Why MOSFET is used as a power transistor on the aim-tti ql355tp

[namster]

Hi ,
in most linear psu a BJT transistor is used as power transistor as i know the mosfet have many disavantage , so why on the aim-tti QL355 two mosfet are used ?

[David Hess]

The major disadvantage used to be cost.  For a given power rating, power MOSFETs used to cost more than bipolar transistors.  But now I think economy of scale has made up for that.

The other disadvantage is high and less precise gate threshold voltage but that does not apply in an application like this where a higher supply voltage is available for the control circuits.

[Seekonk]

Very low drive current is needed. Makes rest of circuitry cheaper.

[David Hess]

Very low drive current is needed. Makes rest of circuitry cheaper.

High current drive is still required for good AC performance.

[namster]

but is there any advantage in performance for using a more expensive transistor ?  , i think that  a BJT power transistor like 2n3055 have better value for money in linear application ?

[Kleinstein]

Some of the older MOSFETs (e.g. old IRFP250) had reasonable good linear performance.  The very low gate current can be handy in the floating regulator variant, as one does not have to worry about the base current flowing through the shunt.

The 2N3055 is usually fast enough, but the SOA is not that great for high power use, especially with the cheaper modern ones. When driven hard the MOSFET can be faster and may help with stability, thought there are other traps for the young players, especially if more in parallel are needed.

[duak]

I wondered if the MOSFET used here was some special linear rated part so I downloaded the data sheet.  No, it isn't and the data sheet doesn't show a DC SOA.  It does seem to show a fairly complex SABRE model with three types of MOSFETs of varying characteristics in parallel but I don't use SPICE or SABRE so I don't know if a simulation shows if the part is fit for purpose.  It does have an interesting active current limiting or balancing circuit composed of Q17 & Q21.

It wouldn't be my first choice for a pass transistor.  Are these supplies considered reliable?

[namster]

@Kleinstein
so the mosfet has an advantage for the precision of current measurement throught shunt , the QL355 have a resolution of 1mA ,such précision can't be reached with a BJT , in such Sziklai  or darlignton configuration a base current is summed to load current ! , What do you think ?

@Duak
https://pdf1.alldatasheet.com/datasheet-pdf/view/52834/FAIRCHILD/HUF75639G3.html
The DC SOA Is given in page 3 Figure 5 isn't that ?

[Kleinstein]

to make the bjt based circuit fast, it may need a low value for R1. Depending on the circuit the this current could also be a problem. With a MOSFET one has a little more freedom around the shunt. I would today prefer an audio BJT.

[duak]

@namster, on the data sheet I used, fig 5 on p.4 doesn't show the DC SOA; it only goes to a 10 ms pulse.  I've attached a copy of the data sheet I used.  It superficially appears the same as at your link.

[David Hess]

but is there any advantage in performance for using a more expensive transistor ?  , i think that  a BJT power transistor like 2n3055 have better value for money in linear application ?

A power MOSFET has a speed advantage and does not suffer from gain and bandwidth droop at high currents but usually a faster bipolar transistor is fast enough when closed loop frequency response is taken into account and even if the bipolar transistor is current derated to avoid Ft and hfe droop, it is still less expensive.  Check out the design of high performance audio power amplifiers for a discussion on these issues when comparing MOSFET and bipolar power output stages.

The insignificant gate current of a power MOSFET means current sensing can be done on the drain side however in the highest performance applications, it is better to do the current sending on the source or emitter side anyway.  Even if collector side current measurement is required, a smaller MOSFET or JFET can be used to drive a bipolar output transistor.

[Hydron]

The PSU in question is not a cheap item - I doubt the difference in cost between BJT and MOSFET comes into the equation, so they will have just used their preferred solution.

[namster]

@Duak
the 10ms can give an idea of DC SOA , the mosfet can easly work with 35V Vds and 3Amp output current
@David Hess
the AC performance tend to be complicated for me , i only know that for a stable regulation circuit the phase margin have to be large , do you have any book or lectures about opamp AC performance  ?
 

[duak]

@namster, you probably know that MOSFETs can suffer from Electro-thermal Instability when operated in linear mode.  It is probably why the device manufacturer does not show the DC SOA.  Jay Diddy & I (plus others) have posted on this subject: https://www.eevblog.com/forum/projects/please-let-me-know-if-my-very-high-power-load-has-a-fatal-flaw/msg3075882/?topicseen#msg3075882

The chosen device may have the DC SOA you have extrapolated from the data sheet.  My experience with data sheets is that extrapolation can be risky.  If the device manufacturer does not test for a condition or even show a predicted characteristic then a conservative designer risks failure by expecting that characteristic.  Sometimes we accept the risk and then compensate and test to see if there is a problem.  I have done that and have had very few failures.

Others have pointed out the speed advantages of MOSFETs.  I have pointed out some possible problems and what appears to be a solution.  We do not know what the designer knows and exactly why the circuit is the way it is.

[namster]

@duak
from my point of view the SOA does not present a problem because the performance of the mosfet performance are much higher then  the work required , the max current of this PSU is 3A the manufacturer used two of these mosfet with current balancing so in the worst case the safe operating area will never be reached, I drew the operating area in red, maybe I'm wrong

[David Hess]

@David Hess
the AC performance tend to be complicated for me , i only know that for a stable regulation circuit the phase margin have to be large , do you have any book or lectures about opamp AC performance  ?

Operational amplifiers have difficulty driving the large input capacitance of a power MOSFET while a similarly sized bipolar power transistor will be used with a smaller transistor to drive it.

Linear power supply and regulator design is very closely related to audio amplifier design so check out Designing Audio Power Amplifiers (2011) by Bob Cordell and Audio Power Amplifier Design Handbook by Douglas Self.

[H713]

I still have a hard time figuring out why the 2N3055 is still as common as it is, given that it isn't all that cheap and has pretty poor performance.

FETs can have superior SOA to BJTs in some applications. At high voltages, BJTs become increasingly useless in this regard. If it weren't for their stupidly high price, lateral MOSFETs would make excellent series pass transistors since they're pretty well immune to secondary breakdown. Modern switching FETs do suffer from something similar to secondary breakdown, however, using high voltage FETs with a high on resistance can help get around this. In my testing, the FQA8N90C-F109 did not exhibit any secondary breakdown at 75V, and they seemed to fail right around the 280W mark. That said, I have found the DC SOA charts in switching FET datasheets to be less than trustworthy.

[namster]

@David Hess
i would be happy to read them thank you
@H713
the 2n3055 was cited as an example only , i agree that for high voltage and current the mosfet is the right chose as long as we drive it correctly

[David Hess]

I still have a hard time figuring out why the 2N3055 is still as common as it is, given that it isn't all that cheap and has pretty poor performance.

The 2N3055 part number has become the "junk bin" of power transistors which did not meet higher specifications.  If you want a more consistent part, then something like the 2N3771 through 2N3773 is more appropriate but it will cost more.  Tektronix had at least three difference "classes" of 2N3055 which they used in different applications, and that was 50 years ago!

If it weren't for their stupidly high price, lateral MOSFETs would make excellent series pass transistors since they're pretty well immune to secondary breakdown.

Lateral MOSFETs have great performance but it comes with lower current density than even vertical MOSFETs and that extra die area directly adds to their cost.  I was not aware that they were even produced anymore except as part of DMOS integrated circuits.  Maybe we can get the audio guys interested in them to drive up production and lower price.

That said, I have found the DC SOA charts in switching FET datasheets to be less than trustworthy.

I have had the same experience.  If the datasheet for a vertical MOSFET does not explicitly show the thermal instability region, then I assume it is just as bad if not worse than the secondary breakdown of a bipolar transistor, which they often are.

[H713]

Exicon still manufactures lateral MOSFETs, but they are costly. They can be purchased through Profusion in the UK. As far as I know, there is no US distributor. The only thing driving that market it audio- they work great in power amps.

[exe]

I still have a hard time figuring out why the 2N3055 is still as common as it is, given that it isn't all that cheap and has pretty poor performance.

I believe modern 2n3055 equivalents are not as bad at all. Definitely more than enough for a linear power supply, which needs a closed-loop bandwidth of some tens (hundreds at most) of kHz. I've built a prototype of a linear power supply using sziklai with 2sta1943 and db139 (afaik) and it was fast (blackdog's power supply). In my measurements, 2sta1943 and tip3055 show pretty much identical performance (measured at 2A collector current), with the later slightly better. That's why I conclude tip3055 is a viable choice.

so the mosfet has an advantage for the precision of current measurement throught shunt , the QL355 have a resolution of 1mA ,such précision can't be reached with a BJT , in such Sziklai  or darlignton configuration a base current is summed to load current ! , What do you think ?

Depends where to put shunt and where's the load. I don't see problems if shunt is in series with the load. With a mosfet it's possible to have accurate current measurements even if the mosfet is in-between the load and the shunt because drain current is the same as source current. With bjt one have to put the load and the shunt somewhere after emitter.

[David Hess]

so the mosfet has an advantage for the precision of current measurement throught shunt , the QL355 have a resolution of 1mA ,such précision can't be reached with a BJT , in such Sziklai  or darlignton configuration a base current is summed to load current ! , What do you think ?

Like exe says, put the current shunt on the output lead.  For increased AC stability reasons, it is advantageous to have resistance in series with the output capacitor and with some clever circuit design, the output current shunt can be this resistance.