During the Conference, Ryan Fobel’s DropBot caught fire, briefly. There was no outward indication of why this happened, so we arranged a session to lift the lid on the broken system and see if we could figure out where the problem originated.
Ryan started the session by speculating where the problem occurred. His best guess was that it happened somewhere in the circuit that generates the high voltage necessary to drive the electrodes that move the droplets around. This voltage starts from a source of 12V DC, and is driven up to ~300V AC [confirm this?].
Upon opening the case, the first thing that we noticed was the insulation over the wiring from the 12V power source was almost completely burned off. Clearly, this was the primary source of the smoke that was emitted by the DropBot.
Ryan then removed the PCB card that has the voltage amplification circuit, and upon inspection we discovered that a power mosfet that is critical for this amplification was blackened, and had lifted off of it’s SMT pad. This was the best indication that we were able to find for the source of the problem. The mosfet in question is used by an integrated circuit to ‘pump’ up the voltage. Essentially, the IC sends a square wave to the gate of the mosfet at 10KHz [confirm this] and voltage is amplified by the following circuitry. The theory is that somehow, the device was ‘stuck on’ or otherwise forced to allow more current that it was rated to handle.
This theory has an issue, in that a part of the circuit is fed back into the IC, which monitors the actual current that is going through the mosfet. If the current through the mosfet is greater than a reasonable threshold, the system shuts itself off. In other words, this mosfet should not have failed because there is a fail-safe built into the circuit. The IC in question showed no signs of burning or other damage from visual inspection. The only other possible explanation for this is that the same point in the circuit which is sensed by the amplifying IC is also attached to one of the pins on the microcontroller that is driving the dropbot logic. This additional connection was made by Ryan as a redundant check on this point in the circuit as an additional safeguard. Could the connection of another device to this point in the circuit have compromised the sensing ability of the IC?
[Someone has pictures of the hardware? Andy?]