Help understanding SOA (Safe Operating Area) Current for a MOSFET?

[LooseJunkHater]

Referenced supervisor I.C: https://www.ti.com/lit/ds/symlink/lm5069.pdf
LM5069 Component Calculator: https://www.ti.com/tool/LM5069QUICK-CALC

Background: I'm currently using KiCad to design a circuit based around the LM5069 I.C. The circuit will have an adjustable over-current protection limit + adjustable over-voltage protection limit. I'd like the circuit to be able to stop an 80v 10a fault (with greater than 10a stopped by a fuse, and greater than 80v stopped by a TVS, and reverse-polarity protection via a p-channel MOSFET/"ideal diode").

TI provides a "component calculator" to help choose the appropriate components for the LM5069. The component calculator requests me to input the values of the "SOA (Safe Operating Area) Current" and I've noticed most FET's don't appear to be able to stop an 80v 10a fault; is that correct?

___________

Question: Attached is an image of the SOA comparing three FET's. Is my understanding correct, that none of these FET's can stop a 80V 10A fault for 10ms or slower? Is 80v 10a (800w) a realistic fault to be stopping?

[macboy]

The SOA curves show the permissible simultaneous drain current and drain to source voltage. Typically, a switching FET is either fully off (near zero current, high drain source voltage) or fully on (high current, very low drain source voltage determined by current and RDSon). Those curves primarily show you under what conditions you can operate the device when it is not fully on or fully off. That can include the time spent transitioning from on to off or back.

When using a FET to stop a larger current, make sure to account for any possible inductive kick when you suddenly stop the current. The voltage from that might exceed the voltage rating of the FET.

[ArdWar]

Is my understanding correct, that none of these FET's can stop a 80V 10A fault for 10ms or slower?

Yes, and probably no single MOSFET in the world can take that much brunt either.
Well, maybe one of those IXYS linear MOSFETS. You can also parallel multiple MOSFETs, but then you run into problem where you need to current balance hot MOSFETs running in linear mode.

Is 80v 10a (800w) a realistic fault to be stopping?

Yes, maybe, but not common for sure. You certainly can design for it but assuming you aren't doing something unconventional here that much level of fault can only happen when the output is dead zero ohm shorted *and* the MOSFET is expected to current/inrush regulate that instead of going into fault mode and outright shutting off the circuit. To avoid that kind of frankly outrageous project scope you may want to do risk analysis first regarding what kind of fault that can happen, what kind of fault the circuit is expected to protect against, and how the circuit is expected to react to that fault.

[T3sl4co1l]

Referenced supervisor I.C: https://www.ti.com/lit/ds/symlink/lm5069.pdf
LM5069 Component Calculator: https://www.ti.com/tool/LM5069QUICK-CALC

Background: I'm currently using KiCad to design a circuit based around the LM5069 I.C. The circuit will have an adjustable over-current protection limit + adjustable over-voltage protection limit. I'd like the circuit to be able to stop an 80v 10a fault (with greater than 10a stopped by a fuse, and greater than 80v stopped by a TVS, and reverse-polarity protection via a p-channel MOSFET/"ideal diode").

A few comments already:
- How will a fuse "stop" current, with this circuit of equal rating being in use?
- TVS don't fire at exactly the rating.  You need more like 30-50% headroom, depending on type and required surge capacity.  60 or 70V would be a safe maximum with this controller (with corresponding SMxJ62(C)A or 68 diode) for modest surge ratings.
- What is the surge capacity, what requirements if any are you working with?  Or what application so that we can reasonably anticipate such?
- Note that the usual PMOS reverse polarity circuit doesn't work well for rapid reversal i.e. hot-plugging; this can be a problem in some applications.

An SMAJ would probably be fine for the TVS, to handle hot-plug inrush, if electrolytic caps aren't doing it already; and no TVS might be required at all for lower level (say, 500V 20Ω+) IEC 61000-4-5 surges, if already handled by same.  ISO 7637-2 pulses 1 and 2a would probably want an SMCJ or better, and, a simple test model can be set up to show which size diode is required.

It wouldn't be adequate at all for 24V automotive load dump, and I'd want either a surge stopper controller (usually have enable and current limiting anyway, basically a similar design, just wired differently), or a series switch in front of it, or a stonking great TVS, to handle that situation.  All depends on application and requirements.

TI provides a "component calculator" to help choose the appropriate components for the LM5069. The component calculator requests me to input the values of the "SOA (Safe Operating Area) Current" and I've noticed most FET's don't appear to be able to stop an 80v 10a fault; is that correct?

Well, "most" is an extremely loose word.  Do you mean FETs in general (including all the SOT-23s etc. in existence would be a clear "no"!), those rated the bare minimum (80V 10A), those rated reasonably (100-200V, Rds(on) low enough for, whatever is acceptable for Pd), or what?  I don't know what kind of selection/range you've been looking over, here.

The key question is, for how long?

The curves you give, all show at least 100us, and two of them are near 1ms.  I certainly can't say they won't handle 800W ever.  But maybe it's not long enough.

Equivalently, how much inrush energy do you need to supply, or momentary fault energy to deliver before breaking the circuit.

Indeed you could strap off the timer pin completely (actually maybe not, I didn't check if it can be disabled) and use an extraordinarily massive transistor (or several in parallel with matched Vgs(th) and other considerations) to handle the 800W continuously.  It's not an exorbitant amount of power in the grand scheme of things, but it will be some expense ($40+ of parts, plus heatsinking), and a modest controller chip like this isn't really made to drive such load (but, you could add your own gate drive booster circuit if you really wanted).

Tim

[Solder_Junkie]

I have built several protection boards using an LT4368 and back to back MOSFETs. They work really well. The IC is rated to 100 Volts input and 80 Volts output, in my applications I am protecting radio equipment that runs up to 35 Amps at 13.8 Volts and yet shorting the supply with a 3 Amp fast blow fuse just shuts off the supply without blowing the fuse, the shut off is that fast. The MOSFET I use is a DMT6002, sadly not rated to 80 Volts, however there are plenty around that will cope with higher Voltages.

The LT4368 is tiny, you need a good head magnifier or better still a binocular microscope.

Analog Devices have a forum and are happy to help "home users" and were great at solving a problem with switch on surges tripping the over current. If you download LT Spice (free), there is a ready to use model of the LTC4368 you can play with. The circuit I used is the same as that shown in the data sheet. The switch on surge fix (if needed), was to add a 3.9K resistor between the sense pin and the input side of the current sense resistor, and a 220 nF cap across pins 8 and 9 (sense and V out). The fix using these components delays an over current surge by approx 1 mS.

https://www.mouser.co.uk/datasheet/2/609/ltc4368-3126108.pdf

SJ

[Doctorandus_P]

I just had a peek at the datasheet of the LM5069 and now your post makes more sense to me. (I and others) agree that the FET's in your selection can't handle a dissipation of 800W for very long, but you are also not trying to dissipate this for very long. The graphs you show suggest it is all right as long as you switch the FET fast enough (somewhere between 10us and 100us). When the FET is on, the voltage over it is low, and thus dissipation is also low. when the FET is off, there is no current, and thus dissipation is also very low. It is the very short interval during switching that there is both current though the FET and voltage over the FET that it's dissipation is high. Apparently this IC can drive the gate with around 100mA (typical) and that is not very much for a gate driver. So you may get into trouble here, depending on the capacitances of the FET used.

A TVS is also not a magic component. Without some external current limiting, it will glow until it pops and releases the magic smoke. Also note that the LM5069 datasheet suggests to use this IC for input voltages only up to 24V. That gives a decent margin for spikes and surges. But a fuse may be quick enough to stop the current before the TVS is damaged. But the current needs to be high enough to pop the fuse quick enough.

[T3sl4co1l]

Apparently this IC can drive the gate with around 100mA (typical) and that is not very much for a gate driver. So you may get into trouble here, depending on the capacitances of the FET used.

Feature not a bug -- notice the pull-up is from an internal charge pump, 10s µA available; weak anyway because internal, but it's not like it needs to be much given the soft-start or current limiting behavior.  While turn-off is faster, it's still on the slow side as switching goes -- because fast switching under fault conditions is a sure way to blow up your chip (or to require additional features or process steps to tolerate inputs transiently well outside GND/VDD), or at least require far tighter layout than otherwise.  Supply switching on the order of 10s of µs is pretty much blazing fast for most applications, so it's perfectly suitable to use ~mA gate drive here.

A TVS is also not a magic component. Without some external current limiting, it will glow until it pops and releases the magic smoke. Also note that the LM5069 datasheet suggests to use this IC for input voltages only up to 24V. That gives a decent margin for spikes and surges. But a fuse may be quick enough to stop the current before the TVS is damaged. But the current needs to be high enough to pop the fuse quick enough.

Just a rough orders-of-magnitude feel on this:

Even for very large TVSs, and very fast fuses, this is likely only true at low current ratings, and low voltages.

For example, an automotive fuse might have to pass 500A fault current for 10ms before opening, and say the TVS is dropping 30V in this condition (say it's a 18V nominal part).  There are "15kW" TVSs available, but not for this duration, it's off by an order of magnitude -- they're rated for 10/1000µs pulses, not 10ms+.  Assuming energy scales as sqrt(t) (not a great assumption in this time scale, but maybe not terrible either), at least triple is needed or 45, say 50 or 60kW type rating.  Or three in parallel (of course, paralleling TVSs is a bit dubious; they will share, avalanche has positive tempco, but they still won't share perfectly, and maybe 4 or more is required).

For PCB level stuff, an SMCJ5.0A should blow a 1206 chip fuse well enough, maybe even SMAJ depending on overvoltage and available fault current.  SMC is kind of laughably big for a 5V board though, so you can see the disparity already, and it just gets worse as voltage goes up.

---

On a related note, there used to be these combination TVS-polyfuse components available, so the TVS chip heats the PTC, clearing it fast enough to, well, not explode.  They're rarely stocked (few suppliers, or obsolete now?) though.  Interesting product, of obvious value and use, but just not popular enough (or economical?), or compared to more conventional methods (like this thread uses ), I guess.

Tim

[LooseJunkHater]

When using a FET to stop a larger current, make sure to account for any possible inductive kick when you suddenly stop the current. The voltage from that might exceed the voltage rating of the FET.

Such as by using a diode to absorb any inductive kick after the shutoff?

You can also parallel multiple MOSFETs, but then you run into problem where you need to current balance hot MOSFETs running in linear mode.

How exactly would I current-balance FET's running in parallel? If I have the FET's on the same heatsink, would that work as the RDS(on) values would change based on temperature?

A few comments already:
- How will a fuse "stop" current, with this circuit of equal rating being in use?
- TVS don't fire at exactly the rating.  You need more like 30-50% headroom, depending on type and required surge capacity.  60 or 70V would be a safe maximum with this controller (with corresponding SMxJ62(C)A or 68 diode) for modest surge ratings.
- What is the surge capacity, what requirements if any are you working with?  Or what application so that we can reasonably anticipate such?
- Note that the usual PMOS reverse polarity circuit doesn't work well for rapid reversal i.e. hot-plugging; this can be a problem in some applications.

My idea is to use ~12a fuse and its purpose to be if the FET fails and doesn't stop a short, as well as if the TVS triggers continuously (such as too high of an overvoltage), it will then blow the fuse. For the TVS, I'm thinking of a 90v TVS (absolute max of the LM5069 is 100v). I'd imagine that if 90v was continuously applied to an 80v TVS, it would draw enough current to blow the fuse?

An SMAJ would probably be fine for the TVS, to handle hot-plug inrush, if electrolytic caps aren't doing it already; and no TVS might be required at all for lower level (say, 500V 20Ω+) IEC 61000-4-5 surges, if already handled by same.  ISO 7637-2 pulses 1 and 2a would probably want an SMCJ or better, and, a simple test model can be set up to show which size diode is required.

I have no idea what any of this means.

Well, "most" is an extremely loose word.  Do you mean FETs in general (including all the SOT-23s etc. in existence would be a clear "no"!), those rated the bare minimum (80V 10A), those rated reasonably (100-200V, Rds(on) low enough for, whatever is acceptable for Pd), or what?  I don't know what kind of selection/range you've been looking over, here.

The key question is, for how long?

The curves you give, all show at least 100us, and two of them are near 1ms.  I certainly can't say they won't handle 800W ever.  But maybe it's not long enough.

Say 80v 9a is being drawn through the mosfet continuously and then suddenly a short-to-ground occurs; would the FET be able to easily stop that short, or would it simply blow up? Continuously, the FET would only dissipate ~0.45w of power (if the FET has an RDS(on) of 0.05ohm and 9a current flowing through).

Indeed you could strap off the timer pin completely (actually maybe not, I didn't check if it can be disabled)

I read most of the datasheet; it appears that the timing capacitor is used for many different functions and likely cannot be disabled.

Also note that the LM5069 datasheet suggests to use this IC for input voltages only up to 24V.

Where are you seeing this...?

[LooseJunkHater]

Just a rough orders-of-magnitude feel on this:

Even for very large TVSs, and very fast fuses, this is likely only true at low current ratings, and low voltages.

For example, an automotive fuse might have to pass 500A fault current for 10ms before opening, and say the TVS is dropping 30V in this condition (say it's a 18V nominal part).  There are "15kW" TVSs available, but not for this duration, it's off by an order of magnitude -- they're rated for 10/1000µs pulses, not 10ms+.  Assuming energy scales as sqrt(t) (not a great assumption in this time scale, but maybe not terrible either), at least triple is needed or 45, say 50 or 60kW type rating.  Or three in parallel (of course, paralleling TVSs is a bit dubious; they will share, avalanche has positive tempco, but they still won't share perfectly, and maybe 4 or more is required).

Tim

For the TVS, I'm thinking of a 90v TVS (absolute max of the LM5069 is 100v). I'd imagine that if 90v was continuously applied to an 80v TVS, it would draw enough current to blow the fuse? Would this TVS withstand the fault for long enough to blow the fuse, but NOT damage the TVS? https://goodarksemi.com/docs/datasheets/transient_voltage_suppressors/P6KEx.pdf

[T3sl4co1l]

My idea is to use ~12a fuse and its purpose to be if the FET fails and doesn't stop a short, as well as if the TVS triggers continuously (such as too high of an overvoltage), it will then blow the fuse.

Got it.

For the TVS, I'm thinking of a 90v TVS (absolute max of the LM5069 is 100v). I'd imagine that if 90v was continuously applied to an 80v TVS, it would draw enough current to blow the fuse?

It might not draw any current at all at 90V.

Check the datasheet: P6KE110A is rated for 94V nominal working voltage.  No other part is rated to handle 90V continuously, so this is the lowest part that would be acceptable.  Max breakdown is 116V, at one measly mA.  It's rated to clamp 3.9A at 152V max, with a 10/1000µs waveform (equivalent to somewhere around 500µs square pulse).  It'll last even shorter at higher currents, and needless to say, at real, practical fuse-clearing fault currents, maybe single to dozens of µs, and no fuse on Earth even notices something's wrong in that kind of time scale.

And with such high peak clamping voltage, it wouldn't even provide cross-wiring protection, say in a 120VAC context.  The controller is still just as dead; if this would be in equipment that's serviced at the board-replacement level, I wouldn't even bother putting in a TVS or crowbar for this purpose; just let the controller blow up, it'll leave a smaller burn mark than the TVS will, heh.

Whereas the P6KE75 is rated 60V nominal and 108V peak at 5.6A, still nowhere near enough to clear a fuse but enough to handle modest transients, and a larger diode would handle most e.g. automotive transients.

Similar values are had for the SMAJ series, https://www.littelfuse.com/media?resourcetype=datasheets&itemid=13c2a823-03b8-4d1f-9ddc-9b44670aed9d&filename=littelfuse-tvs-diode-smaj-datasheet though ratings are a little lower as it's a 400W diode.  SMBJ would be the direct P6KE surface mount equivalent, I suppose.

Like I said. TVS are not precise elements.

An SMAJ would probably be fine for the TVS, to handle hot-plug inrush, if electrolytic caps aren't doing it already; and no TVS might be required at all for lower level (say, 500V 20Ω+) IEC 61000-4-5 surges, if already handled by same.  ISO 7637-2 pulses 1 and 2a would probably want an SMCJ or better, and, a simple test model can be set up to show which size diode is required.

I have no idea what any of this means.

Well, uh...

Do you want to know, then?

Say 80v 9a is being drawn through the mosfet continuously and then suddenly a short-to-ground occurs; would the FET be able to easily stop that short, or would it simply blow up? Continuously, the FET would only dissipate ~0.45w of power (if the FET has an RDS(on) of 0.05ohm and 9a current flowing through).

What do you mean "drawn through"?  If it's saturated*, it's dissipating only Pd = Rds(on) * Id^2, causing temperature rise of Pd * RthJA (for whatever is between J and A in the situation: nothing, PCB, heatsink, jet of boiling freon..).  Under load-short conditions, Vds rises suddenly, Id peaks until the controller gets it back under control, and some time later, the timer runs out and it turns off.  If, through all of this, under worst-case conditions (max ambient temp,  Tj never rises above the maximum limit, then it passes, no problems.

*Voltage saturation, i.e. Rds(on) region.  As opposed to FET [current] "saturation", the regrettably common use of this term.  Hence this disclaimer.

Not sure if your assumption of "0.45W" is a typo (that should be V).

Current flows "through" a component, and voltage "drops" across it.  Power wouldn't usually be said to flow, and one might wonder if that means power dissipated by the device instead of flowing to a load.  (This might just be language quirks, I don't know how this would be phrased in equivalent German.)

Tim

[Doctorandus_P]

Also note that the LM5069 datasheet suggests to use this IC for input voltages only up to 24V.

Where are you seeing this...?

Oops, my bad, I did not read carefully enough.

[LooseJunkHater]

It might not draw any current at all at 90V.

Check the datasheet: P6KE110A is rated for 94V nominal working voltage.  No other part is rated to handle 90V continuously, so this is the lowest part that would be acceptable.  Max breakdown is 116V, at one measly mA.  It's rated to clamp 3.9A at 152V max, with a 10/1000µs waveform (equivalent to somewhere around 500µs square pulse).  It'll last even shorter at higher currents, and needless to say, at real, practical fuse-clearing fault currents, maybe single to dozens of µs, and no fuse on Earth even notices something's wrong in that kind of time scale.

And with such high peak clamping voltage, it wouldn't even provide cross-wiring protection, say in a 120VAC context.  The controller is still just as dead; if this would be in equipment that's serviced at the board-replacement level, I wouldn't even bother putting in a TVS or crowbar for this purpose; just let the controller blow up, it'll leave a smaller burn mark than the TVS will, heh.

Whereas the P6KE75 is rated 60V nominal and 108V peak at 5.6A, still nowhere near enough to clear a fuse but enough to handle modest transients, and a larger diode would handle most e.g. automotive transients.

I knew so little about TVS diodes, so thank you for all of this! I guess a TVS won't do exactly what I thought. However, what about using a crowbar circuit to protect the LM5069? With the crowbar, it'll still blow the fuse (and quite quickly I'd assume) but not scorch the PCB?

Well, uh...

Do you want to know, then?

Yes please lol

What do you mean "drawn through"?  If it's saturated*, it's dissipating only Pd = Rds(on) * Id^2, causing temperature rise of Pd * RthJA (for whatever is between J and A in the situation: nothing, PCB, heatsink, jet of boiling freon..).  Under load-short conditions, Vds rises suddenly, Id peaks until the controller gets it back under control, and some time later, the timer runs out and it turns off.  If, through all of this, under worst-case conditions (max ambient temp,  Tj never rises above the maximum limit, then it passes, no problems.

*Voltage saturation, i.e. Rds(on) region.  As opposed to FET [current] "saturation", the regrettably common use of this term.  Hence this disclaimer.

Not sure if your assumption of "0.45W" is a typo (that should be V).

Current flows "through" a component, and voltage "drops" across it.  Power wouldn't usually be said to flow, and one might wonder if that means power dissipated by the device instead of flowing to a load.  (This might just be language quirks, I don't know how this would be phrased in equivalent German.)

Maybe I'm misunderstanding things then. At a constant drain-source of 80v, 9a, and an RDS(on) of 0.05ohm, wouldn't the FET only produce ~0.45w of heat? I'm not referring to the controller limiting gate-source voltage to increase the resistance of the FET.

[ArdWar]

Maybe I'm misunderstanding things then. At a constant drain-source of 80v, 9a, and an RDS(on) of 0.05ohm, wouldn't the FET only produce ~0.45w of heat? I'm not referring to the controller limiting gate-source voltage to increase the resistance of the FET.

Uh, if MOSFET's Vds is 80V then the load is a short. MOSFET drops 80V while conducting 9A means it will dissipate 80V * 9A = 720 Watt.

If you meant the MOSFET only conduct 9A while supplied from 80V source, working fully ON, and the bulk of the voltage is dropped by load. Then the MOSFET's Vds will be 9A * 0.05 Ω = 0.45V. It will dissipate 9A² * 0.05Ω = 4.05W

Please not that Rds_on also goes up as the device heats up, making it in reality dissipate even more power than you might think.

[T3sl4co1l]

Ok, well let's go through each one then.  A lack of research, or method to do same, seems relevant, so let's take that direction.

An SMAJ would probably be fine for the TVS, to handle hot-plug inrush, if electrolytic caps aren't doing it already; and no TVS might be required at all for lower level (say, 500V 20Ω+) IEC 61000-4-5 surges, if already handled by same.  ISO 7637-2 pulses 1 and 2a would probably want an SMCJ or better, and, a simple test model can be set up to show which size diode is required.

Keywords keywords keywords; let's go through them one at a time, I suppose.

You did say you don't know any of it!  I'm probably taking this too literally, but this too is a good lesson to be precise: if you say "not anything", I can only assume you literally don't know anything [of these].

So, let's see what's in here...

SMAJ:
Top links: Littelfuse datasheet, product page.  Also results: abbreviations (one with different capitalization, ignore that; one with medical scope, ignore that too), hits from Bourns, Diodes Inc., more suppliers, an Amphenol RF adaptor (SMA is also a type of coaxial connector, and a shorthand name for the DO-214AC SMT package hence the SMAJ family being that style), etc.

Looks like a TVS (transient voltage suppressor).

TVS:
Well, I just gave the abbreviation.  You can search if you like.  The most common type is based on the zener diode, mostly just a big enough die to handle desired peak power, and maybe some process optimizations to ensure handling the peak current.

There are other types of transient suppressors: snapback diodes (for low voltages), SIDACs (mainly for signal line surge, esp. analog phone line), and non-silicon types like MOVs (most often, mains surge), spark gaps (various), and more.

You may find old databooks a valuable collection of information.  Motorola's appnotes are quite good:
Full download: http://bitsavers.trailing-edge.com/components/motorola/_dataBooks/1991_Motorola_TVS_Zener_Device_Data.pdf
archive.org: https://archive.org/details/bitsavers_motoroladaTVSZenerDeviceData_27981216
TVS diodes haven't changed since then, they're based on the same old technology and don't have much room to optimize.  Mainly they're in all manner of SMT packages and arrays, and the chip size per package is what's improved (so a SOD-123F like SMF4L can handle as much power as used to require an SMA).

Since this type of TVS diode is based on zeners, all the information about zeners is relevant; given that TVS are generally looser tolerance (5 or 10%).  Conversely, zeners can be expected to handle some amount of peak power, at least assuming they don't fail sooner due to manufacturing defects -- if they're not rated for it, you don't know for sure, and that's the main distinction between zener and TVS product lines.

As you can see from the section discussing pulse ratings of zeners, Motorola at least was aware of this potential application and did some testing on their zener parts.  (Relevant Motorola products are now under the NXP and WeEn names.)

hot plug inrush:
ADI article on hot-swap circuits. They make parts to handle this condition automatically, so it's quite beneficial for them that they have such a high PageRank here.  Ditto TI and STM.

Some results from Stack.  We can apply the critical review process recursively here; note that Stack answers are entirely volunteer generated, so they tend to be poor quality on average, and popularity-based scoring (upvote system) is a measure of popularity, not technical accuracy or depth.

This one for example, https://electronics.stackexchange.com/questions/340202/why-does-hot-plugging-blow-stuff-up-and-how-to-prevent-it more or less introduces the idea, though Voltage Spike's plots are a bit erroneous (the voltage has to peak later than the current, in the "too low" case).

This one stops short of demonstrating an explanation, https://electronics.stackexchange.com/questions/545179/hot-plugging-power-kills-switch-mode-converter, but the reference goes on to: https://www.analog.com/media/en/technical-documentation/application-notes/an88f.pdf
etc.

electrolytic cap
...An electrolytic capacitor.

Perhaps combining with the above would be relevant, e.g. https://www.google.com/search?q=inrush+surge+electrolytic+capacitor
Results about inrush current, makes sense.  Oh hello there EEVblog forum results.

Sometimes the inrush current itself is the problem (sparking of switches, connectors; destruction of transistors).  Others it's the voltage (as in the AN88 above).  Adding "damping" to the search we find a top hit,
https://recom-power.com/en/rec-n-inrush-current--a-guide-to-the-essentials-119.html?0

IEC 61000-4-5 surge
Wikipedia: "IEC 61000-4-5 is an international standard by the International Electrotechnical Commission on surge immunity"
So of course, download links straight from the horse's mouth, which are rather expensive of course.  With some additional prodding, you likely can find "free" copies of it anyway.  Being a standard, the language is of course rather thick, opaque; but it is specific, and this one has fairly limited scope.

This hit is an alright introduction: https://www.emcstandards.co.uk/files/61000-4-5_immunity_to_surges.pdf

Hmm, there used to be a high-PageRank-ing appnote from ST, not sure why it doesn't... oh, it's not served from them for some reason, but it is in fact showing in the list,
https://emcfastpass.com/wp-content/uploads/2017/04/surge_overview.pdf

The basic building block is the 8/20 / 1.2/50 µs "combined" wave generator, an RLC network that delivers a surge wave of such timing depending on the load (short/open circuit).  The basic source impedance is 2 ohms (Vpk(oc) / Ipk(sc), not a true impedance but good enough of a hand-wave for the purpose).  So a surge of say 1kV is understandably quite a wallop (500A peak).

When I say "20Ω+", there are other standards which use the same generator but connect a resistor in series to increase its impedance.  Which will put the waveform closer to 1.2/50, but they may still call it "8/20" as a type rating rather than a descriptive one (i.e. what the exact waveform was under test).  This is common on long distance data lines, telecom, etc., basically the resistance represents the resistance of a long run of cable (various IEC, ITU, ANSI, etc. standards I don't have offhand); on railway standards such as EN 50121-3 where the surge is expected to have lighter severity; and many others.

I'm not actually sure, but I assume this is also how TVS diodes are rated for such specific currents; that, or varying the generator's voltage output, but the standard only prescribes specific voltage options and one may have to interpolate between them anyway (using resistors), if the equipment doesn't provide continuously variable output by itself.  I've not rented/operated a generator before so I don't know what's typical.

ISO 7637-2 pulses
Another standard.  ISO, ANSI, IEC, EN, BS, ITU, etc. are all industry or government standards associations.  This one specifically concerns automotive use, where inductive switching transients, contact bounce and alternator load dump can all create notable changes or excursions on the supply.
Top hit: ADI article giving LTspice models. Nice.
Links to buy the standards.
Test houses advertising capability.
Appnotes describing it.

Critical review exercise: spot the differences between appnotes and other resources, and see which ones are missing what points.

Appnotes in general are awful, so it can be a good exercise to read them carefully and figure out what's just flat-out wrong, and what's been omitted.  Comparing to the standard itself would be helpful of course, which, if you want to go down the rabbit hole, you can again probably find a copy of somewhere.

LTspice: a free circuit simulator.  Download and install and give these a try, see what the waveforms actually are; add a load, see how they change, say with capacitor, LC filter, TVS, etc. loads.  Even if you don't need these standards specifically, having example surge waveforms at all may prove illuminating.

SMCJ: SMC (DO-214AB) package TVS, you get the idea.

"simple test model": aha, we found this along the way.  LTspice is an excellent way to test parts, given a suitable surge network / generator setup.  Cross-referencing models and their measurements with standards is a good exercise.

Maybe I'm misunderstanding things then. At a constant drain-source of 80v, 9a, and an RDS(on) of 0.05ohm, wouldn't the FET only produce ~0.45w of heat? I'm not referring to the controller limiting gate-source voltage to increase the resistance of the FET.

Apply dimensional analysis: ohms times amps is volts.  Volts times amps is watts.  You need volts squared in there to get the power!

Tim