Second Life Physics Scripting - Pinball #4

I need to make the ball a little smarter when it got stuck on the lower slope. I got a little crazy and added something to try and keep it at a constant velocity of 0.42M (the size of the ball), since anything faster than this can cause it to move through a wall across a frame rate. This is the final script.

    float movebump = 0.2;
    float maxspeed = 0.42;
    float speeddiff = 0.02;
    // http://lslwiki.net/lslwiki/wakka.php?wakka=llApplyImpulse
    default
    {
      state_entry()
      {
        llSetTimerEvent(1.0); // generate a timer event every 1 second
      }
       
      timer()
      {
        //vector ra = <llFrand(movebump),llFrand(movebump),llFrand(movebump)>;
        //llOwnerSay("ball moving "+(string)ra);
        //llOwnerSay("ball moving mag "+(string)llVecMag(ra));
        vector vel = llGetVel();
        float velmag = llVecMag(vel);

        // check for anything above or below the maximum velocity
        if ( (velmag>0) && ((velmag>maxspeed+speeddiff) || (velmag<maxspeed-speeddiff)))
        {
            //llOwnerSay("oldvelmag = "+(string)llVecMag(vel));
            // need to slow or speed up the ball
             float magdev = maxspeed / velmag;
             vel = vel * magdev;

             // take that new velocity and apply it to the ball
             llApplyImpulse(llGetMass()*vel,FALSE);        
             //llOwnerSay("newvelmag = "+(string)llVecMag(vel));
        }
        else if (velmag==0)
        {
            // need to give it a random bump!
            vector ra = <llFrand(movebump),llFrand(movebump),llFrand(movebump)>;
            //llOwnerSay("ball moving "+(string)ra);
            llOwnerSay("ball vel 0 move bump = "+(string)llVecMag(ra));
            // give the ball a little wiggle
            llPushObject(llGetKey(), ra, <0,0,0>, FALSE);
        }
        //llOwnerSay("ball vel "+(string)llGetVel());
        // give the ball a little wiggle
        //llPushObject(llGetKey(), ra, <0,0,0>, FALSE);       
      }
    }

I think there are still problems with it, because it should have been a problem in that it would never slow down and move down the board in the way I "thought" I coded it, but it does work, but it also still gets stuck briefly. I had to add the check for 0 velocity at the end because it was getting divide by zero and that else (velmag==0) piece of code was all I really needed in the first place. I'll have to revisit it, but at least it doesn't get stuck indefinitely any longer, but it still gets stuck a lot more than I'm willing to live with.