Dumb down MOSFET Selection?

Can someone dumb this down enough for me to understand?
I have some electronics background (a 100 level class in college), but it seems most of the that has faded over time.

For my first project I have a mega2560, and would like to run the radiator fan in my car with it. I am looking for variable speed control through a digital out (pwm).
The fan probably draws 15-20 amps at full speed. I have no idea what it can spike to on start up.
This is a 12v system.

What mosfet do I need, and more importantly, how do I know what the right mosfet for me is?

In searching, I have found there are n and p types. What's the difference, and why would a dummy like me care?

I can use a heat sink, and really I wouldn't mind running a cooling fan either.

As I understand it, an H-bridge is another option... but I don't need the reverse capabilities...

I'm thinking of the mosfet as a 'fast reacting relay'. Is that fairly accurate?

Anything is appreciated, even if just links to info.

thanks,
jason

Just tried to confirm the specs on my fan, and it appears it is a little more power hungry than I thought...

The only real drawback of the Mark VIII fan is that it draws a hungry 30 to 40 continuous amps, and can momentarily spike up to a 100-amp draw at startup

Read more: http://www.popularhotrodding.com/tech/1005phr_project_fox_cooling_system/viewall.html#ixzz2SA8HeIfU

I am looking for variable speed control through a digital out (pwm).

Why do you need speed control? Auto fans simply operate on/off. Given the the rotational inertia of the fan and the thermal inertia of the radiator how does pwm provide an advantage?

If you can live with on/off a solid state relay is an easy & reliable answer.

Joe

Sorry, I can't advise you on the FET selection. However, every car I've seen uses a simple mechanical relay to turn the fan on and off - even the ones controlled by the ECU ultimately drive it via a relay.

Not ALL vehicles run simple relays.
The fan I have flow a lot of air at max speed, and as such, makes some noise.
If I don't need full power from the fan at idle, then why make the extra noise and deal with the constant cycling on-off of the fan?

Thanks for the objective input, but I want variable fan speed.
From what I understand, mosfets are the way to go.

Anyone have input on how to select a MOSFET?

There are a lot of power MOSFETS with amazing power handling ability. The IRF 3205 handles 110 amps, and can operate, I believe, at 175 deg. C. Since MOSFETS have very low on resistance, when operated at saturation, I would think the heat sink requirements might be modest. I suggest you buy a few (they're pretty cheap), and try running your fan full bore for 15 minutes. If the MOSFET survives that test, it'll probably work fine for your application.

Good luck.

I wonder if a fan running at under full speed get the air up to enough velocity to drive through the core of the radiator.

JimboZA - I'm sure at low levels (5-10% maybe), there wouldn't be much point in having the fan on...
I plan to bring the fan on at 20-30%, but I'll have to see how the fan likes that.

Jrdoner - thanks for the recommendation of the irf3205.
It is an N-type, as opposed to p-type. Can you explain what that means?

Is there a disadvantage to having the MOSFET oversized?
The irf3205 is rated for 55v... Is there anything to be gained by finding something rated in the 20v range?

JimboZA:
I wonder if a fan running at under full speed get the air up to enough velocity to drive through the core of the radiator.

The relationship between fan speed and cooling is highly non-linear and even a tiny amount of air flow can produce significant cooling. In other words, the higher the air speed, the less benefit you get from further increases in air speed. However, what's far more important than fan speed is to ensure that you don't get any hot air recirculation - you need to thoroughly seal the hot and cold sides of the heat exchanger from each other. A 10% reduction in air flow makes almost no difference, but a 10% hot air recirculation immediately reduces your cooling by 10%.

If you decide to use a MOSFET, remember to put it + heatsink in the path of the air flow from the fan. :slight_smile:

N-channel means it is sinking current from the fan (low side drive), vs P-channel which sources current to the fan (high side drive).

A logic level part can be driven direct from Arduino output.
Low Rds means it will heat up very little with large current flow: Power dissipated by the part = IIRds, so 20A x 20A x 4.5mOhm = 1.8W.
So a part like this with some heatsinking would work well

A part like this might be easier to mount a heat sink to, but will also dissipate more heat.
20x20x.0065 = 2.6W

If you use a P-channel device, then you need a buffer in front of the part - the buffer would turn on and pull the Gate low to turn on the MOSFET, a pullup resistor to +12V would turn off the device.

vristang:
Is there a disadvantage to having the MOSFET oversized?
The irf3205 is rated for 55v... Is there anything to be gained by finding something rated in the 20v range?

55V is the standard voltage for an automotive part - as I understand it this covers both 12V and 24V systems and a factor of
two is added to cope with noise spikes, inductive ringing, etc.

Lower voltage MOSFETs will have lower Rds(on) for a given gate capacitance that higher voltage parts, all
other things being equal, due to the shorter drain current path.

Over-spec'ing a MOSFET will typically mean more gate capacitance to drive - so you will either have to
up the gate-driver to a higher current one or put up with slower switching (which with PWM may mean
more heat dissipation - switching losses are not improved by lower on-resistance, but are improved by
faster switching MOSFETs (or lower frequency PWM).

You guys are helping me a lot. Thank you.

Here are the options I'm considering now...

  1. IRF3205PBF
    http://www.mouser.com/Search/ProductDetail.aspx?qs=2r01AXMCG3Ng8ivZ6KPoqw%3D%3D

  2. IRF3205ZPBF
    http://www.mouser.com/Search/ProductDetail.aspx?qs=2r01AXMCG3ONs%2F8RgU3KsA%3D%3D

  3. IRF3205LPBF
    http://www.mouser.com/Search/ProductDetail.aspx?qs=2r01AXMCG3NT2yLcO1Kc%2Fw%3D%3D

  4. AUIRF3205
    http://www.mouser.com/Search/ProductDetail.aspx?qs=cPwVhw1VIU7wnd84RQYJCA%3D%3D

Number 2 has the lowest Rds (6.5), and second highest heat dissipation at 170W.
Seem like a good fit?

Currently, I plan on mounting the Arduino under the passenger seat in the cabin of the car.

I greatly appreciate the input!

jason

If you are going to use a mosfet to PWM a fan that takes that amount of current, you need to use a mosfet driver chip, such as TC4420 or MCP1406. Although the Arduino can drive logic-level mosfets directly, it can't drive high-current mosfets fast enough to prevent them from dissipating lots of power as they turn on and off.

Those chips have voltage ratings of 18V and 20V respectively, so you should protect the driver (and hence the mosfet gate) from transients using an R-C network, perhaps with a TVS diode in parallel with the capacitor. You can feed the Arduino from the same protected supply.

Regarding mosfet selection, it is simplest to switch the low side of the fan using an N-channel mosfet. Also, you can get N-channel mosfets with better characteristics (e.g. lower Rds(on)) than P-channel mosfets. You can connect several mosfets in parallel if it helps keep the heat generation to a low level.

The mosfets you've linked to are not bad, but there are even better ones, e.g. http://uk.farnell.com/nxp/buk952r8-60e/mosfet-n-ch-60v-120a-to-220ab/dp/2254216. Bear in mind that if you use a mosfet driver chip, you are not limited to using logic-level mosfets.

A few weeks ago I read (it could be on this forum) that the TO-220 could not have 100A through the pins, even though the mosfet was specified for that amount of current.

A relay would be the most obvious choice.
If you want to use PWM, you better use a mosfet of 200A. Or use 4 of one of those you selected in parallel.
A mosfet driver as dc42 wrote about is something you really need in this situation.

I'm not sure the fan can handle a PWM signal.

dc42:
The mosfets you've linked to are not bad, but there are even better ones, e.g. http://uk.farnell.com/nxp/buk952r8-60e/mosfet-n-ch-60v-120a-to-220ab/dp/2254216. Bear in mind that if you use a mosfet driver chip, you are not limited to using logic-level mosfets.

Spec on the linked mosfet
Continuous Drain Current Id: 120A

Erdin:
A few weeks ago I read (it could be on this forum) that the TO-220 could not have 100A through the pins, even though the mosfet was specified for that amount of current.

For my skill level, it would be nice to just use a mosfet to drive the fan. I will take it on if I have to though...

Is a continuous current of 100A going to be just as hard on the mosfet as a momentary spike of 100A.
Most likely the fan won't take more than 35-40A at full power.
The 100A just comes at start up, which may be reduced by starting the fan at a slow speed (25%)?

Thanks again for all the great help.

Erdin:
A few weeks ago I read (it could be on this forum) that the TO-220 could not have 100A through the pins, even though the mosfet was specified for that amount of current.

I wondered about that myself... having worked a fair bit on big-ass electrical installation jobs and seeing the size of conductors needed to handle 100A, I thought seeing TO-220 packaged parts being rated for hundreds of amps seemed a bit ridiculous!

Even with PWM you will still see the initial current draw, that is why you over spec parts so that they can deal with it. the initial start up current is nothing to do with the speed, as far as the fan is concerned as soon as it gets power it is coming on it knows nothing about the fact it is about to go off again and how long it will be on for initially.

Remember that there is a specification for vehicle stuff and that includes load dumps and spikes of up to 150 V are seen as normal should they occur and it is your problem to deal with that. 55V on 12V on but make sure you put protection in, for 24V I'd be tempted to try higher. It is also expected that the vehicle could see 24V for some seconds (jump starting from a recovery vehicle) so deal with that too.

BulletMagnet83:

Erdin:
A few weeks ago I read (it could be on this forum) that the TO-220 could not have 100A through the pins, even though the mosfet was specified for that amount of current.

I wondered about that myself... having worked a fair bit on big-ass electrical installation jobs and seeing the size of conductors needed to handle 100A, I thought seeing TO-220 packaged parts being rated for hundreds of amps seemed a bit ridiculous!

Actually TO220 package can take 100A, just not for very long! Its the gold bonding wires to the chip itself that
will fail first, and at very high currents the magnetic force may break them before they vaporize for all I know!

vristang:
Spec on the linked mosfet
Continuous Drain Current Id: 120A

The more important aspect of the spec (and the reason I pickled that one) is the low Rds(on), because you need a very low Rds(on) to keep the power dissipation manageable.

In view of the high current, I suggest you use at least 2 mosfets in parallel, driven by a mosfet driver IC.