Another Decimus patch

By | February 24, 2008

This one was pretty stupid really. The motor drivers are H-bridges made from complementary pairs of MOSFET (the Zetex ZXMC3A16DN8 (datasheet). These need a proper driver chip to ensure good turn-on and turn-off times because of the gate capacitance. for the driver, I used the common Maxim MAX4427 (datasheet). There are several alternative equivalents. However, instead of using the motor supply (battery positive) for VDD, I used the +5V logic supply … stupid.

IMG_2428 The reason this is stupid is that the 4427 driver is itself a complimentary MOSFET pair. VDD is connected to the source of the high-side transistor. The idea is that the driver chip can take the gates of the H-bridge pair all the way to ground and the battery voltage. By limiting the maximum voltage that the driver can provide to the H-bridge, the high-side transistor on the bridge was never really turned off and its gate voltage was always at least 2V. The result of that was that current could pass through one side of the bridge even without a load connected. All that energy ended up as heat and everything quickly got very hot. I only noticed when I could smell the first tantalising odour of escaping magic smoke. That made it doubly stupid because I had already seen that, with the motor driver active, the supply current had increased from 128mA to 230mA. So 100mA was passing through the bridges – even when the drive was at 10%. As the PWM duty cycle went up, the current increased and the chip got hotter…

IMG_2429 Now I have patched the board again to eliminate that problem. The motor drivers do not seem to have suffered any lasting damage but I may yet end up replacing them. Another easy fix but how much nicer it would have been if I had been paying attention when drawing out the circuit.

4 thoughts on “Another Decimus patch

  1. Augusto

    Hi,
    are you using flyback diode?
    Also you use two pwm for control the max4427 or feed one pwm on one input and use the other for control the direction?

  2. Peter Harrison Post author

    I am relying on the intrinsic diodes in the driver MOSFETS.
    The pwm scheme on this mouse uses one pin for direction and the other for variable pwm

  3. Peter Harrison Post author

    I would not like to say. The safe bet would be to add them probably.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.