Hi, another thread about selecting an n-channel, logic level MOSFET...
I've read a lot of posts, usually ending with @crossroads applying a filter on digikey... anyway I saw a lot of people talking about IRL540 (expens-ish) but making a search on farnell (I'm in France) I came across this one PSMN2R7-30PL datasheet and it seems to me like a real good one...
100A / 30V / Rds (on) 0.0023 Ohm / Vth 1.7V and it seems quite responsive td(on) 46nS / td(off) 74nS / Pd 170W
and there is it's 15% more expensive brother PSMN2R0-30PL datasheet almost same characteristics but lower Rds (on) 0.002 / higher power dissipation Pd 211W and a little bit slower td(on) 63nS / td(off) 111nS
So, here are my questions: knowing that my usual applications would be PWM driving leds or motors 12v/24v and no more than 150W (the largest motor I have 24V) with a nano
what do you think about this MOSFET? it seems good or did I miss something?
in your own humble opinion is the 15% extra cost worth it for the 2nd one?
is there any interest into putting an optocoupler between a logic level MOSFET and the MCU?
what exactly means "Power Dissipation (Pd)" on the datasheet? is it the maximum power it can dissipate due to Rds? ==> Ids^2*Rds ?
what is the difference between "td(on) turn-on delay time" and "tr rise time"?
on the time "Dynamic characteristic" chart for turn-on and off times it's stated "Rg(ext)=4.7Ohm; Vgs=4.5V" but I've seen that most of the time the current limiting resistors are of about 150Ohm.. so I guess that bigger the resistance, bigger the times am I right? how do I check that this will be compatible with my PWM frequency? (I use PWM frequency library to get the nano on 20kHz pwm)
they don't state any pull-down resistance... I guess putting it will do no harm?
most Vgs for logic-level on datasheets are 4.5v, but I believe the nano will put 5 on the digital pins and then again I read somewhere that AVR chips logic is 4.5v... so, I'm confused... I guess it doesn't really matters...
Go with the faster one, you want to transition between On and Off as fast as possible, the parts have higher resistance in the middle area and that's where heat gets generated.
Optocoupler will slow the transition down and make more heat.
Yes, exactly. But also power with higher Rds when transistor is not yet fully on.
How long it takes to respond to a control signal (delay time, propogation delay) vs the time to actually transition at the output.
Drive the gate hard to ensure it switches quick, overcoming the Gate capacitance. This means low resistance from an arduino pin, or an actual MOSFET driver.
Pull down resistance just keeps the part off while the Arduino boots up and the IO pins are floating. 10K should be fine, and not interfere with Arduino driving the pin.
Arduino high output is guaranteed to be >= 4.2V with Vcc = 5V and Iout <= 20mA.
If you need 4.5V and higher current, then you need to buffer the output.
With opto couplers be aware that standard ones are extremely slow switching, which is no good for driving
MOSFET gates with PWM.
Fast logic-speed opto-isolators and rf-isolators are available. Isolation is useful if there is any change of MOSFET
failing at high power, which typically blows up the MOSFET and its gate driver chip.
MOSFETs are typically cheap compared to the driver circuitry and PCB board area needed to support
them until you get to screw-terminal package sizes.
Be careful with the rated "wattage" it assumes you can keep the temperature of the transistor junction below ~120 °C. Think of a 150 watt light bulb, its that amount of heat you will have to get rid of (with a heatsink). That said, the power you would dissipate with a typical 12V or 24 volt motor is likely much lower.
Regarding the opto isolator. Yes they tend to be slow, but the if the output switches fast (to switch the MOSFET quickly), the delay may not matter. You will just have to consider it in your PWM parameter selection.
If you want to run a 24V motor, you will need at least a 50V Mosfet. The reason is that when using PWM to control the speed of your motor the FET is switching of and off. During the off period the motor back-emf will take the drain of your FET to at least 24V above the power supply, but could be more depending on the motor load. So for safety you will, as stated, need a 50V device as a minimum but preferably a 100V part. It is also good practice to place a reverse connected diode across the motor to snub the back emf.
Do not use an opto-coupler, it is not necessary but, you may well need to boost the current output of your Arduino so that you can correctly drive the FET. If you plan to run the FET at a relatively low speed you may get away with it but, the gate transitions will be slow which will cause the FET to heat up.
Power dissipation in any semiconductor device depends on its ability to remove unwanted energy (heat) from the active element. In any device there are thermal boundaries that waste energy has to negotiate which affects this ability. Each boundary will have a thermal gradient. The magnitude of the gradient will depend on the amount of energy that is has to cope with.
As an example, a typical TO220 device will have two boundaries one where the silicon is bonded to the metal substrate of its package and another in the silicon itself.
Then there is the interface between the device and its heatsink (assuming one is used) which is another boundary. If an insulating washer is used, this is another boundary. Each of these will also have a thermal gradient. It is quite possible for the silicon in a power transistor to be 10 to 20C higher than the heatsink if you are trying to dissipate a lot of power. Also, the energy that you are trying to dissipate has to get through the metal interface of the package.
So, a low Rds on is a good place to start but, do not try to dissipate 200Watts in a TO220 device as it will not cope. The generally accepted maximum for this package is about 50W and for a TO247 about 115W.
A logic level device is the way to go but, read the data sheet(s) carefully as most are not fully enhanced (turned hard on) at 5V but are generally 70 to 80 % there.
Thank you for ypur comment you all! I just want to add a comment about John's comment.
IRL540 seems quite popular somehow... But line to line specs it's not exceptional. The one I proposed on the first post of this thread is faster, cheaper and has a lower Rds...
be aware about ebay/china items (and I'm quite happy about most of the items I get there) but I got 10 IRL540 that turned out to be counterfits... Not even logic levels... Not even constant in there behaviour. Using a transistor tester (the one based on the avr chip) I got Vth varying between 3.98 to 4.8!! Volts. I was even able to see the one at 4.8v dimming the leds compared to the others at <4v.... That's why I'm getting my next stock from farnell....
NOSUM:
... you may well need to boost the current output of your Arduino so that you can correctly drive the FET. If you plan to run the FET at a relatively low speed you may get away with it but, the gate transitions will be slow which will cause the FET to heat up.
How do I know if I need to boost the arduino pin? I thought that avoiding this was the point of logic-level MOSFETs. And if needed, how could I boost it? With a trasistor like 2N700? or 2N440?
So I could potentially need 3 levels of tension on mi circuit? Like 5v for the arduino, 10~12v for driving the MOSFET and 24v for driving the motor
IRL540 seems quite popular somehow... But line to line specs it's not exceptional. The one I proposed on the first post of this thread is faster, cheaper and has a lower Rds...
Its popular because its popular, its certainly nothing special. Its fairly high voltage, so its on-resistance
is compromized by that for low voltages (high voltage means higher on-resistance, all else being equal).
For a 12V circuit there are far better choices. For a 48V load its not unreasonable, but there are still far
better choices...
People often list a device such as this in a circuit design because its easy to get hold of, and for no better
reason, and so the device gets to far outlive its normal lifetime (which for most semiconductors is a few
years before better variants come out)
There are literally 1000's of new MOSFETs out every year, its a fiercely competitive market.
Any comments about this? When I thought I got it sorted out, I lost it
NOSUM:
... you may well need to boost the current output of your Arduino so that you can correctly drive the FET. If you plan to run the FET at a relatively low speed you may get away with it but, the gate transitions will be slow which will cause the FET to heat up.
How do I know if I need to boost the arduino pin? I thought that avoiding this was the point of logic-level MOSFETs. And if needed, how could I boost it? With a trasistor like 2N700? or 2N440?
So I could potentially need 3 levels of tension on mi circuit? Like 5v for the arduino, 10~12v for driving the MOSFET and 24v for driving the motor
The reason you my need to boost the current output from you ARDUINO is because the gate source junction of the FET is very capacitive. As we all know, to charge a capacitor quickly requires a lot of current depending of course on the capacitors value. If you plan to turn your FET on and off quickly you will need to increase its current drive. At low frequencies it is not an issue.
And what would be low frequency? Is there a formula? Or a line to check on the datasheet? My plan is to control a motor by PWM at 20kHz ~ 30kHz (more or less the same I use for dimming leds... audible reasons for the motor, flickering reasons for the leds)
And in the case of needing a boost, how do I achieve that? Using a transistor I guess? Like BC547?
That is fairly fast for switching a motor.
With regard to calculating how much current you may need to switch on a FET quickly you need to establish the following.
a) The rise time (rate of change) of the output signal from your Arduino.
b) The gate - source capacitance of the FET you intend to use in Farads (not micro or nano as the sum will be wrong).
c) The maximum voltage you will apply to the gate of the FET.
The sum is now I = C x (dV / dT) Where I = Current, V is Volts and T is Rise time or Rate of change.
Capacitance is expressed in Farads. T is in Seconds.
Now plug some numbers in and see how you get on.
I did a quick sum based on the IRL540 which has a gate capacitance of 2.2nF and a rise time of 13nS (which is the only number I could track down) and driving the gate to 5V I ended up with 850mA. This is a substantial amount but, it only lasts as long as the rise time so should be fairly short.
Now, the current output of you Arduino is limited to 20mA, if I remember correctly, which means that the rise time on the gate of the FET will be much longer. This scenario also applies to discharging the Gate on negative transitions.
The period for 20KHz is 50us. At minimum PWM the pulse width could be 500ns. With just 20mA driving the FET, the turn on period could be a substantial part of that period. Increasing the current capability of the ARDUINO output will take you closer to its native rise time.
You must add all of these periods to the quoted response times for you chosen FET
See attachment for a suitable way to increase the current output of your ARDUINO where V1 is your PSU, V2 is the Arduino output and R1 is your motor load.
Thank you NOSUM.
Just to get it clear, why 2 transistors instead of only one? An wouldn't I need a pull-down on the gate?
<>
for anyone else reading this.... what NOSUM suggested is called a "bipolar totem-pole driver" basically the PNP charges the Gate and the NPN discharges it. More info here starlight: Dimming a 12V LED strip with a mosfet and PWM
Other related terms: Class B amplifier; push-pull amplifier;
Still I wonder how would the circuit react to an asymmetrical NPN/PNP pair... since most PNP seem to have lower hFe.... ?????
jorgemarmo:
Thank you NOSUM.
Just to get it clear, why 2 transistors instead of only one? An wouldn't I need a pull-down on the gate?
<>
for anyone else reading this.... what NOSUM suggested is called a "bipolar totem-pole driver" basically the PNP charges the Gate and the NPN discharges it. More info here starlight: Dimming a 12V LED strip with a mosfet and PWM
Other related terms: Class B amplifier; push-pull amplifier;
Still I wonder how would the circuit react to an asymmetrical NPN/PNP pair... since most PNP seem to have lower hFe.... ?????
Everyone uses gate driver chips which are MOSFETs internally (faster), one cheap chip, does everything, no need to
build a discrete gate driver circuit. My favorite low-side MOSFET driver is MIC4422, plenty of current
ability, you can even use it as a class-D audio output stage by itself! Favorite because it will work from 5V to
18V and is DIP or surface mount and powerful enough for any MOSFET of IGBT you can imagine.
Also gate driver chips usually have very useful protection features like under-voltage shutdown which
can save you from an expensive power-excursion event.