The problem may be due to the discontinuities introduced by the
switches. The DAE solver assumes the system to be at least continuuos.
It should be better to control the swith, with a discrete time signal
instead of the step blocks with are continuous time.
Using discret time control insures the solver is stopped when the event
arises and restart just after.
You can generate a discrete command for the switch using a selector
block and an "event at time" block.
Serge Steer
Le 02/06/2013 05:04, dragaoerrante a écrit :
Hi people. I started using Scilab just some moths ago, so I'm not well
experienced on it. I'm trying to simulate a PID control system to be used on
a DC-DC buck converter. The controller and the converter models seems to be
ok, but when I run a simulation to increase or decrease load on the fly the
simulations crashes even before any load has beend added or removed.
The error message says: IDA_ERR_FAIL: Error test failures occured too many
times (= MXTEF = 10) during one step or occured with |h| = h_min
I also noticed that it popped up on a PWM block that is as simple as a
sawtooth generator and a comparator. This block have beeing used several
times without any issue.
Another thing I have to say is that this problem stops occuring if a
simulate the system with a steady load value.
Before running the XCOS simulation file it is necessary to run the sce file.
I would appeciate so much if someone could help me with this.
Best regards,
Leo
Buck_5V_01.zcos
<http://mailinglists.scilab.org/file/n4026755/Buck_5V_01.zcos>
Modelo_Buck_PID_5V_SL_02.sce
<http://mailinglists.scilab.org/file/n4026755/Modelo_Buck_PID_5V_SL_02.sce>
--
View this message in context:
http://mailinglists.scilab.org/Xcos-simulation-IDA-ERR-FAIL-tp4026755.html
Sent from the Scilab users - Mailing Lists Archives mailing list archive at
Nabble.com.
_______________________________________________
users mailing list
users@lists.scilab.org
http://lists.scilab.org/mailman/listinfo/users
_______________________________________________
users mailing list
users@lists.scilab.org
http://lists.scilab.org/mailman/listinfo/users
