MOSFET heat

This is the place for any magazine-related discussions that don't fit in any of the column discussion boards below.
User avatar
Nepo
Posts: 7
Joined: Thu Jan 01, 2009 2:40 am
Location: desert wasteland
Contact:

MOSFET heat

Post by Nepo » Thu Jan 01, 2009 4:32 am

I am building a controller for an automotive fan, the fans I’ve looked at range from 9 to 16 Amps. I’m using a IRL540PBF MOSFET and PWM signal from a PIC to adjust the speed. Heat is not an issue I have dealt with much in the past, and I'd REALLY like some advice about the heat from the MOSFET.

Ideally it would be nice to seal up the project box containing the circuit and keep the size of the box fairly small. I put together a test circuit with a good sized heat sink and have a smaller motor ( about 4.5 amp I believe) and even with this smaller motor it gets a little warm.


Here is a picture,
http://www.wiredvoid.com/MiscPics/FNCT_Sink.jpg

I was going to use a shorter heat sink, but the thought of it in a sealed box in the summer heat is starting to make me nervous. Is there any problem with using two MOSFETs in parallel? (with smaller heat sinks)

What would you do?
Thanks!

dyarker
Posts: 1713
Joined: Fri Aug 22, 2003 1:01 am
Location: Izmir, Turkiye; from Rochester, NY
Contact:

Re: MOSFET heat

Post by dyarker » Thu Jan 01, 2009 6:30 am

Nepo wrote:I was going to use a shorter heat sink, but the thought of it in a sealed box in the summer heat is starting to make me nervous. Is there any problem with using two MOSFETs in parallel? (with smaller heat sinks)
You still need to get the heat out of the sealed box. How about a cast aluminum box, mount the MOSFET to the box using thermal conductive, electrical insulating hardware; then the box is a big heat sink.

Cheers,
Dale Y

User avatar
Viking
Posts: 87
Joined: Mon Apr 12, 2004 1:01 am
Location: Aylesbury, England
Contact:

Post by Viking » Thu Jan 01, 2009 6:45 am

Hello Nepo,
Assuming that you are driving your IRL540 fully into conduction and assuming that it is conducting for 100% of the time, then the proximate dissipation would be as follows:-
4.5A 1.56W
9A 6.25W
16A 19.7W
Assuming a 40 degree C maximum ambient temperature and limiting the maximum temperature rise to 70 degree C, for 16A a heatsink with a 1.5C/W rating would be required.
At less than 100% (say 80%), the dissipation would be 80% of this, but then need to add switching loss, so would bring this back up to the 100% figure, will depend on switching frequency, the lower the better as far as dissipation is concerned.
MOSFET will run higher than the 70 degree C, but I set 70 as upper limit as plastics start to complain above this temperature.
All the above assumes that there is good contact between device and heatsink. If heatsink is inside an enclosure, its dissipation performance reduce considerably, so consider having the heatsink on the outside of the enclosure, but watch out then as if the device is electrically connected to the heatsink, heatsink will not be at ground potential.
Regards
Rob
LIFE….the crappy bit between birth and death

User avatar
Viking
Posts: 87
Joined: Mon Apr 12, 2004 1:01 am
Location: Aylesbury, England
Contact:

Post by Viking » Thu Jan 01, 2009 7:04 am

Hello Nepo,
Forgot to give an answer to parallel devices. With two device, dissipation is halved, so at 16A, the dissipation is around 10W, so a 3C/W heatsink would be needed to give the 70 degree C limit. MOSFETS are easy to drive in parallel, but watch out that you don't increase dissipation by reducing the switching speed due to the increase capacitance of the gates.
Regards
Rob
LIFE….the crappy bit between birth and death

User avatar
jwax
Posts: 2151
Joined: Mon Feb 09, 2004 1:01 am
Location: NY
Contact:

Post by jwax » Thu Jan 01, 2009 7:20 am

Nepo- Is it possible to put the heatsinks in the airflow of the fan?

User avatar
Nepo
Posts: 7
Joined: Thu Jan 01, 2009 2:40 am
Location: desert wasteland
Contact:

Re: MOSFET heat

Post by Nepo » Thu Jan 01, 2009 9:31 am

Thank you Viking , very informative.
dyarker wrote:
You still need to get the heat out of the sealed box. How about a cast aluminum box, mount the MOSFET to the box using thermal conductive, electrical insulating hardware; then the box is a big heat sink.

Cheers,
I have an aluminum box lying around, It would be great to use it as a sink, but the whole box would be connected to the source. What would I use for/ where do I find "thermal conductive, electrical insulating hardware"? Using the case as the sink seems Ideal.
If it was mounted inside the cab it could have airflow, but I was hoping to mount it in the engine bay.

User avatar
haklesup
Posts: 2955
Joined: Thu Aug 01, 2002 1:01 am
Location: San Jose CA
Contact:

Post by haklesup » Thu Jan 01, 2009 11:03 am

What would I use for/ where do I find "thermal conductive, electrical insulating hardware"?
Silicone or mica spacers plus zinc oxide heat sink compound are standard hardware wherever transistors are sold. In most cases, it is preferable to mount them electrically isolated as it avoids a lot of potential (pun) problems. Search on "Transistor Accessories" in the digikey site
http://search.digikey.com/scripts/DkSea ... 20mounting

Keep in mind that determining the box's thermal rating C/W is difficult and won't be rated for the box. You can just find out if it is sufficient by measuring the transistor temperature on the prototype (get a thermocouple and DMM that can measure temp). If its too hot but not way too hot, you can likely add fins to the outside of the box at the hot zone to increse surface area. In heatsinks, surface area is number one, thermal conductivity of the metal is second. Al rivals Cu in this case.

If you need to make one or just a few. a CPU heatsink makes a good start as they are usually very efficient and usually have the rating spec on the box. However if you need to make a bunch, what you buy at a computer store today probably will be a different model next month.

Nice thing about heatsinking is that you can just grossly overestimate what you need and it often does not effect prototype price very much. Too much heatsink will not harm electrical performance just volume pricing and physical fit.

Also is 40C ambient in an engine compartment on a sunny summer day a reasonable starting point? Your worse case may be a bit higher. Easy enough to measure ahead of time.

User avatar
Viking
Posts: 87
Joined: Mon Apr 12, 2004 1:01 am
Location: Aylesbury, England
Contact:

Post by Viking » Fri Jan 02, 2009 4:16 am

Yes, probably 40 degrees is a bit in the low side. Would probably be closer to 60 degrees.
So if limiting temp rise to 80C, then for a 20C rise, single device would need a 1C/W heatsink and two devices a 2C/W heatsink.
With a die-cast box and the heatsink bolted to the side, with two devices bolted to the inside with electrically insulating spacers, final construction should be reasonably compact.
Regards
Rob
LIFE….the crappy bit between birth and death

User avatar
Nepo
Posts: 7
Joined: Thu Jan 01, 2009 2:40 am
Location: desert wasteland
Contact:

Post by Nepo » Fri Jan 02, 2009 7:06 am

I pulled an aluminum project box out of the closet and ordered some insulating spacers. I have made some progress and I learned some new techniques to add to my mental tool box. Thanks for taking the time to help me, I really appreciate it.
:D

Bigglez
Posts: 1282
Joined: Mon Oct 15, 2007 7:39 pm
Contact:

Re: MOSFET heat

Post by Bigglez » Fri Jan 02, 2009 1:54 pm

Nepo wrote:I’m using a IRL540PBF MOSFET and PWM signal from a PIC to adjust the speed.

Is there any problem with using two MOSFETs in parallel? (with smaller heat sinks).
The beauty of PMOSFets (compared to bipolar transistors)
is that they are very easy to place in parallel!

The device that you selected is a good performer
but not perfect. It will require a heatsink if used
above approximately 3,4Amps. Say 3 amps
for safety. Divide your known load current by this
to determine how many FETs to parallel.
For 12A use four, for 20Amps use seven.

Each Fet will not require a heatsink of it's own.
(Great savings in hardware...)

Driving these FETs may require a buffer stage.
What frequency is your PWM signal?

User avatar
dtief
Posts: 59
Joined: Wed Aug 20, 2008 4:13 pm
Contact:

Post by dtief » Fri Jan 02, 2009 7:40 pm

Another alternative: Use a bigger FET like HUF75344G3. RDs On .008Ohms. That would dissipate almost 10 times less power. Near 3W at 20 Amps. They are only $2.27 @Digi-Key.

The gate drive current would be higher.

Bigglez
Posts: 1282
Joined: Mon Oct 15, 2007 7:39 pm
Contact:

Post by Bigglez » Fri Jan 02, 2009 11:49 pm

dtief wrote:The gate drive current would be higher.
I thought FETs were voltage (not current) operated?

User avatar
MrAl
Posts: 3862
Joined: Fri Jan 11, 2002 1:01 am
Location: NewJersey
Contact:

Post by MrAl » Sat Jan 03, 2009 12:57 am

Hello,


It's a pretty well known fact these days that MOSFETs require
high current drive to work as well as possible. Yes, they turn on
with a voltage, but that's a little misleading as the current requirement
is often very high. A better description i think would be to say that
they require a 'power burst' drive followed by a 'holding' voltage.

But even that is a little over simplified when connecting MOSFETs in
parallel. The gates should be decoupled with individual resistors,
lowest values possible. Circuit layout management should include
thoughts about stray inductances...equalizing among devices.
At times, even threshold voltages should be matched.
Low impedance drive helps the turn off transition too.

MOSFETs are pretty amazing devices but are not quite as simple
as they are sometimes portrayed, having characteristics that
include four distinct turn on intervals and four distinct turn off intervals.
LEDs vs Bulbs, LEDs are winning.

User avatar
dtief
Posts: 59
Joined: Wed Aug 20, 2008 4:13 pm
Contact:

Post by dtief » Sat Jan 03, 2009 7:41 pm

The gate has significant capacitance, especcially with the bigger (current) FETs. Since we are talking about a switch mode power control, the FET has to be turned on & off very fast. Charging & discharging a capacitor very fast requires current to flow.

Dave

Robert Reed
Posts: 2276
Joined: Wed Nov 24, 2004 1:01 am
Location: ASHTABULA,OHIO
Contact:

Post by Robert Reed » Sat Jan 03, 2009 7:54 pm

Right on, and not even to get into Miller effect! :eek:

Post Reply

Who is online

Users browsing this forum: CeaSaR and 10 guests