Prev: Inscribed triangle in delaunay trianglation mesh
Next: classregtree
From: Marlene Ferrier on 19 May 2010 04:07
Hi,

I am new to simulink and I am facing the following pb:

I am trying to model the displacement of a mass bouncing between soft shock stops after experiencing a strong and brief shock/acceleration.-quite similar to the "bouncing ball" demo available on the help except that the acceleration in a one time pulse rather than a constant (+ in my case, the mass is attached to a "frame" by 2 springs)

To model the shock pulse I multiply one 'up' ramp with one 'down' ramp shifted by the shock duration (about 1e-3s). The ramp functions need a "sample time" that I usually choose to be somewhere between 1e-6 and 5e-5s.

To calcultate the mass trajectory, I use 2 continuous time integrators, as well as a retroaction to model damping, springs ...
All sample times are 'inherited'.

I played around with various sample times and find a significant discrepancy between the results. And the smallest sample time does not seem to give the best result.

I suspect some sampling mismatch but I am not quite sure

Any idea on what could go wrong would be deeply appreciated

Thanks
From: Marlene on 19 May 2010 07:51
Quick follow up on my previous post.

I set the sample time to 0 in my source (ramp signal) & I don't get those crazy discontinuities anymore..I believe this is related to the continuous time integrator I used in my simulink model.





"Marlene Ferrier" <marlene.ferrier(a)gmail.com> wrote in message <ht0676$oq9$1(a)fred.mathworks.com>...
> Hi,
>
> I am new to simulink and I am facing the following pb:
>
> I am trying to model the displacement of a mass bouncing between soft shock stops after experiencing a strong and brief shock/acceleration.-quite similar to the "bouncing ball" demo available on the help except that the acceleration in a one time pulse rather than a constant (+ in my case, the mass is attached to a "frame" by 2 springs)
>
> To model the shock pulse I multiply one 'up' ramp with one 'down' ramp shifted by the shock duration (about 1e-3s). The ramp functions need a "sample time" that I usually choose to be somewhere between 1e-6 and 5e-5s.
>
> To calcultate the mass trajectory, I use 2 continuous time integrators, as well as a retroaction to model damping, springs ...
> All sample times are 'inherited'.
>
> I played around with various sample times and find a significant discrepancy between the results. And the smallest sample time does not seem to give the best result.
>
> I suspect some sampling mismatch but I am not quite sure
>
> Any idea on what could go wrong would be deeply appreciated
>
> Thanks
 | 
Pages: 1
Prev: Inscribed triangle in delaunay trianglation mesh
Next: classregtree