Can I safetly run this SMD mosfet at 3 amps?

Thread Starter

coinmaster

Joined Dec 24, 2015
502
Ohai thar. I want to run a PMZ390UN at 3 amps 1 volt which is 3 watts. I really don't know too much about cooling SMD parts, it says the thermal characteristics are 50k/Wwand then below that it says 670k/w.
It seems that the 50k/w parameter is for "junction to solder" but how does that differ from a normal connection to the PCB?
I guess in short what I want to ask is, can this be done safely with air cooling and maybe some extra copper on the pin layout?
 

ian field

Joined Oct 27, 2012
6,536
Ohai thar. I want to run a PMZ390UN at 3 amps 1 volt which is 3 watts. I really don't know too much about cooling SMD parts, it says the thermal characteristics are 50k/Wwand then below that it says 670k/w.
It seems that the 50k/w parameter is for "junction to solder" but how does that differ from a normal connection to the PCB?
I guess in short what I want to ask is, can this be done safely with air cooling and maybe some extra copper on the pin layout?
The SMD MOSFETs on PC motherboards are usually somewhere in the direction of 80A - the data sheet should state what yours can do.
 

crutschow

Joined Mar 14, 2008
34,286
Depends upon what you mean by "run".
If you are using the MOSFET as a switch (Vgs sufficient to fully turn it on) then it's dissipation will just be the ON resistance (from the data sheet) times the square of the current, not 3W.
 

kubeek

Joined Sep 20, 2005
5,794
Right off the bat at the first page it says less than 2.5W, and those are marketing wank numbers that you cannot achieve in practice.
The 2.5W number assumes that the tiny pads of the chip are kept solidly at 25°C - you would need an aluminium core pcb, thick copper plate, or similar stuff cooled to that temperature in order to get that - not going to happen.

I would suggest at least a DPAK to get rid of that heat without any elaborate heatsinking, assuming the device will be at room temperature. If not, I´d use some at least 20x20mm copper plane on both sides of the package connected to the drain, with ample vias to transfer heat to both sides of pcb.

vertical in still air probably means the package is floating in 25°C air that does not move.
 

ian field

Joined Oct 27, 2012
6,536
Right off the bat at the first page it says less than 2.5W, and those are marketing wank numbers that you cannot achieve in practice.
The 2.5W number assumes that the tiny pads of the chip are kept solidly at 25°C - you would need an aluminium core pcb, thick copper plate, or similar stuff cooled to that temperature in order to get that - not going to happen.

.
Apparently; some semiconductor ratings assume a theoretical "infinite heatsink" and are not going to happen in the real world.

Come to that - I doubt that the source leads on some SMD MOSFETs can carry the advertised MOSFET current.

They probably get away with it because of PWM operation, I've seen seriously oxidised source leads in some applications.
 

kubeek

Joined Sep 20, 2005
5,794
Come to that - I doubt that the source leads on some SMD MOSFETs can carry the advertised MOSFET current.
They probably get away with it because of PWM operation, I've seen seriously oxidised source leads in some applications.
Some mosfet datasheets say something along the lines of "250A (160A*) *package limited" so the die can handle it, but the bond wires and leads cannot.
 

ian field

Joined Oct 27, 2012
6,536
Some mosfet datasheets say something along the lines of "250A (160A*) *package limited" so the die can handle it, but the bond wires and leads cannot.
There's also the distinction between continuous and pulsed, which is allied to what I said.
 

Thread Starter

coinmaster

Joined Dec 24, 2015
502
Can I get away with 1.5A @ 1v or or am I just kidding myself? I really need the low capacitances, gate charge, and low GS Vdrop.
 
Last edited:

ian field

Joined Oct 27, 2012
6,536
Can I get away with 1.5A @ 1v or or am I just kidding myself? I really need the low capacitances, gate charge, and low GS Vdrop.
You seem to be considering 2 things at once that have nothing to do with each other.

Some data sheets have a SOAR curve that pretty much maps out what you can get away with in terms of voltage and current. The maximum current rating isn't the only limiting factor - you have to ensure the current doesn't develop too much voltage across RDSon, VxA=W.

You only need worry about gate capacitance in fast switching. The output resistance of the driver and the gate form a C/R time constant. The time it takes to charge and discharge is time the MOSFET spends in linear region dissipating excess heat.
 
Top