In your first example,
>> dVdx(0,5E-6,[0.02]) I
note how the first input 0 which is the xval is not used in your calculation (with yval equal to %E-6 and cst equal to 0.02)
g = 4*cst(1) - (4*yval)/cst(1)^2 ;
In your second example,
>> dVdx(5E-6,0.02)
xval is 5E-6 and yval is 0.02 for
g = 4*xval - (4*yval)/xval^2 ;
But the variables and their values aren't the same as your first example. If we substitute then we get
(1) g = 4*0.02 - (4*5E-6)/0.02^2 = 0.03
(2) g = 4*5E-6 - (4*0.02)/5E-6^2 = -3.2e+09
So you are using different values for each calculation (2 vs 3 inputs) and so will get different answers.
