MOSFET gets hot and burns

sagor1 · May 8, 2022

If it is oscillating, you cannot read the gate voltage properly with a voltmeter. That 5.9V may in fact be 11.8V at a 50% duty cycle. Same for the Drain voltages, if it is oscillating, you cannot read the true voltage peaks.
The heating issue may be that one MOSFET gate is not dropping to zero fast enough. Why both don't overheat is still strange however. It may be possible something else is broken or miswired.

Revibes · May 8, 2022

sagor1 said:
If it is oscillating, you cannot read the gate voltage properly with a voltmeter. That 5.9V may in fact be 11.8V at a 50% duty cycle. Same for the Drain voltages, if it is oscillating, you cannot read the true voltage peaks.
The heating issue may be that one MOSFET gate is not dropping to zero fast enough. Why both don't overheat is still strange however. It may be possible something else is broken or miswired.

So what can be done now? Or should I change my oscillation IC ? Like getting a new one

ronsimpson · May 8, 2022

Revibes said:
but end to end winding is about 12-14V

It is hard to understand your meter. Revibes is right.

Revibes said:
but end to end winding is about 12-14V.

This gives me the idea there is A.C. present.

Please send a picture of the transformer.

How do you know the phasing on the primary of the transformer?

Revibes · May 8, 2022

ronsimpson said:
It is hard to understand your meter. Revibes is right.

This gives me the idea there is A.C. present.

Please send a picture of the transformer.

How do you know the phasing on the primary of the transformer?

Here is the transform er, both primary and secondary side. Pls what do u mean by phasing ?I also included the oscillation circuit I draw video

sagor1 · May 8, 2022

Something confusing in your pictures. Your MOSFETs are attached to the heatsinks via mica washer insulators. By the way, heatsinks are way too small to start with.
The Drain is usually connects to the heatsinks via the screws and back mounting surface of the device. It may be possible that one of your screws is not connecting properly and the mica washer/insulator is preventing the drain from making contact. You are not using the Drain pin, just relying on the connection to the heatsink via screw and back plate. Remove the mica insulators, but keep the heatsinks isolated. If this is the case (poor Drain contact), then both MOSFETs will start to get hot, as there is not enough cooling with those heatsinks.

sagor1 · May 8, 2022

PS: The way you soldered the wires to the pins (Source), you may have overheated those devices. A failed MOSFET will usually show as a short circuit. The one that is heating up quickly, I would check first... A shorted MOSFET will also explain the low voltages you may be seeing.

danadak · May 8, 2022

I cant emphasize enough that layout of a prototype is super important.
You have in your point to point wiring added parasitic L, have a ground system
that is not "stiff" with lots of L in it, all this can cause transistors to oscillate and
burn up in a few mS.

No scope is a serious drawback. Like trying to perform brain surgery as a blind man.

This is barely adequate but better than nothing -

Look at post # 2 here -

[SOLVED] - Typical beginner microcontroller questions

Hi, I'm doing preliminary research/reading about using a microcontroller. I see I have a lot more to read before a lot of the fun of e.g. making LEDs blink, but still. The attached picture is if it helps to visualize the basic idea - it's a block diagram concept, not a complete circuit...

www.edaboard.com

Layout :

PCB Designing considerations for Switching Mode Power Supplies

Read my article on important PCB designing considerations for a SMPS you may need to practice. We are referring a 230V AC to 5V DC 2A power supply which will be really interesting and useful because, this is equivalent to a 10W USB mobile phone charger.

chandulanethmal.github.io

https://www.edaboard.com/threads/layout-considerations-for-this-full-bridge-smps.319934/

Regards, Dana.

Revibes · May 8, 2022

sagor1 said:
Something confusing in your pictures. Your MOSFETs are attached to the heatsinks via mica washer insulators. By the way, heatsinks are way too small to start with.
The Drain is usually connects to the heatsinks via the screws and back mounting surface of the device. It may be possible that one of your screws is not connecting properly and the mica washer/insulator is preventing the drain from making contact. You are not using the Drain pin, just relying on the connection to the heatsink via screw and back plate. Remove the mica insulators, but keep the heatsinks isolated. If this is the case (poor Drain contact), then both MOSFETs will start to get hot, as there is not enough cooling with those heatsinks.

Thanks sir, have read the MOSFETs drain that heats up within some few seconds and the heatsinks with continuity but it does not read buzz up, like being not connected at all , why the one the other one that dont get hot at all's drain is reading continuity well with its heatsinks. My question now is do MOSFETs get so hot quickly and burns out in some few seconds of connection if not properly connected to or not connected to an heatsink at all? Won't it work little like some few minutes before it starts heating up?

sagor1 · May 9, 2022

Check the MOSFET between Drain an Source in "Diode Check" mode on the meter. It should conduct in one direction only. If it conducts in both directions (by switching meter probes), then the MOSFET is blown.
Did you fix the design to include the Gate to ground resistor? Have you reduced the gate resistor value?
Do you have a way to measure frequency coming out of the controller chip? Some meters have a Hz function. .

danadak · May 9, 2022

If you do not have access to PCB facilities at least do something like this
where you have a large ground plane, and can keep leads short if you plan
component placement carefully -

If a transistor does not have low L connections, well bypassed, it will oscillate, many into
the RF region, and can self destruct in 10's of milliseconds. Your meter will never see or
catch this.

Bypassing, always use a combination of bulk and ceramic disk bypassing. Not all caps
have good ESR, so polymer tantalums for bulk, and .01 and / or .1 uF Ceramic caps the way
to bypass.

Sanyo OsCon is a polymer tantalum technology.

Regards, Dana.

Revibes · May 16, 2022

danadak said:
If you do not have access to PCB facilities at least do something like this
where you have a large ground plane, and can keep leads short if you plan
component placement carefully -

If a transistor does not have low L connections, well bypassed, it will oscillate, many into
the RF region, and can self destruct in 10's of milliseconds. Your meter will never see or
catch this.

Bypassing, always use a combination of bulk and ceramic disk bypassing. Not all caps
have good ESR, so polymer tantalums for bulk, and .01 and / or .1 uF Ceramic caps the way
to bypass.

View attachment 137017

Sanyo OsCon is a polymer tantalum technology.

Regards, Dana.

Thank u sir. Have applied most of the advice u guys have me, and it was really helpful. Have retested the oscillation now , one MOSFETs is not burning hot anymore withing some few seconds now. Both MOSFETs are working , but still getting hot in some 10secs time frame. I believe is my gate to source voltage that is the problem, cause am still applying that 5V. So my question now is ,pls how can I increase this 5V coming from my oscillation output? So as to be able to fully turn on and off the MOSFETs.? Pls what idea of circuit can I use?

Revibes · May 16, 2022

sagor1 said:
Check the MOSFET between Drain an Source in "Diode Check" mode on the meter. It should conduct in one direction only. If it conducts in both directions (by switching meter probes), then the MOSFET is blown.
Did you fix the design to include the Gate to ground resistor? Have you reduced the gate resistor value?
Do you have a way to measure frequency coming out of the controller chip? Some meters have a Hz function. .

Yes sir I have included the gate to source resistor and the circuit is pretty working fine now Thank u sir. Have applied most of the advice u guys have me, and it was really helpful. Have retested the oscillation now , one MOSFETs is not burning hot anymore withing some few seconds now. Both MOSFETs are working , but still getting hot in some 10secs time frame. I believe is my gate to source voltage that is the problem, cause am still applying that 5V. So my question now is ,pls how can I increase this 5V coming from my oscillation output? So as to be able to fully turn on and off the MOSFETs.? Pls what idea of circuit can

Pommie · May 17, 2022

You need MOSFET gate drivers. A good choice would be the MCP1404.

Mike.

danadak · May 17, 2022

Your schematic shows a deadtime R of 1.5K ? Its range is recommended to be 0 - 500 ohms.

So start with 500 ohms.

The 3525 operating specs seems to convey 1 nF of input C (not counting miller) but your FETS
are 4 nF.

So first and foremost, LAYOUT, GROUND BUSS critical. Then think about a gate driver or a better
control chip.

Regards, Dana.

sagor1 · May 17, 2022

You need either a MOSFET driver, or use Logic Level MOSFETs. Typical ones are the IRL540 (100V rating) and the IRL620 (200V rating).
However, their RDSon is much higher resistance, so you may end up with heat again, depending on how much cirrent you are passing through it.
The IRL540 may be the best choice, it has a low ON resistance, but can only tolerate 100V D-S, which may be ok for a 12V switching circuit.
EDIT: There may be many other, better, MOSFETs to use, I just quoted a few that I'm familiar with. Maybe others here can recommend other MOSFETs.
You need to increase the heatsink size as well.

Revibes · May 17, 2022

How

sagor1 said:
You need either a MOSFET driver, or use Logic Level MOSFETs. Typical ones are the IRL540 (100V rating) and the IRL620 (200V rating).
However, their RDSon is much higher resistance, so you may end up with heat again, depending on how much cirrent you are passing through it.
The IRL540 may be the best choice, it has a low ON resistance, but can only tolerate 100V D-S, which may be ok for a 12V switching circuit.
EDIT: There may be many other, better, MOSFETs to use, I just quoted a few that I'm familiar with. Maybe others here can recommend other MOSFETs.
You need to increase the heatsink size as well.

How about making a circuit that could boost my SG oscillation output

Revibes · May 17, 2022

Revibes said:
How
How about making a circuit that could boost my SG oscillation output

Like boosting the 5V oscillation output to about 10v with an npn and PNP transistor switching

sagor1 · May 17, 2022

If your pin 13 is +12V, you should be getting at least 10V out of the outputs. I don't know where you are getting 5V?
Have you measured the frequency of the output? If it is running ok, then odds are that the drivers are ok, and driving the MOSFETs with 10V or better. You can measure the 10V with a scope. You cannot use a multimeter to measure the output driver voltage if it is oscillating, as a meter cannot follow high frequency signals properly.
If your 5V is a DC meter reading, that may be the "average" of the pulses, which means you are probably getting 10V pulses, just that you cannot measure them. At that point, it comes down to how much current you are passing through the MOSFETs and the heatsinking you are providing.
Even at 10A current with 0.04 ohms, you have 4W of heat (with 50% duty cycle, 2W). You still need a proper heatsink to remove that heat.
Note that some heat may be normal, depending on pulse width, current (resistances) and how fast it is switching (whether gate can drop enough to fully turn OFF)

Revibes · May 18, 2022

sagor1 said:
If your pin 13 is +12V, you should be getting at least 10V out of the outputs. I don't know where you are getting 5V?
Have you measured the frequency of the output? If it is running ok, then odds are that the drivers are ok, and driving the MOSFETs with 10V or better. You can measure the 10V with a scope. You cannot use a multimeter to measure the output driver voltage if it is oscillating, as a meter cannot follow high frequency signals properly.
If your 5V is a DC meter reading, that may be the "average" of the pulses, which means you are probably getting 10V pulses, just that you cannot measure them. At that point, it comes down to how much current you are passing through the MOSFETs and the heatsinking you are providing.
Even at 10A current with 0.04 ohms, you have 4W of heat (with 50% duty cycle, 2W). You still need a proper heatsink to remove that heat.
Note that some heat may be normal, depending on pulse width, current (resistances) and how fast it is switching (whether gate can drop enough to fully turn OFF)

What have observed is a steady rise of the heat in just about 10secs, the heat sink bcame so hot to the touch, the Frequency is okay is 50hz, but I also notice another thing, the output from the transformer is way too high, I try reducing it via the pot resistor at pine 1 of the IC, but it won't even go down. Is reading about 260V AC, can this also be the issue?

Revibes · May 18, 2022

danadak said:
Your schematic shows a deadtime R of 1.5K ? Its range is recommended to be 0 - 500 ohms.

So start with 500 ohms.

The 3525 operating specs seems to convey 1 nF of input C (not counting miller) but your FETS
are 4 nF.

So first and foremost, LAYOUT, GROUND BUSS critical. Then think about a gate driver or a better
control chip.

Regards, Dana.

Ok sir I will try that, thanks so much sir.

MOSFET gets hot and burns

Active Member

New Member

Well-Known Member

New Member

Attachments

Active Member

Active Member

Well-Known Member

New Member

Active Member

Well-Known Member

New Member

New Member

Well-Known Member

Well-Known Member

Active Member

New Member

New Member

Active Member

New Member

New Member

Similar threads

Privacy & Transparency

Privacy & Transparency