I discovered mosfets would be much better for switching my LED 25W lights rather than a noisy relay so bought some IRF40N mosfets off ebay.
Done my research, hooked em up, 1 mosfet per light with a resistor on the gate to limit current flow and a pull-down resistor to earth.
Has worked flawlessly not only switching them on & off but also dimming, only problem I had to remove the power circuit off the breadboard when it caused problems and started melting it, also put the mosfets on a heatsink.
Thought I could do the same with my DC fan invention and control it just like my lights, the mosfets rated for about 33A, the fan 80W / 12V = 6.66A. The fan being a 10inch radiator fan mounted on an adjustable TV bracket arm.
Anyway, I wanted to test it first with the mosfet simply set high/low (on/off), Fan turns on, but won't turn off, touching the mosfet and heatsink which I located directly behind the fan to take advantage of the air flow, and it is hot as after only about 1 minute run-time.
I took the mosfet back to the test bench, and my conclusion is that it is burn't out (shorted), current flows from drain to source regardless what is applied to the gate.
So long story short, any recommendations on what mosfet I should be using?
Also any further info to help me from looking stupid in future would be great...
You need a logic level FET, the one you have needs 10V to turn it on and as you only have 5V it is not fully turned on. This means that the FET is running in a linear mode and will get very hot.
Solution use a logic level FET that will turn on fully with 5V. Look at the data sheet for the FET voltage they give when quoting the drain source resistance. Ignor the gate threshold value in the data sheet, it is not what you think it is.
You don't pay attention to the current rating, unless you want to melt things, its usually represents
the extreme thermal limit of the chip with infinite heatsink - a thoroughly practical regime of use
except for low duty-cycle pulse operation.
The way to choose a MOSFET is by the on-resistance, Rds(on). The power dissipation when on is
I-squared-R, so for instance 10A and 30 milliohm MOSFET dissipates 10x10x0.03 = 3W (which would
need a heatsink). These days on resistances down to 1 milliohm or so are obtainable and heatsinking
can often be avoided.
As for a specific MOSFET there are 10,000's (quite literally) - if you use a good electronics distributer,
and such companies have search tools to help you identify the cheapest suitable components.
Awesome, I was only thinking in terms of current carrying capacity for some reason... Will read the data sheets properly and calculate the heat dissipation and have another go at it with some new fets...
It might not make much difference on such a small motor, but for the hell of it I'll start the fan like a soft-starter keeping the current low.
For the record - the fan blades on a radiator fan seems to provide good air volume up close, but at a distance not so great, and its noisy since it has like 10 blades... So i'll see how it sounds at lower speeds but may have to seek out replacement blades... probably just steal one off a AC fan...
What current are the LEDs using? You may well get away with no need for active cooling, or even a heatsink.
44 milliohm is pretty poor for a MOSFET - and you don't need anything like 100V for 12V circuit - high voltage MOSFETs have a lot larger on-resistance (all things being equal).
Think 5 milliohm 30V, your 6.7A fan would mean 0.2W in the MOSFET - no heatsink needed. You'd
need a free-wheel diode across the fan since its an inductive load, most big rectifier diodes will handle 7A peak
pulses.
44 milliohm means 6.7A dissipates 2W to get rid of with the ancient IRF40N even if it were logic-level..
Are you using PWM? At what frequency? You may need to consider a MOSFET driver too to improve switching
speed.
You also need a protection diode across the motor to prevent the inductive kick from the motor from zapping the MOSFET - the diode goes across the motor in the opposite direction of the current flow through the motor (and as others have pointed out, use a logic level MOSFET so it is turning on completely).
Sorry didn't realise I had more replies, I havn't rushed into getting the fan going yet, the LED lights still working just fine, they are on a big heatsink which I can ditch once getting some better mosfets.
The LEDs are just 7W actually, dimmable and suitable for 12/24V battery systems... So @ 12.5V = 0.56A max.
I understand the PWM output to be 500Hz, I'm assuming a mosfet driver will provide a higher frequency than the arduino can supply?
I'm just about to read up a bit more on PWM, about to start playing with PLC's at TAFE/trade school soon so I'm sure this will be relevant.
Yea I am guilty of choosing a common mosfet on ebay and hoping it would cover most situations.. but that's ok, now I'm learning something...
I'm assuming a mosfet driver will provide a higher frequency than the arduino can supply?
Not sure what you mean or if it is just not expressed correctly.
A mosfets driver does not PROVIDE a frequency it responds to one he Arduino sends.
There is an extra amount of dissipation as the FET turns on. This is due to the fact that gate of the FET looks like a capacitor and during the time it takes to charge up the FET is operating in the linear region. The faster you drive it the more excess dissipation you will get, so don't switch faster than you need to in order to keep the heat generated to a minimum.
Grumpy_Mike:
Not sure what you mean or if it is just not expressed correctly.
A mosfets driver does not PROVIDE a frequency it responds to one he Arduino sends.
There is an extra amount of dissipation as the FET turns on. This is due to the fact that gate of the FET looks like a capacitor and during the time it takes to charge up the FET is operating in the linear region. The faster you drive it the more excess dissipation you will get, so don't switch faster than you need to in order to keep the heat generated to a minimum.
Its the opposite if there's any load, the saving in switching-losses in the primary circuit outweighs the
gate dissipation, which is only 0.25W per MHz for a logic-level 10nF gate anyway, and you won't be
running at MHz!!! Gate driver losses are fCV^2, spread across the driver and internal gate resistance of
the FET.
MarkT:
Its the opposite if there's any load, the saving in switching-losses in the primary circuit outweighs the
gate dissipation, which is only 0.25W per MHz for a logic-level 10nF gate anyway, and you won't be
running at MHz!!! Gate driver losses are fCV^2, spread across the driver and internal gate resistance of
the FET.
You're misunderstanding. Mike's point has nothing to do with the energy required to load and unload the gate, but the energy the MOSFET itself dissipates during the brief period of time it is in the linear region during the transition between cutoff and saturation.
And there is a small correction I should make to what he wrote as well. His words give the impression that the heat dissipation only happens during the OFF->ON transition. In fact, there is a pulse of power dissipation during every transition, both OFF->ON and ON->OFF.