Mosfet getting hot (140° F)even when in spec

Discussion in 'The Projects Forum' started by Kuhny1, Aug 26, 2016.

1. Kuhny1 Thread Starter New Member

Jul 2, 2015
20
3
So im using the IRLB8721PbF. Which is rated at 30v at 60 amps.

I'm trying to make a laser harp which is using a homemade experimental galvo. All I use is a permanent magnet on a shaft and a electromagnet that has a resistance of around 3 ohms. So, theoretically, it should draw around 4 amps max at 12v, right?

That's well within spec of this fet. So it shouldn't be getting redicuosly warm. I do have a flyback diode in parallel with the EMagnet also. I guess if that wasn't there though, the mosfet would just instantly die.

I am PWMing this mosfet too at the atmega328 standard 980hz for position of the "galvo".

My power supply can supply 18 amps btw. So even if my calculations we're wrong, there still isn't enough power to blow it.

If anyone can help me figure out why this is heating up so fast, I would appreciate it

Thanks!
Data sheet:

2. Papabravo Expert

Feb 24, 2006
10,338
1,850
Depending on how you are switching the device it may be spending an inordinate amount of time in the linear region. The power is i squared r, or (4*4)*3 = 48 Watts. Look in the datasheet to see how many degrees C the temperature will rise per watt. Now 150° F is not all that high in terms of semiconductor temperatures; it is only 65°C

3. Kuhny1 Thread Starter New Member

Jul 2, 2015
20
3

So it seems like I should expect a case temperature of around 110°C on the case temperature? Thats seems ok but i think that may just melt a bread board, guess i better put a sink on it. I was looking at the 2.3 watts per degree on the Junction to Case temperature. Is that right?

4. Papabravo Expert

Feb 24, 2006
10,338
1,850
Look carefully at the units. They should be °C/watt which means that at 48 watts and 2.3 °C/watt you should expect a 110 °C temperature rise from case to junction.

5. Kuhny1 Thread Starter New Member

Jul 2, 2015
20
3

I did say 110°C. Or are you saying to add that temperature on to what it is currently?

6. joeyd999 AAC Fanatic!

Jun 6, 2011
2,756
2,914
If you are not using a heat sink, you need to use junction to ambient, not junction to case.

7. Kuhny1 Thread Starter New Member

Jul 2, 2015
20
3
Oh... well, I guess I really need a heat sink. This thing would fry itself pretty quickly if that's that's the case lol.

Ok thanks I'll stick a beefy heat sink on

8. SLK001 Well-Known Member

Nov 29, 2011
848
240
I believe that, like Papa said, your are spending too much time in the linear region. The Rds(on) for your device at Vgs=4.5V is 12mΩ. If you are passing the max of your power supply, that would be 12*12*0.02 or 2.88W. At your current draw of 4A, it would be less than 1/3 of a watt.

absf likes this.
9. wayneh Expert

Sep 9, 2010
12,362
3,220
And show your circuit. You may have something slowing the switching and causing heating more than necessary.

Haha, SLK001 beat me to it.

10. Kuhny1 Thread Starter New Member

Jul 2, 2015
20
3

Ok here it is

Sorry if it is low quality but its what I have made in short notice lol.

File size:
29.8 KB
Views:
22
11. ronv AAC Fanatic!

Nov 12, 2008
3,391
1,446
Are you sure about the 3 ohms for the coil?

12. OBW0549 Well-Known Member

Mar 2, 2015
1,385
929
Hmmm... this is a head-scratcher, for sure. I see no reason for your MOSFET to be running hot; in fact, by my calculations it should be running stone-cold, with a power dissipation of barely a quarter of a watt (4A * 4A * 0.016Ω) static dissipation with 100% on the PWM, plus whatever the switching losses are. These I would expect to be minimal, given the low PWM frequency.

I'm not very familiar with the Atmel parts, but are you sure you've got the pin driving the microcontroller configured correctly as an output?

I can't think of anything else that could be causing this, other than perhaps a wiring error.

13. Kuhny1 Thread Starter New Member

Jul 2, 2015
20
3
Well, I measured 2.8ohms on the multimeter. Close enough to 3 so i rounded there.

Nov 29, 2011
848
240

Nov 23, 2015
2,067
230

16. SLK001 Well-Known Member

Nov 29, 2011
848
240
Also, you might want to put a 10k from the gate to ground to be sure that the FET is pulled to gnd.

17. ronv AAC Fanatic!

Nov 12, 2008
3,391
1,446
So it's in the ballpark.
Max on resistance of the FET with logic level drive = 16 milliohms
4.3 amps squared X.016 ~ .3 watts.
62C / watt junction to case ~ 18C or 64 F rise plus 76F for the room ~140F.
60C is okay, but uncomfortable. You can add a little heat sink if you like, but it will run a long time at 60C.
I think the diode is okay.

wayneh likes this.
18. Kuhny1 Thread Starter New Member

Jul 2, 2015
20
3
Im using the arduino IDE for for programming the AVR, and the analogWrite() function automatically sets the pin to an output.

I have narrowed it down to the coil just by connecting an LED and needless to say it ran cold just fine, so i guess I can assume that, that narrows out the AVR side, but then again, its not an inductive load. I havn't tried with a motor yet and i will try that and see if it gets warm. I have also tried putting 2 FETs in parallel to try to spread out the power but that didn't really seem to help.

I should mention,

Before i ran into this weird problem, I had wired the FET up backwards ie. when the FET was LOW which was more than 90% of the time, it completed the circuit and then there was a nice puff of smoke from the coil, and the FET was completely cool even after a near dead short of ruffly 10+ amps (50 amps under rating ). I thought i fried the FET as well, so I connected the LED to the output and noticed my issue (wired backwards) and then messed around with some tests to see if it was ok and it seemed fine. I have ordered 4 of them so, had i busted it i would've just replaced it. (And Yes, i have tried my other replacements too, they also heat up rapidly).

Im stumped as it stands right now lol.
I will put all 4 FETs in parallel if I have to.

19. Kuhny1 Thread Starter New Member

Jul 2, 2015
20
3
Ah!

Sorry forgot to draw that in...
yep i have a 10K in there. learned that the hard by frying my first coil
Check my last post, not only did i forget the pull down i also had it wired inverted.

Thats all fixed now. Still a lot of heat dissipation though for whatever reason

Updated schematic in the attachments

File size:
30.9 KB
Views:
12
20. Papabravo Expert

Feb 24, 2006
10,338
1,850
How much current can the microprocessor pin source and sink? This current needs to charge and discharge a pretty big capacitance which will slow things down a bit.. Try to look at the waveform with an oscilloscope if you can. You might be greatly surprised at how slow it is and how much time is spent in the linear region.

JWHassler likes this.