I thought the defining aspect of a dynamic rollover was that you reached a position where there is not enough cyclic control to allow you to recover from the situation - no matter how quickly or far you push the cyclic away from the roll, you will not be able to stop the rolling motion.
The key part to understanding this is to think about the way the thrust of the rotor is acting when you lean the helicopter over. Most of the thrust will be acting upwards, but some will be acting sideways. If this is combined with the force on the skids from the ground, which is acting in the oppositte direction trying to keep the helicopter from sliding, then a rolling moment will be generated that tries to tip the helicopter over. Note that although a rock or a tie down chain will assist this process, it can happen without these.
Now beyond a certain point, there is not enough cyclic control to counteract this rolling moment. As Nick mentioned, because an articulated head provides positive cyclic control even in a zero lift situation it is less suceptible to this effect (you don't run out of control as quickly).
Note in particular, that the two instictive reactions that a pilot might have - use the cyclic to correct the roll, or pull collective to lift off - are not useful in this situation. The cyclic input can't correct the situation, and because the rotor's lift is what is causing the problem in the first place, pulling collective only makes it worse.
If you really want to get a good "feel" for the effect, buy one of the cheap fixed pitch micro electric RC helicopters. You can get it tipping over and play around with the dynamic rollover effect. Also, the blades don't tend to break, and they aren't anywhere near as expensive to replace as real ones.
Daniel