How to calculate the power dissipated in Mosfet?

[Okada]

How to calculate power dessipated in Mosfet ?

How to calculate power dessipated in Mosfet ?

Rds(on) is < 0.18 Ohms

Mosfet can handle 40A Current ?

Mosfet is STE40NK90ZD

If 40A current is flowing through it with VBUS voltage 325V DC (H-bridge for Inverter) then what will be the power dessipated by Mosfet ? How to find out voltage drop across Drain and Source ?

 

[KlausST]

Re: How to calculate power dessipated in Mosfet ?

Hi,

this is a frequently asked question here in this forum.
(And a lot more often in the internet)

Therefore this question is already anwered many times.

***
A simple search in this forum gives a lot of answers.
Even in the box "Similar threads" below you find related links.

Klaus

 

[Okada]

Re: How to calculate power dessipated in Mosfet ?

I already searched and read the other pages but they doesn't show how to find out Vds (drop across drain and source). I need to know this to calculate power dissipated.

 

[KlausST]

Re: How to calculate power dessipated in Mosfet ?

Hi,

I already searched and read the other pages but they doesn't show how to find out Vds (drop across drain and source)

this is as simple as can be: Just Ohm´s law.
And for sure it`s written in every MOSFET power dissipation documentation!

V = I * R, Here: V_DS = R_DS_ON * I_D,

or directely as power: P = I * I * R = I^2 * R. Here: P_D = I_D^2 * R_DS_ON.

Klaus

 

[BradtheRad]

Re: How to calculate power dessipated in Mosfet ?

V * A = W. When I wanted to do this, I could think of no other way than to use my oscilloscope to read voltage across the mosfet when it was turned on. I had to be able to recognize the part of the cycle when it was conducting.

I also needed to find Amperes through the mosfet when it was turned on. This required a low ohm resistor in series with the mosfet. I had to hook up my oscilloscope to measure voltage across the resistor. I had to recognize the part of the cycle when it was conducting.

This was a lot of work. I did the work because I realized I should confirm that the mosfet was being driven properly as my circuit operated, therefore I had to observe it when it was turned on. Then I also had to compare that to when it was shut off.

 

[Okada]

Re: How to calculate power dessipated in Mosfet ?

I am really sorry for opening this thread. Thi squestion was related to Pure Sinewave Inverter Design Help thread in Microcontroller section.

Anyway as I have opened this thread I will continue about Mosfet power here only.

My Inverter will be designed for 5 KW and Output voltage will be 230V.

So, P = VI

I = P/V = 5000 / 230 = 21.7 (approx 22A rms)

Ipk = 1.4142 * 22 = approx 31 A

My Rds(on) = <0.18 Ohms

So, Mosfet power dissipation = I^2R = 31 * 31 * 0.18 = 173 Watts ?

Edit, Sorry Rds(on) is 0.14 Ohms

So, PD mosfet = 31 * 31 * 0.14 = 135 watts ?

I am using STE40NK90ZD mosfet.

At 25 degree C it can dissipate 600W according to datasheet. So, can I use this mosfet safely ?

I will be using a DC fan with PWM for cooling the heatsink.

 

[dick_freebird]

Re: How to calculate power dessipated in Mosfet ?

If you want help with your homework (and I'd assume
this is the case, an engineer doing his job would know
how to ***** a book and work a calculator) then you
need to add some details.

{mods - why the verb "c_r_a_c_k" becomes asterisks,
seems like a valid question}

For starters, 325V @ 40A is not a DC condition that
makes any sense to anyone. This has to be about a
switching power converter. PWM converters have a duty
cycle and an inductor-defined triangular current
waveform. Without knowing L you can't calculate
the min and max "on" current of the (and there's more
than one in most modern designs other than the
cheapest) FET, so can't calculate the "on" period
conduction loss. You could approximate with the
40A being called an average but you don't know
how much of that runs through the one FET you are
asking about (for how long, rather) to say what
average conduction loss is, there. You could guess
at I^2*Rds(on)*duty for the high side and
I^2*Rds(on)*(1-duty) for the low side but you
say nothing about duty cycle, which of course
also depends on whether there's a transformer
(H-bridge implies so) or whether this is really a
half-bridge with only inductor and so on; analysis
becomes very different depending on whether you
are simply "smoothing" with an inductor, or pushing
power through a core.

Without knowing FET Cgd and gate drive resistance
you can't calculate the drain slew-interval and it's
hard to get, other than by simulation and requiring
a semi-complex (or provided) FET model and switch
node stray C and designed L, how the drain voltage
and drain current integrate over the rising and falling
edges, so no luck for you on switching losses. Which
can be what takes you from "calculated safe" to
"blowed up real good" on the bench.

But I would bet that if you availed yourself of search
engine goodness, or even referred to the course text,
you would find appropriate formulae that use simple
available parameters to approximate these, to your
professor's satisfaction. Why you are asking us instead,
I can only assume involves laziness. Harumph.

 

[Okada]

Re: How to calculate power dessipated in Mosfet ?

This is not school work. I am working and I am designing a 5 KW Inverter. The design is almost complete.

Sorry, I did some mistake in the calculation.

Actually I will be using 3 mosfets in parallel and hence current will divide.

So, it is 31 A / 3 = 10.33 A per mosfet

So, Power dissipated in each mosfet is

Pd = 10.33 * 10.33 * 0.18 = 19.22 Watts

So, can I use this mosfet safely ?

Find the circuit of the 5 KW inverter I am designing. It is not yet complete (the dc-dc converter section).

Duty cycle of SPWM signal varies between 86% and 98%.

Here is my Original thread.

https://www.edaboard.com/threads/355929/

Find attached circuit which I have designed. All saftey measures are taken while building the device.

I have two doubts.

One is what current rating diodes I have to use for D1 and D9 and what should be the insulated copper wire guages for primary and secondary windings of the switching transformer.

 

[d123]

Re: How to calculate power dessipated in Mosfet ?

Hi,

Probably won't help, it's from snva168e, about selecting components for a SEPIC, and PD in MOSFET...

 

[Warpspeed]

Re: How to calculate power dessipated in Mosfet ?

I am really sorry for opening this thread. Thi squestion was related to Pure Sinewave Inverter Design Help thread in Microcontroller section.

Anyway as I have opened this thread I will continue about Mosfet power here only.

My Inverter will be designed for 5 KW and Output voltage will be 230V.

So, P = VI

I = P/V = 5000 / 230 = 21.7 (approx 22A rms)

Ipk = 1.4142 * 22 = approx 31 A

My Rds(on) = <0.18 Ohms

So, Mosfet power dissipation = I^2R = 31 * 31 * 0.18 = 173 Watts ?

Probably not. your peak current into the load might be 31 amps, but each mosfet in the bridge is not conducting continually with pure dc 31A current.

The current waveform through each mosfet will sometimes be zero, and sometimes be a steeply rising current ramp. You need to calculate or preferably measure the true rms current through each mosfet at whatever duty cycle is appropriate.

Edit, Sorry Rds(on) is 0.14 Ohms

So, PD mosfet = 31 * 31 * 0.14 = 135 watts ?

It will certainly be Irms x Irms x Rdson, for the conduction losses, but not necessarily 31 x 31 Amps.

You also need to add switching losses to that, which will be higher at higher switching frequency.

I am using STE40NK90ZD mosfet.

At 25 degree C it can dissipate 600W according to datasheet. So, can I use this mosfet safely ?

How are you going to keep the junction temperature at only 25C at full flat out power ? Its just not possible.
Its going to run a lot hotter than 25C, that means the power rating must be de rated.
It might be rated 600 watts at 25C, at 150C the rated power dissipation will be zero.

If you run the junction temperature up to say 100C, the power rating falls to 40% of 600 watts. or only 240 watts.

The second curve below is the correct one. I cannot seem to get rid of the first one.

 

[BradtheRad]

Re: How to calculate power dessipated in Mosfet ?

It's a good idea to run some tests to see whether your mosfet achieves the low spec for Ron.

Of course I'm just one voice and you don't have to take my word, but from my own test results I found I was not driving the mosfet to the low R value listed in the specs. (Perhaps it required a higher bias voltage than my circuit operated at.) The spec was a fraction of an ohm. I got nowhere near that low. The mosfets were purchased 16 years ago when reliability was normal to expect.

 

[Warpspeed]

Re: How to calculate power dessipated in Mosfet ?

I am slowly working up towards something like this myself right now, but its a very slow long term project.

The way I am doing it is to first build a low power experimental circuit, in my case only 750 watts. Wind a suitable transformer for that power level, and use only four individual mosfets. Gradually start off at very low power and measure everything, get the control system working. Make sure all the actual measurements agree with circuit theory. All the waveforms must be text book clean, if not, find out why and fix it.

Next I wind my large multi kilowatt magnetics and run that at 750 watts to confirm everything is still working o/k. Gradually add more mosfets, and by that time I should have a pretty good idea about final heat sink requirements, and maximum temperature rises, maximum ripple currents and be reasonably happy its not going to blow up.

The whole thing may require ten rebuilds and re thinks along the way, but it will never have been a complete smoking ruin that goes bang when first powered up.

 

[Okada]

Re: How to calculate power dessipated in Mosfet ?

You need to calculate or preferably measure the true rms current through each mosfet at whatever duty cycle is appropriate.

Duty of SPWM signal will vary because feedback is used for maintaining constant output voltage and this is being done by varying PWM duty. So, I will take max PWM duty which in my case will be 98% because max duty = max power dissipated. So, If I calculate for max duty then it will work for lower duties safely.

 

[KlausST]

Re: How to calculate power dessipated in Mosfet ?

Hi,

Consider the folowing three situations:
* conductive loss: integral of drain_current x drain_voltage
* switching loss: at rising edge and falling edge
* body diode power dissipation, depends on inductive load.

Klaus

 

[Okada]

Re: How to calculate power dessipated in Mosfet ?

Ok. Actually this i snot my office work. I am designing this Inverter for personal use. I will be using 12V 140Ah ABB batteries. At home I don't have an oscilloscope which can measure more than 60V and I cannot test this circuit in office. Next month I am going to buy a 100 Mhz Rigol Oscilloscope for this purpose. So, it will take another one month for me to do more testing. Currently I am doing all caluculation theoritically because I want to put the best components in the design. I have to buy these components from mouser as these are not available locally and mouser shipping cost is 40 USD and hence I can't change components too much.

I will take into consideration what all the members have advised.

There will be two separate heatsinks for the mosfets and on each heatsink a Corsair High Static Pressure 120 mm Fan will be mounted directly above mosfets using piller nuts (stand offs). Fans will be running all the time and it will keep the mosfets cool.

 

[KlausST]

Re: How to calculate power dessipated in Mosfet ?

Hi,

For me it seems the heatsink and fan combination is more than enough.

Klaus

 

[dick_freebird]

Re: How to calculate power dessipated in Mosfet ?

At home I don't have an oscilloscope which can measure more than 60V

With a 10Mohm probe you would add almost no error by
making yourself a (say) 99Kohm / 1000 ohm resistor divider
and using that as your HV probe (suggest some thought to
safety and so on). Resistor power dissipation would be
maybe 4W (at 600V, scaling the output to your 60V) so
break up the large one into sensible segments that let you
use available resistors (bearing in mind that resistors need
cooling too). 1Kohm*10pF makes a new 100MHz corner,
probably no issue for your present activity.

If you were willing to deal with hand work you could
make it out of (100) 1K resistors all alike and enjoy some
volume discount (at the cost of your time and solder).
4W spread across 100 resistors is trivial unless you pack
them all together, too tightly.

 

[dick_freebird]

Re: How to calculate power dessipated in Mosfet ?

Actually I must correct myself; the 99K/1K would be a
100:1 divider. You could of course use the 90K/10K
point of the string for 10:1 (at some cost to BW) or
you could use a 9K/1K (at the cost of needing higher
power devices - much higher, since you'd presumably
spread 10X power over 1/10 the resistors). I expect
that you could pick a variation on this theme that is
acceptable for both component cost and transient
waveform fidelity.

 

[Audioguru]

Re: How to calculate power dessipated in Mosfet ?

Guess what? The fan will not keep the Mosfets cool. It will keep their temperature lower than without the fan but they will still get hot.
Guess what? The Rds rises as the Mosfets get hotter that increases the Rds more that increases the temperature more that increases Rds more that increases the temperature more.... thermal runaway.

 

[Warpspeed]

Re: How to calculate power dessipated in Mosfet ?

1544458]How to calculate power dissipated in Mosfet ?

Its only possible to make a very rough guess, because many of the factors you need to know to do it properly will not be known to sufficient accuracy.
If you have access to a wide band digital oscilloscope with maths capability, that is the best way.

If like most of us you have just very basic test equipment handy, all you can do is measure the actual power loss in a working circuit.

A fairly simple methods might be to just measure the overall efficiency. Power in minus power out equals loss as heat, and most of that will be into the mosfet heatsink.

Run it for sufficient time to reach a stable heat sink temperature, maybe half an hour, then measure the heatsink temperature rise.
Replace the mosfets with an aluminium clad power resistor.
Run the resistor up with either an adjustable dc power supply, or a variac to reach the same heat sink temperature rise as the mosfets.
Measure the dc or ac watts and you will know the mosfets dissipated that same amount of power.

That will also tell you the thermal resistance of the heat sink, and from that you can calculate the mosfet junction temperature.

It sounds like a lot of messing about, and it is, but its about the only way to do it with minimal basic test equipment.