Replacing bipolar transistors with FETs

[bcdonadio]

Hi!

I've built two power supplies based on this project. They're working wonderfully, but those 2N3055 are simply enormous and require a big heatsink. I've been thinking in replacing them with some MOSFETs, but I'm not sure exactly what should be changed in the circuit. Can someone give me some clues?

If possible, would be good to use some VNP10N that I have (got hundreds of those). They're omnifets, which (if I understood correctly) are MOSFETs with additional protection circuitry in the same die. Here's the datasheet of them: **broken link removed**.



Thank you!

 

[dick_freebird]

MOSFETs in a linear supply where they are dropping all
the voltage, will dissipate roughly the same heat. Only
they may not like it, since many are optimized for on
resistance and voltage blocking but don't have a square
SOA. Comes down to die power density and the thermal
path.

They are also liable to present a higher input capacitance
which might want some compensation tweaking, but present
less of a current load / higher DC impedance to the loop
amp. And threshold voltage of 5-10V to be well driven,
might eat up more headroom than you'd like between
predriver and load.

 

[goldsmith]

Hi!

I've built two power supplies based on this project. They're working wonderfully, but those 2N3055 are simply enormous and require a big heat dissipater. I've thinking in replacing them with some MOSFETs, but I'm not sure exactly what should I change in the circuit. Can someone give me some clues?

If possible, I would want to use some VNP10N that I have (I got hundrends of those). They're omnifets, which (if understood correctly) are MOSFETs with additional protection circuitry in the same die. Here's the datasheet of them: VNP10N06.pdf.

Hi bcdonadio
First of all let me confess that the circuit which you have attached above is one of my first PSUs which i built successfully in my first days of electronics and it is nice to see it again .:wink:

Anyway
If your problem is 2n3055 and it's legendary face , :wink: and much dissipation across it i afraid that is a linear PSU and you have no other choice . you can not run from dissipation in a linear regulator . you can replace transistors together but it won't change anything about dissipation . furthermore mosfets are not good for this case why ? because a bjt will be triggered by 0.65 volt but threshold voltage of a mosfet is more than this and it means more dissipation rather then BJT in linear region when we're attempting to make a variable PSU . so what's the solution ? simply you can use SMPS instead of Linear PUS .
Best Wishes
Goldsmith

 

[crutschow]

.....................
If your problem is 2n3055 and it's legendary face , :wink: and much dissipation across it i afraid that is a linear PSU and you have no other choice . you can not run from dissipation in a linear regulator . you can replace transistors together but it won't change anything about dissipation . furthermore mosfets are not good for this case why ? because a bjt will be triggered by 0.65 volt but threshold voltage of a mosfet is more than this and it means more dissipation rather then BJT in linear region when we're attempting to make a variable PSU . so what's the solution ? simply you can use SMPS instead of Linear PUS .

The threshold voltage of the MOSFET has no effect on the transistor dissipation since the gate draws no current. It's the drain-source voltage drop times the drain current that determines the dissipation and that's the same as with a BJT for a given output current and voltage.

 

[goldsmith]

The threshold voltage of the MOSFET has no effect on the transistor dissipation since the gate draws no current

Hi crutschow
Of course it has ! just think about this : if we have a mosfet with VTH=4 volts ( as an example ) and if our input voltage is 15 volts . and if we want out put be around 0-15 it is impossible because it will be about 0-11 volts ( unless we prepare the bias condition with an auxiliary path which can not be harvested by the circuit which has been posted above by the original poster ) . it means more dissipation isn't it ? i was referring to this issue .
Best Wishes
Goldsmith

 

[crutschow]

Hi crutschow
Of course it has ! just think about this : if we have a mosfet with VTH=4 volts ( as an example ) and if our input voltage is 15 volts . and if we want out put be around 0-15 it is impossible because it will be about 0-11 volts ( unless we prepare the bias condition with an auxiliary path which can not be harvested by the circuit which has been posted above by the original poster ) . it means more dissipation isn't it ? i was referring to this issue .

OK, I understand. If you replace an NPN emitter follower with an N-MOSFET source follower, you need a higher voltage to drive the gate. To avoid that, you would have to use a P-MOSFET (but that would affect the loop compensation, of course).

 

[goldsmith]

Hi crutschow

OK, I understand. If you replace an NPN emitter follower with an N-MOSFET source follower, you need a higher voltage to drive the gate.

Precisely !

To avoid that, you would have to use a P-MOSFET (but that would affect the loop compensation, of course).

P mosfets are usually dealing with lower speed and more dissipation because of their on state resistance which at maximum out put conductance is more than N types .
Effect on loop compensation ? can you explain it's effect perhaps via an example please ?

Best Regards
Goldsmith

 

[Audioguru]

You show the defective original power supply that is a kit in Greece and is a project at Electronics-Lab. Many of its parts are overloaded and are operating with a supply voltage higher than their maximum allowed voltage. The maximum output voltage at 3A is not 30V, it is about 25V with lots of ripple.

I fixed the circuit and there are a few long threads about it at Electronics-Lab. I upgraded the overloaded transformer, replaced the over-voltage opamps, replaced the over-heated driver transistor, replaced many resistors with larger ones and used two output transistors to share the heat.

The circuit is a linear design. If you want low dissipation then make a switching power supply.

 

[crutschow]

..................................
Effect on loop compensation ? can you explain it's effect perhaps via an example please ?

If you use a P-MOSFET, then the drain is the output and you have the added gain in the loop from the transconductance of the FET (as opposed to an emitter or source follow which has a gain of about 1). This added gain can cause loop instability and likely will require added compensation circuitry to stabilize the loop.

 

[goldsmith]

If you use a P-MOSFET, then the drain is the output and you have the added gain in the loop from the transconductance of the FET (as opposed to an emitter or source follow which has a gain of about 1). This added gain can cause loop instability and likely will require added compensation circuitry to stabilize the loop.

Hi crutschow
Thanks for the description . yes you're right .

Respectfully
Goldsmith

 

[Audioguru]

The original circuit uses emitter-follower output transistors that do not invert. The opamp that drives them has negative feedback from the output. The output circuit has a voltage gain of 30V/11.2V= 2.68.

A common-source Mosfet inverts so of course the output stage will be an oscillator because the opamp will have positive feedback instead of negative feedback.

 

[crutschow]

................................

A common-source Mosfet inverts so of course the output stage will be an oscillator because the opamp will have positive feedback instead of negative feedback.

That's true, of course. But even if you invert the polarity of U2 to compensate for this, you may still have instability problems due to the added loop gain from the common-source MOSFET.