Maker Pro
Maker Pro

Unrealistic SMD power dissipation ratings

A

Adam Seychell

Jan 1, 1970
0
Surface mount power devices such as D-pack often have power
dissipation ratings over 80W power at Tj=25²C, but almost always these
devicea will be mounted on a PCB where the junction to ambient thermal
resistance shoots up to around 50²C/W, meaning practical power
dissipation is 1 or 2 watts. So My question is if the design requires
more than say 2 watts dissipation does that normally mean using a
through hole component with heat sink ?
When the number of paralleled SMD power MOSFETs devices become
impractical is that when through hole and heat sinks are considered ?

The application is for synchronous current doubler rectifier having an
output of 50 amps at 2.7 volts. The Ron required to limit power
dissipation at 1 watt per MOSFET would mean many paralleled devices. A
9 mOhm MOSFET can pass only 10.5 amps RMS for 1 W dissipation.
 
R

Robert Baer

Jan 1, 1970
0
Adam said:
Surface mount power devices such as D-pack often have power
dissipation ratings over 80W power at Tj=25²C, but almost always these
devicea will be mounted on a PCB where the junction to ambient thermal
resistance shoots up to around 50²C/W, meaning practical power
dissipation is 1 or 2 watts. So My question is if the design requires
more than say 2 watts dissipation does that normally mean using a
through hole component with heat sink ?
When the number of paralleled SMD power MOSFETs devices become
impractical is that when through hole and heat sinks are considered ?

The application is for synchronous current doubler rectifier having an
output of 50 amps at 2.7 volts. The Ron required to limit power
dissipation at 1 watt per MOSFET would mean many paralleled devices. A
9 mOhm MOSFET can pass only 10.5 amps RMS for 1 W dissipation.

For SMD, one could use 2 Oz copper for better heatsinking, a 1 mil
laminate thickness from top layer to bottom layer which is
sweat-soldered to a copper heatsink or heatpipe.
The laminate material in this case is Kapton and, due to the thickness
is called "flex".
Otherwise, use TO-220 (same package size as D2-pak) and clamp the tabs
onto similar heatsinks.
 
G

Graham Holloway

Jan 1, 1970
0
Robert Baer said:
For SMD, one could use 2 Oz copper for better heatsinking, a 1 mil
laminate thickness from top layer to bottom layer which is
sweat-soldered to a copper heatsink or heatpipe.

Can I suggest that you pepper the area under the tab with plated vias. It
makes a big difference. You might have to experiment to find out how much.
The laminate material in this case is Kapton and, due to the thickness
is called "flex".
Otherwise, use TO-220 (same package size as D2-pak) and clamp the tabs
onto similar heatsinks.


Graham Holloway
WPS Electronics
 
B

Boris Mohar

Jan 1, 1970
0
Surface mount power devices such as D-pack often have power
dissipation ratings over 80W power at Tj=25²C, but almost always these
devicea will be mounted on a PCB where the junction to ambient thermal
resistance shoots up to around 50²C/W, meaning practical power
dissipation is 1 or 2 watts. So My question is if the design requires
more than say 2 watts dissipation does that normally mean using a
through hole component with heat sink ?
When the number of paralleled SMD power MOSFETs devices become
impractical is that when through hole and heat sinks are considered ?

The application is for synchronous current doubler rectifier having an
output of 50 amps at 2.7 volts. The Ron required to limit power
dissipation at 1 watt per MOSFET would mean many paralleled devices. A
9 mOhm MOSFET can pass only 10.5 amps RMS for 1 W dissipation.

There are surface mount heatsinks for these. Use 2oz copper it the rest of
the board permits it.



Regards,

Boris Mohar

Got Knock? - see:
Viatrack Printed Circuit Designs http://www3.sympatico.ca/borism/
 
N

N. Thornton

Jan 1, 1970
0
Surface mount power devices such as D-pack often have power
dissipation ratings over 80W power at Tj=25²C, but almost always these
devicea will be mounted on a PCB where the junction to ambient thermal
resistance shoots up to around 50²C/W, meaning practical power
dissipation is 1 or 2 watts. So My question is if the design requires
more than say 2 watts dissipation does that normally mean using a
through hole component with heat sink ?
When the number of paralleled SMD power MOSFETs devices become
impractical is that when through hole and heat sinks are considered ?


To diss 80w in a smd at Tj = 25C I would imagine you'd need liquid nitrogen cooling.

Regards, NT
 
M

Mac

Jan 1, 1970
0
To diss 80w in a smd at Tj = 25C I would imagine you'd need liquid nitrogen cooling.

Regards, NT

The OP misspoke. The spec. is usually with Tcase at 25C.

--Mac
 
N

N. Thornton

Jan 1, 1970
0
Mac said:
The OP misspoke. The spec. is usually with Tcase at 25C.


Even then, I'm thinking LN2 would be needed.

Regards, NT
 
W

Winfield Hill

Jan 1, 1970
0
Graham Holloway wrote...
Can I suggest that you pepper the area under the tab with plated vias. It
makes a big difference. You might have to experiment to find out how much.

There's an app note that quantitatively details the efficacy of this
technique, showing it works surprisingly well. Sorry, I've forgotten
which app note. :>(

Thanks,
- Win

whill_at_picovolt-dot-com
 
O

onestone

Jan 1, 1970
0
Graham said:
Can I suggest that you pepper the area under the tab with plated vias. It
makes a big difference. You might have to experiment to find out how much.

This works extremely well, IIRC an old Linear tech app note had some
calcs on this, but can't rememeber, I last used it around 1995 when I
was working on a 36W PSU for a transceiver using the LT1370, hence why I
think it might have been an LT app note. The PSU was in a wholly
enclosed armoured box (for mining purposes), with no possibility of a
fan or heat sink on the PCB. the PCB itself was used as the heat sink,
It worked like a charm, in the Australian desert, with cab temperatures
often hitting 70C with no engine running. The PCB was 4 layers. The area
immediately around the switcher was almost a sea of vias, there were
radiating areas on the top or bottom layer, but never both in the same area.

Al
 
N

Nico Coesel

Jan 1, 1970
0
Mac said:
The OP misspoke. The spec. is usually with Tcase at 25C.

In that case, the manufacturor is merely specifying the thermal
resistance between the junction and the case. Perhaps they printed a
nice derating graph somewhere further on for the engineers who are
unfamiliar with Mr. Ohms thermal law.
 
M

Mac

Jan 1, 1970
0
In that case, the manufacturor is merely specifying the thermal
resistance between the junction and the case. Perhaps they printed a
nice derating graph somewhere further on for the engineers who are
unfamiliar with Mr. Ohms thermal law.

Yes, indeed. Please note that I didn't go and read the datasheet for this
part. I'm just going on the datasheets I've seen.

--Mac
 
M

Mac

Jan 1, 1970
0
Even then, I'm thinking LN2 would be needed.

You mean to keep the case at 25C while dissipating 80 Watts? Maybe. But no
one is saying that this is a requirement. They are just saying that if you
can keep the case at 25, then you can dissipate 80 Watts. Since no one is
going to be able to keep the case at 25 with that kind of dissipation,
they'll have to keep the dissipation under 80 Watts. I have to agree that
it borders on dishonest to spec these devices this way, but it seems to be
the industry standard.
Regards, NT

Mac
--
 
Top