In article , Mike Young wrote: [....]
As others have said, it is the result of how the part is made.
In your case, the simple freewheeling diode should not be used. You need bidirectional control.
OR: You need to give up on this single wheel idea and just add another wheel and handle bars.
[....]
Unless the surface this is operating on is smooth and fairly level, you will need to be able to make pulses of power much greater than the average running power. You are going to have to have a higher voltage, bigger motor and bigger MOSFETs than you may at first suspect. The pulses of power are short enough that the heat sinking requirements are not so bad.
Your H bridge needs to be able to go to 4 states: "forwards", "backwards", "open" and "short".
When the bridge goes to the "open" state, the diodes across the MOSFETs will conduct briefly as the current in the motor comes to a stop.
In the "short" state turns on, lets say, both bottom transistors. This shorts out the terminals of the motor. This will create a strong braking effect. In many situations, pulsing this braking effect will be a better way to slow down than reversing the motor.
You also should place a small inductance in series with the motor. The iron in a DC motor isn't very good at high frequencies. Any high frequency power that goes to the motor will just end up as extra heat.