Why is the (second) if statement in the for loop not working?

Hi, after many tries i still not understand why the second if statement inside the for loop isn't working. The picture shows the result of a simulation which should match with the result of my matlab code.

[Image Analyst]

The code doesn't run. In this line:

i_Last(i) = 0;

what is (the badly-named) "i"?

[pm96]

That's a mistake. I'm going to edit the post and delete it.

[David Goodmanson]

HI pm96,

the second if statement doesn't work because for positive i_Amp and positive tau_Last which you have, the function

i_Last = i_0+i_Amp*(1-exp(-(t/tau_Last)));

is an increasing function of t. In the second if statement you are checking for a decreasing function. So it always fails.

If you want to reproduce the curve it looks like you will have to cycle between

-A1*exp(-(t/tau_Last)) + b1;   % increasing
and
 A2*exp(-(t/tau_Last)) + b2;   % decreasing

with appropriate A1 A2 b1 b2.

[pm96]

Hi David,

thx for your explanation. Meanwhile i made some changes and i tried to implement your improvements. But got stuck again. Now my signal i_Last just starts to swing after it reaches i_og for the first time.

[the cyclist]

I see two problems:

First: If you zoom in to the early part of your plot (and plot your data with '.-' to show the individual points), you'll see that your first "correction" (when i==5) to prevent going past the upper bound actually puts the new point below the lower bound.

Second: You always do your conditional tests against the "uncorrected" trajectory. After i==5, every point is above the upper bound, so that is the clause that is entered.

I haven't looked how to correct these flaws.

[pm96]

Hi cyclist,

thx. I actually recognized that and i'm still trying to find a solution for that. So far without success. Hope to find a solution soon or maybe one of you know how to solve that.

Answer 1

[the cyclist]

I don't know if this is perfectly correct, but it is certainly closer:

Where you have

i_Last(i) = Faktor*(i_0+(i_Amp*(1-exp(-(dt/tau_Last)))));

I instead put

i_Last(i) = (i_0+Faktor*(i_Amp*(1-exp(-(dt/tau_Last)))));
i_Last(i+1) = (i_Last(i)+Faktor*(i_Amp*(1-exp(-(dt/tau_Last)))));

This does two things:

1. calculates i_Last forward one step, so that the if statement will calculate against the "corrected" trajectory
2. does not multiply the current value of i_Last by Faktor (because that's just the current position)

[pm96]

I tried to implement your idea with i_Last(i + 1) but i got the error, that i can't plot (t,i_Last) because the vectors of t and i_Last aren't the same length (i_Last is 1x202 and t is 1x201). Tried to fixed that but still not working.

[pm96]

ok, i solved that. it was the wrong condition for the length of the for-loop. it should be T1*SR not T1*SR+1. And SR should be > 10000 so the signal gets more clear. Thx all for your help.

[the cyclist]

Oh, right, I forgot that I had also adjusted the for loop. Glad it worked out.