How many users of Octave and Scilab are there?

I see that Matlab/Simulink is very popular at universities.

How many of you are using Octave or Scilab open source products for simulation?

How well do they work?

Thanks,

Peter Way

Reply to
qfufsxbz
Loading thread data ...

I did use Scilab a bit. I tried using Scilab to solve a significant motion profile generator problem. I found that Scilab's debugging didn't work very well. HMI programming is awkward. I went back to C and C++. I haven't used the Scicos part which is what you really should be asking about since it directly competes with SimApp.

Peter Nachtwey

Reply to
pnachtwey

OTOH, I use Scilab all the time.

I stopped using Simulink the second time I had a big simulation break thanks to a MatLab upgrade, so I'm out of the habit now with Scilab.

The debugging has gotten miles better in recent versions, although the move to 5.0.x was huge (they migrated from Tcl to Java) and they're still shaking the bugs out.

Reply to
Tim Wescott

I didn't notice the migration to Java. I will look into it. That may restart my interest. Right now I am writing a hydraulic servo simulator in java. I had considered doing it in Scilab but switched to Java when my previous project didn't work too well.

Peter Nachtwey

Reply to
pnachtwey

Is simapp or Scilab widely used in control engineering? As far as I know, in China, we mostly use Matlab and Simuboth in both industry and academics. Scilab is used in some embedded systems to run some complicated algorithms.

Reply to
workaholic

I can get Matlab with all the toolboxes and extra bits I need to do control engineering with about the same amount of money as I could spend to get a nice, reliable, late model used small car. Then the next year I can get it all upgraded for the price of a major repair to the car.

I can get Scilab for the price of a download, it does nearly everything that Matlab does plus plenty that Matlab doesn't, and upgrades cost nothing.

Since I don't have a big corporation with more dollars than sense backing me, guess which one I use.

How much does a Matlab seat with all the control systems trimmings cost in China?

Reply to
Tim Wescott

But I think it is nearly impossible for one with no practical numeric computation knowledge and adetp programming skills to directly do simulation in C/C++, even with a numeric library... What will you do if you have to solve a stiff ODE problem?

Reply to
workaholic

There should be a price for ignorance. Pay to have it done, buy Matlab, or invest in yourself and learn what you need. I chose the latter. I had calculus and differential equations is college and a few terms of numerical analysis. That is all I needed. The rest is learned from the school of hard knocks.

Write a RK4 program is easy. Solving a system of first order linear and non-linear equations is easy. One should be able to do a simple RK4 in 15 minutes in C.

Reduce the sample intervals. The stiff equations are not as big a problem as what to do when a limit is hit. For instance an actuator hits the end of the cylinder or the pressure go to 0 and can't go lower. The limits are usually reached in the time between two samples. Then I divide the sample time into the time before and after the limit was hit. A more brainless approach is to divide the current sample time in to smaller equal sample times. RK4 is easy to restart after the leaving the limit.

There are much fancier methods than RK4 but RK4 is easy to make work well enough and efficiency is not a problem unless you are trying to do real time simulations.

What is funny is that I saw a page

formatting link
had some useful information that I could have used a few years ago. It would have helped. I learned on my own and can now easily determine when systems are unstable. It is a problem for us because we have a real time simulator built into our motion controller so engineers can test at their desk. We must limit the damping factor and natural frequency the customer can simulate so our simulator doesn't go unstable. This limit is rather high so it doesn't affect any practical applications. We don't have the extra time to play games with reducing the sample intervals.

Peter Nachtwey

Reply to
pnachtwey

Well, I have to admit that nowadays majority of engineers who cannot write numeric programs on their own, or if they have to, it requires a lot of time because they have never really tried it out when taking numeric analysis courses in universities, someone still cannot do that, though with a M.Eng degree. I don't know how many people like you are there in U.S.

Reply to
workaholic

PolyTech Forum website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.