Can MOSFET be Used as Pass Element in Linear Regulator?

[Mechatrommer]

i usually see pass element using power bjt, but... which one will make up a better linear regulator? setup like in the picture. mosfet or bjt? whats the pro and con? thanks.

[Mechatrommer]

my question is also applicable to this setup, from LM317 datasheet. can i change bjt power elements (LM195's) with suitable mosfets? and change 2N2905 with a suitable gate driver?

[danadak]

This might be of assistance -

http://www.ti.com/lit/an/sboa046/sboa046.pdf

Regards, Dana.

[pelule]

In your application the MOSFET is used in linear mode.
Even a mosfet is designed for switching it can be used also in linear mode (as in many electronic loads).
But you have to ensure, the part is allways operaring in its Safer Operating Area (SOA).
This SOA will at the end define the part you select.
PeLuLe

[Gyro]

my question is also applicable to this setup, from LM317 datasheet. can i change bjt power elements (LM195's) with suitable mosfets? and change 2N2905 with a suitable gate driver?

Just take note that this particular circuit is using the current/overtemp limit feature of the LM195 ICs to provide short circuit protection - they are not bog standard bipolar transistors.

[Zero999]

Well, an N-MOSFET on the high side will definitely have a higher drop-out voltage than a Darlington pair, unless a charge pump is used to drive the gate at a higher voltage than the output.

[Mechatrommer]

the reason i ask is because i have a suspicion mosfet is more difficult to drive (driver side) in linear mode like in the setup i showed. and it produces more noise on the regulated output. got more heated up compared to bjt working at the same power dissipation, easily damaged etc... no?

Well, an N-MOSFET on the high side will definitely have a higher drop-out voltage than a Darlington pair, unless a charge pump is used to drive the gate at a higher voltage than the output.

in bjt, its a emitter follower setup, so with the nmosfet, it will be source follower. the linear output of the power nmosfet will be set at much lower than its input, around 5-10Vds, making charge pump unnecessary, yes?

[MosherIV]

i usually see pass element using power bjt, but... which one will make up a better linear regulator? setup like in the picture. mosfet or bjt? whats the pro and con? thanks.

As others have pointed out, both are in linear mode of operation.
I like to think of them as powerful variable resistors in this scenario. It does not matter which is used, they both have to dissipate the same amojnt of power.

Considering FETs are trickier to bias/configure compared to the BJT for the same operation, it is therefore easier to use BJTs.

When used in the first schematic, a design trick you can do with BJT is to put a switch in line with the base of the power transistor, the switch can turn off the power transistor by cutting off the base current. This cannot be easily done with FET because disconnecting the gate means it is free to float and therefore continue in linear operation (but now uncontrolled), the FET gate must be pulled to 0V to turn it off.

[danadak]

In the boosted regulator example the MOSFET has much higher Vth than
the Vbeon of a bipolar, so the series R to develop turnon has to be much
higher, thereby wasting power.


Regards, Dana.

[David Hess]

MOSFETs work fine in linear applications but have some disadvantages:

1. At high voltages MOSFETS have lower current density than bipolars so require a larger die making them more expensive.
2. Vgs is higher than Vbe and has a lot more variation.
3. MOSFETs optimized for switching applications may lack specifications for linear operation.

[Whales]

I tried doing this a few weeks ago with an IRL40B212

Notes:

• I just tied the op-amps output directly to the mosfet
• Even with the gate shorted to zero, the fet leaked too much current in its fully off state.  Output was around 3-5V (of possible ~12V) even with a 1K load.

[Mechatrommer]

thanks all guys, it seems concluded that, with proper control, mosfet can be made as pass element in linear mode regulator just as good as bjt.. i guess, from understanding the replies here... as i'm currently designing my linear regulator psu, i bought TIP142 darlington bjt. its rated 10A, but big TO-247 package. and i also have IRF540 nmosfet rated 28A, at normal (smaller) TO-220 package, so from amperage per surface or footprint area (and i believe cost saving), mosfet is more advantageous, i need 2 of TIP142 for 20A rating which took more space on the PCB. but as my linear psu design already get involved using the bjt, i will continue with bjt, maybe on revision 2 next time, i will consider using mosfet. thanks all.

(ps: to be more accurate, my current psu design is following exactly like the LM317 setup above, with paralleled bjt's using 2x TIP142. yes its a normal bjt with no over temp protection, but i think i will take care that in the design such as limiting the power (current) consumption, or maybe temp sensor on the heatsink, fwiw)

[Mechatrommer]

I tried doing this a few weeks ago with an IRL40B212
Notes:

• I just tied the op-amps output directly to the mosfet
• Even with the gate shorted to zero, the fet leaked too much current in its fully off state.  Output was around 3-5V (of possible ~12V) even with a 1K load.

thats a 250A mosfet! maybe the excessive leakage is something common with this kind of very high amperage transistor?

[langwadt]

thanks all guys, it seems concluded that, with proper control, mosfet can be made as pass element in linear mode regulator just as good as bjt.. i guess, from understanding the replies here... as i'm currently designing my linear regulator psu, i bought TIP142 darlington bjt. its rated 10A, but big TO-247 package. and i also have IRF540 nmosfet rated 28A, at normal (smaller) TO-220 package, so from amperage per surface or footprint area (and i believe cost saving), mosfet is more advantageous, i need 2 of TIP142 for 20A rating which took more space on the PCB. but as my linear psu design already get involved using the bjt, i will continue with bjt, maybe on revision 2 next time, i will consider using mosfet. thanks all.

for linear operation the current rating isn't very relevant, the limit is temperature rise from power dissipation

for getting the heat into a heatsink TO-247 is better than a to-220

[VEGETA]

thanks all guys, it seems concluded that, with proper control, mosfet can be made as pass element in linear mode regulator just as good as bjt.. i guess, from understanding the replies here... as i'm currently designing my linear regulator psu, i bought TIP142 darlington bjt. its rated 10A, but big TO-247 package. and i also have IRF540 nmosfet rated 28A, at normal (smaller) TO-220 package, so from amperage per surface or footprint area (and i believe cost saving), mosfet is more advantageous, i need 2 of TIP142 for 20A rating which took more space on the PCB. but as my linear psu design already get involved using the bjt, i will continue with bjt, maybe on revision 2 next time, i will consider using mosfet. thanks all.

(ps: to be more accurate, my current psu design is following exactly like the LM317 setup above, with paralleled bjt's using 2x TIP142. yes its a normal bjt with no over temp protection, but i think i will take care that in the design such as limiting the power (current) consumption, or maybe temp sensor on the heatsink, fwiw)

Can you post a simple schematic?

I want to design my own PSU but I want it to achieve true 0v output operation, is that possible using your design? Plus, what is the dropout of this design? If it is too small then you won't need any heatsink.