MOSFET Power Dissipation

Discussion in 'General Electronics Chat' started by guitardenver, Aug 6, 2012.

  1. guitardenver

    Thread Starter Active Member

    Jan 24, 2009
    30
    0
    I'm trying to pick the right MOSFET for the job. I really would like to know how to calculate how much current I can put through the drain without the need for a heatsink.

    I have a 12 source.
    The transistor is used as a switch. It will be fully on for hours or more.
    The load will draw max of 3.5 Amps.
    It will operate at a Max temperature of 50 deg C. I'm designing for the most extreme (inside a car in Arizona, CO) There will be stagnant air (Very little air flow.)

    I was thinking about this MOSFET
    http://pdf1.alldatasheet.net/datasheet-pdf/view/101295/ZETEX/ZXMP3A17E6TA.html

    Looking at the graphs:
    Max power dissipation at 50C is .8W
    Rds(on) at 50C is about 1.1 ohms (Vgs = 10V) (Id = 3.2A)

    My calculation:
    P = I^2 *R
    P = 3.5^2 * 1.1
    P = 13.5W

    This is WAY to much to dissipate without a heat sink. Did I do this calculation right? How can a make this more accurate to my conditions? 12V 3.5A 50 deg C?

    Can anyone reference me to a P-channel MOSFET that can do this without a heatsink?
     
  2. mcgyvr

    AAC Fanatic!

    Oct 15, 2009
    4,771
    971
    Rds(on) is .07 ohms when fully on like a switch...(under 1W)
     
  3. guitardenver

    Thread Starter Active Member

    Jan 24, 2009
    30
    0
    At what voltage is this MOSFET fully on like a switch? The datasheet says 20V Max? I only have 12V Max. And that .07 ohm rating is at 25 deg C not 50. What is Rds when Vgs is 12V at case (Not junction) temperature of 50 deg C? The junction temp is the thermal resistance * case temp right?
     
  4. Ron H

    AAC Fanatic!

    Apr 14, 2005
    7,050
    657
    That graph is normalized. At 50°C, Rds(on) will be about 1.1*0.07Ω. Unfortunately, this still results in Pdiss=0.84W. With Rθja=113°C/W, this will result in a junction temperature rise of 95W. Add this to 50°C ambient, and your junction temp will reach 145°C, which is much too close to the absolute max of 150°C for comfort.

    You need a FET with lower Rds(on). Fortunately, many are available.
     
    absf likes this.
  5. Ron H

    AAC Fanatic!

    Apr 14, 2005
    7,050
    657
    10V Vgs is enough for almost any MOSFET. Logic level devices have lower thresholds, but you don't need one, unless you are driving the gate with logic such as CMOS or TTL.
     
  6. guitardenver

    Thread Starter Active Member

    Jan 24, 2009
    30
    0
    Thank you for the reply.

    I picked a new MOSFET:
    http://www.vishay.com/docs/66823/si1441ed.pdf

    Here are my calculations.
    According to the On-Resistance vs. Junction Temperature graph, the Rds, at 150°C the Rds will be (1.48Ω * .034(typical)) = .05Ω

    My load will draw 3.5 amps. So, the MAX power dissipation will be:
    3.5A^2 * .05Ω = .613 Watts

    Next, find what the junction temperature is at .613W by using thermal resistance:
    .613W * 60°C/W = 36.9 increase from the ambient temp
    So, ambient temp(50°C) + 36.9°C = 86.8°C (Junction Temp)

    Since 86.8 < 150(MAX), This MOSFET can dissipate my power draw without a heat-sink right?

    All this takes into account the amount of heat the package can dissipate right? Not just the junction itself?
     
  7. shortbus

    AAC Fanatic!

    Sep 30, 2009
    4,019
    1,542
  8. Ron H

    AAC Fanatic!

    Apr 14, 2005
    7,050
    657
    You missed footnote d on the thermal resistance:
    What is your application?
    Are you driving it with a logic level?
    What is the switching frequency and duty cycle?
     
  9. ErnieM

    AAC Fanatic!

    Apr 24, 2011
    7,395
    1,607
    Your calculations are correct. "Junction to ambient" means the thermal impedance from the silicon to the surrounding air. While I would use the max number of 80°C/W myself you still should be safe there.

    There is another path for heat to escape hinted at by the "Junction-to-Foot" rating. The drain is the bottom of this chip and it is both electrically and thus thermally connected to the mounting pads. One method to get the heat out is to provide a large area of copper on the PCB itself to act as a heat sink.

    Just remember this is an electrically connected mass (so don't short it to where it should not be shorted to).
     
    guitardenver likes this.
  10. Ron H

    AAC Fanatic!

    Apr 14, 2005
    7,050
    657
    I would use 125°C/W. See my previous post.
    I just realized that both FETs are P-channel, so you must be using them as high side switches. You don't need a logic-level MOSFET for a 12V high side switch.
     
  11. guitardenver

    Thread Starter Active Member

    Jan 24, 2009
    30
    0
    Yes I did miss that footnote! I didnt see that it was for t<5s. I will redo my calculations with 125°C/W. Thanks!
    It came out to be a final junction temperature of 127°C. Is that Ok??

    My application is switching a 3.5 amp load inside a car. (It can get really hot in there about 50°C where I will have it).

    I am switching it with 12 volts. I can use a voltage divider to make it ten if necessary.

    In my original post I said this device will be on for hours or more. So the duty cycle is 100%.
     
    Last edited: Aug 7, 2012
  12. guitardenver

    Thread Starter Active Member

    Jan 24, 2009
    30
    0

    Does that mean the new MOSFET I picked is a logic level MOSFET? If so that would be ok too, because I have a 2n3904 being turned on by an MCU which turns on the MOSFET. If I can bypass that gp transistor and turn the MOSFET on straight from the MCU that would be less components on my PCB later.
     
  13. Ron H

    AAC Fanatic!

    Apr 14, 2005
    7,050
    657
    A logic level high side (PMOS) switch is useful if you are switching a load to the same vcc as your signal source (MCU). Since you are switching an NPN which drives the PMOS, you don't need a logic level device, but, in the case of the SI1441ED, you can use it, but you need to be sure that |Vgs| is less than 10V and more than 4.5V when the device is on. An R-R resistive divider (e.g. 1k - 1k) between +12V and the collector of your NPN should work well.
    If you can switch the low side of the load, you can use a logic level N-channel MOSFET and avoid the need for the level translator.
    Do you have a space problem? I'm curious why you don't choose a lower resistance part, like SUB75P03-07
    I confess that I didn't check price or availability of this, but I would think there are better parts than the one you chose.
     
  14. guitardenver

    Thread Starter Active Member

    Jan 24, 2009
    30
    0
    That was also in my original question. I was open to suggestions for MOSFETs. I'm just going down the list on P-Channel MOSFETs on mouser.com. I'm looking for the usual, meaning cheapest one that will reliably do what it needs to do for a long time. So money is a slight concern (I might have to buy 5 or 6 of these). And I just like to be efficient (power efficient, money efficient). A 3 dollar MOSFET just seems a little much? I can get a 5A relay for less than a dollar. But like you asked, I do have a space concern! It's not critical but the smaller it can be the better.

    So any suggestions? I looked at the one you said but Mouser nor Digikey had it in stock. I will keep looking as well.
     
  15. Ron H

    AAC Fanatic!

    Apr 14, 2005
    7,050
    657
    Can you switch the low side, or is the load permanently connected to GND?
     
  16. guitardenver

    Thread Starter Active Member

    Jan 24, 2009
    30
    0
    The load is permanently connected to ground. I can only switch the high side.
     
Loading...