MOSFET heat question

[skibum100]

I'm new to using MOSFET's and i thought I understood the basics through lots of reading in last several weeks. I prototyped my circuit and found that the MOSFET is getting hot to the touch even though I'm using it well below the rated current and voltage.

This is a simple buck converter to drive a 12V 50 Watt light. I have plans to use the circuit to have the ability to:
1.) flash the light at 0.5 Hz (on 1 second, off 1 second)
2.) ability to dim the light (this is why i'm using a buck converter circuit)

I also wanted to use this MOSFET on a higher voltage / wattage light bulb but started with a smaller bulb to validate my circuit. I can add a heat sink but wanted to make sure this is normal before I continued using this particular device.

Thanks in advance....

View attachment 66509

 

[ronv]

The way you have it now it is not a buck converter yet, but maybe you have more to come.
The way it is now the FET is always on. If your numbers are correct it is disapating 1.7 watts.
Temperature rise of 62C/watt puts it at about 130C at room temperature.
That FET has an on resistance of .2 ohms so you might want one with a lower Rds on.

 

[skibum100]

Thanks for the quick reply.

It's a buck converter circuit with Vg tied to ground to keep the switch closed. I was trying to test at 1:1 input output voltage. I'll be using PWM into the Gate.

I'm confused where you got the C/Watt and FET resistance. The data sheet I'm looking at says Rds (on) 0.117ohms and thermal resistance of 40C/watt for junction to ambient.

Do you have any suggestions of FETs that would be better suited for driving a incandescent light bulb? This bulb is 50Watts but I'm looking to drive bulbs up to 300-400 watts (at around 80V)

 

[ccurtis]

According to the late analog guru, Robert Peese, (he calls this the 5 second rule) if you can hold your finger on a hot device for 5 seconds the case temperature is about 85 deg C. Hotter than that, or too hot to touch, dot your finger with saliva and touch the device for a fraction of a second. If the moisture dries quickly, it's probably about 100 deg C. If it sizzles instantly, the temp may be as hot as 140 deg C.

I just wanted to slip that in here. Maybe help to determine if really too hot. Rest in peace, Mr. Peese.

 

[skibum100]

Good information....

The device starts off where I'm able to touch for over 5 seconds, but after a short period of time (1-2 minutes) gets to the "too hot to touch". At one point I did use saliva and was able to touch shortly still without too much pain, I had another MOSFET sizzled and shortly after smoked (it was a different type - and I threw that one out). I'm afraid if I let the device run for 5-10 minutes, it would definitely be at the 140C range.

I'm just trying to understand the data sheets - this device claims 140W dissipation of power, why when I'm at only around 2 watts is it getting so hot, am I misunderstanding the data sheet???

Thanks again...

 

[ronv]

Sorry, I missed the case type and just assumed the 62C/watt. One data sheet I see says 40C another 30C. Both spec. .2 ohm max Rds on, although you are getting less in your test. At 40C/watt that still puts it at 93C, One of the things with FETs is the resistance goes up with temperature.
The 140 watts is with a heatsink that keeps the case at 25C. Maybe we can find a better one.

https://www.electro-tech-online.com/custompdfs/2012/08/sihfp914.pdf

 

[skibum100]

The MOSFET you suggest has almost the same characteristics as the one I'm using...did I miss something?

 

[skibum100]

I think I found one that works

https://www.electro-tech-online.com/custompdfs/2012/08/DS99967BIXTY-TA-TP32P05T.pdf

 

[audioguru]

An old fashioned incandescent light bulb operates normally at an extremely high temperature, maybe 2000 degrees C. Then the resistance of its filament is much higher than it is when cool. So when it is dimmed its filament is cooled then it uses much more current than you think. Then the driver transistor becomes hotter than you think.

 

[skibum100]

Yeah, i understand the changing resistance of the incandescent bulb that is why i was doing this test at steady state. All the numbers I posted are of the buck converter with G tied to ground (FET always ON or 1:1 converter). I know when I start dimming and/or pulsing the bulb that the heat in the MOSFET is going to rise, that is why I was asking for some advice on MOSFET selection. I also did tests using the transistor to dim the bulb, as well as some pulsing, but i wasn't able to run these test that long as the MOSFET heated up pretty quickly (20-30 seconds).

 

[ronv]

I just posted that data sheet so we could both have the same one.
The one you have is much better for Rds on, but only good for 50 volts.

 

[skibum100]

Well I was trying to keep cost down, I did find 2 others that would work but they are $4 and $8 at low quantities...

747-IXTH90P10P https://www.mouser.com/ds/0/Infineon/SPP80P06P H_Rev1.5-45623.pdf
747-IXTP76P10T Could find the data sheet...but Low Rsd, current rating is good as well as voltage rating.

 

[alec_t]

Deleted

 

[alec_t]

At 2W continuous the FET will get hot without a heatsink.

 

[ronv]

If in the long run if you are looking at say 5 amps at 60 volts (300 watts) the first one will be ok. 5^2 X .023 = .575 watts. X 62C/watt = 60C at room temperature. If it is a regular bulb the current will be higher as the average voltage goes down but the duty cycle will also be lower. It may still be pretty hot when dimmed.

 

[ChrisP58]

Can you float the negative side of the lamp? If so, you can use an N-channel mosfet. They are less expensive, available with lower RDS-on values, and are easier to drive.

 

[skibum100]

I think the answer is yes to your question, but I'm not sure what you mean by "float" the negative side. The lamp is driven off a battery system, which has the negative terminal which is not connected to ground, so yes...it is floating...

 

[ChrisP58]

What I was asking is if the negative terminal of the lamp could go to the - cap terminal and one side of the inductor. The other side of the inductor goes to the anode of the diode and the drain of an N-channel mosfet. The mosfet source is grounded. The positive side of the lamp is at +12V, along with the + cap terminal and the cathode of the diode.

Since the mosfet source is at battery negative, a logic level gate can easily be driven by a uC pin (provided you are not switching it at hundreds of KHz.)

Mouser is currently listing US$0.86 for the Infineon IPP16CN10L 100Volt, 0.016 ohm logic level mosfet in a TO-220 pkg.

 

[skibum100]

So I was able to test with the MOSFET. They worked well at full power (MOSFET on 100%) and also worked in pulsing the bulb. One issue I encountered is that when trying to dim the bulb I encountered excess heat in the MOSFET. I did a little more research and found the are 2 power dissipation equations for mosfet's, one for Pconducting and one for Pswitching. The example I found Pswitching is and order of magnitude higher than Pconducting. So finding a MOSFET with low Rsd fixed the power for the conducting phase but not when trying to dim the light. I was running the MOSFET at 30kHz originally but when I encountered this other heat issue I decided to play with the frequency and duration of the on cycle. I went as low as 30 hz, and found the MOSFET heat got better with lower frequency, but the capacitor started to heat up. I imagine this is because the cap is fully changing and discharging each cycle, but need some help / explaination before arbitrarily changing the cap value.

 

[MikeMl]

I think you dont understand that to get rated power dissipation when using any FET or transistor it MUST BE BOLTED TO A LARGE HEATSINK! A TO220 device in free air can only dissipate a couple of WATTS. To get rid of 50W of heat, it must be bolted to a large heatsink with an area of several HUNDRED square inches...