Application board damaged

In the application board, the chip under the control board labeled J1 is burnt and damaged. I need to know the actual cause of this so that I don't repeat the same mistake. Before the damage, following was done:

- Attached 4 servo motors to control the robotic arm

- Powered servo with external 5V

- Connected USB to run the software for positioning servo motors at pre-defined values

- Servos are powered on for a while to keep the arm in a certain position

- When he software was tuned off, servo power was brought to rest and the arm fell down. Very soon, the USB was pulled out, and external power was disconnected.

- Turned on the application board again, but didn't see the LEDs turned on, but got burning smell. At this point, the board was broken. It was noticed that the chip labeled J1 (D20) was split into half.

What could be the potential reason for that?

D20 is a voltage suppressor diode on the 3.3V line and if it got cooked, that means there was more than 3.3V applied on there for an extended period of time.  What kind of load were you driving with your servos?  

Usually broken diodes are fairly simple to replace - if you'd like to send the board in for repair we can quote on that, or we can provide you with a part number if you'd like to try it yourself.

We were driving the weight of an arm.  There are 4 servos connected to the micro-controller.

When we unplugged the USB port, we didn't turn off the servos.  We are also using the external +5 V.  Is that the cause of the problem?

Is your 5v source regulated?  If not, that could definitely cause some trouble when driving a load.

When you say "the weight of an arm", what kind of arm do you mean?

I am not sure if 5V source is regulated.  It is taken from a computer power supply.  But that power supply is quite large and should be able to drive the load.

We were driving the servos on this arm:

http://www.crustcrawler.com/products/arm6.php?prod=11

I still think because we unplugged the USB, either cause Ldi/dt discharge or something with external supply caused the controller to be borken.  Note that we have driven this load before with +9 V supply.  And no problem there.  Something is wrong with external suply or something.