IRF4905 MOSFET getting HOT

Thread Starter

Garry48

Joined Jul 18, 2017
12
I have designed and simulated an Arduino circuit to sequentially pulse six (6) glow plugs.
I'm using two units to heat glow plugs on a V12 static engine. I need a Neg ground system and I have included flyback diodes to prevent issues with the 12V starter motor inductive load.
The arduino is run in Fast PWM mode and interrupt driven. I have the PWM pulse on pin 3 which runs at 1.409kHz frequency with a 1.40% duty cycle.
Each glow plug is cycled at 1/6th the PWM frequency of Pin 3. The duty cycle is also 1/6th of Pin 3.
The 10K pot will allow me to increase the duty cycle of the signal going to the glow plugs.
My breadboard circuit was not tested long enough to determine the MOSFET was getting hot! It heated the glow plug just fine.
When I received my Printed Circuit Boards (PCB's), I built the 5V supply circuit and one glow plug circuit.
The circuit schematic diagram is missing 4 circuits exactly like the two in the diagram.
I'm powering the PCB with a three cell 2200 mAH 30C LiPo Battery (12.6VDC fully charged).
The Green LED on the Arduino illuminates and the Red LED is being pulsed (LED dimly illuminates).
I do not have the glow plug connected in the circuit and I felt the IRF4905 and it was too warm for me to keep my fingers on it.
A typical glow plug is heated with a 1.5V battery and I measured really close to 4 Amp current draw.
I've searched the forums and there are lots of similar issues, but I don't know enough to solve my problem.
I have a 1.409kHz signal on pin 3 with a 1.40 % duty cycle.
Pin 2 should have a signal 1/6th of the PWM signal on pin 3. I measure 234.2 Hz and duty cycle of .25%.
Transistor 2N2222 has 11.1V on the collector, freq 234.2 and 99% duty cycle.
I think the MOSFET turn on voltage must be low.
Have I chosen the wrong MOSFET? What can I do to lower the MOSFET temperature?
Stuffing the PCB and 6 hot MOSFETS won't work in a PLA 3D printed case.
All comments appreciated.

pdf of schematic
 

Attachments

Last edited by a moderator:

ebp

Joined Feb 8, 2018
2,332
If the FET is getting hot without the glow plug in the circuit, something is seriously wrong. The current should be only a few milliamps for the LED.

The peak current into the plug at 12 V is going to be about 32 A, based on your 1.5 V 4 A note, but that is perfectly tolerable for that FET with nearly 12 V gate to source. The dissipation while ON would be about 21 W, which would make it extremely hot without a heatsink, but with a duty cycle of 0.25% the average power is next to nothing. 0.25% is only going to give you about 1.6 A RMS, 0.6 V RMS, which is far below target.
 

RichardO

Joined May 4, 2013
2,270
Some really unlikely possibilities...

Could you have the diodes installed backwards? See if they are hot. This is my best (but not very good) guess.

Maybe the 2N2222' have the collector and emitter reversed causing about a 6 volt "collector to emitter" breakdown. This might cause the FET's to always be partially on. Even if this is the case, the FET's can't get hot without an heavy load. (Note that since you say there is 11.1 volts on the collector of the 2N2222 I don't see how this could be the case, however.)

The FET's could be in backwards so that the body diode is always conducting. There is still that pesky problem of where the excess current is going without the glow plugs connected.

Could you have fooled yourself about the duty cycle and it is over 99%?

But as @ebp said, the LED alone can't draw enough to get the FET's hot. Unless, the LED current limiting resistor is a really low value. I would expect that to blow the LED's, though.


Bottom line. The best I can come up with is that there are two problems. One is that the FET's are on more or longer that you think. The other is that you have an unexpected load without the glow plugs connected.
 

Alec_t

Joined Sep 17, 2013
14,264
The Reset pins of your Arduino are apparently floating. That's a no-no for CMOS circuits generally (but I'm not familiar with Arduinos), as it can cause spurious oscillation. If MOSFETs oscillate rapidly they get hot.
 

Thread Starter

Garry48

Joined Jul 18, 2017
12
The Reset pins of your Arduino are apparently floating. That's a no-no for CMOS circuits generally (but I'm not familiar with Arduinos), as it can cause spurious oscillation. If MOSFETs oscillate rapidly they get hot.
Can you suggest how I should correct this issue? I have not worked with Arduino's previously.
I have made a few PCB's for personal use.
I don't have a good working knowledge of electronic hardware. I'm a software guy.

Thanks
 

olphart

Joined Sep 22, 2012
114
Swap out the 2N2222 for a real FET driver. Simple transistor drives are fine for light loads, but the junction voltage drop will not work well at >1A loads. Had similar - bench testing ran fine at .75A (13V, 125Hz, 75%DC). Mfg testing had problems that I couldn't see until I put on real load ~2.25A - too hot to touch in under 10S. Changed 2N3904 for MIC4126, heat up scarcely noticeable..
 

Thread Starter

Garry48

Joined Jul 18, 2017
12
Some really unlikely possibilities...

Could you have the diodes installed backwards? See if they are hot. This is my best (but not very good) guess.

Maybe the 2N2222' have the collector and emitter reversed causing about a 6 volt "collector to emitter" breakdown. This might cause the FET's to always be partially on. Even if this is the case, the FET's can't get hot without an heavy load. (Note that since you say there is 11.1 volts on the collector of the 2N2222 I don't see how this could be the case, however.)

The FET's could be in backwards so that the body diode is always conducting. There is still that pesky problem of where the excess current is going without the glow plugs connected.

Could you have fooled yourself about the duty cycle and it is over 99%?

But as @ebp said, the LED alone can't draw enough to get the FET's hot. Unless, the LED current limiting resistor is a really low value. I would expect that to blow the LED's, though.


Bottom line. The best I can come up with is that there are two problems. One is that the FET's are on more or longer that you think. The other is that you have an unexpected load without the glow plugs connected.
Thank you RichardO!
My O'scope displays the frequency and % duty cycle. I can see a steady signal, so I'm pretty sure the scope is measuring it correctly.

I lifted the Cathode of my 1N4001 and the MOSFET stopped getting hot! It turns out the silkscreen from the vertical footprint for the 1N4001 is REVERSED! I would have never found it, without your help! Excellent diagnosis from a place far away! Simply amazing! I was totally stumped!
By the way, my breadboard design had omitted the diode, since I didn't think I needed it for the resistive glow plug load!
Once I looked at the footprint on the PCB and saw the diode was reversed, I knew what the problem was. I never thought to check it before ordering my PCB's.

Thanks for your superb 'remote' trouble shooting skills! Outstanding work!
 

Picbuster

Joined Dec 2, 2013
1,047
I have designed and simulated an Arduino circuit to sequentially pulse six (6) glow plugs.
I'm using two units to heat glow plugs on a V12 static engine. I need a Neg ground system and I have included flyback diodes to prevent issues with the 12V starter motor inductive load.
The arduino is run in Fast PWM mode and interrupt driven. I have the PWM pulse on pin 3 which runs at 1.409kHz frequency with a 1.40% duty cycle.
Each glow plug is cycled at 1/6th the PWM frequency of Pin 3. The duty cycle is also 1/6th of Pin 3.
The 10K pot will allow me to increase the duty cycle of the signal going to the glow plugs.
My breadboard circuit was not tested long enough to determine the MOSFET was getting hot! It heated the glow plug just fine.
When I received my Printed Circuit Boards (PCB's), I built the 5V supply circuit and one glow plug circuit.
The circuit schematic diagram is missing 4 circuits exactly like the two in the diagram.
I'm powering the PCB with a three cell 2200 mAH 30C LiPo Battery (12.6VDC fully charged).
The Green LED on the Arduino illuminates and the Red LED is being pulsed (LED dimly illuminates).
I do not have the glow plug connected in the circuit and I felt the IRF4905 and it was too warm for me to keep my fingers on it.
A typical glow plug is heated with a 1.5V battery and I measured really close to 4 Amp current draw.
I've searched the forums and there are lots of similar issues, but I don't know enough to solve my problem.
I have a 1.409kHz signal on pin 3 with a 1.40 % duty cycle.
Pin 2 should have a signal 1/6th of the PWM signal on pin 3. I measure 234.2 Hz and duty cycle of .25%.
Transistor 2N2222 has 11.1V on the collector, freq 234.2 and 99% duty cycle.
I think the MOSFET turn on voltage must be low.
Have I chosen the wrong MOSFET? What can I do to lower the MOSFET temperature?
Stuffing the PCB and 6 hot MOSFETS won't work in a PLA 3D printed case.
All comments appreciated.

pdf of schematic
Schematic looks correct.
Are the fets mounted correctly?
Observe the gate.
You did used transistors to control the gate but a transistor in saturation will produce a collector- emitter voltage.
Try to change the collector resistor to 47K lowering that voltage and moving the gate better to gnd.
I always use N-mosfets to drive P-channels to avoid that problem.
Picbuster
 

ebp

Joined Feb 8, 2018
2,332
Ah, the old PCB CAD library symbol is wrong problem!

Actually, you probably don't need the diodes at all. The glow plugs are, for all intents and purposes, purely resistive. There will be a little inductance in the wiring, but the total energy stored will be very small and probably "eaten" completely during FET turn-off. It would also be perfectly safe to allow the FET to "avalanche" to conduct any energy not dissipated during turn-off.

There is nothing wrong with the gate driver circuit you are using for the frequency and pulse width. However, as I mentioned before, at 0.25% duty cycle the current is going to be much less than 4 amps. A duty cycle of about 1.6% should be about right to yield 1.5 V RMS (because the plug is a heater/resistor it is the RMS current that "counts") with a supply of 12 V.
 

RichardO

Joined May 4, 2013
2,270
I lifted the Cathode of my 1N4001 and the MOSFET stopped getting hot! It turns out the silkscreen from the vertical footprint for the 1N4001 is REVERSED! I would have never found it, without your help! Excellent diagnosis from a place far away! Simply amazing! I was totally stumped!
Glad I could help.

When I was a production pre-test technician -- the first one to power up and test a circuit board -- I had to replace a number of diodes because they were *marked* backwards. After a while you learn not to trust anything.
 

Thread Starter

Garry48

Joined Jul 18, 2017
12
5k_resistor.jpg
Schematic looks correct.
Are the fets mounted correctly?
Observe the gate.
You did used transistors to control the gate but a transistor in saturation will produce a collector- emitter voltage.
Try to change the collector resistor to 47K lowering that voltage and moving the gate better to gnd.
I always use N-mosfets to drive P-channels to avoid that problem.
Picbuster
Picbuster,

I have a simplified circuit in EveryCircuit. When I change the 5K collector resistor to 47k, the current decreases from 2.5mA to less than 300 microAmps. I not understanding what you mean when you stated "Try to change the collector resistor to 47K lowering that voltage and moving
the gate better to gnd".

Can you explain in more detail? I'm very weak on practical application. I do understand some theory. :(

Thanks
 

Attachments

Last edited:

Thread Starter

Garry48

Joined Jul 18, 2017
12
View attachment 159729

Picbuster,

I have a simplified circuit in EveryCircuit. When I change the 5K collector resistor to 47k, the current decreases from 2.5mA to less than 300 microAmps. I not understanding what you mean when you stated "Try to change the collector resistor to 47K lowering that voltage and moving
the gate better to gnd".

Can you explain in more detail? I'm very weak on practical application. I do understand some theory. :(

Thanks
Apologies for the multiple uploads! I did not see them in 'Preview' and now I can't delete the multiples!
 

Picbuster

Joined Dec 2, 2013
1,047
View attachment 159729

Picbuster,

I have a simplified circuit in EveryCircuit. When I change the 5K collector resistor to 47k, the current decreases from 2.5mA to less than 300 microAmps. I not understanding what you mean when you stated "Try to change the collector resistor to 47K lowering that voltage and moving
the gate better to gnd".

Can you explain in more detail? I'm very weak on practical application. I do understand some theory. :(

Thanks
A transistor is a current amplifier when punt vin saturation the emitter -collector voltage is depending on the current.
This current is caused by base current and gain but limited by the collector resistor.( like two resistors in series and measure over one )

The one between gnd and gate provide the voltage that's why I suggest to look at the gate voltage.
If the collector voltage is not 'low' enough than the P-mosfet will not go in saturation.
A fet is voltage controlled to get into saturation you have to apply a certain voltage( see datasheet).
When not in saturation the mosfet will produce a resistance higher ( power(heat) will be defined by I^2 x R[fet at G-voltage) see datasheet]
I do use your circuit with the following changes;
replace transistor with a n channel mosfet 10K to Pmosfet gate resistor 100K Pmosfet irf9388 or sqj407 both sot669 package.
I used to PWM(2KHz) 12V 4Amps with no noticeable temperature raise @ 100%)

Picbuster
 

ebp

Joined Feb 8, 2018
2,332
If you increase the value of the gate-source resistor you will simply slow the turn-off of the FET. This probably makes no appreciable difference for the intended pulse. Slowing the transition does result in higher power dissipation in the FET because it spends longer with both significant current through it and significant voltage across it.

The bipolar transistor is being driven with about 2 mA of base current which is plenty for the static collector current of just a little more than that. As the FET is turning on, the collector current will be higher due to the charging of the the capacitances of the gate.

Again, there is nothing wrong with your gate driver circuit. It will drive the FET entirely adequately for the approximately 32 A peak current and make it switch with speed suitable for the application. The FET you have selected is quite adequate.
 

Thread Starter

Garry48

Joined Jul 18, 2017
12
If you increase the value of the gate-source resistor you will simply slow the turn-off of the FET. This probably makes no appreciable difference for the intended pulse. Slowing the transition does result in higher power dissipation in the FET because it spends longer with both significant current through it and significant voltage across it.

The bipolar transistor is being driven with about 2 mA of base current which is plenty for the static collector current of just a little more than that. As the FET is turning on, the collector current will be higher due to the charging of the the capacitances of the gate.

Again, there is nothing wrong with your gate driver circuit. It will drive the FET entirely adequately for the approximately 32 A peak current and make it switch with speed suitable for the application. The FET you have selected is quite adequate.
ebp,
Thanks for the comment/info. I'll proceed 'full speed ahead' building my project. I really appreciate the help I've received on this forum.
Best Regards!
 
Top