Problem Driving a MOSFET

Thread Starter

colin.t

Joined Jul 6, 2018
16
Hi, I am having problems driving a mosfet on our project. Originally I had a LVD circuit that uses a 555 timer as a Bistable Latch, The output drives a 3v relay via Pin3 and 5v. this in turn drives a 10amp relay. this all works perfectly well. However I would like to upgrade to using a Mosfet. I had copied this circuit from a similar thing on the net. I am using a logic level mosfet BUK964R4-40B,118, this is being driven by a FOD3180. The problem is that when power is applied the 555 and the FOD3180 get very hot, and the Mosfet does not switch. (when I replace the FOD3180 with a 3v relay the Mosfet works perfectly). I have obviously done something wrong and my skill level in electronics is very limited so I was hoping that someone could help me with this. Thank you, Colin

LVD Mosfet_colin.t.png
 

AlbertHall

Joined Jun 4, 2014
12,345
You do not have a current limiting resistor in series with the FOD3180 LED. The current should be 10mA to 16mA and a 270Ω resistor would do the job.
 

Thread Starter

colin.t

Joined Jul 6, 2018
16
Hello Albert, Thanks for your quick reply. That makes sense to me, I will give it a Try. I just hope I have not fried the opto.
Thanks Colin
 

BobTPH

Joined Jun 5, 2013
8,813
Get rid of the resistors on the optocoupler. This has a push pull output and should be connected to ground and V+.

Bob
 

BobTPH

Joined Jun 5, 2013
8,813
With yout 1000 Ohm resistor from +5V to the driver you are limiting the current to 5 mA. So why did you pick a 2A gate driver?

Bob
 

ebp

Joined Feb 8, 2018
2,332
An optocoupled driver is a rather expensive way to control the FET when the driver's output supply is the same as the input supply. Normally you would resort to an optocoupled driver only if necessary for galvanic isolation for safety reasons, such as if the FET controlled a load that was in some way connected directly to AC mains. Sometimes isolated drivers have merit if there is a difference in "ground" between the output and input, as might happen with long connecting wires or in some difficult industrial environments.

Since the circuit, as is, would turn the FET on when the 555 output is low, an inverting driver is necessary. If the switching frequency is low (say 5 kHz or less; 10 kHz might be getting a little iffy, 50 kHz isn't "low"), there is probably no need to turn the FET on and off very fast. You can probably use a simple NPN transistor inverter with a moderate resistance (perhaps 500 to 1000 ohms) between the collector and +5V. If you need higher switching speed, an inexpensive driver can be made by paralleling the inputs and outputs of multiple sections of a CMOS inverting gate such as the 74HC00. A resistor of something in the range of 20 to 100 ohms between the paralleled outputs and the gate would be appropriate. Be sure to use a decoupling capacitor of 100 nF or more very close to the gate package.

If you don't need inversion, the 555 output can drive the FET directly. Again, a resistor of a few tens of ohms in the gate circuit would be appropriate.

A resistor of something in the range of 10k between the gate and source of the FET helps to assure the FET will stay turned off if power to the driver circuit is lost while there is still power for the FET and the load.

===
Elaboration on why switching speed doesn't usually need to be fast if the switching frequency is low:
When the FET switches, it passes though a condition where there is both significant voltage across it and current through it simultaneously. For example, if you were switching a 10 amp load with a 12 volt supply, there would be a brief time when the current would be (say) 5 amperes and the voltage across the FET 6 volts, for an instantaneous power dissipation in the FET of 30 watts. If the switching takes a significant fraction of the total period. the average power dissipation could be quite high. For example, if switching took 5 microseconds and occurred every 20 microseconds, the average power due to that loss during the switching transition could be quite high. If switching took 5 µs and occurred every millisecond, the average switching loss would be 50 times lower - same peak, but much lower average. The more current you can source to or sink from the gate to charge and discharge the capacitance of the gate, the faster the switching, which is why you might need the driver to be able to source and sink hundreds of millamperes up to multiple amperes for really fast switching.
 

Thread Starter

colin.t

Joined Jul 6, 2018
16
An optocoupled driver is a rather expensive way to control the FET when the driver's output supply is the same as the input supply. Normally you would resort to an optocoupled driver only if necessary for galvanic isolation for safety reasons, such as if the FET controlled a load that was in some way connected directly to AC mains. Sometimes isolated drivers have merit if there is a difference in "ground" between the output and input, as might happen with long connecting wires or in some difficult industrial environments.

Since the circuit, as is, would turn the FET on when the 555 output is low, an inverting driver is necessary. If the switching frequency is low (say 5 kHz or less; 10 kHz might be getting a little iffy, 50 kHz isn't "low"), there is probably no need to turn the FET on and off very fast. You can probably use a simple NPN transistor inverter with a moderate resistance (perhaps 500 to 1000 ohms) between the collector and +5V. If you need higher switching speed, an inexpensive driver can be made by paralleling the inputs and outputs of multiple sections of a CMOS inverting gate such as the 74HC00. A resistor of something in the range of 20 to 100 ohms between the paralleled outputs and the gate would be appropriate. Be sure to use a decoupling capacitor of 100 nF or more very close to the gate package.

If you don't need inversion, the 555 output can drive the FET directly. Again, a resistor of a few tens of ohms in the gate circuit would be appropriate.

A resistor of something in the range of 10k between the gate and source of the FET helps to assure the FET will stay turned off if power to the driver circuit is lost while there is still power for the FET and the load.

===
Elaboration on why switching speed doesn't usually need to be fast if the switching frequency is low:
When the FET switches, it passes though a condition where there is both significant voltage across it and current through it simultaneously. For example, if you were switching a 10 amp load with a 12 volt supply, there would be a brief time when the current would be (say) 5 amperes and the voltage across the FET 6 volts, for an instantaneous power dissipation in the FET of 30 watts. If the switching takes a significant fraction of the total period. the average power dissipation could be quite high. For example, if switching took 5 microseconds and occurred every 20 microseconds, the average power due to that loss during the switching transition could be quite high. If switching took 5 µs and occurred every millisecond, the average switching loss would be 50 times lower - same peak, but much lower average. The more current you can source to or sink from the gate to charge and discharge the capacitance of the gate, the faster the switching, which is why you might need the driver to be able to source and sink hundreds of millamperes up to multiple amperes for really fast switching.
 

Thread Starter

colin.t

Joined Jul 6, 2018
16
Thanks to all for the input, I will try the easiest things first and see how it turns out. The application does not require really fast switching. it is switching LED lamps between 3 and 10 amp load. The 555 is adjusted to give a on/off hysterysis of about 0.5 volt. so that the system does not restart when the batteries recover after LVD. I will be in touch in the next day or two. Thanks again, Colin
 

Thread Starter

colin.t

Joined Jul 6, 2018
16
Hi, I have done the easiest thing and put a 330ohm resistor between the 555 and opto. this has reduced the current to about 11Ma. I have also changed the input to pin 8 to the unregulated supply with a 220ohm resistor. the mosfet switches really well now. However, the trimmers VR3 and VR4 (used to change the on/off voltage and hysterysis) have no effect the cut in is at 11.91v and cut off is at 11.71v. I have changed the resistor values on R6 and R7 but still the same. any ideas?

Thanks Colin
 

Attachments

Thread Starter

colin.t

Joined Jul 6, 2018
16
Thanks for the super quick response Bertus. I have made the change (and replaced the original resistors) and it is now working as before (trimmers set at more or less the same resistance as the version with relays). Just one other thing, before changing the place of the 220ohm resistor the opto was running at 35°C and now it is at 52°C. Have you any ideas what could I do to reduce this?
Thanks Colin
 

Janis59

Joined Aug 21, 2017
1,834
Generally any mosfet will like to overheat in few cases
1) the wild oscillation had been started: kill it by gate resistor, or by gate ferrite ring
2) The FET stays half-opened, ie gate spannung is too low. Must be average 16...18...20 V except few non-standard devices. If that stays below 12V thats it.
3) The inductive load gives bold backshoots at state change, apply the proper dampfer between D and S.
 

Thread Starter

colin.t

Joined Jul 6, 2018
16
Hi Janis, thanks for the reply. I may not have been clear on this. It is the optocoupler that is a bit hot, not the MOSFET. The MOSFET is running at about 38°C (ambient is 28°C) while drawing around 4 amps. The FET is switched with a gate voltage at 1.35v supplied by the Optocoupler. My concern is the temperature of the Optocoupler at around 60°C, is this acceptable? what can I do to reduce it?
Thanks Colin
 

ebp

Joined Feb 8, 2018
2,332
60°C is acceptable, but seems rather more than expected.

"The FET is switched with a gate voltage at 1.35v"
1.35 V where? Gate to source? That is far too low for proper enhancement of that FET, and doesn't even meet the maximum spec for gate threshold voltage, which is measured at drain current of 1 mA. You need to be driving it with at least 3 volts for reasonable enhancement. With the driver powered with 12 volts, if the gate-source voltage really is only 1.35 V the driver must be damaged. Since there is no DC current into the gate, the driver output should be over 11 volts.

You most certainly do not require "average 16...18...20 V" or even 12 volts at the gate. 5 V gate to source is entirely adequate for the application. Only very slight reduction in ON resistance is achieved by higher drive voltage. The reality is that very few modern FETs get much benefit from drive of more than 10 volts or so. Many drivers have internal regulators to limit the gate drive to 12 volts.

The temperature rise of the FET is more than would be expected with adequate gate drive with only 4 A of drain current. The expected power dissipation, ignoring switching loss (which is virtually zero for your ap) should be around 60 to 70 mW, which would cause only 3-4°C temperature rise of the actual die. The package temperature rise would be a fraction of that. The fact you are seeing about 10°C rise says the FET isn't being driven adequately - so back to that 1.35 V. When the FET is properly driven, you should measure no more than about 20 mV between drain and source.

Your schematic does not show a decoupling capacitor for the output of the driver. This is essential to assure clean switching. Use at least 100 nF on the shortest leads possible as close to the IC as possible.
 
Last edited:

BobTPH

Joined Jun 5, 2013
8,813
If the opto is overheating it is because you are putting too much current through the LED. Do you have a current limiting resistor?

Everything necessary to fix this was given to you in posts 3 and 4.

Bob
 

ebeowulf17

Joined Aug 12, 2014
3,307
Thanks for the super quick response Bertus. I have made the change (and replaced the original resistors) and it is now working as before (trimmers set at more or less the same resistance as the version with relays). Just one other thing, before changing the place of the 220ohm resistor the opto was running at 35°C and now it is at 52°C. Have you any ideas what could I do to reduce this?
Thanks Colin
When you moved the 220ohm resistor, did you also connect pin 5 to ground?

Your very first schematic showed pin 5 connected to ground indirectly through a resistor, but your most recent schematic showed no pin 5 connection at all. @bertus' update showed the proper connection, which is a direct connection from pin 5 to ground.

I don't know if this would explain any of your troubles, but it's worth a shot.
 

Thread Starter

colin.t

Joined Jul 6, 2018
16
Hi All. My mistake (see updated schematic). I do have opto pin 5 direct to ground. I have 1.35v on the LED in the Optocoupler (if I increase the resistance more it does not switch). I have a full 12v arriving at the FET gate with a 330Ohm resistor between the opto pin 6 and the FET gate, the FET is switching perfect and does not get hot. It is the OPTOCOUPLER that is too hot? that is what I am concerned about.
 

Attachments

Kjeldgaard

Joined Apr 7, 2016
476
I've had some thoughts about this somewhat special '555 circuit, and will give other readers the chance to use my idea.

To get fixed trigger voltages on the TRIG and THR inputs, it is enough to keep the CVOLT input at a fixed voltage:
HiLo555Det_1.jpg
The supply for the '555 is directly from the 12V supply, keeping the CVOLT on 6.2 Volt with the Z1 Zener diode and giving the Zener diode a slightly higher work current with Rz.

The R2 and R4 values must be recalculated according to the modified voltage on CVOLT.
 

ebeowulf17

Joined Aug 12, 2014
3,307
Hi All. My mistake (see updated schematic). I do have opto pin 5 direct to ground. I have 1.35v on the LED in the Optocoupler (if I increase the resistance more it does not switch). I have a full 12v arriving at the FET gate with a 330Ohm resistor between the opto pin 6 and the FET gate, the FET is switching perfect and does not get hot. It is the OPTOCOUPLER that is too hot? that is what I am concerned about.
Sounds like you're getting close. Thanks for the clarifications. Everything you described, except the excessive heating, sounds right to me. However, you've missed a key point raised several posts ago. This is also supported by the datasheet:
Your schematic does not show a decoupling capacitor for the output of the driver. This is essential to assure clean switching. Use at least 100 nF on the shortest leads possible as close to the IC as possible.
From the datasheet:
4F95447A-4D51-42E9-8220-892F00DA5053.jpeg
 
Top