Practical differences between Transistor and MOSFET

[Youkai]

In another thread someones suggested to me using an H-bridge/MOSFET setup to control a DC motor. I'm trying to understand the differences between a transistor and MOSFET and why you would choose one over the other.

Please correct me if I'm wrong but my understanding of the "for dummies" difference between MOSFET and a transistor is this: Transistor - all 3 pins on the same voltage; MOSFET - switch pin is very low voltage, circuit pins can be significantly higher voltage up to whatever the MOSFET is rated for.

This would mean that with something like an Arduino you could use a MOSFET to control an arbitrarily large piece of equipment requiring practically any voltage needs. Where as with an Arduino-transistor setup you are limited to an equipment with a maximum of 5v since that's what the Arduino/Transistor are running at. Is that correct?

What if the thing I'm trying to control is a 6v DC motor that I'm happy to have run at the same 5v the Arduino runs at? In that case is there a difference between MOSFET and transistor? Will one heat up faster than the other?

[rstofer]

Actually, none of those ideas are quite correct.

I can use an N-channel MOSFET at the bottom of an H-Bridge and I can use an NPN transistor just as well.  Same for the upper device:  I can use a P-channel MOSFET or a PNP transtor.

The transistors are easier to turn on; for the bottom device, the base needs to be 0.7V higher than the grounded emitter.  For the upper device, the same thing happens in reverse, the base needs to be 0.7V lower than the emitter tied to the upper rail.  In both cases, for the MOSFET, the Vgs (Voltage gate to source) often needs to be several volts.  There are logic level devices that reduce this Vgs but never to just 0.7V.

The big difference is in VceSat (of the BJT) and RdsOn of the MOSFET.  The MOSFET will drop less voltage thus generating less heat.  They are preferred for switching heavier currents.  The BJT will have a VceSat of at least 0.2V, perhaps much more at high current, so is not the preferred switching device for high current.

P-channel MOSFETs are not preferred (over N-channel) but using an N-channel device in the upper position requires a gate voltage some volts higher than the upper rail voltage.

Google for 'difference between BJT and MOSFET' for more information.

Many times you will need to use a BJT to control the high voltage on the upper device gate.  Google for 'Arduino MOSFET H-bridge'

[Ian.M]

I wouldn't say an ordinary bipolar junction transistor is *easier* to turn on, as it will draw base current, and if the Arduino cant provide enough to keep it fully saturated at the maximum load current, the voltage drop across it (Vce) will be excessive and it will overheat, and at the same time the load wont get enough voltage.

A MOSFET only draws gate current while actually switching, to charge or discharge the gate to channel capacitance.   As long as the logic supply voltage is much greater than the MOSFET's gate threshold voltage (the minimum voltage to start to turn it on), many smaller MOSFETs can be driven direct from a logic output.

[abraxa]

While not directly related to the H bridge question you originally had, I just would like to point out that BJTs and FETs are fundamentally different devices: BJTs are controlled by current, FETs by voltage.
If it helps, a simplified view is to think of BJTs as current amplifiers (collector current = base current * B; B being the static current amplification factor from the datasheet; emitter current = base + collector current) and FETs as voltage-controlled resistors (there is virtually no current flowing into the gate as the gate acts like a tiny capacitor; neither drain nor source are electrically connected to the gate).

[james_s]

I would simplify it and say a transistor is current-driven, while a mosfet is voltage-driven. That is the most fundamental difference from a usage standpoint.

[Ian.M]

In its usual operating range, a BJT (bipolar junction transistor) can be treated as a current driven black box model and this is useful for 'back of the envelope' design. However the fact that its h_FE varies dramatically with collector current and from device to device of the same type, should give an alert observer a clue that there is something more complex going on. 

It turns out that its actually voltage driven and the current you thought controlled it is actually an unwanted side effect of the controlling voltage being applied across the base-emitter diode junction.  There is a reasonably accurate mathematical model for the behaviour of a BJT that actually corresponds to what's happening to electron flows through its layers of semiconductor, but the maths of the Gummel-Poon model isn't user friendly, and the parameters you need to know to apply it aren't usually found in datasheets.

[james_s]

True, but I think for the typical layman or beginner describing a BJT as current driven is accurate enough, and makes it easier to understand.

Another factor I forgot to mention is forward drop. Mosfets behave a bit like a resistor when fully on, the voltage drop depends on current through them, while BJTs behave like a diode and have a fairly constant voltage drop.

[Zero999]

The big difference is in VceSat (of the BJT) and RdsOn of the MOSFET.  The MOSFET will drop less voltage thus generating less heat.  They are preferred for switching heavier currents.  The BJT will have a VceSat of at least 0.2V, perhaps much more at high current, so is not the preferred switching device for high current.

It's true that MOSFETs have lower on losses, than BJTs, at low voltages, but at high voltages, above a few hundred volts, the reverse is true: BJTs start to have lower on losses, than MOSFETs, which is why BJTs are used in fluorescent tube ballasts, and the IGBT (a type of BJT) is common in industrial variable frequency drives.

As far as the 50VDC circuits a beginner will work on, MOSFETs are generally the better option, unless the current is very low: <1A or so.

[Ian.M]

@James_S,

As long as the student is made aware that its lies-to-children, I've got no problem with saying "A BJT behaves like a current driven black box: put I_b into the base and the collector current Ic=h_FE*I_b in its linear region of operation".  Heck, I usually use that simplification myself to rapidly choose component values when breadboarding.   

However if you don't then go on to explain that its large signal current gain h_FE is rather uncertain, and you need to study the datasheet to get an idea of the range it can vary over between devices and with respect to Ic, and that you have to over-drive the base considerably when using it as a saturated switch, you are setting up a novice for disappointment as they are going to have a hard time making BJT circuits that work reproducibly and reliably.

[wraper]

but at high voltages, above a few hundred volts, the reverse is true: BJTs start to have lower on losses, than MOSFETs, which is why BJTs are used in fluorescent tube ballasts

It's because of lower cost, not efficiency. With 2 cheap BJTs you can make a simple Royer oscillator with minimum count of components required and no driver IC. If you go for highest efficiency, you still need to use MOSFETs. For example, no half decent computer psu uses BJTs, only low power bottom of the barrel crap.

[mtdoc]

Oh no!  Not the current driven versus voltage driven BJT debate again.   As entertaining and educational as the debate can be, it's been done here before. For example this thread.

[Youkai]

Ok, so there is a lot of information here but from a couple of the posts I get the impression that generally speaking: MOSFET is better for these hobby applications. Is that correct?

[Brumby]

It depends on the application, but from my observations, people do seem to like using MOSFETs for switching roles.  The H-bridge is a classic.

[sleemanj]

It's true that MOSFETs have lower on losses, than BJTs, at low voltages, but at high voltages, above a few hundred volts, the reverse is true: BJTs start to have lower on losses, than MOSFETs

Not sure what you are trying to say there....

If you have a "few hundred volts" across (D S or C E) your mosfet or BJT in it's on state, I'd say you have a pretty serious problem wouldn't you?

Is it not that the (ideal) losses in a MOSFET, in it's fully static on state are described by it's RDSon, and the current passing through it...

  W = I * I * R

the voltage across the mosfet in it's on state therefore is a product of the current, unless something is really wonky. 

And in the saturated BJT the voltage across is also somewhat current dependant.

The voltage the rest of your system runs at is of no more interest to a MOSFET or BJT in the on state than it is to a resistor, or a switch.

At high currents a BJT may have an edge over a MOSFET (but this all comes down to picking good parts, you could probably just pick a better MOSFET, or you could probably pick a better BJT, the BJT might be cheaper, the MOSFET might be smaller....), at high voltages, I think it doesn't make sense to talk about losses since we can only assume the devices are off and any loss is down to leaking where it shouldn't be leaking and you picked the wrong thing.

Of course, this is all assuming low frequency (100% duty) gate/base driving.

[agehall]

Ok, so there is a lot of information here but from a couple of the posts I get the impression that generally speaking: MOSFET is better for these hobby applications. Is that correct?

I don't know what is better, but I like MOSFETs for stuff I build. First of all, cost is really never an issue for me, so even if a MOSFET I need for a project costs me $1, that is fine for me. I pick components that make my life easier or that I want to play with for some other reason.

I find MOSFETs easier to deal with because I don't need to worry about any biasing, base currents or things like that. With a MOSFET, I just make sure Vgs(th) is in the range I need it to be and that it can handle the load and off I go.

Is this the best approach? Probably not. Is it a simple and working approach for a hobbyist? Sure...

[wraper]

Not sure what you are trying to say there....

If you have a "few hundred volts" across (D S or C E) your mosfet or BJT in it's on state, I'd say you have a pretty serious problem wouldn't you?

When you go for BJTs rated for high voltage, CE voltage drop on them in fully open state still remains about the same as for low voltage parts. On the other hand, RDSon of MOSFETS rises dramatically when you go for high voltage parts. Therefore their advantage diminishes. MOSFETS are still more efficient to use in SMPS powered from 230V. Though max rated voltage is limited compared to BJT. However when you need higher voltage, there are IGBTs. So BJTs are basically dead for switched power electronics. Still BJTs have some place for linear.

[Zero999]

It's true that MOSFETs have lower on losses, than BJTs, at low voltages, but at high voltages, above a few hundred volts, the reverse is true: BJTs start to have lower on losses, than MOSFETs

Not sure what you are trying to say there....

If you have a "few hundred volts" across (D S or C E) your mosfet or BJT in it's on state, I'd say you have a pretty serious problem wouldn't you?

Well I would have thought that was obvious, but forgot that this is the beginners section!

Is it not that the (ideal) losses in a MOSFET, in it's fully static on state are described by it's RDSon, and the current passing through it...

  W = I * I * R

the voltage across the mosfet in it's on state therefore is a product of the current, unless something is really wonky. 

And in the saturated BJT the voltage across is also somewhat current dependant.

The voltage the rest of your system runs at is of no more interest to a MOSFET or BJT in the on state than it is to a resistor, or a switch.

At high currents a BJT may have an edge over a MOSFET (but this all comes down to picking good parts, you could probably just pick a better MOSFET, or you could probably pick a better BJT, the BJT might be cheaper, the MOSFET might be smaller....), at high voltages, I think it doesn't make sense to talk about losses since we can only assume the devices are off and any loss is down to leaking where it shouldn't be leaking and you picked the wrong thing.

Of course, this is all assuming low frequency (100% duty) gate/base driving.

That's true but doesn't change the fact that my previous statement about BJTs being more optimal, than MOSFETs in high voltage applications, is true.

Of course, what I neglected to say was that BJTs with high voltage ratings, have lower on losses, compared to MOSFETs, everything else being equal.

but at high voltages, above a few hundred volts, the reverse is true: BJTs start to have lower on losses, than MOSFETs, which is why BJTs are used in fluorescent tube ballasts

It's because of lower cost, not efficiency. With 2 cheap BJTs you can make a simple Royer oscillator with minimum count of components required and no driver IC. If you go for highest efficiency, you still need to use MOSFETs. For example, no half decent computer psu uses BJTs, only low power bottom of the barrel crap.

You're right about the Royer oscillator, although they can easily be made with MOSFETs too. BJTs are chosen, not only due to efficiency but because they don't have a gate sensitive to high voltages, which would need protection.

400V is about the crossover point. The main reason for choosing MOSFETs, over BJTs in computer switched mode power supplies, is they're easier to drive, as they don't need a high current and a faster switching, hence suited to lower power levels.

Not sure what you are trying to say there....

If you have a "few hundred volts" across (D S or C E) your mosfet or BJT in it's on state, I'd say you have a pretty serious problem wouldn't you?

When you go for BJTs rated for high voltage, CE voltage drop on them in fully open state still remains about the same as for low voltage parts. On the other hand, RDSon of MOSFETS rises dramatically when you go for high voltage parts. Therefore their advantage diminishes. MOSFETS are still more efficient to use in SMPS powered from 230V. Though max rated voltage is limited compared to BJT. However when you need higher voltage, there are IGBTs. So BJTs are basically dead for switched power electronics. Still BJTs have some place for linear.

Yes, that's largely true and I must point out that an IGBT, is a type of BJT!

[wraper]

Yes, that's largely true and I must point out that an IGBT, is a type of BJT!

If you look at construction, they are more like a type of MOSFET.

Another way to compare IGBTs and MOSFETs is to say that an IGBT is, essentially, an N-channel power MOSFET on top of a p+ type substrate. 

[james_s]

That's interesting, thanks. I've used IGBTs on a few occasions but I'm not nearly as knowledgeable about them as I am with mosfets and BJTs.

Yeah it's certainly possible to build a Royer oscillator using mosfets, I have a number of flyback transformer drivers I built with mosfets.

[wraper]

That's interesting, thanks. I've used IGBTs on a few occasions but I'm not nearly as knowledgeable about them as I am with mosfets and BJTs.

Yeah it's certainly possible to build a Royer oscillator using mosfets, I have a number of flyback transformer drivers I built with mosfets.

Yes you can, but you will spend more on MOSFETs and will need more additional components. CFL ballasts usually are built with cheap clones of MJE13001-MJE13005, even branded stuff.

[David Hess]

Ok, so there is a lot of information here but from a couple of the posts I get the impression that generally speaking: MOSFET is better for these hobby applications. Is that correct?

I would not really agree.  The low base-emitter forward voltage drop of a bipolar transistor makes for simplified and predictable biasing; there is no worry about trying to drive standard MOSFETs on a 5 volt or lower supply voltage which will require low threshold voltage parts.  This literally goes double with a bipolar transistor h-bridge since diagonal transistors can be driven with the same signal at any supply voltage if a brake function is not required as shown below.  (1) If a brake function is required, then a third signal will be needed which is still simpler than four separate inputs.

Due to their minority carrier mechanism, bipolar transistors operate at higher current densities and especially so at higher voltages making them less expensive than MOSFETs.  (2) IGBTs essentially add this capability to MOSFETs combining high current density and high voltage operation.

At low powers though, losses will be irrelevant with either device and cost is unlikely to be a factor.  Use whatever you are most comfortable with.

(1) I prefer to use Darlington optocouplers in place of the drive transistors shown to keep high currents away from the controller but with some care about the return currents through the shared ground, this is not strictly necessary.  The same problem exists with MOSFET h-bridges.

(2) I don't think low voltage MOSFETs have caught up in this respect but economy of scale in production has made them practically as economical in low voltage designs.

[Youkai]

Hmm. Well then I guess I'll wait until I have a better idea of what all is going into the final product and work up a circuit diagram. Then I can see what you all think will work best based on the circuit.

[IanB]

I'm trying to understand the differences between a transistor and MOSFET and why you would choose one over the other.

Since this thread title bothers me every time I read it, let me just expand it a little:

I'm trying to understand the differences between a bipolar junction transistor and a (metal-oxide-silicon) field-effect transistor and why you would choose one over the other.

A MOSFET is a transistor, just a different kind of one.

[ez24]

 

[mrkev]

Ok, so there is a lot of information here but from a couple of the posts I get the impression that generally speaking: MOSFET is better for these hobby applications. Is that correct?

Both BJT and MOSFET are frequently used, because they both have advantages and disadvantages in different applications.

• MOSFET - Generaly, low R_DSON means less of a voltage drop when current is running throught. That makes them ideal for switching. The have high inpuit impedance, which means very low current into the gate, that is why (certain types, especially low noise J-FETs) are used for input in amplifiers. Disadvantage is that you need some voltage on gate to switch them properly (look up V_Gth and "logic level" FETs) and in some cases, that can be a problem (if you need something to work from 1V, like something that is powered from a single AA cell). For some MOSFETS, you need more than 5V to switch them on, that is true especially with power MOSFETS and transistors that have higher D-S breakdown voltage. Also, the gate treshold can very much vary from piece to piece (even within the same type), for example in datasheet of commonly used low-power 2N7002 is that V_Gth can range from 1V to 2,5V.
• BJT - You need only about 0,6V on base to turn them on (of course with propper current) and because this value doesn't vary (it does with temperature, but not that much), you can somethimes use them as rough and simple comparator. Collector current is (mostly) proportional to the base current (h_21E, h_FE or ?) and, in some cases, you can achieve higher amplification, there are even BJT darlingtons that have very high amplification. BJTs that are capable of driving higher voltages (with low current) are fairly cheap (f.e. almost everything that starts with MJE).