EEVblog Electronics Community Forum
Electronics => Beginners => Topic started by: ITman496 on February 17, 2015, 09:28:20 am
-
So on a drone I'm making, it has two battery packs made with 18650 cells. Their C-rating isn't too spectacular, so I have two packs in parallel. In the interest of safety, each pack has a small computer in it monitoring various things, so if anything goes wrong, I can know about it.
What I want to do is make a circuit to isolate the battery pack in case it starts to dip, so the second battery doesn't drain into it and make whatever is going wrong even worse. Or more importantly, one pack can just barely supply enough rated current to land the drone, but not if its dumping into a shorting out second battery, then I'd crash.
This circuit obviously has to be very low resistance, and inherently hard to trigger. I'll be having a dedicated micro watching the battery data and decide if one should be isolated or not. (high temp, cell imbalance, etc)
The tricky bit is that upwards of a hundred amps has to flow, and I cant burn too much power. And then even tricker, current has to flow both ways. As I learned, in a multicopter esc, they tend to use the batteries as a current sink to engine brake against to slow the propellers down faster. I had one diode-d once and when it attempted to motor brake, it essentially instantly exploded because the voltage on the now isolated supply rail went through the roof.
So this is the circuit I designed. It uses two chunky p-fets, possibly four if I parallel two sets of these to half the resistance. It defaults to being on when there is voltage on the rail, and can be turned off by the microcontroller. So in the condition where it would turn off, the battery (left) would be lower voltage then the rail, (right)
Its annoying that all fets seem to have body diodes. I wish they didn't.
So, is there any issues with this design? This time I'm asking 'before' I send an osh park order off.. I learned my lesson from last time..
Thank you!
(https://dl.dropboxusercontent.com/u/4963246/octocopter/v2/IMG_20150217_042449_256.jpg)
-
As I understand FETs have reverse diodes because reverse voltages tend to kill them anyway. Might not be as true as I think.
I would have suggested diode OR'ing them until I read about your ESC braking >:D. Is there any particular reason you can't switch to some lipos? 18650s are indeed generally designed for capacity, not C rating, unless they're the more exotic LiFe and co from power tools.
Perhaps a couple of relays would be called for? They're power needs would be insignificant compared to the copter's draw.
-
You need diode OR ing, or if the drop is too much, you need an ORing controller. There are ASICs for this, it is unnecessary to do this stuff with a micro. Check power management from TI and Linear.
-
I am using 18650's to get a massive flight time. Their power density is near double what normal lipos are. So I have enormous packs of them to make the C-rating work out.
I was looking at relays but they are just too heavy for the current flow, I think? Do you know of one that can be lightweight and flow at worst case 100A? (which would occur if it was trying to land with one battery isolated)
Does reverse voltage on a fet kill it? Or does this setup mitigate that?
Diode oring wouldn't allow me to esc brake though, would it?
-
Out of curiosity: what are your flight times like?
Yes plain diode OR'ing will have the problem you describe. NANBlog mentioned power-managment/OR-controller ASICs which may be what you are after.
Reverse voltages are fine on FETs as long as they have reverse diodes to get rid of them AND you're not going to cook them with the power. If the latter, place high-power diodes in parallel with the FET, facing the same way as the internal diode. I presume low-voltage schottky ones would be better so you can guarantee the current prefers them over the FET.
I've only dealt with lower current relays, but I presume you can get better ones that are not too heavy. Most will be at this current level (industrial situations?) but there ought to be better options.
-
Flight times are on the order of 30-40 minutes of actual flight, not just stationary hovering, while carrying a 4k camera and all the payload, gimbals, etc. The battery packs are 6S 11P with NCR18650B's. and 2 of those in parallel, making a 6s 22p pack.
Does my double fet, backwards to each other arrangement combat the frying of fets from reverse voltage? Or do I need additional diodes?
-
all fets have reverse diodes because it isn't a separate diode, it is side effect of how the fet is made. for many fets that reverse diode is rated for the same current as the fet.
You can do it with a single fet per battery, use a Pchannel fet "in reverse" on the positive side of each battery, the reverse diode will do diode and when you pull the gate low to turn on the fet it'll conduct equally well in both directions
If the batteries don't have a common negative you can use an Nchannel fet in each negative in the same way, Nchannel fets are generally lower rdson
roughly something like this
http://imgur.com/GqfCSi5 (http://imgur.com/GqfCSi5)
-
I think your circuit would work, but it's worth checking what real-life losses are. Build it and test it with something small like a cheap brushed motor, then try it on your craft if you don't encounter any bugs.
. for many fets that reverse diode is rated for the same current as the fet.
Make sure to check, especially if the FETs are expensive. If the data-sheet does not have the details I wouldn't trust them without external diodes, but it's up to you.
EDIT: Kudos for 30-40 minute flight time with a payload! Pictures or it didn't happen.
-
Once my refurbishment/mods are done, you'll have your pictures. ;)
I am going to breadboard some of the stuff before getting a pcb made. This is the fet I want to use, if you are curious.
http://www.newark.com/vishay-siliconix/sum110p04-05-e3/mosfet-p-channel-40v-110a-to-263/dp/96K5342 (http://www.newark.com/vishay-siliconix/sum110p04-05-e3/mosfet-p-channel-40v-110a-to-263/dp/96K5342)
Here is a cad model in the meantime.
(https://dl.dropboxusercontent.com/u/4963246/octocopter/v2/nearfinal3.jpg)
-
Also be wary: if one pack fails and you disconnect it you start putting the remaining pack(s) under more stress, making them more likely to go too.
-
One pack does have enough of a C rating to fly the drone, it just really abuses it and the flight time would be rather sad. But it's enough to land, hopefully!