Transistor VS MOSFET for linear power supply

[Prithul0218]

What are the advantage/disadvantage of using MOSFET/transistor as the main pass transistor of a linear power supply?

I see most professionally made power supplies use transistor, why do they avoid MOSFET? Can I just replace the transistor of a linear power supply with a MOSFET with similar ratings?

[rstofer]

I think you will find that MOSFETs make excellent switches but poor linear elements.  If it was simple to pull out the transistor and drop in a MOSFET, there wouldn't be any transistors used for pass elements.  But, no, we use a 55 year old 2N3055 or even several in parallel.

Here's a technical paper on the subject.  It doesn't say you can't do it but it does point out the complications.

https://www.infineon.com/dgdl/Infineon-ApplicationNote_Linear_Mode_Operation_Safe_Operation_Diagram_MOSFETs-AN-v01_00-EN.pdf?fileId=db3a30433e30e4bf013e3646e9381200

Google has more material.

[Jwillis]

Mosfets are better suited for switch mode operation because when operated in their linear region  MOSFETs are subjected to high thermal stress due to the simultaneous occurrence of high drain voltage and current, resulting in high power dissipation.When the thermo-electrical stress exceeds some critical limit, thermal hot spots occur causing the devices to fail. To prevent such failure, MOSFETs operating in the linear region require high power dissipation capability and an extended forward-bias safe operating area.Thermal hot spots can also trigger a second breakdown which is also destructive to the mosfet.
Some Linear Mosfets are designed to handle these conditions by suppressing positive feedback of elector-thermal instability and extending the forward bias safe operating area.They'er built much tougher than a standard mosfet .But this also increases the price as much as 10 times the price of a standard mosfet or a BJT.
BJT function much better in linear operation because unlike a mosfet they are current controlled.So simple current limiting is all that is needed to control a BJT.This makes for a simpler and cheaper linesar circuit .

[spec]

What are the advantage/disadvantage of using MOSFET/transistor as the main pass transistor of a linear power supply?

I see most professionally made power supplies use transistor, why do they avoid MOSFET? Can I just replace the transistor of a linear power supply with a MOSFET with similar ratings?

Hi Prithul0218,

That is a good question, because at first glance you would imagine that power MOSFETs are far superior in all respects to power bipolar junction transistors (BJTs).

The disadvantages of power MOSFETs compared to an ordinary power BJTs are:

• Massive parasitic capacitances, sometimes as high as 15nF, which vary with voltage and current. This makes designing with MOSFETs difficult as the parasitic capacitances act as integrators and limit the frequency response of the overall circuit (frequency response is important in PSU design)
• MOSFETS have a very high frequency response, but this is a double edged sword because, coupled with the high parasitic capacitances,  MOSFETs are susceptible to parasitic oscillations, normally around 4MHz
• Higher voltage drive than a BJT. You can turn a power BIT on hard with about 1.8V, but at least 4V would be required for a power MOSFET. The MOSFETs with a lower drive voltage tend to have correspondingly higher parasitic capacitances
• Generally poorer safe operating area (SOA)
• Generally, lower maximum junction temperature
• Higher cost

These are also the reasons, especially the parasitic capacitances which cause distortion, why you don't see MOSFETs used much in high end audio power amplifiers either.

But another reason that MOSFETs are not used that much in PSU is tradition- production designers like to stick to well tried and tested circuits whenever possible and the old workhorse NPN 2N3055 and PNP MJ2955 have a lot going for them, and they only cost around £1UK, and 45p if you buy 25 or more.

Having said that though, there is a trend towards using MOSFETs in PSUs. And the Nitride power FETs look promising.

MOSFETs are used extensively in switch mode PSU because of their high frequency response and and low ohmic on resistance, giving a compact and efficient PSU. But you have to go through all sorts of antics to get them to work. Luckily there is a host of standard chips and other components, especially capacitors and inductors, that have been developed over the years to help use power MOSFEts in SMPSUs and other applications.

Of course, power MOSFETs do have many advantages which have not been described here.
 

[MosherIV]

Hi

In a linear, what ever the pass device is, it must be capable of dissapting lots of power. Worst case is when the psu output voltage is at near min and max current, the pass device must drop near max voltage and pass the max current - ie this is the max power it MUST dissipate.

FETs need to have the gate voltage removed rapidly (pulled) when swtching off or they will randomly drift half on/off.

Transistor can be switched off by removing the base current, ie you can add an output on/off switch by putting a switch to the base of the pass transistor and the switch does not need to be capable of switching the full rated psu current.

[Prithul0218]

Thank you everyone. I got my answer. I apologise for not doing enough research before asking the question.

Thanks again.

[spec]

Thank you everyone. I got my answer. I apologise for not doing enough research before asking the question.

Thanks again.

No sweat from me- glad you got the answers you wanted.

[Zero999]

What are the advantage/disadvantage of using MOSFET/transistor as the main pass transistor of a linear power supply?

I see most professionally made power supplies use transistor, why do they avoid MOSFET? Can I just replace the transistor of a linear power supply with a MOSFET with similar ratings?

A MOSFET is a type of transistor. It stands for Metal Oxide Semiconductor Transistor.

I know what you mean though. People often assume BJTs when they talk about transistors, without specifying the type of transistor.

FETs need to have the gate voltage removed rapidly (pulled) when swtching off or they will randomly drift half on/off.

Transistor can be switched off by removing the base current, ie you can add an output on/off switch by putting a switch to the base of the pass transistor and the switch does not need to be capable of switching the full rated psu current.

That's true, but bear in mind a BJT will turn off very slowly if the base current is merely interrupted. To turn a BJT off quickly, the base voltage should be taken below the emitter voltage, reversing the base current.

Mosfets are better suited for switch mode operation because when operated in their linear region  MOSFETs are subjected to high thermal stress due to the simultaneous occurrence of high drain voltage and current, resulting in high power dissipation.When the thermo-electrical stress exceeds some critical limit, thermal hot spots occur causing the devices to fail. To prevent such failure, MOSFETs operating in the linear region require high power dissipation capability and an extended forward-bias safe operating area.Thermal hot spots can also trigger a second breakdown which is also destructive to the mosfet.
Some Linear Mosfets are designed to handle these conditions by suppressing positive feedback of elector-thermal instability and extending the forward bias safe operating area.They'er built much tougher than a standard mosfet .But this also increases the price as much as 10 times the price of a standard mosfet or a BJT.
BJT function much better in linear operation because unlike a mosfet they are current controlled.So simple current limiting is all that is needed to control a BJT.This makes for a simpler and cheaper linesar circuit .

Generally poorer safe operating area (SOA)

Are you sure? BJTs are typically more susceptible to secondary breakdown than MOSFETs. Look at a few data sheets and you'll see what I mean.

[mvs]

What are the advantage/disadvantage of using MOSFET/transistor as the main pass transistor of a linear power supply?

It depends on the design of the PSU. Even RDSon vs VCEsat may play a role, since in some power supplies pass transistors may also work as switches.

Liv has made very nice and efficient lab power supply with multilevel output stage based on MOSFETs, where it is the case.

[Kleinstein]

The SOA is a difficult topic for both BJTs and MOSFETs. Up to some 100 V there are BJTs made for linear audio amplifiers with a good SOA. Much above 200 V it gets really difficult with BJTs.  For the MOSFETs some old types and high voltage ones tend to have a good SOA. There are special individual tested MOSFETs for linear operation - but at a high price. SOA tested BJTs are not that rare. At high voltage (e.g. 300 V) it might be easier (though still difficult) to find MOSFETs with a useful SOA

Another problem with MOSFETs is that there speed depends quite a lot on the current. At low current they are rather slow, while at high current they can be much faster. This can make the compensation tricky or require quite some standing current, like usually used in MOSFET based audio amplifiers.

Another positive feature of BJTs is that there current gain usually goes down at very high currents and this provides a kind of peak current limiting. With MOSFETs the trans-conductance usually goes up with current, so that rather high peak current may happen quite fast.

To speed up the turn off of an BJT does not need to drive the base below the emitter level - it is sufficient to reverse the current direction, e.g. by shorting the base to emitter. During the turn off delay the base voltage is still positive.

Usually it takes a different circuit to use an BJT or MOSFETs. Though there are a few circuits that could work with either a MOSFET or darlington transistors.

[Zero999]

You're right, V_BE doesn't need to be reversed, just the base current.

As far as the SOA of BJTs vs MOSFETs is concerned, it's a pain to find BJTs which can dissipate a significant amount power at high DC voltages, but with MOSFETs, it's much easier. Take the FQL40N50 for example. Its SOA at DC goes all the way up to 500V at 9A*. Now see if you can find a BJT who's SOA extends to even 90mA at 500V. Note that I found that MOSFET fairly quickly and there are many others with similar SOA ratings.

*Of course it won't be able to do that in real life, unless it's cooled with liquid nitrogen.

http://www.mouser.com/ds/2/149/FQL40N50-112136.pdf

[Kleinstein]

I would net get too thrilled about the SOA curve of the FQL40N50. The SOA curve does not show any sign of thermal instability effects. It also shows a transient thermal response curve.  There are quite a few MOSFET data-sheets that just use the transient thermal response curve to calculate a kind of pseudo SOA curve that ignores possible thermal instability. So not all SOA curves shown are real.

So finding a MOSFET with a promising SOA is easy, the trouble is to know if the curve is real FB_SOA and not just a different way to show P_tot for different puls length form transient thermal response.

The FQL40N50 might still work - from a quick look I would consider it boarder-line. So maybe worth a try for less power, but still be prepared for a failure.

Still it's my feeling that above some 200 V it's getting difficult with BJTs.

[Zero999]

I would net get too thrilled about the SOA curve of the FQL40N50. The SOA curve does not show any sign of thermal instability effects. It also shows a transient thermal response curve.  There are quite a few MOSFET data-sheets that just use the transient thermal response curve to calculate a kind of pseudo SOA curve that ignores possible thermal instability. So not all SOA curves shown are real.

So finding a MOSFET with a promising SOA is easy, the trouble is to know if the curve is real FB_SOA and not just a different way to show P_tot for different puls length form transient thermal response.

The FQL40N50 might still work - from a quick look I would consider it boarder-line. So maybe worth a try for less power, but still be prepared for a failure.

Still it's my feeling that above some 200 V it's getting difficult with BJTs.

I agree about the FQL40N50.

There's lot of information in the thread linked below on MOSFETs for linear HV PSUs. The general consensus is at high voltages, they're better than BJTs and the older types are more likely to be good for linear operation, than the newer ones. T3sl4co1l tested the IRL740 in linear mode at 300V, with good results.
https://www.eevblog.com/forum/projects/looking-for-yours-opinions!-hv-stabilized-power-supply/msg1355890/#msg1355890
Before we get carried away, the original poster didn't state the voltage. Below 50V or so, the SOA is a less of an issue.

[Wolfgang]

What are the advantage/disadvantage of using MOSFET/transistor as the main pass transistor of a linear power supply?

I see most professionally made power supplies use transistor, why do they avoid MOSFET? Can I just replace the transistor of a linear power supply with a MOSFET with similar ratings?

The story is quite complicated. As stated in other posts, its a SOAR issue. At higher voltages, bipolar transistors and *switching* MOSFETs show a hot-spot effect called second breakdown. To be on the safe side, there are linear MOSFETs from IXYS that are specified for use with high dissipation and high voltage simultaneously. They make good PSU pass transistors.
A lot of manufacturers try to cut corners and use old switching MOSFETs at a fraction of their power rating. Its not guaranteed to work, and the consequences of a broken down pass transistor in a PSU can be quite catastrophic.

[not1xor1]

What are the advantage/disadvantage of using MOSFET/transistor as the main pass transistor of a linear power supply?

I see most professionally made power supplies use transistor, why do they avoid MOSFET? Can I just replace the transistor of a linear power supply with a MOSFET with similar ratings?

The story is quite complicated. As stated in other posts, its a SOAR issue. At higher voltages, bipolar transistors and *switching* MOSFETs show a hot-spot effect called second breakdown. To be on the safe side, there are linear MOSFETs from IXYS that are specified for use with high dissipation and high voltage simultaneously. They make good PSU pass transistors.
A lot of manufacturers try to cut corners and use old switching MOSFETs at a fraction of their power rating. Its not guaranteed to work, and the consequences of a broken down pass transistor in a PSU can be quite catastrophic.

AFAIK secondary breakdown concerns only BJTs, mosfets problems are different (thermal runaway in linear mode).

I had saved a few tips from IRF, regarding switching mosfets use in linear region. Unfortunately the links are dead.
In any case they suggested to select high voltage mosfets, with high Rdson and a slowly rising (i.e. not-steep - please suggest a more appropriate English term) Idrain-vs-Vgs curve.

[xavier60]

In hard switching applications, MOSFETs generally require large Gate drive current because the Gate driver has to fight the large current  of the Miller capacitance being charged and discharged due to the Drain being made to slew so quickly.
As a series pass element, although there can be rapid load current fluctuations, the voltage across Drain and Source doesn't change all that quickly so the Miller current is very small.
MOSFETs in linear power supplies are typically driven by an ordinary op-amp.

[MiDi]

I had saved a few tips from IRF, regarding switching mosfets use in linear region. Unfortunately the links are dead.

Is AN1155 one of those notes?

[Jwillis]

Are you sure? BJTs are typically more susceptible to secondary breakdown than MOSFETs. Look at a few data sheets and you'll see what I mean.

That is true but power MOSFETs have parasitic PN and BJT elements within the structure and as the internal geometry shrinks due to hot spotting beyond the thermal runaway stage some power MOSFETs can exhibited a failure mode resembling secondary breakdown.I was simply trying to be brief in my explanation.
Not to say it can't be done but I would use a mosfet that is rated much higher than the BJT your attempting to replace.Or ,as suggested by others, to use a Mosfet designed for linear applications .

[not1xor1]

I had saved a few tips from IRF, regarding switching mosfets use in linear region. Unfortunately the links are dead.

Is AN1155 one of those notes?

they were short web pages.
it looks like infineon deleted lots of old IRF web pages and app notes.

In any case here is Using HEXFET MOSFETs as Linear Amplifiers
BTW I've got also Using HEXFET MOSFETs as variable loads, just let me know if anybody is interested.

Answer ID 214  |   Updated 04/08/2009 12:43 AM
Question
We would like to design a linear power amplifer with N channel HEXFET - Power MOSFETs.

In various forms of literature we find many designs with complementary HEXFET - Power MOSFETs but only N channel devices are employed.

The output voltage is +200V/-30V and the current is 0.5A. Because, the N channel devices are better than P channel devices for high voltage applications we want to use the N channel FETs in the output stage. But the drive stage is difficult to design. Help.
Answer
IR HEXFET® MOSFETs are designed to be used as Switches. This does not preclude their use in linear applications but as the device technology advances to provide ever better performance as switches with lower Rds(0n) and gate charge figures, this makes the devices ever less suitable for linear applications. A lower Rds(on) equates to an output characteristic with a steep slope, which makes it difficult to maintain a suitable gate bias to keep the drain current operating point stable. As the drain current is deliberately varied in response to an applied input voltage, this changes the dissipated power within the part, changing the junction temperature and thus the value of Rds(on), i.e the slope of the output characteristic. If the device is being used in audio amplifier applications, other than class D for which most are ideally suited, the frequency of Rdson variation will not be seen as a fast variation as the thermal inertia will average them out to produce an average value for Rdson. This is the value you would use to solve the thermal equlibrium equation to determine if the Tj achieved in the application will be within permitted limits or not.

As higher voltage MOSFETS have higher Rds(on) values and thus a more shallow output slope these will be more suitable for linear applications.
All our current Application Notes are written around switching operations and we have none dealing directly with linear amplifier circuits.

With N channel devices, as the MOSFET is turned on by taking the gate voltage positive with respect to the source, you will have to arrange a gate bias supply capable of going positive of +200V by around a maximum of 10V if you want to control the MOSFET completely between full on and full off. As in your application the required output voltage also has to go negative you will have to use a split supply so that the effective gate voltage range is 0V to +210VDC for full range control.

P channel devices are controlled by taking their gates negative with respect to their sources, but the unavailability of high voltage P channel HEXFET® MOSFETs means they are not available with Vds(br) ratings above around 150V.

[joeyjoejoe]

As a beginner, I found this video a very nice overview on MOSFETs in linear mode :

[Wolfgang]

What are the advantage/disadvantage of using MOSFET/transistor as the main pass transistor of a linear power supply?

I see most professionally made power supplies use transistor, why do they avoid MOSFET? Can I just replace the transistor of a linear power supply with a MOSFET with similar ratings?

The story is quite complicated. As stated in other posts, its a SOAR issue. At higher voltages, bipolar transistors and *switching* MOSFETs show a hot-spot effect called second breakdown. To be on the safe side, there are linear MOSFETs from IXYS that are specified for use with high dissipation and high voltage simultaneously. They make good PSU pass transistors.
A lot of manufacturers try to cut corners and use old switching MOSFETs at a fraction of their power rating. Its not guaranteed to work, and the consequences of a broken down pass transistor in a PSU can be quite catastrophic.

AFAIK secondary breakdown concerns only BJTs, mosfets problems are different (thermal runaway in linear mode).

I had saved a few tips from IRF, regarding switching mosfets use in linear region. Unfortunately the links are dead.
In any case they suggested to select high voltage mosfets, with high Rdson and a slowly rising (i.e. not-steep - please suggest a more appropriate English term) Idrain-vs-Vgs curve.

Sorry for oversimplifying. I do know that the "second breakdown Term" only applies to bipolars. For MOSFETs its the "Spirito effect".
The practical outcome is the same, however; at high voltages and currents a hot spot forms on the chip that finally melts a whole into the die and that was it for your transistor, be it a MOSFET or bipolar type.

[not1xor1]

AFAIK secondary breakdown concerns only BJTs, mosfets problems are different (thermal runaway in linear mode).
 [...]

Sorry for oversimplifying. I do know that the "second breakdown Term" only applies to bipolars. For MOSFETs its the "Spirito effect".
The practical outcome is the same, however; at high voltages and currents a hot spot forms on the chip that finally melts a whole into the die and that was it for your transistor, be it a MOSFET or bipolar type.

Actually I've just found another document mentioning secondary breakdown regarding MOSFETs so I was probably wrong  :
Using Trench Power MOSFETs in Linear Mode

[Wolfgang]

Thanks,

yeah, I've seen another paper using "second breakdown" also for MOSFETs.

http://www.ixyspower.com/images/tech-papers/Article_Linear_Power_MOSFETs_2007.pdf

has some info about MOSFETs that are hardened against this type of failure.