Converting MOSFET headline specs into realistic ratings

[Simon]

I'm considering a MOSFET. I have attached an extract from datasheet. I'm trying to work out how much current this MOSFET can drive continually. The claim is it will take 17 A. I am looking to run 10 and continuous. Figure 8 seems to be saying that I can run easily at 10 A. I assume the voltage referred to is the voltage across the MOSFETs and not the total supply voltage therefore I don't see a problem with 10 A. Is that correct?

[T3sl4co1l]

Datasheet link, http://www.irf.com/product-info/datasheets/data/irlr3410pbf.pdf

Switching or linear?

SOA says you can go up to about 10V at 10A, so at least it won't explode suddenly!

More importantly, what's the thermal resistance in your end use?  How much can you afford?

Examples: this is a DPAK, so you can mount it bare on the board (minimal pad), which I guess would be the 110 C/W figure they give.  Which is 1.36W at 175C max Tj, 25C ambient, and much less for higher ambient.

Any means of spreading heat (big thermal pours, via-in-pad, heatsinking the board (top, bottom or both), solder-on heatsinks that straddle the device, etc.) will increase power dissipation, up to the thermal limit of the device itself (however, when they say Pd @ Tc = 25C, they mean all points of the case simultaneously held at 25.0C with nucleated boiling liquid, so, you're unlikely to achieve the full 79W rating even in the best of circumstances).

At Rds(on) max 0.2 ohm (notice Rds(on) rises substantially with temperature; the 0.105 ohm figure is at Tj = 25C only), with a minimal footprint, you can do no more than 2.61A (DC or RMS).  Which is around 3.69A in an average switching application (i.e., 50% duty cycle, say in a buck converter delivering that output current, with a generously large inductor so current ripple is small).

FWIW, you might as well use a SOT-89, -223 or SO-8 device if your current is that low -- you only waste space (and possibly money) with a DPAK.  Even if your current is higher (which it is!), a smaller device may be more economical, because it's impractical to try getting lots of power from a DPAK.

Tim

[Simon]

I'm looking to work in continuous mode full on driven by 5V, switching may be an option but I'd reduce the current allowed in that case. This would be the low side of a H Bridge.

I wanted to use something surface mount but not as large as a D2PAK

[T3sl4co1l]

So like, motor driver?  Is it really not switching [at some reasonable frequency]?...Are you sure it doesn't need to be?

Note that Rds(on) is worse at 5V, and if switching speed is much concern, it will be worse as well (less voltage to deliver the required current through internal and external gate resistances).

I'm guessing the die inside that IRLR3410 is simply small.  It doesn't sound like an older device.  Beware that, because DPAKs and etc. are rated based on massive (and impossible) heatsinking, they always have disproportionately high Id(cont) headline ratings, given the Rds(on).  I would suggest searching primarily by Vdss, Rds(on) and package type; that way, you know that the selected devices will be able to handle the current at low enough dissipation, not just some wanking number they pulled out of their...

Tim

[suicidaleggroll]

According to the datasheet, at Vgs=5V and Id=10A, Rds=0.125R.  That means it would have a 1.25V drop from drain to source and would have to dissipate 12.5W continuously.

It has a junction-to-ambient thermal resistance for PCB mount applications of 50 C/W, which means you'd get a rise of about 625 C, give or take 

Surely you can find a unit with a better Rds(on) at 5V than that, especially for non-switching applications.  Single digit milliohms should not be hard to find and would drop your voltage drop and power dissipation to an almost negligible level.

[Simon]

Yes your right it will be hopeless. I'm driving an actuator up and down, current requirement was for a 1A motor but I wanted to make a reusable project so thought if I aim for more amps I can reuse it more often and get some boards made.

[Simon]

This looks way better: http://www.onsemi.com/pub_link/Collateral/NTD5804N-D.PDF but I'm a bit concerned about figure 5 as it says "normalized" resistance and is should figures of 1 or more, is that percent ? as it's under 10mR on the specs

[PSR B1257]

figure 5 as it says "normalized" resistance

That's not uncommon.

is that percent ?

No, it's a linear factor.

What voltage do you apply to the H-bridge?

[suicidaleggroll]

"Normalized" means "relative to 25C" (or wherever the value is 1).  It's a multiplier to show you how Rds(on) changes with temperature.

That one has a max Rds(on) of 12m at Vgs=5V and Id=10A.  You're looking at about 100mV drop at 10A and about 1W dissipation.  The junction to ambient thermal resistance is 106 C/W on FR4 with the minimum pad size and no heat sink.  With an operating temperature of up to 175 C, you could use it in ambient temperature of ~70 C.  Of course at 175 C the Rds is 75% higher, so power dissipation would increase, etc.  It's an iterative process.

You can go lower though, a quick search found this one:
http://www.mouser.com/ds/2/200/irlr3114zpbf-52067.pdf

Rds(on) of 5.2m at Vgs=4.5V

I don't know if it meets the rest of your specs though.

[Simon]

That looks pretty good, now I'll have to find a matching P channel