Help! Mosfets getting really hot

pbateman

New Member
Hi there.

As the title says, I am having trouble with some mosfets getting really hot, like 115 C.

I am using a picaxe (14m) with the pwmout command to drive the mosfets. The mosfet that I am using is a STP35NF10 from STmicroelectronics. The digikey part # is 497-2645-5-ND. The datasheet can be found here:

http://rocky.digikey.com/scripts/ProductInfo.dll?Site=US&V=497&M=STP35NF10

The mosfets are rated at 100V and 40A.

I am using 2 of these mosfets to power a blower motor in an automobile. The max current draw is about 10A.

I use the pwmout command with a period of 49 and the duty varies from 180- 200. This allows for operation at 20000 hz so that there is no audible whining from the motor.

The tabs of the mosfets are connected to an aluminum sheet as a heat sink (3x5 inches). When the motor is running, the sheet gets very hot and will boil water in the vicinity of the mosfets.

I originally used only one of these mosfets, but it got so hot that it desoldered itself from my circuit board, so I added the second one. I have also run the pwmout command at a lower frequency (12000hz) but there was an audible whine at lower speeds and the mosfets still got just as hot.


So after that longwinded explanation, does anyone have any suggestions on what I can do to get cooler temps. Am I using the wrong mosfet for my application?


Thanks in advance for your assistance!
 

Rickharris

Senior Member
HOT means high current - if you think the motor is inside the FET parameters then there is something else miss wired. Remember motor draw a lot of current when they start up - if you stop start them this adds up.

if no other choice - then more FETS to spread the load.
 

Michael 2727

Senior Member
Sounds like your MOSFETs are running in Linear Mode.
As opposed to Saturation Mode where they switch ON
and OFF, fully, quickly and cleanly. Much less Heat.
In Linear Mode MOSFETs behave more like a BJT (transistor)
and are prone to similar heating effects.

What is the Gate switching voltage 10V or 5V ?
What is the gate setup you are using ?

Many MOSFETs have a 10V gate rating, they will
start to switch @ 5V, and at very low load levels
will not get that hot, but as you draw more load the
gate Voltage needs to be increased or they will get HOT.


You could try running your Picaxe @ 5.7V, the extra
0.7V may help.

Or go for a true Logic Level switching MOSFET type.

That link provided was not the full data sheet, go to
St Micro and get the full Data Sheet -
http://www.st.com/stonline/books/pdf/docs/7330.pdf

From the Full Data Sheet - Table 3.
Gate threshold voltage VDS = VGS, ID = 250µA, 2V-Min, 3V-Typ, 4V-Max.
( suggests they Start to turn ON @ 3V )
VGS = 10V, ID = 17.5A
(and when @ 10V Gate you will be able to draw 17.5A)
 

goom

Senior Member
It is quite possible that the MOSFETS do not have time to turn on fully at the frequencies that you are using. They need a high current to turn on initially because of the gate capacitance. You could check if this is the problem by turning them on continuously (set the output pin high rather than pulsing), and see if they get as hot.

I belive that there are specific MOSFET driver IC's which get around this problem. Perhaps the gates could be driven at 12V by a puch-pull transistor setup. Has anyone ever tried this?

MOSFETS with lower ON resistance are available, but may more expensive. Adding more MOSFETS in parallel will definitely help a lot. Double the number, and the power dissipation in each will reduce by a factor of 8 (I think).
 

pbateman

New Member
Hi guys,

Thanks for the ideas, keep them coming ;)

I will try running the mosfet continuously rather than pulsed. That is a really good idea.

Previously before I even knew I had a heating problem, I tried using an optocoupler between the picaxe and the mosfet, but the PWM signal would not pass through the coupler. Now I just run the mosfet gate directly from the picaxe.

If it does turn out that the mosfet runs much cooler when receiving a steady signal instead of a pwm signal would it make sense to use a transistor in between the picaxe and the mosfet? I could use the transistor to switch the 12V from the vehicle to the mosfet gate instead of the 5V from the picaxe.

Thanks again!
 

Michael 2727

Senior Member
Try this link -
http://www.picaxeforum.co.uk/showthread.php?t=7785
From the original Page of the User Projects - Miscellaneous.
Multi Purpose PWM Supply 12V DC 10A.

Here I used an IRF540 MOSFET running @ 4KHz PWMOUT.
20KHz is still relatively slow as far as MOSFETs go, many
app notes suggest 4KHz to 20KHz for motor driving, though
I personally prefer the lower end.

I ran a Main and Rear auto windshield wiper motor from
this supply at once, then a 100W Spotlight = 8.33 AMPs
and the MOSFET on a reasonable heatsink runs barely warm.

@ Pbateman, you are using Back EMF protection Diodes I hope ?
I used a 1N5408 (3A) and it does get warm.
Don't rely on the internal MOSFET diode.
 
Last edited:

leftyretro

New Member
Heat dissipation at higher frequency will always be more then at a lower PWM switching frequency even when the duty cycle is the same, say 50%. The transition period between on to off and off to on is a source of heat dissipation, so everything else being equal a lower PWM frequency is better then a higher one because there will be fewer heat generating transitions at the low switching frequency. Of course a lower switching frequency can conflict with other desired circuit parameters, so all one can do is carefully study the device data sheets and then either add more or different devices or more heat sinking or both.

Lefty
 

moxhamj

New Member
Resistance is 0.03 ohms so at 10A W=I^2R is only 3 watts which is so little heat it almost wouldn't need a heatsink. Something isn't right and it probably is either the gate voltage or the frequency. If it needs 10V to turn on then it will need a transistor to drive the gate. Two 1k resistors and a BC547 would do the trick (plus a 10V supply) and as a bonus the 1k pullup could be reduced to give drive currents up up to 100mA into the mosfet gate which will help overcome the gate capacitance at high frequencies.

First test is to determine if it is the frequency or the gate voltage that is the problem.
 

goom

Senior Member
Dr Acula, wouldn't it need a pair of transistors, on to turn the MOSFET on, and one to turn it off (i.e like a push-pull amplifier)? I would assume that the ability to turn off quickly is just as important as turning on from a heating point of view.
 
Last edited:

moxhamj

New Member
You can certainly use a push/pull and it does use less power. But a pullup resistor will work too - it just wastes a bit of power when it is on but the power is very small compared with the power being used by the circuit being switched. BC547. 1k into the base. Emitter to ground. 100 ohms between a 10V supply and the collector. Connect the collector to the base of the mosfet. When the transistor is off the gate is held high by the 100 ohm resistor. When the transistor is turned on the gate is pulled low and 100mA flows through the 100 ohm resistor. The gate is effectively pulled high and low by 100mA which is more than a picaxe can supply. The drive current could go up to 800mA with a BC337 and a lower resistor.
 

pbateman

New Member
Thanks again and again for all the useful info.

I tried powering the mosfet gate directly with 5v instead of the pwm signal and the mosfet still got very hot, so i switched to 12v and the mosfet ran very cool, as in no noticeable heat generation. Looks like we may have found the problem.

Looks like I will need to use a transistor to drive the mosfet gate with 12v.

I also did a bit of digging and found the following part from ST - STP16NF06L. The digikey part # is 497-2765-5-ND and the spec sheet is here:

http://rocky.digikey.com/scripts/ProductInfo.dll?Site=US&V=497&M=STP16NF06L

The spec sheet says that the Rds is only .1 ohms max with a Vgs of 5V. It sounds like this would be a direct drop in and I wouldn't need to use a transistor to drive the gate.

Can anyone see why this mosfet would not work for my situation?

Also, on a side note, on my current mosfet, the tab on the to220 package is connected to the drain. This seems very silly since the drain is at a high voltage state when the mosfet is off and makes it very difficult to attempt to use the case as a heatsink. Why would the source not be connected to the tab? Am I missing something here?


Thanks so much again. You guys have been tons of help!!
 

Michael 2727

Senior Member
A MOSFET would be the best way to go,
just a matter of finding one that works well
at only 5V.
They do make Logic Level MOSFETS, designed
to fully switch at 5V, they may be a little hard
to locate readily and they cost silghtly more.

As to the casing Tab, Drain, Source, it's usually
the easiest way to manufacture the device. It may
not be the best way but would be the cheapest.
Unit price seems to govern everything these days.
 

pbateman

New Member
great info

Thanks to everyone who replied. All of your suggestions have been very helpful.

Goom - a special thanks for the last link that you posted. Wish I would have found that page before I started.

Based on the info there, I have decided to go with the irf3706 mosfet. With 14V and 12A max current @ 20000hz the mosfet should only generate a total of 1.8W - .8 due to switching and 1 due to resistance.

I will update you guys after I get the new mosfets.

Many thanks again!
 
Top