Mosfet Info Needed for Submarine Project

[AUTO XX]

Hey guys, it's been a looong time since I posted anything but I have run into a snag I can't seem to figure out.

I'm making a propulsion system which basically amounts to strapping a couple of trolling motors to my legs and hoping it works...
Each motor is 30lb thrust max, ~30A draw but I want to run them with a waterproof reed switch between my knees (need my hands free for the speargun or salvage diving).
The switch of course can't take any real current so a couple of mosfets (two is better for me because of redundancy, it could be a long swim back to the boat...) rated for the current should be okay eh?

I figure something that can handle >13v on the gate would be ideal so I don't have to have a BEC that could fail and I could just use the 3S lipos (12.6v max) to run both the gate and the motors.

http://www.irf.com/product-info/datasheets/data/irfp4232pbf.pdf
I found these ones about a week ago, way overrated on the voltage but they can also handle ~20V on the gate. Would something like this work?

My sled of batteries will be 8X5000mAh lipo batteries paralleled (40Ah X ~12v) so ~40 minutes at full draw. I will also be including the original speed controllers in the powerpack so I can adjust it for 5-30lb as needed but they will be in the sealed pack so adjusting on the fly would be difficult.

The next option is to use a couple of brushed motor ESCs and then sealed switches to varied resistors using a servo controller.

I'm still debating whether I should print the housings in PLA/ABS or just fab them up out of aluminum/stainless, I just got a TIG so either way is good practice.
This will mostly be used for free diving (I can stay down for ~6 minutes at 100' so no worries on the duration) but could be handy for SCUBA if it works well.

 

[Sigurthr]

What you want is an ultra low RdsON n channel enhancement fet with a VgsMAX of >15V. There are tons. Might I recommend Fairchild, they tend to have better RdsON selection.

 

[AUTO XX]

Would it be able to handle >30A? I don't mind paralelling a few of 'em
The one I listed will do 42A constant or 60A max but I think I read somewhere that it needs a Vmin of 80V.

 

[MarioMaster]

There's a simple alternative if you want to use cheaper mosfets. Ideally as others have said, you want the lowest RDS(on) value possible, with plenty of current overhead.

As for controlling gate voltage, you can use a simple resistor voltage divider to turn on the mosfet with your main battery voltage. Voltage on the gate is the ratio between the two resistors. You'd get half your battery voltage in this instance.

     Reed Sw   10K                10K
+Vbat───o/ o──/\/\/\/─────┬─────/\/\/\/─────Gnd
                          │
Gate──────────────────────┘

 

[Sigurthr]

Yep, I'd use the ascii circuit Mario posted, but replace the right hand resistor with 100k. This will put 90% of the voltage on the gate, which is right around 10V, where you want it for a fully on fet.

I searched through my Fairchild Semiconductor catalog for you and found some nice ones you can choose from that are inexpensive and ideal for your application.

FDH055N15A − N-Channel PowerTrench® MOSFET 150V, 167A, 5.9mΩ
FDH055N15A PDF Datasheet N-Channel PowerTrench® MOSFET 150V, 167A, 5.9m?

FDP020N06B − N-Channel PowerTrench® MOSFET 60V, 313A, 2mΩ
FDP020N06B PDF Datasheet N-Channel PowerTrench® MOSFET 60V, 313A, 2m?

 

[AUTO XX]

Damn you guys are good. Well I'll order them before I head out camping. >160A lol, THAT is plenty of overhead...

I'll waterproof the whole works but leave the heatsink exposed to the surrounding water, should be able to dump a fair amount of heat lol

 

[Sigurthr]

The 2mOhm FET would dissipate only 1.8Watts under continuous 30A load. The 5.9mOhm fet; only 5.4W of heat. So you shouldn't need a ton of heatsinking, especially if you're going to have it connected to the chassis.

 

[AUTO XX]

The 60v rated fet looks too good to be true, damn that's a huge max current...
There won't be any "chassis" to mount it to, it will just be floating around in a backpack, likely mounted in a 3D printed box that water can flow through just for durability sake. With that little heat though it really wouldn't be bad to mount it in with the batteries so at least it is dummy proof and a little more streamlined... The flowing water (even through the silicone "bag" should pull any heat away in no time anyway. Needs further testing

Thanks a million for the info, I'll post the completed project and hopefully footage of it in use *even if it is of me skipping across the water and slamming into the far bank*.
With 60lbs of thrust I'm hoping to be able to get up on step :crackup:

*Edit - Holy Hera! ~$50 for shipping for 4 chips? WTF over? The fets are only ~$20!!!

Got a line on a different company? Preferably one that has some (not insane) shipping options other than fedex...

Edit#2 - Found these on digikey, lower amp specs but still way overkill, would these work? http://www.digikey.ca/product-detail/en/FDP020N06B_F102/FDP020N06B_F102-ND/3163340

 

[Sigurthr]

You can request a sample of up to 10 chips for free (free shipping) if you register on Fairchild site with a business email. (Or a dispostable.com email, shhh).

But use the MOSFET model number in digikey or mouser, don't buy direct from Fairchild unless you're ordering thousands.

Edit; ya the one you linked looks good too!

 

[DashApple]

Would it be able to handle >30A? I don't mind paralelling a few of 'em
The one I listed will do 42A constant or 60A max but I think I read somewhere that it needs a Vmin of 80V.

80V will probably be its drain to source voltage .

The voltage on the gates are usually 15V / 20 / 25 and 30V in some cases .

As others said the minim voltage to get the FET to turn fully on is around 10V .

The lower the voltage rating of the mosfet the lower the RSd on value in most cases .

If the mosfet has a RDs of 0.002 Ohms and 30A though the mosfet , that's I^2*R , 0.002*30^2 = 1.8 Watts

Edit , Sig pointed it all out , didn't notice till now

 

[AUTO XX]

What does drain to source voltage mean?
I ordered the 0.002ohm mosfets but fedex is giving me grief... You Americans don't know how good you have it when it comes to your postal service.

 

[Sigurthr]

Vds is the Drain-Source voltage, it is the maximum voltage the transistor is safely able to switch. It can switch any voltage lower than VdsMAX, but higher than that value will blow up the transistor.

 

[AUTO XX]

Okay, I must be slower than I thought.
The ascii circuit posted shows the vpos going through the reed switch (and a 10kohm resistor) and then connecting to the ground (through a 100kohm resistor).
Am I missing how this is supposed to be wired? You can't short ground to +V or your little resistors will go poof...
Or is the negative side resistor a high watt, connected to the source pin?
I haven't built a circuit in a while and only a few with fets so that is probably where the confusion comes from.
Also, doesn't the gate voltage have to be higher than the source? I just read that, don't know if it applies.

Maybe a bit more user friendly diagram would be easier for my little brain if it's not too much trouble.

I have 100k resistors ordered so there is no panic on time

Thanks guys.

 

[MarioMaster]

The two resistors form a voltage divider. The total resistance of the resistors is high enough to limit current to a very low value.

     Reed Sw   10K                10K
+Vbat───o/ o──/\/\/\/─────┬─────/\/\/\/─────Gnd
                          │
			  │
			  │   Q1
			  │    	
			  │    │■───────(M)─── +Vbat
			  │    │■─┐	
			  └────┘■◄┴───┐
				      │
				      │		
				      ┴ Gnd

 

[Sigurthr]

Don't worry, that's a perfectly logical question.

This is where a full schematic really shines. So, I drew one up for you!

http://i.imgur.com/3pcR079.png

You normally cannot place a resistor directly across a battery or power supply without it going up in smoke, but what determines if it will end in fire or no harm is the value of the resistor. I = V/R, I = current and I * I * R = power that the resistor has to withstand. In our example the battery sees 110k ohms (100k + 10k)... I = 12.6 / 110,000. I = 0.000114 amps, or 114uA (micro amps). 114uA * 114uA * 110,000 = 0.00144 Watts or 1.44mW (milliwatts). A standard resistor is 250mW, so you don't have to worry about the resistor going up in smoke at all.

Now, as for how this circuit works: A N channel MOSFET is a special switch that senses how much voltage is applied between the Gate and the Source. When enough voltage is applied the switch turns on completely and acts as a short between Drain and Source. Source must be connected to ground. So, apply enough voltage between Gate and Source and it connects the negative side of the motor to ground, completing the circuit. So, why bother with the resistors, why not just place the Gate on the other side of the reed switch? The reason is that the Gate to Source sensor acts like a capacitor and holds whatever voltage you place across it until you discharge that capacitor through a load or resistance. If you didn't use the resistors the Gate would still hold and sense 12.6V even after you disconnected the Reed Switch and the motors would never turn off. The 10k resistor isn't 100% necessary but it makes the circuit more reliable and efficient. The 100k resistor is how we discharge the capacitor in the Gate sensor safely once the reed switch removes power, it's how the mosfet turns off reliably.

Two resistors in series between a voltage and ground with the middle junction leading outward to something is called a Voltage Divider. The voltage output from the middle junction is the ratio of the values of the resistors used. In this case the ratio is 10:100. The smaller the resistance of the resistor closest to the voltage source is, the closer to the input voltage the middle point junction will be. The smaller the resistance of the resistor closest to ground is, the lower the voltage at the middle junction will be. In our circuit with 10k and 100k the middle point will be 10% less than the input voltage. This means that it will be 11.34V for a fully charged 3S lipo and 9.18V for a nearly discharged lipo.

The general rule of thumb is you want >8V at the gate of a standard N channel MOSFET if you want the mosfet to turn on fully. If you apply a voltage to the gate that is above the datasheet value of VgsTh (gate threshold) but below the Fully On voltage then the MOSFET acts like a resistor (a dimmer switch) and instead of wasting almost no power and generating almost no heat it acts like a space heater and generates a tremendous amount of heat. So when using a fet to switch power on and off you want to make sure the gate always sees >8V (most ppl state 10V as fully on no matter what, so they use that) or as close to 0V as possible (technically anything under VgsTh is good) so you avoid the Space Heater state (called Linear Region).

There are special MOSFETs labeled as TTL Level (5V turns fully on) and Logic Level (5V or 3.3V turns fully on) or Small Signal (a very low voltage turns fully on) for applications where you don't have >8V available.

If we used 10k and 10k resistors the Gate would see HALF of whatever voltage the battery pack puts out (5.1V to 6.3V), which might turn the fet on reliably when fully charged, but when the lipo discharges there's a very good chance that the fet would no longer turn on fully. Instead of switching normally it would burn up and explode the next time you turned the reed switch on. [Non laymens disclaimer: 5.1V is well within the linear region of most VgsTh = 4V power fets for high Ids loads. I didn't check the curves for his specific fet, but chances are I'm not far off, it's better practice to design in a working overhead than play close to the limits.]

Edit* forgive the typo in the image of "NPN MOSFET". No such thing exists, it should be "NPN / N-ch MOSFET" as both types of transistors would operate identically in this schematic. I whipped it up in a hurry and didn't change the label fully.

Oh, and note you can run multiple motors in parallel as long as the current draw from them doesn't exceed the MOSFET's maximum Ids (drain current) value.

Edit 2: Mario beat me to it but see the underlined passage about why NOT to use two 10k resistors in this application.

 

[AUTO XX]

I now fully understand what "gobsmacked" means.
I am in awe and wonder from the most descriptive explanation I have read in my 31 years.

I pride myself on being able to explain complicated things simply but that was just beautiful.

Thank you both very, very much.