a complex impedance that includes the interface and the transmission line.

Edgar J. Kaiser Certified Consultant

Please login with a confirmed email address before reporting spam

Posted: 6 years ago 07.01.2019, 09:31 GMT-5

Thomas,

everything in your model seems to be linear and you seem to be interested in the dynamic steady state. So you might consider to do a frequency domain study which is far more efficient and will directly provide complex quantities.

Cheers Edgar

-------------------
Edgar J. Kaiser
emPhys Physical Technology
www.emphys.com
Thomas, everything in your model seems to be linear and you seem to be interested in the dynamic steady state. So you might consider to do a frequency domain study which is far more efficient and will directly provide complex quantities. Cheers Edgar

Please login with a confirmed email address before reporting spam

Posted: 6 years ago 07.01.2019, 10:31 GMT-5

Thomas,

everything in your model seems to be linear and you seem to be interested in the dynamic steady state. So you might consider to do a frequency domain study which is far more efficient and will directly provide complex quantities.

Cheers Edgar

Yes the frequency study is definitely the end game but I wanted to validate the behaviour at the interfaces in the first place at to more easily targets errors in my simulation, since I'm rather new to COMSOL. I've improved the probing points and put them closer to the interface and it seems I don't get any interface effect so far, as seen below where it seems I'm back at the low freq / big wavelength point: https://i.ibb.co/YN51TP8/close-probes.png

What was I expected? decayed magnitude depending on the difference of impedance at the interface. Alu-Alu should have seen no difference, Alu-Iron should have a seen a big one

Here are the probes: https://i.ibb.co/t2vdc3t/probe-location.png

Since we are talking power transfer at interface should I also take a look at the stress along with the strain I'm already visualizing?

Thanks for your tips

>Thomas, > >everything in your model seems to be linear and you seem to be interested in the dynamic steady state. So you might consider to do a frequency domain study which is far more efficient and will directly provide complex quantities. > >Cheers >Edgar Yes the frequency study is definitely the end game but I wanted to validate the behaviour at the interfaces in the first place at to more easily targets errors in my simulation, since I'm rather new to COMSOL. I've improved the probing points and put them closer to the interface and it seems I don't get any interface effect so far, as seen below where it seems I'm back at the low freq / big wavelength point: ![https://i.ibb.co/YN51TP8/close-probes.png](https://i.ibb.co/YN51TP8/close-probes.png) What was I expected? decayed magnitude depending on the difference of impedance at the interface. Alu-Alu should have seen no difference, Alu-Iron should have a seen a big one Here are the probes: ![https://i.ibb.co/t2vdc3t/probe-location.png](https://i.ibb.co/T2vDc3T/probe-location.png) Since we are talking power transfer at interface should I also take a look at the stress along with the strain I'm already visualizing? Thanks for your tips

Note that while COMSOL employees may participate in the discussion forum, COMSOL® software users who are on-subscription should submit their questions via the Support Center for a more comprehensive response from the Technical Support team.