I've been reading some basic tutorials on using MOSFETs as switches. Have I got this right for an N type enhancement FET?
When VGS > VTH, current can flow into the drain (ID).
FETs get hot due to this drain current.
So if VGS < VTH, the FET shouldn't get hot. Right?
Am I right in thinking the heat in the FET is a function of power (P = I2 R) and ID and the RDS(on)
So if the load the FET is switching is relatively small, and/or only on for a small amount of time, again a FET getting hot is less of a consideration.
I'm asking because I've been comparing FETs in different packages and wondering about TO92 not having built in heat sinks whereas TO220 do. Those that I've looked at so far have RDS(on) a lot larger for the TO92 but will only handle a very small percentage of the ID and power.
(This question isn't related to a specific circuit at the moment.)
Yes. Usually, the data sheet characteristic you look for is RDS(on). This is the effective resistance of the Mosfet at the quoted voltage Vgs (say 5 volts for a "logic" level Mosfet). That resistance causes a heating effect I^2 R.
If the Mosfet is "partially conducting" because the Vgs is above zero (or Vth) and below this minimum threshold quoted for RDS(on) , it will also get warm. This can especially be the case where it is being driven by higher frequency PWM which is not sharply rising and falling, prolonging the time the Mosfet is neither fully on nor fully off.
TO92 parts tend to be old designs, newer MOSFETs come in TO220 or SMD packages. Those newer parts have generally much better specifications.
A MOSFET when on behaves very much like a resistor, so you can easily calculate the heat dissipation that way.
But especially when using PWM make sure the gate voltage can rise and fall fast, as otherwise the part spends a significant time in partial-on state and heat goes up. The gate behaves like a small capacitor (the gate and total capacitance you can find in the datasheet). As a rule of thumb, the larger the current a MOSFET can switch the larger this capacitance. This is where gate drivers come in the picture.
From the formula and generally for all components, higher current higher heat but in transistors there is also switching which is heat generation impact
I've found graphs of how case temp changes with ID, but they're not on all datasheets.
If you're not dealing with PWM and/or gate voltages around VTH, is there anything else typically on datasheets that will indicate when a switched load will start resulting in a heat problem for the FET?
Most datasheets will give you a few gate voltages and corresponding on resistance. Don't try to operate the MOSFET at a gate voltage below the lowest explicitly listed. Just being above the threshold voltage doesn't mean the MOSFET is switched on properly.
So if VGS < VTH, the FET shouldn't get hot. Right?
For the sake of discussion, under WHAT circumstances could this occur ?
If the drive signal is 0V, then an N-channel would be completely OFF and a P-channel would be completely On.
Likewise a P-channel would only be above 0V but below Vth if the OFF signal is not high enough, which should
not occur in a proper design. An adequate voltage source and pullup resistor should prevent this.
Also, it seems that your question would only be an issue if the choice of device had specifications that were
marginal for the application which could be avoided by making a suitable choice for the device.
Remember that the Vgs threshold voltage is the point where the mosfet just begins to conduct. You'll need a higher voltage to turn it fully on.
The heat dissipated by the mosfet is also equal to the voltage drop across it times the current flowing through it, both of which can actually be measured under load. You can also see whether the mosfet is too hot to touch. Just a bit warm is probably ok.
All of the mosfets in TO-92 packages have pretty poor RDSon values. They will be several ohms. The little SOT-23 surface mount parts will be maybe 30-40 milliohms, so are far better able to handle current. To use them for breadboarding, you would need to get some SOT-23 to SIP adapter boards, and solder carefully.
raschemmel:
If the drive signal is 0V, then an N-channel would be completely OFF and a P-channel would be completely On.
Both would be completely OFF.
An N-channel needs a voltage VGS > VGS, TH, for most n-channel MOSFETs it's somewhere between 1 and 4 V), while for a p-channel MOSFET it needs to be VGS < VGS, TH, for most p-channel MOSFETs it's somewhere between -1 and -4 V.
I remember you need something like a JFET to be fully on at 0V.
The vast vast majority of power MOSFETs are enhancement mode, which are off with zero gate voltage,
but they can be made as depletion mode, where they are on with zero volts.
JFETs are always depletion mode, MOSFETs can be either enhancement mode or depletion mode, but
the latter are rare as for typical switching applications you want the switch to default to off.
And either can be p-channel or n-channel, the latter being about 3 times superior in performance by
the physics of silicon.
Thus for power electronics most devices are n-channel enhancement-mode, so must be MOSFETs.
When you pick a power MOSFET, the important specs are:
breakdown voltage
on-resistance and the associated Vgs values
gate capacitance
power dissipation of the package
The threshold voltage confuses newcomers, ignore it.
An N-channel needs a voltage VGS > VGS, TH, for most n-channel MOSFETs it's somewhere between 1 and 4 V), while for a p-channel MOSFET it needs to be VGS < VGS, TH, for most p-channel MOSFETs it's somewhere between -1 and -4 V.
I remember you need something like a JFET to be fully on at 0V."
So [-4V <---P-ch---- -1V][1V----N-ch-----4V]>
"The vast vast majority of power MOSFETs are enhancement mode, which are off with zero gate voltage,
but they can be made as depletion mode, where they are on with zero volts.
JFETs are always depletion mode, MOSFETs can be either enhancement mode or depletion mode, but
the latter are rare as for typical switching applications you want the switch to default to off.
And either can be p-channel or n-channel, the latter being about 3 times superior in performance by
the physics of silicon.
Thus for power electronics most devices are n-channel enhancement-mode, so must be MOSFETs."
Too many modes for my comfort...
"When you pick a power MOSFET, the important specs are:
Breakdown voltage On-resistance Associated Vgs values Gate capacitance Power dissipation of the package
or
"B.O.A.G.P."
"Buy Only Authorized Genuine Products"