How about the LTC4365 from Linear Tech?
This IC and similar ones would still require a TVS, in this case a bidirectional one clamping at >-40V so we're back to step 1. Several other surge stoppers from LT say they don't normally need TVS but ISO 7637 pulses say otherwise. I'm not sure on how they plan on dealing with the most demanding -600V pulses... maybe input filtering should take care of them?
It's not just load dumps, it's all the other low energy crap there is that must also be stopped.
To summarize:
-My device has a nominal 24V input, maximum current in very worst case is 10A (probably it could be reduced to less than 5A).
-This input line needs to be reverse-polarity protection (-28V for an indefinite amount of time)
-Working voltage range is 10V to 32 V DC
-There are devices connected to this line, which are the ones who get the most current, that have an absolute maximum supply voltage of 40V
-So any input protection must clamp to +40VDC or less in any circumstance
-Must handle pulses defined in ISO 7637-2 and load dump conditions in ISO16750-2. That means -600V pulses and such.
-Must be cheap-ish
-Can just disconnect itself during transient
My current options:
-Reverse polarity with a power diode with >600V reverse voltage rating + load-disconnect circuit from
http://www.ti.com/lit/an/snva717/snva717.pdf adapted to 24V
-Pros: cheap-ish, simulations say it's realiable
-Cons: many parts, needs HV PMOS
-Surge-stopper IC's from linear or maxim
-Pros: integrated, uses NMOS
-Cons: i think they'd still need TVS even if they say they don't. They say you can absorb those low energy pulses with capacitors and such but you do the math and simulate according to ISO test procedures and nope.