What size MOSFET do i need to drive about 6A 12V load? Will small TSOP6 package do the job without overheating?

What size MOSFET do i need to drive about 6A 12V load? Will small TSOP6 package do the job without overheating?

I was thinking of using this one: http://www.tme.eu/en/details/irlts6342trpbf/smd-n-channel-transistors/infineon-irf/ to PWM leds drawing about 6A, says that it can handle 8.3A but I'm worried about heating.

• Have you done the math yet? – Ignacio Vazquez-Abrams Aug 1 '16 at 19:02
• Figure out what the power dissipation is in the package, then find the junction temperature spec in the datasheet. You can then estimate the temperature of the part under full current. – Voltage Spike Aug 1 '16 at 19:08
• Under static on conditions you'd just look at the Rds-on spec relevant to the gate voltage you'll be driving it with. But since you'll be PWMing it you may need to take the turn-on and turn-off time into consideration depending on how you're driving it and how fast your PWM frequency is. – brhans Aug 1 '16 at 19:11
• Sure it can without overheating. With sufficient heatsinking and careful consideration of how you are driving it. – Passerby Aug 1 '16 at 19:11

Running the numbers: Power = I^2 x R. With a resistance of 0.022 ohms and 6 amps of load, you get 0.792 watts of power dissipation.

At this point, you consult the datasheet to determine the junction to ambient temperature rise, in terms of degrees/watt. For this part, it's a maximum of 62.5 C/W. Multiply that by the power dissipation we found out above, and you get ~50C.

So, you're looking at about a 50C rise, which puts your part in the 70C range. That's OK.

It's probably fine. The only thing to watch out: at 70C, the current rating is not 8.3A, but 6.7A. So you don't have a lot of headroom.

EDIT: As Brian Dohler correctly points out, you will be PWMing the MOSFET, which causes switching losses and will increase the temp above what I stated here.

So you may need a heatsink. On those little packages, they are meant to be attached to a heavy PCB by a bottom pad (if you flip them over, they're shiny), which will act like a heatsink.

As this pad is often inaccessible, you could consider mounting something to the plastic top of the chip. While the plastic is not a good thermal conductor, it beats nothing, and you don't need a lot of heatsinking in any case.

• This does not take into account switching losses. Calculate switchING losses as fVIt_switch, where t_switch is your Q_millerI_gate. To minimizing switching losses, use a higher gate current and lower pwm frequency. This will have other effects though. Be careful not to use too much gate current. – Brian Dohler Aug 1 '16 at 23:30

No you can't.

The power dissipation is too much for the device in that package. For worst case estimation you need to use graph 4. The Rdson at max junction temperature, that is also the value used for package current rating,
is 1.7 times the one at 25C. Including switchin losses you will be around 1.7 W

Please note the Rth J-A written on the datasheet is :

• Too optimistic (a small package like that will have ~ 30 C/W just between junction and board)
• it is referring to a board with 1 squared inch of thick copper layer to spread the heat.

On top you need to consider the worst ambient temperature the devices can meet. What is the usage and where it will be located ?

I suggest to use a mosfet in DPAK with ~ 10 mOhm or less to handle the power dissipation.