ODE45 exact algorithm

Hi, iam trying to implemet ode45 algorithm. There is a good algorithm in a book named "Numerical analysis with matlab". I implemented the code and it gives me the same results as ode45 but the length of my solution at my algorithm and ode45 is differente also ode45 changes it step much more rapidly than in my algorithm and it seems to be more efficient . I changeg the tollerance, hmin and hmax

I am sure that the difference is in error corection. I have opened ode45 matlab code and i cant understand exactly how they adapt step size at every step.

I also opend matlab referanced at ode45 mathwork.

Can any body tell me how ode45 change h.

The algorithm I implemented is in attached files, maybe someone will find it usefull...

Thanks allot.