I am rather new on Arduino, having bought my first kit somewhere later last year.
I am a mechanical engineer, but have a vivid interest in all other engineering disciplines and I enjoy the electronics immensely.
Right, so I am busy building myself a simple solar voltage regulator with PWM control.
For this application I see no need to mess around with PWM frequencies and I am using pin 10 on my Pro-Mini which is supposed to have a 490 Hz frequency.
Now obviously I need to learn all the tricks to make sure the MOSFET runs as cool as possible and so far I observed the following:
The MOSFETS I have in my toolbox, (GPW30N06 and IRFZ44N) does not have much of a current rating at 5V switching and in any case it draws quite a bit of current from the pin. Swtiching it directly the MOSFET ran very hot.
That is when I discovered that it is usually a good idea to use some kind of driver to switch the MOSFET.
With my limited spares, I had 2 options:
Use a plain transistor which can up the current and voltage to switch the gate, but it seems that transistors are not that quick with rise and fall times. I have not done this yet, but will try later today.
What I did however was to use was a 4N33 optocoupler which can give quite low rise and fall times, but it is also rather demanding on the input current from the controller if one wants to achieve that. However, even with this limitation, it made a big improvement in MOSFET temperature, but I think there is still more room for improvement.
I will evaluate the above by fixing a TMP36 to the heat sink and read it with the Arduino when I try the different options.
I do understand that most of the losses of the MOSFET takes place in the rise and fall times, and the more there are of those cycles, the more demand there is on quick rise and fall times.
So my question is this: 490 Hz PWM frequency is adequate for this application, but it is still roughly a factor of 1000 longer than the rise and fall times of the MOSFET, so IMO one does not really move in a critical zone here.
If the above is true, then a transistor should do the job pretty well and that would be the most convenient to use. Is this observation correct?
Look at the data sheet of those FETs. They are not logic level FETs and so you are not turning them on fully. The lack of gate voltage is causing you excessive dissipation not the rise time of the gate signal.
[To me fast switching is when you push GaNFETs to the edge - several MHz, but that
might not be relevant here!!]
Standard power MOSFETs when driven by a suitable MOSFET driver chips will typically
switch in between 50 to 200ns, depending on size of the MOSFET. That's the normal
notion of fast switching - so keeping switching time to 1% of total time means limiting
the switching frequency to about 50kHz (for 100ns per switching event) or 25kHz for
200ns. Put another way you can get ultrasonic switching frequencies without excessive
switching losses using MOSFETs.
If you don't use a proper MOSFET driver you won't be getting anything like those
switching speeds, MOSFET gate is a large capacitive load and needs 100's of mA or
more to switch promptly.
The moral is switch a MOSFET with a dedicated MOSFET driver chip - there are 1000's
of driver chips available, quite literally.
Use lots of decoupling on a MOSFET driver (10uF ceramic minimum I suggest). Keep the
wiring between a MOSFET and its driver short and free of loops (run gate and source wires
parallel or twisted pair).
Grumpy_Mike:
Look at the data sheet of those FETs. They are not logic level FETs and so you are not turning them on fully. The lack of gate voltage is causing you excessive dissipation not the rise time of the gate signal.
Thanks!
At least I could gather that from the data sheets, that is why I switched it through the opto-coupler and I do get over 10V on the MOSFET gate. That said, my Arduino is struggling a bit not being able to keep a full 5V on the opto-coupler pin so this is not at all ideal.
According to my measurements, I do get the 0.04 ohm resistance if the MOSFET is switched permanently on.
I have not tried a logic level MOSFET yet. The datasheet I have from one does (NTE2986) does not really indicate the MOSFET resistance, so it is rather difficult to determine power losses.
MarkT:
If you don't use a proper MOSFET driver you won't be getting anything like those
switching speeds, MOSFET gate is a large capacitive load and needs 100's of mA or
more to switch promptly.
koosjr:
I have not tried a logic level MOSFET yet. The datasheet I have from one does (NTE2986) does not really indicate the MOSFET resistance, so it is rather difficult to determine power losses.
Its clearly given as a maximum of 28 milliohms at 5V gate drive - search DS for "Rds(on)"