Member Since: July 3, 2009

Country: United States

  • I was called in as a consultant to diagnose random resets of a custom engine control unit for a diesel to natural gas conversion kit on a commercial generator. The genset was housed in a storage container like you see on the back of a semi truck. It had a large diesel motor and our system intercepted the existing control module signals to/from the engine sensors and control systems. When the conditions were acceptable, our system would force their control module to minimize the amount of diesel being injected while operating our own natural gas solenoids. The diesel at that point was basically only used to initiate a pilot ignition while most of the combustion was natural gas. The engineers were experiencing difficult-to-reproduce failures which appeared to be ECU resets (on our control module). One failure was apparently quite exciting as it caused a large backfire which blew the doors of the storage container open. I verified all of the wiring and signals looked correct and we were about ready to go into the firmware looking for a culprit when somebody asked if the issue could be caused by the fact that the microcontroller was in a socket instead of soldered down (this was an early prototype unit). Once we moved to a soldered-down version of the ECU the problem was solved. The vibrations from the engine must have been enough to cause the chip to lose electrical contact from time to time. Later on we also found another issue where wires to a solenoid were run in a wiring harness next to sensitive sensor outputs. The solenoid was being driven by a MOSFET switching a +15VDC PWM signal, which made for some nice noise glitches. When the microcontroller happened to sample the ADC inputs at just the wrong moment, bad things happened. We solved the problem with some series resistance and ferrite beads. Sharp edges bad, yo.

  • I like how you were wearing your Colorado safety shoes. We have similar protective equipment in California which I utilize often.

  • Keep in mind that the datasheet specs the minimum gap between 2 conductors as 0.4mm. Conductors spaced closer together than 0.4mm may short together. The Atmega328 in TQFP shown in the video has pin to pin gap of 0.35mm to 0.5mm from Atmel’s datasheet.

  • Very cool board, but I’m not sure that this will work reliably with 1.8V devices. From the referenced Philips application note the low voltage should be at least 1V higher than the threshold Vgs value of the MOSFET. The BSS138 has a maximum Vgs threshold of 1.5V (1.3V typical), so this circuit should work down to 2.3V, but 2.5V is safer. At 1.8V there may still be enough juice to turn the MOSFET on, but I’m not sure I would trust the results.

No public wish lists :(