Transistor vs FET

[alank2]

I've been using a base drive resistor and npn transistor for a long time in designs, but I'm transitioning to SMD and trying to reduce part count and simplify, and I am wondering if I should consider looking at a logic level fet instead.

I'm driving a LED backlight at 175mA, a LED at 20mA, and a piezo at 20mA...

One reason I've stayed away from FET's is that I read their gate is more static sensitive and I figured they would be easier to damage and hence that a transistor might be more durable.

How do you guys decide?

[c4757p]

Transistor vs FET

Field Effect Transistor.........

Yes, the gate is more sensitive. Also, when I use FETs I usually don't save any parts: a BJT gets a base resistor, and a FET gets a resistor from gate to source to keep it turned off if its driver tristates (okay, that's a MCU problem only, in general). The gate is a capacitor, and will hold charge if there's nothing to discharge it.

Lately, I've been using these a lot. It's a BJT with resistors integrated: a base bias resistor, and a pulldown resistor from base to emitter. Just one part, and quite robust and easy to use. Of course, they cost more too.

For higher currents, however (I'd consider this at 175mA), the MOSFET will be more efficient, if you get one with nice, low resistance. BJT V_CE(sat) doesn't go all that low, typically about 0.2V for moderate current, whereas MOSFETs with R_DS(on) well below an ohm are easy to find.

[BravoV]

Is your bipolar transistor driving the led in linear mode (eg. current regulation) or in fully saturation mode (as a switch for on/off the led) ?

If the transistor plays role in current regulation loop (aka linear mode), imo you must be careful when you plan to switch to mosfet especially the "tiny" logic level one, as they're not designed to dissipate too much heat. Only good for switching either on/off state.

CMIIW 

[David Hess]

For saving parts where space is at a premium I might do this but in general bipolar transistors are less expensive than MOSFETs and are more flexible if you want to run them in common base mode because of their low Vbe which is handy for constant current level shifters.

[Jeroen3]

For switching purposes you'd be better of with a N-channel logic level mosfet.
- Negligible gate current when active.
- Has mOhm on-resistance.
- Doesn't increase bom cost.

However, most mosfets tend to self-destruct when not they're not opened fast enough.

[mikerj]

However, most mosfets tend to self-destruct when not they're not opened fast enough.

This is not a MOSFET problem, it's simply an application issue.  If you are switching high current loads then you don't want your switching element to be spending any significant time in a linear operating region, it needs to be fully on or fully off.  A BJT is just as likely to fail if the power dissipation is greatly exceeded due to slow switching.

[jucole]

Lately, I've been using these a lot. It's a BJT with resistors integrated: a base bias resistor, and a pulldown resistor from base to emitter. Just one part, and quite robust and easy to use. Of course, they cost more too.

I didn't know those existed - thanks c4757p! ;-)

[Jeroen3]

This is not a MOSFET problem, it's simply an application issue.

Yes, a thing to note when using with microcontrollers. Enable the output buffers.

[alank2]

What I _really_ don't want is to have to repair something later because of a design issue.  Are BJT's more robust or tough for this type of task, or not really?

I will be controlling them from an Atmel uC output pin (in output mode, usually 20mA max or 40mA max) though I prefer to not run them over 5mA if given a chance.

[David Hess]

What I _really_ don't want is to have to repair something later because of a design issue.  Are BJT's more robust or tough for this type of task, or not really?

I would not say that they are more robust.  The MOSFETs include a free drain to source freewheeling diode which protects against reverse voltages.  Some bipolar transistors, especially darlingtons, have this added.  Both types of transistors may be rated for avalanche switching inductive loads.

I think the largest practical difference are their drive requirements at higher power levels and the ease of paralleling MOSFETs in switching applications.

In some applications like synchronous rectification and wired-or power switching, you can operate the MOSFET backwards to advantage.

[mikerj]

I just posted a reply to this thread and it's disappeared, very odd.

Anyway, either a BJT or a MOSFET will be adequate for driving such a light load from a port pin, though a MOSFET will give you a lower voltage drop and consume no drive current from the pin under steady state.

One thing you need to remember is that under reset the IO pins of most (all?) micros are tri-stated, so you must provide a suitable pull-up or pull-down to drive the MOSFET into your desired default state.  Failure to do this could easily see the transistor being partially switched on and power dissipation will greatly increase.

[alank2]

I think the largest practical difference are their drive requirements at higher power levels and the ease of paralleling MOSFETs in switching applications.

I would probably PWM the backlight on the LCD display (LED backlight) and its maximum current is 175mA.  I recall something about mosfets heating during the transition so you need to use that to calculate the heat...

[David Hess]

I think the largest practical difference are their drive requirements at higher power levels and the ease of paralleling MOSFETs in switching applications.

I would probably PWM the backlight on the LCD display (LED backlight) and its maximum current is 175mA.  I recall something about mosfets heating during the transition so you need to use that to calculate the heat...

Power dissipation during switching applies to any transistor.  In low power applications it can usually be ignored.

Depending on the input and output voltages, a linear regulator can be competitive with a switching regulator at low currents like 175 milliamps and it avoids any EMI issues.

[Jeroen3]

I recall something about mosfets heating during the transition/quote]
Refer to any dummy load topic on this forum for details about that.

[wagon]

Lately, I've been using these a lot. It's a BJT with resistors integrated: a base bias resistor, and a pulldown resistor from base to emitter. Just one part, and quite robust and easy to use. Of course, they cost more too.

I didn't know those existed - thanks c4757p! ;-)

Also called 'digital transistors'. 

[Sigmoid]

So you're saying MOSFETs aren't significantly less robust than BJTs? For some reason I always thought they were extremely prone to death by static discharge.

[David Hess]

So you're saying MOSFETs aren't significantly less robust than BJTs? For some reason I always thought they were extremely prone to death by static discharge.

When I think of ruggedness I think of forward biased safe operating area and avalanche capability but power MOSFETs have quite a bit of gate capacitance which helps protect them.  Small signal bipolar junction transistors can easily be damaged by static discharge which reverse biases their base-emitter junction.

[dannyf]

I have worked with mosfets since the 1980s and have yet to damage one from static.

[Phaedrus]

So you're saying MOSFETs aren't significantly less robust than BJTs? For some reason I always thought they were extremely prone to death by static discharge.

CMOS circuits (mosfet based) are much more susceptible to ESD damage than older TTL circuits. That's probably where you heard it. But power MOSFETs are pretty robust. In off-line SMPS, in my experience, BJTs tend to blow up slightly more often, though that may be because most of the BJTs used in my industry are in the PFC stage, which is often less robust than the PWM or rectifier stages.