Linear mode operation and mosfet FBSOA question

[1001]

After some bad experiences with chinese dc current loads I decided to design one. In the pic the SOA of the MOSFETs I intend to use, TO220 package, 300W @ 25 ° C, Vds 200V, Id 54A; my question is:
Is it safe to push up to 100Vdc @ 1A if properly cooled? In this case it would dissipate 100W and there would be a theoretical margin of 200W.

[Kleinstein]

Judging from the shown SOA diagram the 100 V 1 A could be OK. However SOA diagrams with modern FETs are not always that reliable. Many are just calculated from the thermal impedance curve, ignoring a possible thermal run away.

With a modern 200 V device I would be somewhat careful. The SOA seems to just show power limits. A 300 W Ptot rating for a TO220 case is nothing to take too serious. Usually the rule of thumb is no more than 50 W from a TO220 if cooling is good. So any rating of more than about 80 W for an TO220 makes me think  .

So using this FET at 100 W is likely ending up in smoke.

[bktemp]

What Kleinstein said.
If there is no derating at higher voltages in the FBSOA curve (less Ptot than at 1/10 voltage), the shown DC SOA curve is most likely not correct/useable.

Also notice the conditions for the SOA curve: It says @Tcase=25°C, single pulse. This may imply Tjunction is also at 25°C.

Operating a heatsink at <0°C (which is necessary if you want to keep Tc at 25°C at high power levels) is impractical, that's why you end up with something like max 50W for TO220 and 100W for TO247 regardless of the insane high values in the datasheet.

[digsys]

What is the part# ? Package ? Never trust ANY PowerFET SOA curves in Linear mode. Most are just made up or approximate. I've destroyed 100's devices,
of several types, to create my own realistic SOA curves, and they are NOTHING like specs, not even white papers. The biggest killers are - getting the heat fast enough
away from the die to bond wires, then case to outer heatsink - PLUS there's hot-spotting issues, which derate the figures. The only SURE method, is to destroy a few
yourself and make your own charts. If you pass on the part #, I may have some test results.

[bktemp]

If you pass on the part #, I may have some test results.

Do you have by chance some results comparing older mosfets to their improved versions like IRF540 <-> IRF540N or IRPF150 <-> IRFP150N?
The older ones should handle linear operation better than the new versions because of the larger chip area, but the improved ones do have a lower thermal resistance between junction and case.

[tszaboo]

If you pass on the part #, I may have some test results.

Do you have by chance some results comparing older mosfets to their improved versions like IRF540 <-> IRF540N or IRPF150 <-> IRFP150N?
The older ones should handle linear operation better than the new versions because of the larger chip area, but the improved ones do have a lower thermal resistance between junction and case.

My experience is this: Old FETs have worse Tjc, new FETs cant handle linear operation.
Newer FETs also have much steeper transfer characteristics, meaning that it is much more difficult to make them work in the linear region.
1001:
You cannot really put more than 75W into TO220. And for that you need something close to 0.5 K/W Rtjc, and water cooling. And the graph is just for SOA, not FBSOA. For sure if you put anything like 75W and 100V on this FET, it will thermal runaway and blow up. One hint, if you need large FETs and linear region, check out IXYS.

[digsys]

Do you have by chance some results comparing older mosfets to their improved versions like IRF540 <-> IRF540N or IRPF150 <-> IRFP150N?
The older ones should handle linear operation better than the new versions because of the larger chip area, but the improved ones do have a lower thermal resistance between junction and case.

I didn't run full destructive testing on any "older" devices, as I wanted the very best I could find. (IXYS rated pretty high. They are about the last maker of true Linear
FETS). And you're right, there were many more "far" better FETs in the past with SOA suited for linear operation, and funnily enough, a lot of that was due to the slower
dI/dT .. dV/dT rise times that could be achieved, so they had to use very large single dies). All (most?) the newer ones have up to 100s individual FETs, all thermally
bonded together. The only "bad" part about this is hot-spotting ! It may be solved soon enough though. The TO247 package has the best die > case thermal transfer.
I'd never bother with TO220 for anything linear. I'll check tomorrow if the ones you mention are in my spreadsheet.
PS: NANDBlog just pipped my post :-) I run the IXYies at max 25W each, in the TO247 package with ONLY convection cooling and a small heatsink. That give me plenty of
headroom.

[Rerouter]

digsys, by chance would you have tested any smaller SMD mosfets? I'm kinda hoping i can get away with underrating the wattage by a factor of 5 (10W mosfet at 2W).

[Mechatrommer]

If you pass on the part #, I may have some test results.

Do you have by chance some results comparing older mosfets to their improved versions like IRF540 <-> IRF540N or IRPF150 <-> IRFP150N?

i smoked a pair of IRF540 quite an instant in my latest project ... here ... iirc running at 15V 4A each (60W) but that could be counterfeit china, could be not... (20cent each probably counterfeit) albeit tucked to the cpu grade heatsink (heatsink havent got hot even a bit)... then i changed to bjt TIP142 (125W rated) from trusted local source ($1+ each). running at 15V 5A (75W) per bjt until the heatsink is quite hot to the touch, then one of it showed intermittent damage (i have mosfet/bjt/power drive damage detect circuit to cutoff current) now it still run initially but quickly show damage symptom before being cut off... my current hypothesis is you need a very good heat transfer or case temperature detection for proper derating. my thermistor is at the other heatsink's end so i guess temperature reading is lagging so is power derating calculated by the mcu, so cut off is done after damage has occured. i should put thermistor very close to the power elements and embed lower power derating curve into the mcu i guess, this certainly require experimentation...

[digsys]

digsys, by chance would you have tested any smaller SMD mosfets? I'm kinda hoping i can get away with underrating the wattage by a factor of 5 (10W mosfet at 2W).

I have a few favorite SMD devices, that I use for e-bridges, up to 25A, but NOT for linear mode. rds-on is my critical spec here. NXP NextPower in LFPAK are impressive.
ie PSMN1R8-40YLC is one I use. My "big one" is an AUIRFS 8409-7P at only 0.5mR ! Again low rds-on
In D2Pak packages, my best hard working devices are - "above", IRFS3004-7P, IPB017N06N3 G, AUIRFS3004 .. hopefully you may find something useful here.

[1001]

Thanks for the answers, you have confirmed what I was thinking.
IXYS mosfet are too expensive, so if I use IRFP250 mosfet (TO247) and real world heatsink with fan could I push it up to 100V@0.5A (50W) and live happy? 

[digsys]

... my current hypothesis is you need a very good heat transfer or case temperature detection for proper derating. my thermistor is at the other heatsink's end so i guess temperature reading is lagging so is power derating calculated by the mcu, so cut off is done after damage has occured. i should put thermistor very close to the power elements and embed lower power derating curve into the mcu i guess, this certainly require experimentation...

That's why I use those eyelet types that you bolt directly to the package !with a thin smearing of arctic silver (your VERY best friend !!), I even bolt copper finned strips
to the TOP of the package as well, to get the heat away from the TO220 / 247 as fast as damn possible !! It's incredible how fast the thermal gradient increases.
I've had to explore this in great detail, as I never use fans or ANY other type of active cooling. That certainly sorts out the real men :-)

[digsys]

Thanks for the answers, you have confirmed what I was thinking.
IXYS mosfet are too expensive, so if I use IRFP250 mosfet (TO247) and real world heatsink with fan could I push it up to 100V@0.5A (50W) and live happy?

Well, I use TO247s on my e-load, www.pbase.com/digsys/image/158741424 , www.pbase.com/digsys/image/158741422 , which we use for loading solar EV
battery packs, and I can push the one pictured to 2KW (1 minute), 2.5KW (~10 secs) without a problem. Remember to ADD turbulence !! I use 2 fans - push/pull.
We even couple up a few units, to reach 20KW+, I double up on the fans. A well planned air flow can make magic ! :-)

[bktemp]

IXYS mosfet are too expensive, so if I use IRFP250 mosfet (TO247) and real world heatsink with fan could I push it up to 100V@0.5A (50W) and live happy? 

It depends on your heat sink, but in general yes. IRFP250 are useable up to maybe 100W at low temperatures.
For test thermal performance, use a heatsink with either a copper heat spreader, a heatpipe or at least a very thick aluminium below the mounting position. And avoid any additional insulating material between package and heatsink and apply the pressure as even as possible (preferably an external clamp instead of a single screw in the mounting hole).

[1001]

...and I can push the one pictured to 2KW (1 minute)

2000W per 24 bjt, 83W per mjl3281? what is the voltage on the load?

[1001]

from this video of our boss: EEVblog #281 - BK Precision 8500 Electronic Load Teardown https://youtu.be/AHu0MGEagSo,
I understand to not push more than 38W for each IRFP250, what do you think about this?

[tszaboo]

Thanks for the answers, you have confirmed what I was thinking.
IXYS mosfet are too expensive, so if I use IRFP250 mosfet (TO247) and real world heatsink with fan could I push it up to 100V@0.5A (50W) and live happy? 

They are not that expensive, at least for the system point of view. For example, they have the SOT227 transistors, safely rated for 600V, 50A. And you can easily put 500W into them if you have the heatsink for it. All this for 20 odd dollars. So if you design a load with it, instead of 7-10 TO220, with its own opamp, resistor, mounting, heat spreader, fuse, everything, you have one transistor, which is for sure going to work.

[Kleinstein]

Only 38 W for each of the IRFP250N is really conservative.

The hard part is something like if there is some kind of inductance that can force the electronic load to absorb some spike. Different from a normal linear supply it can be difficult for a electronic load to turn off on overload: with an inductive source the voltage would rise to much if current is forced down fast. So it is a good idea to have the capability to absorb quite some extra energy. The MOSFETs are not that expensive anymore - added robustness can be worth a lot. Arguing about failure under  warranty or from exceeding the specs is something one should avoid, even if this means making the unit more robust than needed to just meet the specs.

Also operation at higher voltage is more demanding on the FETs. This is the critical part that is missing in some of the diagrams. So it not only the power that counts.

At 50 V or similar one should be able to allow higher power. So with different software one might make the same circuit (or maybe changed shunts / emitter transistors) to work with higher power at low voltage too. Even for just 50 V, the IRFP250 can still be a good choice.

[T3sl4co1l]

Rule of thumb: more than 50W for a TO-220, 100W for a TO-247, etc., is getting into risky territory.  It may be fine, or you may need specialty materials and methods (better insulators, heat spreaders, water cooling).

The Tc=25C power rating of the device, is measured under boiling Freon.  When they say "Tc" (case temperature), they mean the whole case!  It is a meaningless figure.

Expect to spend about double the RthJC on an air-cooled heatsink, and double again on thermal interface material.  This basically brings a meaningless "300W" figure down to a more sane 60W, hence the 50 and 100W ballparks.

Heat dissipation, and 2nd breakdown, track fairly well with respect to die size.  Older FETs are advantageous for this reason: I tested (to destruction) an IRF740 at much higher power level than the datasheet suggests.  (The gotcha is, they are rated for maximum RthJC, but clearly, the typical RthJC can be much lower.  Can a measured RthJC be relied upon?  Who knows.  You better have a good QC program in place...)  Meanwhile, I tested a new production 6N60 (I forget what exactly, FDPF6N60 or something), which failed at something like 105% of ratings, in other words, spot on.

Taking them apart, the IRF740 die was easily triple the area of the FDPF6N60.

2nd breakdown isn't an issue for low voltage parts, or at low voltages; for high voltages, it depends on type.  Most modern, high density switching transistors are not suitable.  Older transistors largely don't care: just make sure DC is on the SOA.  The newest high voltage transistors (SuperJunction) apparently are not susceptible to 2nd breakdown, and are often seen with straight DC SOA curves.

As for "linear" transistors: yes, they're more expensive.  So what?  As kinds of parts go, transistors are one of the cheapest components you put into any project.  They're surely beaten out by resistors, but you'll easily spend more on capacitors, or transformers, in a typical project (given that I'm talking about projects a little more involved than a mere current sink).  Particularly relevant here: the heatsinks, and all the mounting hardware, and all the labor to stick them together.  Another five bucks for a good transistor is a stupidly small cost on top of hours of labor.  Stay real about your priorities!

Tim