I am looking for a good FET that I can directly control with a GPIO pin of the ESP32.
After shopping around for a bit, I found the NTD4858N at Digikey. It needs to produce little heat, support 12 VDC drain to source, and up to 10A of current.
From what I can tell, this meets all that criteria, and should trigger the gate at 3.3V easily. I already know NPN transistors are not on the table, as I am replacing a TIP120, and it gets way too hot. I have never tried N-channel MOSFETS, but as long as it is controlling DC power, from what I have read today, they work great when heat is an issue.. so long as the drain-source remains DC current.
Can anyone provide any comments on this (or similar) FETS before I place the order? Any other gotchas I need to be mindful of?
It doesn’t look like a good choice for 3.3v. Check out figure 3 on page 4. The on state resistance Rds goes off the charts at about 3.5v. What this means is that your voltage will drop and will get hot... how hot don’t know.
Power dissipation in watts is calculated as P= I^2 x R. As R goes up your heat will go up exponentially.
It would be a good one for 5V but not recommended for 12v 10A switched with 3.3V logic.
Bjerknez:
I use FQP30N06L. Works perfect with ESP32 and 3.3v generally. And 5v off course.
Then you are being lucky, its not even properly rated for 5V logic, let alone 3.3V.
Logic level FETs for 5V are rated for 4.5V gate drive, to allow for variation in power supply voltages and
such. The FQP30N06L is rated for 5.0V only.
What you may not realize from the datasheets is the large variation between devices for gate
voltages, meaning the performance below the rated gate voltage will vary substantially between
batches and even over the lifetime of a device due to ion-migration in the gate oxide.
For 3.3V operation you'll only find surface mount MOSFETs in my experience.
Another way to handle the issue is to drive the MOSFET at 5V (or even 12V), using a gate-driver
chip - gate drivers take logic inputs (almost all are 3 / 5 volt compatible), and provide high current
gate drive to a MOSFET, improving switching times as well as boosting voltage.
Another way to handle the issue is to drive the MOSFET at 5V (or even 12V), using a gate-driver chip - gate drivers take logic inputs (almost all are 3 / 5 volt compatible), and provide high current gate drive to a MOSFET, improving switching times as well as boosting voltage.
Do you mean something like this? Have you tried them with high currents and 12V, and found they don't get too hot? The datasheet indicates that it has a high RON compared to the 'good' part of the RDS of FETs in the 5V trigger range. However, that might be lower than RDS if the FET is operating at 2.7V. Can someone vouch these things perform better given the conditions?
I have used it with ESP32, and Wemos D1 Mini, and i have also tryed PWM on power led and a 12v fan. It works fine...
Did you run components at 12V, 10A though? The TIP120 technically works too, but it produces way too much heat. I might order the FET you suggested, to run in a comparison test, but RDS seems like it could be a problem here too, according to the datasheet, as others have pointed out.
I'd be looking at a different approach here. As mentioned above: only surface mount MOSFETs for 3.3V gates, and those tend to come in SOT23 package supporting no more than 6-8A continuous current (and at that require a large copper pour for heat dissipation in the PCB).
Heat goes up with the current squared, so to keep heat under control you want a low on resistance. There are lots of MOSFETs that will switch 10A without breaking a sweat, if you apply 12V to the gate - which is easy as you have 12V available. Just add a transistor to switch the gate with your 3.3V signal and you're good to go.
Or if you want to PWM your load, probably better to use a gate driver IC instead. Most work fine with 3.3V signals and can happily switch a 12V gate voltage.
Bjerknez:
I'm not understand. "The Guy With Swiss Accent" on Youtube said that this mosfet is fine for 3.3v and 5v...?
I have used it with ESP32, and Wemos D1 Mini, and i have also tryed PWM on power led and a 12v fan. It works fine...
No, it works well enough at the moment for your purposes, and you are being lucky.
The datasheet says what the device is guaranteed to do, across the full temperature range and for
every device in every batch manufactured and over its lifetime - its the contract from the manufacturer
that they test for.
Random person on Youtube is not a source of reliable information.
Many devices will perform better than the datasheet, but that's being lucky.
MOSFETs in particular have a lot of variation in gate voltages - a big manufacturing spread (as much as +/-1V
for some devices).
Pardon me I just watched some Tiger King episodes, so I might sound like a redneck here..
But..
I found out the biggest spec to look for in a transistor is its power tolerance. I guess it's pretty obvious, but you can find N-Channel transistors that aren't bad... no heat sinks needed for 12V 10A. The thicker ones do better, even if they have the same thresholds to go from 0 to 1, its not just about the threshold. You want the beefier THT types for that. 120+ watt tolerance ones don't get very warm. Might have been obvious to some of you, but I didn't know how key that was. Voltage thresholds, and watt ratings are the keys!
Anyone making some homemade ventilators might find that helpful! Thought I would share.
Careful with the PWM signals though, that increases heat pretty bad. I'm talking about a linear voltage signal in.. DC power ratings are reasonably accurate.
Trust me, a "120W tolerant" TO-220 or it's big brother TO-247 will get VERY hot when you try to have it dissipate just 10% of that power with no heat sink. If it doesn't simply burn out after a few seconds, it'll get hot enough to desolder itself.
On resistance combined with current gives you the power dissipated. If ~1W or more, a TO-220 needs a heat sink.
brento:
Careful with the PWM signals though, that increases heat pretty bad.
That just shows you're driving the gate too slow, too big series resistor and/or too little current ability - like from an Arduino pin. Use a gate driver.
