Maker Pro
Maker Pro

MOSFET getting hot and making noise

scotth87

May 8, 2017
10
Joined
May 8, 2017
Messages
10
I'm working on building the PCB for a circuit that I have already breadboarded successfully. Part of it is a high voltage power supply, which has a MOSFET in it. Unlike the breadboarded version, the MOSFET on the PCB is making a high pitched noise which changes pitch if anything touches the heat sink mount as well as getting rather hot. Does anyone know what could be causing this? I have tried switching out the transistors, 555 timer, and inductor to no avail.

The power supply seems to be putting out the correct voltage, however, when I'm measuring it, my multimeter was occasionally switching to VAC for some reason. It isn't doing that anymore at the moment though.

Any help would be greatly appreciated.
 

Harald Kapp

Moderator
Moderator
Nov 17, 2011
13,700
Joined
Nov 17, 2011
Messages
13,700
changes pitch if anything touches the heat sink mount as well as getting rather hot.
Imho a sign of a bad thermal connection between transistor and heat sink.
Another possibility is a short circuit between transistor - heatsink - other potential (voltage).

my multimeter was occasionally switching to VAC for some reason.
Sounds like the AC ripple on top of the DC is very high.


I have already breadboarded successfully
Check any differences between breadboard setup and your actual PCB. Find the differences, analyze along these to locate the faulty point(s).
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,510
Joined
Jan 21, 2010
Messages
25,510
I would check to see if the MOSFET is a lot hotter than the heatsink. If it is, check for thermal resistance issues (loose attachment, lack of thermal grease, etc.). If they're both hot, check that the gate is being driven correctly. If it's neither of these, check that the current through the device is as expected.
 

scotth87

May 8, 2017
10
Joined
May 8, 2017
Messages
10
Sorry if I wasn't clear. I do not have a heatsink attached to the MOSFET. I just meant the part on top of the MOSFET that one would attach a heat sink to. Maybe worth noting that if what I touch the top of the MOSFET with is metal, I can see a tiny spark.

I just hand drew the circuit from scratch based on the PCB and confirmed that it matches the breadboarded circuit.

Is it surprising that the circuit seems to be outputting the correct voltage if the problem is a short? I can't find any shorts on the board.

Should I try replacing the diodes? They are the only components other than the resistors and capacitors that I haven't replaced.
 

Harald Kapp

Moderator
Moderator
Nov 17, 2011
13,700
Joined
Nov 17, 2011
Messages
13,700
if what I touch the top of the MOSFET with is metal, I can see a tiny spark.
That sounds dangerous. What is this circuit for? Let us see your schematic.
Sounds like your building some kind of switch mode power supply. The noise you hear probably comes from a coil/inductor, not from the MOSFET.
Anyway, if the MOSFET gets hot, it will need a heatsink. Otherwise it will die from overheating. But as your MOSFET obviously is attached to some high voltage (sparks!), the heatsink neds to be electrically insulated from the MOSFET to avoid any serious injury due to the high voltage.

We really need to see the circuit.
 

scotth87

May 8, 2017
10
Joined
May 8, 2017
Messages
10
That sounds dangerous. What is this circuit for? Let us see your schematic.
Sounds like your building some kind of switch mode power supply. The noise you hear probably comes from a coil/inductor, not from the MOSFET.
Anyway, if the MOSFET gets hot, it will need a heatsink. Otherwise it will die from overheating. But as your MOSFET obviously is attached to some high voltage (sparks!), the heatsink neds to be electrically insulated from the MOSFET to avoid any serious injury due to the high voltage.

We really need to see the circuit.
Sure, I was thinking of going ahead and posting the circuit anyway. The power supply will be powering the tubes for a nixie clock. The reason I haven't considered a heat sink is because the MOSFET in the breadboarded circuit does not get hot.

schem.JPG
 

scotth87

May 8, 2017
10
Joined
May 8, 2017
Messages
10
That sounds dangerous. What is this circuit for? Let us see your schematic.
Sounds like your building some kind of switch mode power supply. The noise you hear probably comes from a coil/inductor, not from the MOSFET.
Anyway, if the MOSFET gets hot, it will need a heatsink. Otherwise it will die from overheating. But as your MOSFET obviously is attached to some high voltage (sparks!), the heatsink neds to be electrically insulated from the MOSFET to avoid any serious injury due to the high voltage.

We really need to see the circuit.
Good point about the noise, I suppose I don't have any reason to assume I could pinpoint the noise to the MOSFET.
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,510
Joined
Jan 21, 2010
Messages
25,510
the MOSFET in the breadboarded circuit does not get hot.

It sounds like there is some difference between the circuit you breadboarded and the circuit you now have.

Is there output voltage correct?

Does the problem happen even if there is no load connected to the power supply?

What is the input current?

It might be worth checking that D4 is not shorted

Do you have an oscilloscope? I'd check what's happening at both the gate and drain of the MOSFET.
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,510
Joined
Jan 21, 2010
Messages
25,510
BTW, it's the inductor that is singing to you, and my differential diagnosis is avalanche breakdown of the MOSFET or poor gate driving.
 

scotth87

May 8, 2017
10
Joined
May 8, 2017
Messages
10
It sounds like there is some difference between the circuit you breadboarded and the circuit you now have.

Is there output voltage correct?

Does the problem happen even if there is no load connected to the power supply?

What is the input current?

It might be worth checking that D4 is not shorted

Do you have an oscilloscope? I'd check what's happening at both the gate and drain of the MOSFET.

The output voltage is the same on the breadboard and the PCB circuit. I have it set to ~180V which has the trim pot at the same setting.

There is no load connected to either circuit at the moment. I stopped at this stage of soldering the PCB to check if the power supply was functioning properly.

I had not checked the input current. Something is definitely wrong there. I am at about 50mA on the breadboard circuit and it starts at about 150mA on the PCB and continues to rise. It seems to pretty much stop rising at about 190mA.

I don't read a short at D4. I was worried that i have the pads on the PNP transistor too close together and even though I checked over and over and wasn't reading continuity, removed that and soldered leads to another transistor on the breadboard. The result was the same.

I do not have an oscilloscope. Is that something that could be had affordably? I've only seen the bench-top kind.

With regard to your differential diagnosis. Could you explain what those are? I honestly have no idea what those things mean. Unfortunately I do not know nearly enough to have designed this circuit on my own so am troubleshooting fairly blindly.

Obviously there are some differences between my circuits. It is frustrating that so far I haven't found anything. I started checking voltage differentials and got some different readings but didn't see anything different around where the readings were taken. It seems like it would be less frustrating if I wasn't able to get the correct output voltage. I'm more used to something either working or not. I will keep looking tomorrow.

In case anyone is bored or generous enough to try to decipher a shot of the PCB layout, I have attached it. The only relevant components to the left are the power input followed by a diode followed by an on off switch. I highlighted a ground trace green that was passing through two pads and was hard to see.

pcb1.png

Thanks everyone for the quick replies.
 
Last edited:

Harald Kapp

Moderator
Moderator
Nov 17, 2011
13,700
Joined
Nov 17, 2011
Messages
13,700
Maybe worth noting that if what I touch the top of the MOSFET with is metal, I can see a tiny spark.
No wonder. The dran of the MOSFET is at high potential as this is is a step-up regulator. You need to take care that this part is well isolated from other parts. That counts for the whole high voltage part of the circuit.

With regard to your differential diagnosis
Circuit 1: o.k.
Circuit 2: not o.k.
Therefore there has to be a difference.

In case anyone is bored or generous enough to try to decipher a shot of the PCB layout
There are a few incedibly thin traces:
upload_2017-5-9_7-5-53.png
Try to reinforce them by a piece of wire soldered at least between the endpoints of the thisn traces.
Also, the ground is routed as an uncomfortably thin trace on the pcb. Wherever you have pulselike signals, especially in combination with high voltages, I recommend using a ground plane. As this is not possible in the existing layout, I recommend you reinforce teh ground connections either by soldering pieces of wire in parallel or by removing the solder stop over these traces and put a generous amount of solder on top of the traces. Both methods will not look good but will help to reduce the impedance of the ground.
 

scotth87

May 8, 2017
10
Joined
May 8, 2017
Messages
10
No wonder. The dran of the MOSFET is at high potential as this is is a step-up regulator. You need to take care that this part is well isolated from other parts. That counts for the whole high voltage part of the circuit.


Circuit 1: o.k.
Circuit 2: not o.k.
Therefore there has to be a difference.


There are a few incedibly thin traces:
View attachment 33914
Try to reinforce them by a piece of wire soldered at least between the endpoints of the thisn traces.
Also, the ground is routed as an uncomfortably thin trace on the pcb. Wherever you have pulselike signals, especially in combination with high voltages, I recommend using a ground plane. As this is not possible in the existing layout, I recommend you reinforce teh ground connections either by soldering pieces of wire in parallel or by removing the solder stop over these traces and put a generous amount of solder on top of the traces. Both methods will not look good but will help to reduce the impedance of the ground.
That image is a little misleading. For some reason KiCad distorts the scale on smaller traces unless you are zoomed in all the way. Those traces are half the width of the other traces, which is still small. The larger traces are 0.5mm and the smaller ones are 0.25mm. I attached a picture of the actual PCB for clarity.

Could these problems really just be caused by lack of sufficient grounding? I will give tinning the associated traces a shot. Thanks!

IMG_20170509_105329941-2.jpg
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,510
Joined
Jan 21, 2010
Messages
25,510
The main problem is that the voltage control doesn't work right. It controls the frequency rather than the duty cycle.

As a simple mod, I would disconnect the stuff connected to pins 4 and 5, then connect what was connected to pin 5 to pin 4.

This will pull the 555 into reset as the voltage rises above the set limit.
 

scotth87

May 8, 2017
10
Joined
May 8, 2017
Messages
10
No wonder. The dran of the MOSFET is at high potential as this is is a step-up regulator. You need to take care that this part is well isolated from other parts. That counts for the whole high voltage part of the circuit.


Circuit 1: o.k.
Circuit 2: not o.k.
Therefore there has to be a difference.


There are a few incedibly thin traces:
View attachment 33914
Try to reinforce them by a piece of wire soldered at least between the endpoints of the thisn traces.
Also, the ground is routed as an uncomfortably thin trace on the pcb. Wherever you have pulselike signals, especially in combination with high voltages, I recommend using a ground plane. As this is not possible in the existing layout, I recommend you reinforce teh ground connections either by soldering pieces of wire in parallel or by removing the solder stop over these traces and put a generous amount of solder on top of the traces. Both methods will not look good but will help to reduce the impedance of the ground.
Well I tinned all of the ground traces, which didn't make much of a difference, so I drilled a few holes in an unused filled copper space and connected the ground to it at the power input as well as two points along the ground loop. Voila, no more noise or heat and less than 40mA input current. 180V output.

Thanks! I guess that teaches me for not taking a ground plane seriously enough.
 
Last edited:

scotth87

May 8, 2017
10
Joined
May 8, 2017
Messages
10
The main problem is that the voltage control doesn't work right. It controls the frequency rather than the duty cycle.

As a simple mod, I would disconnect the stuff connected to pins 4 and 5, then connect what was connected to pin 5 to pin 4.

This will pull the 555 into reset as the voltage rises above the set limit.
Would you still recommend this in lieu of my above comment? Is there anything else I should check to confirm it is functioning properly other than input current, output voltage, and heat? Thanks!
 

scotth87

May 8, 2017
10
Joined
May 8, 2017
Messages
10
Thin traces, as I said;)

Still heed my warning about the high voltage. Please!
Would putting some kind of Corona dope on there be sufficient, or do you mean having the whole circuit on the same board is a bad idea period?
 

Harald Kapp

Moderator
Moderator
Nov 17, 2011
13,700
Joined
Nov 17, 2011
Messages
13,700
An insulating cover like corona dope (I admit I had to look up this material) will protect you from the high voltage, but will also be a thermal insulation. Probably not the best idea when the transistor becomes hot under load.
I recommend you place the high voltage components in a separate space of the pcb, reserved for the high voltage components only. Keep a distance of >= 4 mm between the high voltage parts and the low voltage parts.

As you have a common ground there is always a risk of electric shock in case of a failure of a component. The whole circuit should be completey packaged into an insulating housing. Use well insulated wires in case any signals need o leave the housing. No electrically conductive part of the circuit should be touchable.
In case an insulating housing is not an option, use an earth connection to provide a path for fault currents (a path that goes not through your body, that is).
Make this as foolproof as possible.
 

scotth87

May 8, 2017
10
Joined
May 8, 2017
Messages
10
An insulating cover like corona dope (I admit I had to look up this material) will protect you from the high voltage, but will also be a thermal insulation. Probably not the best idea when the transistor becomes hot under load.
I recommend you place the high voltage components in a separate space of the pcb, reserved for the high voltage components only. Keep a distance of >= 4 mm between the high voltage parts and the low voltage parts.

As you have a common ground there is always a risk of electric shock in case of a failure of a component. The whole circuit should be completey packaged into an insulating housing. Use well insulated wires in case any signals need o leave the housing. No electrically conductive part of the circuit should be touchable.
In case an insulating housing is not an option, use an earth connection to provide a path for fault currents (a path that goes not through your body, that is).
Make this as foolproof as possible.
Would it not be a good idea to at least coat the traces? That shouldn't affect any heat issues, should it?

The circuit will be fully enclosed in a wooden box. The only wire that will leave the box is the power cord.

Thanks so much for the tips!
 

Harald Kapp

Moderator
Moderator
Nov 17, 2011
13,700
Joined
Nov 17, 2011
Messages
13,700
Would it not be a good idea to at least coat the traces? That shouldn't affect any heat issues, should it?
It won't harm, but is not necessary when the circuit is fully enclosed in a wooden (isolating) box.
Save the money for other purposes.
 
Top