MOSFETs - Safe Operating Area

[joeyjoejoe]

Saw a nice video explaining this, so I get the concept.

However, some I notice some SOA charts do not have a DC rating.

For instance, the IRF 5305PbF doesn't.

https://www.infineon.com/dgdl/irf5305pbf.pdf?fileId=5546d462533600a4015355e370101993



What does this mean? It's not rated/certified/intended to be used for constant linear operation? Or they just haven't bothered?

(An aside : I notice International Rectifier/Infineon doesn't seem to have DC on any of the datasheets for the (very small sample set) 3 MOSFET's I chose. I went and picked 4 ON Semiconductor datasheets and they all had DC specified in the safe operating area.)

[thm_w]

Its not rated to be used in that way. You can find some good discussion here: https://www.eevblog.com/forum/projects/igbt-vs-mosfet-for-dummy-load/
http://www.digikey.com.au/Web%20Export/Supplier%20Content/microsemi_278/pdf/microsemi-power-an-make-linear-mode-work.pdf?redirected=1
https://www.infineon.com/dgdl/Infineon-ApplicationNote_Linear_Mode_Operation_Safe_Operation_Diagram_MOSFETs-AN-v01_00-EN.pdf?fileId=db3a30433e30e4bf013e3646e9381200

It could fail at less than 10x its rated current for example.
But some loads are using IRFP250, which has no DC SOA rating as well. So they have either done a lot of internal testing or discussed with the manufacturer.

[joeyjoejoe]

Awesome, thanks - that first one sure is lengthy!

So, is there a way to find mosfet's that are good for linear operation, short of reading tons of data sheets? I feel like I know the answer but might as well ask

[thm_w]

The way is to use the same parts Agilent/Rigol/BK have, and derate them in the same way (IRFP250, IRF540, etc. run at <30W each?)

Or buy purposefully linear rated FETs, some are not too expensive if you need guaranteed operation:
http://ixapps.ixys.com/family.aspx?i=5
https://www.microsemi.com/product-directory/mosfet/732-linear

[joeyjoejoe]

Gotcha.

Looks like ON Semi has some nice ones for linear use.

2SJ652 (P Channel)


NTP5864N (N Channel)


Kind of nice as a hobbyist, I don't have to sweat the difference between a MOSFET that's 1.40$ versus 2$. (Or 3.25CAD$ in this case!) Just derate to make up for my lack of the ability or willingness to test.

[digsys]

A few years back I spent weeks on blowing up dozens of different FETs, creating my own DC SOA curves. It was for a new generation UL drop-out linear
P/Supply series I was making. Comparing my curves against manufacturers - it was clear that MOST the SOA curves are either approximate, or total guesswork.
The INXS were pretty close, maybe a bit over-pec'd. Others were either better or worse, sometimes by a large margin. I did find some white papers on the
whys/hows etc. I'll dig them up and post them when I have a min. I settled for ~40V @ 5A per device as my guaranteed standard.
Hint: The MIC5156 / MIC5158 is the ideal device for driving several in parallel :-)

[Zero999]

Saw a nice video explaining this, so I get the concept.

However, some I notice some SOA charts do not have a DC rating.

For instance, the IRF 5305PbF doesn't.

https://www.infineon.com/dgdl/irf5305pbf.pdf?fileId=5546d462533600a4015355e370101993



What does this mean? It's not rated/certified/intended to be used for constant linear operation? Or they just haven't bothered?

(An aside : I notice International Rectifier/Infineon doesn't seem to have DC on any of the datasheets for the (very small sample set) 3 MOSFET's I chose. I went and picked 4 ON Semiconductor datasheets and they all had DC specified in the safe operating area.)

What voltage and current do you need?

If it's under 30V or so, a BJT might be better.

[MiDi]

Keep in mind that the given SOAs are @25°C die temperature, you hardly could estimate the SOAs for higher temperatures.
IXYS gives for their linear mode rated MOSFETS another SOA for higher temperature, 75°C if I remind correctly.

[BravoV]

An example from IXYS linear fet, IXTN46N50L datasheet, at the SOA chart, sometimes they called "Extended" FBSOA, watch carefully the highlighted areas I made, most common switching fets (or almost all) didn't make this kind of details.

Also watch the bold DC line as they deliberately made it.

[aheid]

I can't recall in which article I read this, but it basically said that if a SOA graph for a FET only has a straight line between R_DS(On) and V_(BR)DSS area for DC, then the SOA graph should not be trusted for DC operation.

Attached is the SOA for the IPP65R095C7, which has the characteristic kink in the graph the article talked about (and also at a more representative Tc).

Now, I'm no expert and I haven't had a chance to test this in practice yet unfortunately so can't say if the above FET is capable of linear loads. But compared to many other SOA graphs, it at least looks much better than many others.

[joeyjoejoe]

If it's under 30V or so, a BJT might be better.

Yup under 30V for sure - "true" current would be around 4A.

[MiDi]

An example from IXYS linear fet, IXTN46N50L datasheet, at the SOA chart, sometimes they called "Extended" FBSOA, watch carefully the highlighted areas I made, most common switching fets (or almost all) didn't make this kind of details.

Also watch the bold DC line as they deliberately made it.

Bit confused both SOAs are for 150°C Tj (junction/die), what has the Tc (case) to do with the SOA?

Edit: Oh, they are doing it on others the same way 

[aheid]

Bit confused both SOAs are for 150°C Tj (junction/die), what has the Tc (case) to do with the SOA?

If Tj is the same for two different Tc's then, assuming the systems are in equilibrium, the lower Tc system must have a better cooling solution and hence is able to dissipate more power as heat.

Alternatively, the absolute temperature of Tj depends on Tc. Higher Tc means less current to reach the same Tj.

[MiDi]

Bit confused both SOAs are for 150°C Tj (junction/die), what has the Tc (case) to do with the SOA?

If Tj is the same for two different Tc's then, assuming the systems are in equilibrium, the lower Tc system must have a better cooling solution and hence is able to dissipate more power as heat.

Alternatively, the absolute temperature of Tj depends on Tc. Higher Tc means less current to reach the same Tj.

  too late here, I should go 
It is the Rthjc which defines how much power at given dTjc...

[KE5FX]

Relying on Tj is problematic for power FETs that are constructed by paralleling multiple dies internally.  In the region below the point where the transfer curves cross over, a given drain current will be achieved at a much lower Vgs as temperature goes up.  The dies are never 100% identical, so thermal runaway becomes a possibility as the hotter dies start to carry more drain current.

To paraphrase Dr. Bronner, "Derate!  Derate!  OK!" 

[aheid]

too late here, I should go 

Np, was at the same stage, you might have noticed the edit