In the application I was talking about the leads themselves got hot. At 0.01Ω its dissipation is 0.1W The wires inside will also suffer some heating. Another possibility is a DC SSR ( Solid State Relay) which are typically much more robust and have much heavier duty leads or you could drop the same exact MOSFET in parallel with the first to divide the heat. Part of the problem is not whether the device is rated for that is the magnitude of current you have flowing through it. 10A is not an insignificant amount of current.

 

Another possibility is a DC SSR ( Solid State Relay)

They look to be about $10 with the rating i need.
For similar rating, mosfets seem to be under $1.

you could drop the same exact MOSFET in parallel with the first to divide the heat.

Would that work in a dual mosfet package?

Should i derate the D-S voltage? or continuous drain current?

 

At some point you're just going to have to audition components and see if they work. Experience is recognizing you have made this mistake before.

 

Fair enough, then assume i WILL have a heatsink.
Next question is, how big a heatsink will enable this chip to carry 10 A continuous?
I experimented with a few heatsink calculators, but didn't get anywhere.

Now you're asking the right question!

Is there a reason you've chosen an SMD MOSFET rather than a through-hole device? Heatsinking for SMD devices is much more difficult, especially if you pick a device with a limited dstasheet - ie no heatsink recommendations and no charts for recommended max case temperature. Charts for '25C' are fairly useless as that's never the case.

Anyway the process is relatively simple, this is the way I work it.

You start with your max dissipation, which you now state is 0.33W based on an Id of 10A continuous and and Rds of 3.3mOhm. For safety I'd actually use a factor of 1.5 as Rds(on) increases with Tj. Working with a max Tj of 120C, fig 13 gives Rds(on) as approx
5mOhm, equivalent to 0.5W. Next, I choose a desired case temperature. If this is an enclosed device, heatsinking can be tricky so I go for a low one, say 50C If there is more flexibility over cooling arrangements, 75C is a common choice, as many manufacturers quote device specs at that case temperature.

So if Tc (or Tmb, same thing essentially) = 50C the junction temp will be Tj = Tc +Th(j-c) * Pd = 50 + 1.82 * 0.5 = 51C (rounding up). Now from the case/mounting base to the actual heatsink there is some form of interface, possibly an insulator but for an SMD this is usually soldered directly to a PCB, so the interface is some form of solder paste. Typically this will be ~0.1mm thick and have a thermal conductivity of around 70W/mK, giving an actual thermal resistance Th(i) of ~0.15C/W which at 0.5W means the heatsink temperature Th = Tmb - Th(i) * Pd = 51 - 0.15 * 0.5 = 50.9C (in this example its insignificant, but at higher powers the interface can be a show-stopper).

So, we now know that the heatsink has to conduct/radiate 0.5W of power at 51C to an ambient of typically 25C, so it needs a thermal resistivity of (Th - Ta)/Pd = (51 - 25)/0.5 = 50C/W
For a through-hole device its a 'simple' matter of looking through catalogs to find something suitable, but for an SMD device its not so easy and some empirical testing is needed, though the calculator here is a good start and suggests around 100mm x 100mm is needed.

 

Is there a reason you've chosen an SMD MOSFET rather than a through-hole device?

I searched Octopart for mosfets with my required current and voltage, low rdsON, sorted by price, and got these guys.
My real requirement is up to 22 V @ 5 A.
So i search for something like 100 V @ 15 A (i tried various).
It seems none of mosfets i found include 22 V @ 5 A in the Safe Operating Range (even when i search on 600 V @ 100 A).
Is 22 V @ 5 A not in the Safe Operating Range of any fets? Is that just a fet thing?
I'm fine with paralleling two mosfets, if that will achieve it.

On your advice, i picked only fat thru-hole fets (same current & voltage specs). Lo and behold, the SOA is bigger But still not seeing 22V @ 5A within SOA. It seems that doesn't exist. The best i found so far goes up to 2.5A at ~20V, so i could parallel two to get 5A.
http://www.gofordsemi.com/u_file/product/742/GOFORD-G700P06J.pdf

@Irving Aside from package style, what parameter will tell me the same thing? Thermal resistance?

 

If Rja = 49 'C/W for ~ 2W thats about 100'C rise above 50'C ambient. So you need a heatsink and choose a design target of a 40'C max rise