When do I want to use a transistor vs a MOSFET?

Thread Starter

santoshreddy21

Joined Feb 15, 2020
1
If I want something to turn on or off I would use a MOSFET. Why would I ever want to use a transistor when I can use a MOSFET? Sorry, this is a really general question. I'm just trying to get the lay of the land.
Thanks.
 

peterdeco

Joined Oct 8, 2019
231
Personally I prefer MOSFETs. No base limiting resistor necessary. My MOSFET of choice is IRLZ44N. Although it's a power type that can switch up to about 50 amps, we never run them higher than about 1 amp. At this level no heatsink is necessary and only 3V on the gate can switch it on enough for our requirements. We buy them by the thousands and cost us only about 35 cents each.
 

dl324

Joined Mar 30, 2015
11,577
Welcome to AAC!
Why would I ever want to use a transistor when I can use a MOSFET? Sorry, this is a really general question.
It depends on a number of things.
  1. How much current you need to handle.
  2. What is driving the switch. If you have a microcontroller operating at 3.3V, you have to use a logic level MOSFET (low \(V_{GS(th)}\)). With BJT's you need to consider available drive current. If you use a darlington configuration, you increase current gain, but you also increase the saturation voltage by a diode drop unless you can split the collectors (not possible on integrated device).
  3. What you have on hand.
  4. Cost. Bipolar transistors to switch less than a quarter amp cost a few cents. N channel MOSFETs will be several times that, P channel even more (10-20X).
  5. Form factor. Inexpensive logic level MOSFETs only come in surface mount package.
  6. Etc.
EDIT: I assumed operation as a switch.
 
Last edited:

ElectricSpidey

Joined Dec 2, 2017
1,258
For me personally…

I use MOSFETs as switches IE: on/off, and I use BJTs when I need an amplifier or voltage follower…etc.

But I also use 907s and 222s as switches all the time, so nothing is written in stone.
 

MaxHeadRoom

Joined Jul 18, 2013
21,451
If I want something to turn on or off I would use a MOSFET. Why would I ever want to use a transistor when I can use a MOSFET? Sorry, this is a really general question. I'm just trying to get the lay of the land.
Thanks.
For logic level interface, the 2n7000 is always a nice replacement.
I almost never need to use bi-polar now.
Max.
 

Attachments

dl324

Joined Mar 30, 2015
11,577
@santoshreddy21 When using MOSFETs, you should be aware that threshold voltages for the same device type vary, possibly significantly.

The maximum threshold voltage for 2N7000 is 3V. That is the voltage that the MOSFET is just starting to turn on. When using as a switch, you'll usually want to operate at a higher gate-source voltage to reduce ON resistance.
1581785652893.png
1581785669252.png
1581785686304.png
You could get a device that turns on at 0.8V, or 3.0V, but most will be around 2.1V.

At Vgs=4.5V, ON resistance is 5.3 ohms maximum.


What I referred to as a logic level device switches at lower voltages:
1581785841093.png
1581785829156.png
1581785905759.png

At Vgs=4.5V, this device has an ON resistance of 33 milliohms (0.033 ohms) maximum.

These devices would be suitable for switching with a Raspberry Pi that has 3.3V I/O's. 2N7000 would be sufficient for Arduino's operating at 5V.
 

ebeowulf17

Joined Aug 12, 2014
3,274
All other factors being equal, I tend towards MOSFETs most of the time now in simple switching applications.

Two key factors that would make me use a BJT in certain situations (both covered in @danadak's linked articles as well) are:

1) Reverse voltage blocking:
Most MOSFETs, at least the ones I'm familiar with, have an intrinsic body diode, essentially an inverse parallel diode built into the device, which means that the MOSFET can act as an effective switch for current in one direction, but it can't control reverse current at all. Quite often, if you need that reverse blocking, it's easier to use A BJT than to jump through all the hoops to make the MOSFET work.

2) Drive voltage requirements:
Although it's not too hard to find MOSFETs that work well at 5V, or even 3.3V, for direct use with microcontrollers, sometimes you need to operate something with a lot less drive voltage. I can't think of a perfect example off hand, but sometimes you've created a circuit where your signal has already gone through some rectifiers, or an LED, and you need that signal to control a switch. After all those diode forward voltage drops, you may only have 1V left to work with. A BJT works fine here, as long as adequate current is available, while there's no MOSFET that does anything useful with 1V gate voltage.
 

dl324

Joined Mar 30, 2015
11,577
Most MOSFETs, at least the ones I'm familiar with, have an intrinsic body diode, essentially an inverse parallel diode built into the device,
All MOSFETs have two parasitic diodes. Unless you have a 4 terminal device (separate substrate connection) the one on the source is shorted in the case.

This image from RCA ICAN-6218 shows the parasitic diodes on a P and N channel MOSFET. They use a PWELL process, but it's similar for an NWELL process.
1581794833071.png
 
Last edited:

crutschow

Joined Mar 14, 2008
25,460
if you need that reverse blocking, it's easier to use A BJT than to jump through all the hoops to make the MOSFET work.
They only works if the reverse voltage is less than the reverse base-emitter breakdown voltage (typically about 5V).

If you want to use a MOSFET to block in both directions, you just put two in series with their sources and gates tied together.
That way there is always one substrate diode reverse biased when off, but they will conduct in either direction when turned on.
 
Last edited:

ebeowulf17

Joined Aug 12, 2014
3,274
All MOSFETs have two parasitic diodes. Unless you have a 4 terminal device (separate substrate connection) the one on the source is shorted in the case.

This image from RCA ICAN-6218 shows the parasitic diodes on a P and N channel MOSFET. They use a PWELL process, but it's similar for an NWELL process.
View attachment 199162
Cool! I had no idea there were two of them, so if course I had to Google it immediately and found the attached image. Thanks!article-2016september-the-significance-of-the-fig1.jpg
 

ebeowulf17

Joined Aug 12, 2014
3,274
They only works if the reverse voltage is less than the reverse-base emitter breakdown voltage (typically about 5V).
Ah, thanks for that clarification. I'm learning a lot today!

I guess I got away without knowing this so far because I'm *usually* either working at 5VDC or less, or at mains voltage, hardly ever anything in between. I'll be sure to keep an eye on that spec and design accordingly in the future!
 

dl324

Joined Mar 30, 2015
11,577
Cool! I had no idea there were two of them
You have a parasitic diode whenever P and N material touch:
1581795402982.png
Picture from Intel 1983 Memory Component Handbook.

This is for an Intel NWELL process. A tap in the NWELL is typically connected to the positive supply which shorts the source diode. It's usually a mistake when it isn't.

In a CMOS process, all of those junctions form a 4 layer diode which can cause the infamous latchup condition.
 
Top