D2PAK,To263 and other surface mount for hi power MOSFET

[xaccto]

I'm asking if someone could please share their experience in using such packages,
specs for some show 300W Power dissipation.

not being experienced I have some trouble comprehending how such packages can
be used to allow such high power dissipation.

I am researching devices for switching either 12V or 24V at high current, possibly
100A, and will use a number of mosfet switches in parallel.
The switching won't be fast, quite slow actually, but should still switch state fast enough to minimize loss

I know how to mount To220 packages, heat sink and so on,
but using surface mount packages confound me for such power levels,
they are a few $ cheaper in some cases, which is why i'm interested in this question.

Thanks.

 

[FvM]

For slow switching applications (< kHz), switching losses can be almost neglected (at least with fast, low impedance gate drivers).
You can reduce on-state losses by paralleling many devices. Switching 100 A without a heatsink is basically possible then.

For true SMD assembly techniques, you have at least these options:

- using copper pour and possibly power planes (with multilayer PCB) for heat transport. Thermal vias can connect the
top layer to bottom layer and power planes. Natural or forced air cooling, possibly mounting of the PCB bottom side to a
heatsink to carry the heat.

Higher thermal conductivity substrates are used for special applications (LED modules, automotive power electronics).

- using small solderable SMD heatsinks "around" the SMD power devices

- using modern DirectFET or similar devices with heatsinks on top. IRF has application notes discussing this techniques.
PCB layout and assembly technique are much more critical than with conventional devices, however.

Of course, the above listed techniques are applicable to fast switching as well, but the transistor number and size has to be
optimized for conduction and switching losses then.

 

[Mr.Cool]

usually when you see 300W capability on a datasheet that it is tested under specific conditions and NO DERATING considered. no doubt your application is different and you may want more than 5 minutes of reliability... so be mindful on how you select your components.

the higher up in power you go the more critical your component selection, layout, partitioning, thermal management becomes.

also.. there is a big difference between a heatsink and a piece of metal.

Mr.Cool

 

[karesz]

Hi Mr Cool,
You must it now!)
Cooling is often tricsy; than is i.e. with 100ft/sec air stream specified or Pdiss as its mounted on infinite cooler, or with a standard +25Cels. ambient...
K.

 

[FvM]

The said application doesn't actually require 300W power dissipation or forced cooling. You can handle 24V/100A with
two paralleled D2PAK MOSFET with 2 mOhm Rdson and 5W conduction losses, each. Low switching frequency is
assumed, as requested.

 

[karesz]

Hi,
The best MOSFETs are for 24V apps with ca 1.5mOhm Rdson_I remember, but cannot find it yet:-(...
If you have 100A to switch_I think you have on a FET 1.5-2W dissipation.
2SK3272-01L,S Automotive FETs has 60V, 80A & 6.5mOhm Rdson...
Thes means; you need a bigger Cooper surface on both (or more) layers, or a smaller heathsink over their as Aavid 7109C (Fischer has some similares too), their Rth=ca.2.5Cels/W by500ft/Min. Depends, of course, on the real ambiant temp...

I think in TO263 (or D2Pack) package is Vishay`s SUM110N04-2m1P the best possibility, hes Rthjunction-ambient is=< 40Cels/Watt.
But 2.1 mOhm has it w. 10V! Ugs... , by 4.5 V it will be 2.4mOhm, with Tjunc= 25 Cels... If Tj >> (realistically up to 100Cels) it will rise width ca. 40%.
I think, some optimistically, but such in direction worst case~2.5-3mOhm is to calculate.
K.

 

[xaccto]

I think its still useful to consider as much power dissipation as possible to survive
fault conditions, until a circuit breaker breaks, or maybe better monitor the current and switch OFF the switch before the circuit breaker....

karez,
thanks for suggestion of some mosfet, it is taking forever researching them.
I think i'm only looking for ones with Rdson of 2.5mOhms or less,
I think that is the bleeding edge of mosfet ATM, though plenty more $ for it:|

I haven't found very much in manufacture app notes on mounting and board layout etc for such surface mount.

thanks pdf re heat sink, will have to check if can get some from digikey or mouser.....

 

[FvM]

You'll find plenty of mOhm FET's at major manufacturers with Vds up to 40V. If you intend a higher Vds,max (it wasn't mentioned
yet), there are only a few remaining . Manufacturers and catalog distributors mostly have web selection tools or catalog
overviews to identify suitable devices.

Your correct however about the overload/short circuit problem. But that's a common issue in power electronics. It can be
handled best by active desaturation detection or current monitoring. Fuses should be present anyway for a possible case of circuit failure.

 

[karesz]

I havent very detailed searched for D2Pack 100A FETs, but the most best Rdson Types arnt in D2Pack, but usually more in TO220... Btw; I think, the values of 2.5-3 mOhms are not so bad
Over 60V is problematic to find such low Rdson, but the newest (most automotive) types up to 30-40V are really good_self in D2Pack not so bead.
In some higher pwr motor controller we applyed the same system as FvM proposed: you must measure the "DC" currents and switch it electronically off if it will be dangerous for the driver stages...
Otherwise: for your efficienter cooling=ask your PCB producer pls for thicker cooper on FR4, usually its not problem to order i.e. with 100um!
Some has thicker cooper layers (up to 3-400um) too, or metall inlays, but they are not so very cheap:-(... Easylie - you can mask out the high-current lines & makes so possibly a thicker tin over their, or "a plus cooper wire" to solder on it_it is simple but efficient way for heat extractions.
K.