That one has room for head, shoulders, and a fat mother in law!

I hope you got it in the surplus bin and paid about 98 cents.

Does it? Say if I have 30A going through, that would be 30 * 30 * 0.1 = 90W
Is this the right way to estimate the power?

 

Yes. I just didn't expect you'd be anywhere near 30 amps after measuring 6 amps.
However, I am grateful for the LOL.

 

Just remember 50W and 100W resistors generally require a LARGE heatsink to safely reach their rated max disspation.

Just because the resistor is big, metal and has little fins doesn't mean it can dissipate 100W all by itself...

 

Just remember 50W and 100W resistors generally require a LARGE heatsink to safely reach their rated max disspation.

Just because the resistor is big, metal and has little fins doesn't mean it can dissipate 100W all by itself...

I'll keep it in mind, thanks!

 

Today I finally got around to taking the measurements, meanwhile I've been "field testing" the temporary mock up with the two MOSFETs and it worked fine all this time. Today I hooked up the power resistor between the ground and the sources and measured both the on-time at the gates and the voltage level at the power resistor. The on-time, seems to be around 25-30us, that seems fast enough? The level at the power resistor was roughly around 2.2V, which means the current with ~0.1ohm resistor is about ~22A! I guess this explains why the single MOSFET unsoldered itself?

 

So I've been using the two MOSFET setup for slightly over a year in the field and everything was fine until a couple of days ago:

To recap, I had a PWM motor control built that varied the motor speed depending on the thermistor voltage. The first design was using a single MOSFET, which very quickly got so hot, it unsoldered itself.
I then added another in parallel as pictured above and this circuit worked for over a year.

Please note that the actual components used are not the ones pictured, here's what I actually have:
- R1 and R2: 27ohm;
- T1 and T2: IRF8734pbf;
- SD1: STPS1545;
Datasheets attached.

A couple of days ago I noticed that the fan continued to spin with the system off and I had to manually pull the fan's plug.
Later that day I had to turn the fan back on and it was spinning for a while and then stopped by itself and I felt a burning smell.
When I took the box with the above circuit apart, I found signs of fire, the board had two black holes where the MOSFETS were before...

This weekend I'm redesigning the entire PWM controller, I'm not using Arduino Mega2560 and it's PWM to control the circuit, so the IN now varies 0-5V.
I also added another MOSFET in parallel to make sure they can handle the current.
The new system was working for a short while until I noticed, while the system was off, the fan came on...
Upon examination I found that one of the MOSFETs now had about 12ohm resistance in OFF state between the source and drain.
Again, this seemed to have happened while I was working on the Arduino sketch with the system OFF.

Any idea why that happened?
Another question I had is regarding R1 and R2. Should I maybe use just one resistor from IN to all three gates connected?
My thinking is maybe because R1 and R2 (in the latest version there is also an R3 for the third MOSFET) are not exactly the same, the ON-OFF transition for different MOSFETS does not happen at the same exact moment so the current distribution is not the same for a moment.

Appreciate any input, need to make this work ASAP without blowing up

 

You have a lot of little things that may be adding up.
You stated you have no heat sink? The little S08 IC has terrible thermals without a PCB board heatsink.
At 22 amps the diode has the same problem.
The 324 can only drive within about 1.5 volts of the positive rail and then only about 20ma. So if possible raise the voltage to it so the fet will turn on harder. You might as well make the gate resistors 150 ohms or so.

 

You have a lot of little things that may be adding up.
You stated you have no heat sink? The little S08 IC has terrible thermals without a PCB board heatsink.
At 22 amps the diode has the same problem.
The 324 can only drive within about 1.5 volts of the positive rail and then only about 20ma. So if possible raise the voltage to it so the fet will turn on harder. You might as well make the gate resistors 150 ohms or so.

Thanks for the reply! That's right, no heat sink at the moment, I'll try adding something.
The diode however stays cold as far as I can tell, I think it's because it doesn't actually have 22A flowing through it (I think) and only has the back flow current, which should be significantly lower.
I think you are referring to LM324 from the schematic I posted in my earlier posts. I'm not using that circuit at the moment, I'm using Arduino and its PWM outputs to drive the MOSFETs.
I checked with the scope, PWM signal now ranges 0-5V. One thing I've noticed though is that the edges of the signal are somewhat roundish, not as square as I remember them from looking at LM324 output.
What else can I do other then adding heat sinks?
What would 150 ohm resistors change?

Perhaps you should start your own thread...

If you read the first post, you'll see that it IS my thread...
Yes, it's old, but I'm back working on the same project so I thought it only makes sense to continue the old thread.

 

This is the exact schematic that I currently have (the one on the right with three MOSFETs):

 

What voltage is Vdd?

 

Pull down resistor on the MOSFET gates.

 

How much current does the motor need? Either way you should not be getting that much power dissipated in the mosfet, unless the mosfet is staying largely in the linear region and dissipation lots of power. What is your gate frequency? Could you post some shots of the gate voltage, D-S voltage and current look like?

 

He has a 22 amp motor with 3 S08 FETs without any heat sink. 0.00167 ohms @ 22 amps is still 0.8 watts @ 120C/watt is about 120C if he was driving it at 10 volts and not 5. His pwm is only 300 Hz. but it probably spends a lot of time switching because of the low voltage and low current available to drive 3 of them in parallel.

 

This one would probably do it without a heat sink. It will still be real toasty.

http://www.mouser.com/ds/2/302/PSMN1R1-30PL-352613.pdf

 

What voltage is Vdd?

Between 13.5V-14.4V.

Pull down resistor on the MOSFET gates.

What value should it be? Around 10K?

How much current does the motor need? Either way you should not be getting that much power dissipated in the mosfet, unless the mosfet is staying largely in the linear region and dissipation lots of power. What is your gate frequency? Could you post some shots of the gate voltage, D-S voltage and current look like?

I measured the motor drawing as much as 22A. Accordingly to Arduino docs, PWM frequency is about 490 Hz.
I'll try to get some measurements.

 

If that is the case, I´d suggest getting some mosfet in TO-220 or TO-247 case. Such tiny dip package cannot bear the dissipation caused by less-than-perfect drive, but if you stick a mosfet with similar parameters in larger package it should do it without overheating even without any heatsinking.

 

The other thing you might be able to do depending on your space is to solder all the drain and source pins to a small piece of copper sheet.
Check your diode when running at 50% duty cycle. That is where it works the hardest.

 

I wanted to get it done today, but looks like it's better if I get a different MOSFET in a larger package.
This one seems good, looks like it can handle 120A?

This one would probably do it without a heat sink. It will still be real toasty.
http://www.mouser.com/ds/2/302/PSMN1R1-30PL-352613.pdf

Would you recommend getting that one?

 

For the today project try going to the hardware store. Most of them sell copper sheet in little squares for less than the shipping on the bigger FET. 1/2" X 1" on each side of your little FETs would probably Keep them from thermal runaway. Buy enough for an square inch or two for the diode.

 

There ain't no sucha thing as a 120 amp (continuous) mosfet. The internal wires would melt. That's for fast pulses only. Revert to the theta junction to case and do your thermal math from there.