QUESTION:
Results from the input file (mirrror_static.inp) for MEMES Training Workshop #5 do not look like those shown in the notes (msm81_ws05.ppt), slide 17).
ANSWER:
The plots are actually very similar up to 56 volts. The solution no longer converges (or doesn't get as far as it once did). The reason why the solution crashes earlier than in previous releases is currently being investigated. In the meantime, try the input file attached to this SOLUTION (mirror_xsient.inp). This is a modified version of the input file supplied with workshop 5 that performs a transient analysis rather than steady state.
Notes from the author of this solution: This is almost ALWAYS the way I solve one of these problems. I do transient analyses to overcome divergence that occurs when structure goes through snap-down.
I did a modal analysis of the beam and mirror elements and determined that the first fundamental frequency was around 400kHz. I then did a transient analysis, ramping voltage to the final value with T_final = 250*T, where T = period = 1/400000 s.
Other changes:
Use default convergence criteria
NSUB,5,10000,5
It can still be difficult to achieve a converged solution for problems like this. Although not included in this example, I sometimes also add damping.
I realize this is NOT consistent with the procedure in the notes but I find doing transient analyses VERY helpful.
|
QUESTION: Results from the input file (mirrror_static.inp) for MEMES Training Workshop #5 do not look like those shown in the notes (msm81_ws05.ppt), slide 17). ANSWER: The plots are actually very similar up to 56 volts. The solution no longer converges (or doesn't get as far as it once did). The reason why the solution crashes earlier than in previous releases is currently being investigated. In the meantime, try the input file attached to this SOLUTION (mirror_xsient.inp). This is a modified version of the input file supplied with workshop 5 that performs a transient analysis rather than steady state. Notes from the author of this solution: This is almost ALWAYS the way I solve one of these problems. I do transient analyses to overcome divergence that occurs when structure goes through snap-down. I did a modal analysis of the beam and mirror elements and determined that the first fundamental frequency was around 400kHz. I then did a transient analysis, ramping voltage to the final value with T_final = 250*T, where T = period = 1/400000 s. Other changes: Use default convergence criteria NSUB,5,10000,5 It can still be difficult to achieve a converged solution for problems like this. Although not included in this example, I sometimes also add damping. I realize this is NOT consistent with the procedure in the notes but I find doing transient analyses VERY helpful. |
||
|