[problems] mosfet easily blown.

[sprooch]

Hello,

I have a problem with a mosfet (IRF2807) in my Boost converter.
It is easily get broken.
I have tried to hold it on the pin to avoid electrolytic. I also put a gate resistor.
The drain to source voltage specs is still according to the datasheet, even I put it lower than it can handle.

Can anyone help me for this?
I have blown 6 of them till now. I still have the last mosfet. I hope I can save this last mosfet.

thanks

 

[CrabMan]

post your circuit diagram for more clarification, what is the switching frequenncy and what current are you passing through the mosfet?

 

[sprooch]

When I tested it (real, not simulation), I used only 11 volt with 6 ampere in the input of boost converter.
Pulse generator of 5 volt 150kHz , 70% Duty cycle.
It worked fine...

temperature was increased in period of time (it did not reach 175 celcius)
I switched on and off the boost converter several times.
and the problems came....

It did not switch anymore.
Drain is shorted with Source....


I attached the boost converter circuit.

 

[max0412]

Is that your actual test bench circuit? If it is you should have a FET driver IC between your function generator and your mosfet. That is one possible explanation as to why your fet is blowing.

A boost converter uses feedback to maintain output regulation, you should be using a control IC. The IC adjust the duty cycle of the mosfet drive.

 

[sprooch]

max0412,


Are there any explanation why it will be blown if we dont use a driver?

Yes, I am using a microcontroller also for this boost converter.
I just want to try the converter in open loop (no feedback), and then put the converter, the driver and the microcontroller together.[/b]

 

[max0412]

Yes your mosfet is large and therefore has a large effective input capacitance. You need to provide a large peak current to the gate of your Fet to turn it on rapidly you also need to provide a low impedance path to discharge the parasitic capacitance to turn it off quickly. If you fail to do this the Mosfet will be operating in it’s linear region and will exceed it’s safe operating area depending on Ids and Vds and maximum allowable time the fet is subject to this stress. The datasheet will show a graph typically labeled SOA.

You can get an approximation of the switching speed of a particular fet using a particular driving source with this equation.

Ton = Qg/Ipk

Qg is in the datasheet Ipk is the peak current your driver can source. So you say you want to drive this FET directly off a PIC.

Ton = 160n/25mA = 6uS -----> Not a good idea.

You should use a mosfet driver. If you don’t have one available you could construct a discrete one using transistors. Even this may not be ideal given the limited output of a PIC 25mA but definitely an improvement until you get a proper drive.

I posted some example discrete Push pull drivers here.Use your fifteen volt input for your drive signal. Try and find a more suitable mosfet as well.

&highlight=

A dedicated SMPS controller would be the better route unless this is just a school project.

 

[sprooch]

I have a driver IR2301 and I have just put the driver before the gate mosfet.

I will look at the Ipk (peak current) of IR2301 and calculate it using the formula Ton.

How much will be the nice "Ton" for the mosfet?

Added after 18 minutes:

max0412,

This is what I calculate from taking the Ipk (I think it is Io+ in the datasheet)

Ton=Qgs/Ipk= 160n/200mA

Ton=0.8us-------> I dont think it is a good value

is that what you mean?

I attached the driver's datasheet also...

 

[FvM]

IR2301 is goog for lower gate capacitance, high voltage transistors. You should be able to source at least 1 - 2A to the gate. For full FET protection, the driver circuit should include a latching current limit, as PWM controllers usual have.

 

[Audioguru]

You used a signal of only 5V to the gate of the Mosfet but it needs 10V to fully turn on. it gets extremely hot when it is not fully turned on.

175 degrees C is its max allowed chip temperature. You cannot measure the chip. If the heatsink tab is 175 degrees C then the chip is much hotter and is melting.

 

[max0412]

You used a signal of only 5V to the gate of the Mosfet but it needs 10V to fully turn on. it gets extremely hot when it is not fully turned on.

I don’t think that’s the problem. The FET in question can pass about 35 or 40 Amps even at an elevated temp with 4.5Vgs.

Since the OP doesn’t state the power output I can only estimate from this statement

When I tested it (real, not simulation), I used only 11 volt with 6 ampere in the input of boost converter.

that the FET will be passing nowhere near that (35 to 40A).

This includes the fact that he is likely incorrectly using a DVM to measure the current. But he does have fifteen volts available so it would be wise to use that as your driving signal. The mosfet is grossly over sized .

Either beef up the driver or select a more suitable fet.

Like I said earlier running it openloop isnt a good idea. The inductor current ramps up at rate determined by the inductance and the input voltage and the on time.The current ramps down based on the Vout V in differential and the off time and inductance .

Vl = L di/dt

So if you aren't generating a sufficient output voltage to ramp your current back down well then your current would keep ramping up---> not good.The switch would fail in a way similar to what you describe. I'm assuming you checked all this in simulation.

IR2301 is goog for lower gate capacitance, high voltage transistors. You should be able to source at least 1 - 2A to the gate

Where are you seeing those current values for that driver.

Heres a decent driver low propagation time and decent pk current and best of all it doesn’t cost an arm and a legg. MIC4427

http://www.micrel.com/_PDF/mic4426.pdf

EDIT

I forgot to add if you have a scope a simple way to see the current waveform through your FET,is to place a small low value resistor between the source and ground.This way you know what your fet is passing current wise,also monitor your Vgs.Definitely use your 15v Input as your gate signal if you do this.You should use this anyways as long as your input voltage will never exceed your fets max VGS.

 

[Audioguru]

Are you guessing that you have a "typical" Mosfet?
What if it is a "minimum" one? Then it won't conduct well and it will melt.

I always use the "worst case spec's". I don't test every device that I buy. I depend on the minimum spec's and everything I make works perfectly.
The IRF2807 is guaranteed only if its gate voltage is 10V, not 4.5V.

 

[sprooch]

The boost converter has to convert 15V 20A (300W) DC to 55V 5A.

I chose that mosfet after I did some calculations and by looking at the datasheet, I can say that IRF2807 will be suitable for the converter.

The driver was here already, so I just took that driver without calculating the Ton. So I guess it is the problem. what do you think?

I have done a openloop simulation in multisim before I started build the hardware.So, in my opinion, the openloop circuit can be done also in the real application. Tell me if I am wrong here, please...

Ton = 160n/25mA = 6uS -----> Not a good idea.

what will be the good value for Ton then?
and I am looking at the mic4426 now.

 

[max0412]

Are you guessing that you have a "typical" Mosfet?
What if it is a "minimum" one? Then it won't conduct well and it will melt.

Lets see the data sheet says this FET is capable of 280A pulses so at the comparative Mickey mouse current pulses the OP is putting through the Fet I doubt it. There are component tolerances but considering he’s not using it anywhere near the extremes and I doubt he plans on banging out a couple thousand units I doubt that’s the reason. If he was attempting to pump say 200A through it yes I would agree, but for the single digit to at most 10 20A pulses I doubt that’s it.

At 175Tj there is only a couple hundred mV difference Vds between 15Vgs and 4.5Vgs.

But you maybe missed where I already suggested he use the 15v for his gate signal.Even at 4.5 vgs your implying the FET has a worse then 50% tolerance spread.

If your seeing that wide of a production spread you may want to look into your suppliers maybe your getting counter fit parts.

Now that he's actually supplied the output power he's still not pushing the extremes of the FET 22Amps peak (nowhere near it).This is about 45% below the data sheet spec for when the FET hits the linear region at 4.5Vgs. You believe further derating is req?

what will be the good value for Ton then?

You have to calculate the min and max duty cycle under worst case conditions ,Then you have to make sure your rise/fall times will let you obtain those duty cycles. Then you have to ensure your mosfet can dissipate the power by calculating conduction losses and switching losses, provide the necessary heat sink.Then you check all this by measuring it.

 

[sprooch]

Hi guys,

I have solved the problem.

I increased the driver Vcc to 20V to go the SOA...

the mosfet never die...


Added after 1 minutes:

one more thing...

Thanks for your help......