Electronics > Beginners
Operation in This Area is Limited by R DS(on)
T3sl4co1l:
Or TLV2372, a good general purpose jellybean without the quirks of LM358. Mind the lower voltage rating, still fine here.
Regarding SOA, please stop repeating misinformation. If hotspotting happens, it happens. If the SOA says full DC operation, it means full DC operation (at the plotted voltage and current, and rated Tc). If you don't trust them, you think they're lying, go ahead and perform the actual test. Generalizations are not useful here: there are even IGBTs available with DC ratings. SOA primarily depends upon power density, which increases with tech node (MOSFET --> BJT --> IGBT) and voltage, but current sharing depends on design just as much, and it seems SuperJunction devices are usually optimized that way, and some IGBTs are as well.
So, you got legacy (lateral?) MOSFETs that were "free from 2nd breakdown", in contrast to contemporary BJTs which often were subject to that limitation (even the power amp rated ones, which usually have a small corner still subject to the limit). I'm guessing that's when all the classic books were written, from which everyone took as unchanging fact that these were the limits. Meanwhile designs continued to evolve, MOSFETs got terrible SOAs, IGBTs were introduced, SOAs improved. Finally today, parts from several generations and optimization goals are currently available.
Tim
tszaboo:
--- Quote from: T3sl4co1l on January 09, 2020, 03:18:47 pm ---If the SOA says full DC operation, it means full DC operation (at the plotted voltage and current, and rated Tc). If you don't trust them, you think they're lying, go ahead and perform the actual test.
--- End quote ---
I did, it was part of my job years ago. They blow up.
magic:
--- Quote from: NANDBlog on January 09, 2020, 02:08:36 pm ---The DC rating is typically 10-100ms rating. Otherwise you get hotspotting. You gotta understand, that a modern FET is a lot of smaller FETs in parallel, sometimes hexagons shaped.
--- End quote ---
IMO the DC rating should be valid for DC. The rating for 10-100ms ought to be labeled 10-100ms :-//
But I did link two appnotes which are frankly quite confusing, by saying things along the lines of "typical plots seen in the datasheets don't tell the full story". Oh well, so why are they even included?
And by the way, I believe I have already corrected myself regarding the part that International Rectifier blew up. It was not rated with high DC SOA; in fact, the datasheet clearly shows second breakdown behavior, both at DC and in timed pulses.
I went for more information to DIYAudio because that's where lots of people who use and abuse MOSFETs hang out. The impression I got is as follows:
Lateral FETs are to the first order indestructible, because their Vth tempco is weaker than Rds(on) tempco for almost any reasonable drain current. But barely anyone makes them anymore, I think Renesas still has one complementary pair which shows up as "special order" rather than "obsolete" at distributors and there are some boutique vendors for audio.
Old-school planar verticals like the classic IR HEXFETs (IRFx(9)yz0) or Fairchild Stripe FETs are prone to thermal runaway when biased with constant Vgs because their combined Vth/RDS tempco remains positive up to many amps of current, but generally free of second breakdown. Somebody tested IRFP240 with 2kW pulses at 100V for milliseconds at a time, others are fanboys of Fairchild and so on.
Fancy low-RDS(on) parts, particularly those using trench gate technology, have serious second breakdown issues even in short impulses and are not recommended.
I know it's just a bunch of unsourced claims, but hopefully they may serve at least as search keywords.
Finally, for everyone's consideration
https://www.diyaudio.com/forums/solid-state/101745-bob-cordell-interview-bjt-vs-mosfet-95.html#post1206694
--- Quote from: Nelson Pass ---I guess I have been lucky, never having problems with
reliability with power Mosfets. I've personally put over a million
of them into the field, and through Adcom maybe twice that.
--- End quote ---
Knowing Pass, it's probably mostly class A amplifiers. But I don't know how much he derates, so there is that.
Jwillis:
--- Quote from: Circlotron on January 09, 2020, 10:55:45 am ---^^ Nup, no good. Gotta have input CM range down to ground. A humble LM358 or LM324 would probably be good enough.
pdf.datasheetcatalog.com/datasheet/stmicroelectronics/2163.pdf
--- End quote ---
Oh Ok . So that would explain why the outputs jump up to 8.3V on the OPA604 and 7.75v on the 741 with a 9V supply .I thought something was wrong but couldn't pin it down. What if I ran them with a negative rail instead of ground ? I don't have any LM324.
--- Quote from: NANDBlog on January 09, 2020, 03:25:09 pm ---
--- Quote from: T3sl4co1l on January 09, 2020, 03:18:47 pm ---If the SOA says full DC operation, it means full DC operation (at the plotted voltage and current, and rated Tc). If you don't trust them, you think they're lying, go ahead and perform the actual test.
--- End quote ---
I did, it was part of my job years ago. They blow up.
--- End quote ---
Yes running Mosfets in parallel in the linear region can be tricky but it can be done. http://www.digikey.com.au/Web%20Export/Supplier%20Content/microsemi_278/pdf/microsemi-power-an-make-linear-mode-work.pdf?redirected=1 . You are correct in many respects as the test in this report indicate . I don't want to have to destroy a whole bunch of transistors just to find out the FBSOA . So if I can keep them in a except-able temperature range things should go Ok. I even considered starting from scratch and using PWM to control parallel Mosfets .I have to admit that using IGBTs was a lot easier.
Vovk_Z:
If in short: two FDL100N50F must withstand 40 V 20 A (with good enough cooling) . Only one can withstand too, but it will be on the edge.
IRFP260 is very good for linear mode, but 100N50F is on the very other level of power capability - it is 10 times more capable.
IRFP260 (IRFP360, 460 etc) can work with 50-100 W per case.
100N50F can withstand 500-1000W per case.
Navigation
[0] Message Index
[#] Next page
[*] Previous page
Go to full version