(I am considering stacking a couple of resistors and a LED in that spot for each channel to provide a visual status of individual gate activation. It will be a useful debugging tool, and besides - I like flashy lights.
)
Nice! Your cab will have quite a light show on the inside. I wish there were room to put in optional pads for these on the board to make it easier to add those, but I don't think there's any way I could fit them in.
Adding series gate resistors to the MOSFETs will slow their slew rate slightly, and help moderate any ringing / transients that may occur at switching.
That's the exact reason I have in mind. I haven't seen any overt effects from that ringing in my testing, but I'll bet it's there, and there's at least one good reason to suppress it even if it's not causing obvious problems: it apparently can cause a lot of EMI emission if left unchecked. I'm running the PWM clock at 500 kHz, which provides very flicker-free lighting and buttery smooth fades, but it's fast enough to be in the radio spectrum.
(There's also a lot of dogma in the Arduino world about needing them to limit the initial surge of current through the Arduino GPIO pin to the gate as it charges or discharges on a switch edge, but after doing the math I think it's mostly superstition. The gate charge in most FETs is so tiny that I don't think enough the overall power dissipation is significant to the GPIO circuitry. That's utterly irrelevant for us in any case since the MOSFET gates aren't anywhere near the GPIO pins - we have optos and TLC5940's between here and there.)
I have not researched the opto's characteristics nor the PWM waveforms that drive them. I know that in high speed applications, a gate-ground resistor allows the gate voltage to bleed off much faster and yield a more "controlled" transition to off. It looks like the 1k resistors between gate and ground are targeted to do just that.
And in this case, I think the gate-ground resistors are actually required to pull the gates to ground when the outputs are off, because otherwise they'd just float. In Arduino applications with the gate connected directly to a GPIO pin, the pull-downs would be optional because the GPIO pin would actively assert 0V when at logic low, but with the optos there'd be nothing to pull to ground when the opto isn't conducting. So the gate would just stay charged and would most likely stay stuck on indefinitely.
I have to plead ignorance on some aspects of MOSFET operation. I know that most - if not all - MOSFET designs include an intrinsic body diode between source and drain. I understand its primary purpose is to prevent an unstable point in the MOSFETs operational range, and is not so much there as a protective device. What I don't know is what kind of transient it can absorb or if it needs an external device to keep it stable. I expect in this application, with power demands so far below operating limits, this paragraph is moot.
I haven't studied them in very much depth either, but I'm pretty sure the integrated diode won't offer much protection against flyback current. I think you'll really want to use a separate diode for each inductive device. Even separately from protecting the MOSFETs, pin cab builders routinely have problems with spurious logic noise on their PC motherboards and other USB devices throughout the system if they don't have good flyback protection on each contactor, motor, relay, etc. This is one of the vexing problems that comes up over and over, and diodes are usually the answer.
I actually thought about putting flyback diodes on the boards themselves, but there are too many problems with that. The space budget is pretty much a showstopper for that, but even if there were space, connecting the voltage supply would be a big problem. Each device's diode has to attach between power and ground for the particular voltage source the device is using. But the outputs are specifically designed to be used with a mix of voltage supplies - you might want 24V for the contactors, 35V or 40V for the knocker, 9V for the shaker, 12V for the gear motor, 5V for lamps, etc. To integrate the diodes on the boards, each output would have to be assigned to a particular power supply voltage, which would have reduced the flexibility of the setup. This is the exact reason almost no one takes advantage of the integrated diodes on the LedWiz. Plus I think there's something to be said for putting the diodes physically close to the device so that the flyback current is dissipated in the shortest possible length of wire, to minimize antenna effects.
Edited by mjr, 01 February 2016 - 07:58 PM.
Top
are all trademarks of VPFORUMS.