BJT and MOSFET- Switching and Drive

[dknguyen]

I'm a bit confused about something about BJTs and MOSFETs. My understanding was that MOSFETs tend to be better for power switches since they have a positive temperature-resistance coefficient, can switch at higher frequencies, lower power drive circuitry, and a resistive conduction characteristic that is low enough such that at most current levels the losses are less than a BJT's diode conduction characteristic (the voltage drop across the FET S-D resistance is less than the BJT's C-E diode voltage drop for the same current).

But here is where I get confused. Now, the BJT doesn't have the base capacitance like a MOSFET has a gate capacitance, and yet it switches slower? BUt then I read about BJT's are often used to drive FET gates because they can supply more current and are can turn on faster- except, I thought the MOSFET could supply more current due to its low S-D resistance compaed to the BJT's C-E diode voltage drop.

I'm a bit confused.

EDIT: From what I have gathered since I posted this, it seems some of the differences involve the output impedances, but I don't know much more than that.

 

[Hero999]

It thought it was because BJTs have a higher output capacitance.

Anway BJTs seem to be better for lower currents, MOSFETs seem to be better for power.

 

[ronsimpson]

MOSFET vs BJT

I am sure you know MOSFETs take voltage to turn on while BJT use current to turn on. A MOSFET, while on, looks much like a low value resistor while the BJT looks more like a switch.

DC power loss in BJT and MOSFET. To compare I will use MPS2222 and a BS170.
The BS170 MOSFET has a D-S resistance of about 1.25 ohms when “closed”, Vgs=7 volts. Here is the ON curves for the MPS2222 transistor. The red dots are for the MOSFET.
The MOSFET with 1.25 ohms on resistance will have
0.625 volts loss at 500mA
0.187 volts loss at 150mA
0.0125volts loss at 10mA
The BJT will have 0.2 volts loss at 500mA with a base current of 50mA. The base current will have to be under 12mA for the BJT to have more DC loss than the MOSFET.

Now to the question of input capacitance.
The MPS2222 has an input capacitance of 30pf. (base to emitter) There is also a base collector capacitance of 8 pF.
The BS170 FET has an input capacitance of 24 pF and an output capacitance of 17 pF and an input to output capacitance of 7 pF.
In many cases the BJT B-E capacitance has little effect because we are charging the input capacitance from 0 to 0.6 volts from a high current source. The MOSFET’s input capacitance must be charged from 0 to 10 volts from a voltage source that has limited current. In the case of a big FET the input capacitance is 2000pF and we need 15 volts on the gate.

One of the differences between MOSFETs and BJTs is turn off delay (speed). If a BJT is kept from turning on hard it remains fast. If you keep several volts across the C-E then the part is fast. If the C-E voltage is small then the transistor gets sticky (slow to open up). In the case of RF and audio amplifiers, care is given to never allowing the C-E voltage to get very small. With the MPS2222 if the base current is high and the C-E voltage is low (0.2 volts) the transistor will take time to open up. A MOSFET does not have this effect. If you need more information, look for information on “storage delay”.

 

[Nigel Goodwin]

You need to choose your device for a number of reasons, reliablity, cost, performance - in domestic electronics MOSFET's are pretty uncommon - probably due to cost reasons?, but certainly where they are used they are no more reliable than BJT, and could well be less so?.

 

[Hero999]

ronsimpson,
Now try comparing a power transistor with a MOSFET, you'll find the latter is much better. Also you haven't looked at the losses in the base current which will be 10% of the power if it's used as a common emitter switch and the switching voltage on the base resistor is the same as the load voltage.


Nigel Goodwin,
Last time I checked MOSFETs were the same price as BJTs.

 

[dknguyen]

I see the part about the BJT having to charge to a lower voltage. I also get now that although the BJT C-E diode voltage drop is higher than a MOSFET's D-S resistance voltage drop for a given current, the BJT is still a diode while the MOSFET is still a resistance, so the BJT can charge up a load capacitance to a higher level much faster since it's like charging through a diode (but the maximum charging voltage will be less than what a MOSFET can get). The MOSFET on the other hand is like charging through a resistor so that the charging voltage can achieve full supply voltage, except it levels out pretty quickly and takes longer to get there.

I was always aware of quasi-saturation, but we never looked into it anymore in class. I didn't know it only applied to BJTs (and IGBTs) and not FETs.

 

[ronsimpson]

Not too many years ago high voltage high current MOSFETs were not an option.
I often drive the base of a BIG transistor with a transformer to reduce power loss. A 1200 volt 20 amp transistor will need 1 amp at .8 volts on the gate so that’s 0.8 watts. On the primary side of the gate drive transformer is 12 to 24 volts at very little current. In this example I am switching 1000 volts at 10 amps and the C-E voltage is very important to me. (color TV set)

Now-days most people use FETs in power supplies. A gate takes 15 volts at 0 current to keep on. Driving 2000pF input cap + the G-S cap at 200khz will take energy. There will have to be a turn on and a turn off every 200khz. Normally I drive the gate with at least 1 amp for the transition time. It takes less power to drive a MOSFET but not much less at high speed.

In a 1000 watt supply I will not fight over 1 watt.

I suggest using MOSFETs for most power applications because of simplify and many of the BJT are no longer available.

In small signal application I will go either way. Today I am building a power supply with an input voltage of 1.8 to 3 volts. At 1.8 volts MOSFETs do not turn on so BJT is the way to go.

 

[Nigel Goodwin]

Last time I checked MOSFETs were the same price as BJTs.

Not as spare parts they aren't!

Usual places you see them are in SMPSU's, although BJT's are far more common - and if you obtain them as spares, or from distributors, the required MOSFETS are always considerably more expensive.

 

[Hero999]

I don't think I've seen a single BJT power transistor in an SMPS, even the older SMPSes seem to use MOSFETs.

 

[Nigel Goodwin]

I don't think I've seen a single BJT power transistor in an SMPS, even the older SMPSes seem to use MOSFETs.

Almost every TV, VCR and satellite receiver uses a BJT, it's rare to see a MOSFET, and in the few places you do they cost a lot more, and seem less reliable.

For that matter, don't PC supplies normally use BJT's as well? - thought so, here's a circuit:.

http://www.pavouk.org/hw/en_atxps.html

 

[Roff]

I see the part about the BJT having to charge to a lower voltage. I also get now that although the BJT C-E diode voltage drop is higher than a MOSFET's D-S resistance voltage drop for a given current, the BJT is still a diode while the MOSFET is still a resistance, so the BJT can charge up a load capacitance to a higher level much faster since it's like charging through a diode (but the maximum charging voltage will be less than what a MOSFET can get). The MOSFET on the other hand is like charging through a resistor so that the charging voltage can achieve full supply voltage, except it levels out pretty quickly and takes longer to get there.

I was always aware of quasi-saturation, but we never looked into it anymore in class. I didn't know it only applied to BJTs (and IGBTs) and not FETs.

The BJT is not a diode from collector to emitter. It looks much like a resistor, but with a little (typically < 100mV) offset added.
The "stickiness", or turn-off delay, of BJTs is due to the fact that, in order to get them into full saturation (low Vce), you need to provide the base with a lot of current (typically 10% to 20% of the collector current). This includes a safety factor, because you don't know what the beta actually is. This results in excess charge being stored in the base, which has to be removed before the transistor will turn off. This can be done by providing a high-current discharge path for the base, such as a smaller NPN switch between base and ground, or a capacitor across the base-current limiting resistor. MOSFETs are much simpler, in concept. You just need to drive the gate with a low impedance voltage source (easier said than done, though).

 

[Hero999]

All PC SMPSes use MOSFETs and I've opened up plenty of monitors that use them.

Older TVs used BJTs and VCRs are no longer made but I'd bet that the new flat screen and plasma TVs use MOSFETs.

 

[Nigel Goodwin]

All PC SMPSes use MOSFETs and I've opened up plenty of monitors that use them.

So you didn't bother looking at the generic ATX PSU circuit I posted?, which (in common with most) uses BJT's.

Older TVs used BJTs and VCRs are no longer made but I'd bet that the new flat screen and plasma TVs use MOSFETs.

You would lose your bet!.

 

[Nima.]

the original question first asked,hasn't been answered yet.why do we use BJT for MOSFET gate drive circuits?
and do you have any website for building mosfet gate drive?