Re: Choosing BJT or MOSFET

[CrossRoads]

At low signal levels, they are pretty interchangeble, NPN & Logic-Level N-channel, and PNP and Logic-level P-channel.
At high power switching levels, MOSFETs are better than BJTs because their lower on-state resistance means that less power is dissipated in the part.

Compare:
12V circuit, with 500mA being switched on/off.
BJT will have ~0.7V drop from Collector to Emitter, rest of circuit has 11.3V drop across it.
MOSFET will have say 50mOhm on-resistance. With 500mA, V=IR so voltage drop is 0.025V. More voltage available for your load - say a motor, or strings of LEDs.

Power dissipated by transistor is P=IV, P=(I^2) * R, P = (V^2)/R. What values do you have:

BJT: P = IV = 0.5A * 0.7V = 350mW
MOSFET: P = I^2*R = 0.5A * 0.5A * .05ohm = 0.0125W

BJTs are current controlled devices - MOSFETs are voltage controlled devices.
May need 50+mA to get a BJT turned on fully for low voltage drop across CE.
MOSFET needs very little. Low Vce parts are obtainable as well for more money.

[fungus]

It's really simple: Use a BJT unless you have to use a MOSFET

(ie. because the power handling or the Vce of the BJT isn't suitable).

[68tjs]

fungus:
It's really simple: Use a BJT unless you have to use a MOSFET

+1
One speeks often of the Mosfet Vdson which is better than bjt Vcesat .
One forget often that only 0,7 V are suffisant for saturate a bjt, but 5V are needed to be sure to saturate a "numerical" Mosfet.
Power Mosfet needs often Vgs >= 8 V. to be able to drive 5 ou 10 Amperes.
Sometime it is necessary to interface electronic and Mosfet with à bjt.
With 3.3V electronic it is more easy to use bjt than Mosfet..

Power Mosfet presents a Vgs parasistic capacitor around 1nF to 4 nF. .To charge or discharge this capacitor electronis I/O have to drive a current..With power Mosfet a resistor is recommanded to protect electronic I/O.

Also don't forget that Vgs threshold means the begining of conduction, few µA for small mosfet or few mA for power Mosfet..

[MarkT]

CrossRoads:
Compare:
12V circuit, with 500mA being switched on/off.
BJT will have ~0.7V drop from Collector to Emitter, rest of circuit has 11.3V drop across it.

BJT saturation voltages can be as low as 50mV, 0.7V is Vbe, not Vce. Admittedly
the wrong BJT might have a Vce of 0.7V at 0.5A, but you can choose a better one.

Check out ZTX450 datasheets for a top-class modern BJT (Vce = 0.1V at 2A collector
current). Mostly modern BJTs (superbeta) are surface mount though...

[retrolefty]

Pretty simple for me. If I have a BJT on hand that will do the job (proper voltage and current rating for the application) I would use it. If I don't have one on hand I would buy a logic level mosfet that fits the application.

There is no real price difference these day between the types for similar ratings with careful shopping. For high current applications the very low Ron resistance (<10mv) available in modern mosfets is a big advantage over BJT, even if the mosfet requires a gate driver to reach it's best ratings.

Lefty

[CrossRoads]

Look at specs & prices for NPN transistors:

This is what Digikey lists for ZTX450:
250mV @ 15mA, 150mA, 67 cents

ZTX450 | DigiKey Electronics

more than 2 times the price of a PN100A
400mV @ 20mA, 200mA device, 31 cents

PN100A | DigiKey Electronics

while a 5mOhm N-channel, logic level MOSFET, AOI516, will blow both away for current rating is just 55 cents, and will have Vds of (.150 * .005) = 0.75mV

AOI516 | DigiKey Electronics

So: there's all kinds of parts, all kinds of prices, do some research and evaluate what is best for your specific application.

[68tjs]

Vce sat (bjt) and Rdson (Mosfet) are not linear with the current.

There is two part in Vcesat : non linear part and linear part (resistor inside the silicium).
So, with bjt, the worst case is given for max current where the effect of inside resistor is maximum.
At lower current Vcesat can be very low.

Rdson :
The best case is given for max current. Be careful a mosfet given for 50milliohms at 20 A, but used at a too small current, can presents a Rdson > 1 ohm.

Use the right type of transistor at the right place and read the datasheets., particularly the conditions of measurements.

[pwillard]

So let's put this out there. Using a MOSFET does not make you more "cool" but it can make your motor driver run cooler.

A good designer uses the right tool or part for the job at hand.

[system]

For a long time, I didn't understand why someone would use a BJT unless they needed drive capability at low voltages. But, as I get used to BJTs I'm starting to see their advantages. Each has strengths and quirks.

BJTs conduct in more ways than MOSFETs do... you can conduct either way between the base and collector on an NPN for example, but the gate and drain of an N-channel MOSFET seem to be more isolated. (Someone correct me if I'm wrong here.) So you might need to watch out for current flowing in unintentional ways when, for example, a pin isn't being driven, or in a fault scenario. OTOH, this behavior can also be useful.

Then there's cost... When searching for low-current switches (like <100mA to PWM an LED), Digikey seems to have lots of TO-92 NPNs for less than $0.50 in single quantities, whereas N-ch MOSFETS are a bit more expensive -- in many cases, closer to $1.00.

The circuit might also be simpler. Check out the first half of each set in the attached image. For a MOSFET, you should usually have a pull-up or pull-down to set a reference voltage when the drive pin is high-Z. With BJT, a high-Z pin should not conduct current, so it should remain off. (More experienced guys -- any qualms with this?) However, most people advocate using a series resistor to limit current draw from gate capacitance with MOSFETS. So you're using two drive resistors instead of one.

Now, referring to the second half of each set, if you're using an NPN-PNP pair (to switch a higher voltage on the positive end with logic HIGH = ON, for example), you may be able to economize further by limiting the current that can flow through the PNP's base by limiting the drive through the NPN. With a 150K resistor on the NPN base, and a gain of 100, it's at the sweet spot for the PNP with no need for a current-limiting resistor on the PNP. Although the success of this approach depends on the actual gain of the BJT, which is sometimes not quite what is specified in the datasheet. You would have to be wary of parts tolerance, or go the safe route and just add one more resistor. The latter is probably a better approach most of the time, but if parts count is the priority, you have the option...

The biggest difference of course is that BJTs are current-controlled, and MOSFETS are voltage controlled. There are applications that make one more suitable (or convenient) than the other, and if you're comfortable with both, you can choose the one that is the best fit instead of shoe-horning the part you have and understand.

[system]

Vce(sat): of a 2c 2n3904 is only 0.2V. With it you could switch 12V @200mA = 2.4W.
http://www.taydaelectronics.com/t-transistors/2n-series/2n3904-npn-general-propose-transistor.html

For more power and still a lot cheaper than a MOSFET, try the 2n5551:
http://www.taydaelectronics.com/t-transistors/2n-series/2n5551-transistor-npn-160v-0-6a.html

[fungus]

SirNickity:
For a long time, I didn't understand why someone would use a BJT unless they needed drive capability at low voltages. But, as I get used to BJTs I'm starting to see their advantages. Each has strengths and quirks.

Cost, physical size, switching speed, one less resistor... I'm sure there's more reasons but they're enough for me.