One more FET as a switch question

Thanks all for being so helpful with my question last night and this morning. Here is one last one before I start building.

I have this cool little key switch that my kid found and we want that to be our main power. I found what I think is the same part online and I think maybe it can just barely handle the power on our robot. We're right at the rating and that is assuming I'm not mistaken about what part it is and that's a 50:50.

So I don't want to push everything through that switch, but it has to cut the main ground to the battery, again that's 7 - 18V. The bot pulls somewhere in the neighborhood of 10-12A at the most at 7V and far less at 18V. Probably in the neighborhood of 80W if everything is locked up at full stall current. Most of the time it will be far far less than that, but those are the worst case with a little safety built in numbers that I'm working towards.

I have a relay that I can use for sure. But I wondered if a transistor solution might be nicer. I have a few NFET transistors laying around from other projects.

I was thinking something like this:

I have this huge transistor with a big honking heat sink on it. I know it can take the current:
FCA47N60F - 47A - 600V. It may be way overkill. Even so I prefer overkill to buying something else.

But I also have RFD3055LE and AOI514 but those are logic level and I've got full battery voltage here so I shouldn't need that right?

Or do I?

It's another case of if I can build with what I got. I've got all kinds of resistors and caps and stuff here. But that's about all I got for FETs. If I can't do it with this then I'll just use the relay.

If it's a bad idea for whatever reason, please school me. That's why I came here today was to learn about switching power. Consider this me turning in my homework.

Allow me to revise those numbers a bit. I'm looking at it now and it's probably more like 110W at the worst case. I've got two motors that combined can pull 10A at 9V at full stall so that's 90, plus I've got 10 small servos in there.

Still nothing compared to the rating on that superfet and it can screw right to the side of the robot chassis as a heat sink.

No clue what you’re guessing at no idea what motor you’re using or any relevant details. My idea of robots can pull way more current than that. Are we talking table top ones?

wolframore:
No clue what you’re guessing at no idea what motor you’re using or any relevant details. My idea of robots can pull way more current than that. Are we talking table top ones?

No a little remote controlled tank. It's not very big. I think the one transistor is so much overkill that I'm not too worried about actually calculating it.

I do know what the motors pull, I looked that up. I know what the servos use. I've added a bit.

I'm not too worried about the power, it's the circuit I'm looking at. I think any of those NFETs would be more than able to handle the currents as long as they can tolerate the voltages.

If the lowest voltage you switch is 7V, you probably need a logic level one. The non-logic level FETs usually want 10V to be fully open.

Do check the maximum rating of the gate-source voltage. E.g the IRL540 can handle no more than 10V, the IRLZ44N is limited to 16V. Your 18V would destroy both. A zener can limit this voltage. The higher the voltage at the gate, the lower the on resistance.

Do also check the on resistance. The IRLZ44N is a fairly decent 22 mOhm at 5V on the gate, but at 12A that's over 3W of heat dissipation, that needs a good heat sink! An easy way to reduce this is to use two MOSFETs in parallel. That cuts the total heat dissipation in half, and per device to 1/4, some 800 mW, so now just a little heat sinking is needed.

Let's see. The AOI514 has Vgs limited to +-20V. So I'm OK there. On resistance at 4.5Vgs is 8.5mOhms. So at 12A I'm at 1.2W. I don't have a frame of reference for how hot that will make it. I could certainly double up and use two. Would you can look at the datasheet and tell me what the power dissipation numbers mean. But it kind of sounds like I can just use that.

At the top of the datasheet:

R DS(ON) (at V GS = 4.5V) < 11.9mΩ

That’s the “max” number as given in the data sheet, and what you should be calculating with. Basically a worst case. Then the heat dissipation at 12A is 1.7W. No serious problem. There are small clip-on heat sinks available for the TO251A package, which I assume you will want to use. For the DPAK consider a stick-on heat sink on the back of the PCB, with a good number of thermal vias. That’ll also do the job.

This 1.7W is really a worst case scenario for you: a part with on resistance at the top of the tolerance, and an exceptionally low gate voltage. Most of the time your gate voltage will be significantly higher than 4.5V, as your battery starts at 7V. At 10V the on resistance is about half the value at 4.5V, and of course that means less heat. For this one I don’t see the need to double up. Just add a small heat sink and make sure there’s some ventilation to get rid of the heat. In reality you probably have less than 1W dissipated there.

The part can handle 46A so that should be OK even with motors stalled for a moment. Of course in case you have 30-40A flowing you need to switch off power soon as otherwise it will go up in smoke :slight_smile:

Delta_G:
I have this huge transistor with a big honking heat sink on it. I know it can take the current:
FCA47N60F - 47A - 600V. It may be way overkill. Even so I prefer overkill to buying something else.

It's certainly not overkill, its fairly inappropriate. 600V MOSFETs have very high on-resistances compared to lower voltages - this one is 62 milliohms on-resistance which means about 10W dissipation for 12A, requiring
a large heatsink

An equivalent modern high-performance 30V device might be 1 or 2 milliohms, and not even require a
heatsink.

Even a pretty standard older 30V device would be about 10 to 15 milliohms dissipating a couple of watts and
get away with a small heatsink.

MarkT:
It's certainly not overkill, its fairly inappropriate. 600V MOSFETs have very high on-resistances compared to lower voltages - this one is 62 milliohms on-resistance which means about 10W dissipation for 12A, requiring
a large heatsink

An equivalent modern high-performance 30V device might be 1 or 2 milliohms, and not even require a
heatsink.

Even a pretty standard older 30V device would be about 10 to 15 milliohms dissipating a couple of watts and
get away with a small heatsink.

And this answer and the one before is the reason I ask here before I build. Thanks guys. I owe somebody some code help. Hit me up when things get complex in C++. I'm so much better on the code side of things.

Thank you so much guys. The AOI514 looks like it is going to be the winner.