So, I'm getting my first ArduCopter dropped off a little later this week and I've ordered a separate switching BEC to prevent any brownouts coming from the ESC's BEC circuit. I'm a little paranoid, plus I like the efficacy of switching hardware more then linear. :)
The main problem with MOSFET's, however, is that when they fail they normally fail in the on or off position. This means that if the BEC blows the copter will crash, but there's a 50% chance of it also killing a chunk of the electronics. I can't just add a fuse because a sudden increase in voltage is unlikely to have a sudden increase in current, so the fuse won't blow.
The thought occurred to me that I can build a little circuit with a fuse and a TVS diode, so that if the voltage exceeds a certain point the excess voltage will be dumped out of the circuit. Ideally, this should trigger a BEC shutdown, but adding a fuse would likely be a good idea in the event the BEC doesn't shut down fast enough to prevent the diode (or wires) from burning up.
Does this sound like something that would be worthwhile to add, or since the copter is going to crash anyway, should I just not worry about it? :)
I have destroyed countless Buck converters, and the upper FET has universally failed open. Of course this may have been preceeded by failing short, but I've never had an oscilloscope attached at the right time to determine the exact sequence of events.
I think a TVS has merit. The fuse may not be necessary, as the converter should go into current limit.
The main reason I have a fuse in there is because if the BEC is truly hosed the over-current protection may not work. Plus, the TVS is going to heat up pretty fast if the BEC is dumping the battery into it, and I'd rather lose a $0.25 fuse then a $2.50 TVS.
I'm not totally sure a TVS is the best thing to use for this scenario, either. I'm not sure how it'll behave with what is basically a constant 10V+ load, since they're really designed for microsecond or less surges.
In my experience, well-designed Buck converters are incredibly reliable. Failure is generally associated with exceeding the ratings, either on the input or the output. If you arrange your protection so that ratings can never be exceeded, then short of implementing power supply redundancy you should have a reliable system.
I do not know whether output current limiting is implemented on all commercial BECs, so I guess it would be risky to presume it. When implemented, current limiting is usually reliable, as failure of the shunt is invariably open, resulting in no output. The resistive element is often in series with the lower FET or diode, or indeed the intrinsic resistance of the lower FET itself. If the lower FET fails short then the output is clamped to ground anyway.
Of course you are quite right - TVS devices are only specified for transient conditions. But they can take a massive hit and survive. I am always a little nervous about 5c fuses blowing on a transient current surge and causing total vehicle loss.
I figure a 7.5A or 10A auto mini fuse should be alright. The BEC that I'm planning on using should switch into current limiting mode for anything above 7.0A, so if it's pumping out more then that something is horribly wrong. Plus, I can always add a couple of caps after the TVS to assist in any surges from the servos.
>The thought occurred to me that I can build a little circuit with a fuse
Google "SCR crowbar"
That'll take care of ya.
Yeah, that is a much more elegant solution to the problem. Thanks!
I haven't done any electrical work in awhile and I completely forgot about those.
My concern with fusing the main output is that potentially a non-essential item such as a servo could stall, blow the fuse and take out your vehicle. If you do want to use a fuse, it may be preferable to separate out power to non-essential items and fuse them separately.