MOSFET Power Dissipation

[K1JOS]

I would like to use a MOSFET such as a STU95N2LH5 in a TO-220 case to switch a 12vdc 10A motor load.    In addition I want to control the MOSFET gate directly using the PWM from a PIC or Arduino so I can control the motor speed as well.  I am trying to figure out how much power dissipation the MOSFET will experience and how much heat sinking I will need.  I have read on-line calculations ranging from simple to complex.  For example, some sites say that when switched on fully this MOSFET has a Source to Drain resistance of only 0.0049 ohm.  So with the 10A source to drain load it should only drop 0.049 volts and therefore be using 0.49 watts of power at 100% duty.  I think 1/2W would need some light duty heat sinking.  On the other hand, other sites say you have to use iterative calculations and using junction temperature and expected ambient temperatures to calculate both resistive and switching losses.

Is there a less technical or more practical way of getting into the right ballpark?

[Pillager]

Is there a less technical or more practical way of getting into the right ballpark?

Yes, there is. Mount a heatsink and give it a try. If it gets to warm, use a bigger heat sink. If not, leave it :-)

And btw.: The most power is dissipated in the mosfet during the transition from ON to OFF and back. It may not be for a long time, but depending on the switching frequency it will be more or less often per second, bla bla ...

But, like I said, just give it a go and keep an eye on the mosfets temperature :-)

[SebG]

The Mosfet is an N-channel so you would need to tie the source to ground and the drain to the motor negative wire ( and the positive wire of the motor to 12V) since the gate voltage needs to be with respect to the source. If you are using a pic powered with 5V it should be ok. However, from the data sheet if the gate-source voltage is 3 volts and the current is 10A then Vds will be about 1V so it will need to dissipate 10W of power.  Arduino, i believe, runs on 3V and it's pins output 3V so it's not good for this. 

In this type of thermal design you need to consider how fast can the PWM output be switched from High to Low or vice-verse.  If the output impedance is low and the input impedance is high with a low capacitance then you will get a faster rise/fall time. Pic powered by 5V should be good to drive the mosfet.

[miceuz]

At 5V gate voltage Rdson will be 0.007 Ohm max, this gives you 0.7W steady state dissipation.

5V gate gate charge will be ~15nC. At 20mA from micro pin you will be able to switch the mosfet in ~750ns. If your PWM frequency is around 1kHz, your switching losses will be negligable - I get 0.015W using a ballpark formula from this article: http://www.eetimes.com/document.asp?doc_id=1225701

I've done similar calculations just a week ago.

[K1JOS]

Thanks people !  SebG says my dissipation will be 10W and  miceuz says it will be closer to my estimate - around .7W.


Help !!!
Who is correct ?

[wraper]

Thanks people !  SebG says my dissipation will be 10W and  miceuz says it will be closer to my estimate - around .7W.


Help !!!
Who is correct ?

Lazy to calculate but SebG is definitely wrong, EDIT: I see he said that figure at 3V gate voltage. So he might be right in those conditions, anyway it is not a good idea to run it with less than 5V gate voltage.

[SebG]

Most Mosfets need 10V between gate and source to be fully on but some logic level mosfets can be fully on at lower voltages like 5V or down to 2V.  The STU95N2LH5 mosfet has a good on characteristic when the gate-source voltage is 5V but anything lower it starts to have a higher resistance.  I looked at one of the graphs on a datasheet that showed at 3V Vgs and 10A Id the Vds was around 1V that is it's drain-source resistance was about .1 Ohms.  That is why it would dissipate 10W with a Vgs of 3V and i recommended using the PIC powered by 5V.

[Bukurat]

SebG is basing  his calculations on a 3.3V pulse, not 5V.

If you are using an arduino development board pick one that runs at 5V., same if you are using an Atmel MCU.

John

[Charles Creations]

Another option is to use a BJT to drive the MOSFET with a 12V signal. The arduino pin will drive the BJT which can apply a full 12V to the MOSFET and give it the quickest full off to full on time. This will limit the power dissipation. I have ran MOSFETs at almost 130KHz switching several amps, and they got fairly warm on a 6 degree C per Watt heat sink. I don't think you should have excessive heating issues even if you don't employ the BJT. However, the BJT offers the best protection for the arduino since some of the larger MOSFETs have greater gate capacitance. A series resistor can alleviate this as well.

[a210210200]

Or you can just get a lower voltage logic level MOSFET if you want to use 3.3V (http://www.irf.com/product-info/datasheets/data/irlb8314pbf.pdf), using a driver for the mosfet is going to be higher performance better and for the micro. You should make sure you don't exceed the microcontroller's driving current level, in our teaching boards we limit it with a series resistor. (This will make the mosfet heat up more as it will slow the gate charging but you don't really want to heat up your microcontroller till it burns out either)

Also it is a good idea to have a relatively high resistance resistor on the gate line to ground to make sure when the microcontroller is tri-stated the mosfet is fully off and can't float up and get damaged by a half on state.

Another thing is that switching a 10A inductive load is not the same thing as switching a 10A resistive load. Also if you can get away with a low PWM frequency then the gate capacitance won't be as big of an effect vs. just having a low on resistance.

Whatever you do you need to be able to both sink/source current to the mosfet quickly when doing PWM as if you just use say one BJT and a pull-up/down resistor your PWM signal is going to get trashed.

You can get mosfet gate drivers but they are meant for high performance switching and a low current motor driver or discrete half bridge would work well as well. (You basically need a half bridge in some shape or form and whatever you do don't turn both sides on at once)

[Seekonk]

I use opto isolators to drive FETs.  300 ohm in series with the LED  and a 750 ohm from the emitter to common.  That gives decent 12V drive to the FET at 490hz.  A single BJT will invert the signal and power the motor till the program is stable.

Where do you get the 10A value for the motor?  Is that STALL current which can be quite high.  Don't forget to use a flyback diode on the motor , fast and at least 20A.  I would use a couple FETs in parallel.  But burning things up is the best way to get an education.