Thank you for your inputs. The DUT does in fact have to meet the requirements I mentioned above per ISO7637-2. The standard lets you test a load dump scenario according to either the "a" waveform (unclamped @ ~200V peak) or the "b" waveform (clamped, peak voltage to be defined by manufacturer). The particular project I am talking about is for a paying customer and they will ultimately make the call whether they think "a" or "b" is appropriate. (A little weight off my shoulders.)
This thread is mainly for my own edification, and possibly to help answer any questions the customer may have if they are not fully educated on the matter. (Fine line to dance on there.) The end-use application is fairly critical, but it's still reasonably high volume with cost constraints, so I thought it reasonable to think twice before throwing $5-6 in chunky load dump TVS diodes at it.
Well, if you have to meet the standard you posted, then that's one less decision for you! I read the IEC60601 spec that you posted--it appears this is a medical-related device in a 24-volt system, like an ambulance or something? That's one of the particular electrical 'horror shows' that I mentioned. I don't know how much power it draws or the typical way of doing this, but it seems to me that for anything using significant power the safest bet would be a regulator that can simply withstand the overvoltage on a continuous-except-for-thermal basis. In addition to not blowing up your device, you also don't want it to damage anything else by being a huge current sink during the surge.
Indeed, it is for ambulance use. The power level is 100-120W. The DUT already uses a LT4356 protection IC, which allows ride-through up to 80V for ~100ms, 60V for ~300ms, and will cut the power completely if the fault/surge lasts longer than those durations. In this way, I believe it to already be fairly robust and will not act as a huge current sink for a large part of the load dump waveform. But due to the Vds(max) rating of the pass MOSFET (100V), can't guarantee it will survive a surge >100V. Of course, it seems simple: Just use a 200-250V MOSFET! The issue is that the additional losses incurred by the increase in Rds(on), due to increase in Vds(max), will not be acceptable during normal operation.
So, to recap, the input protection circuit can handle the bottom "half" (24 - 100V) of a load dump surge waveform via linear regulation, but not the top "half" (100 - 200V). To handle the top half, it seemed reasonable to clamp that via parallel element. Yes, it will cause a current surge, but not as bad as if the entire waveform was clamped via TVS. Still, it costs money, and if vehicles are not really making these kinds of transients, then that is probably over-engineering. The first revision of ISO7637-2 is from 1990, and it is conceivable that the levels were selected based on the types of vehicles that were on the road at the time. But that's only a hunch, which is why I am seeking
manufacturer's standards, and not the ISO type standards.
There are standards for construction of ambulances, see EN1789, but I don't think it is terribly explicit on all details, and who knows about adherence country to country.
As I mentioned, the decision is ultimately in the customer's hands, but I want to be informed on the subject enough to know if their decision ultimately feels right, or if it will be a regret later.
Thanks!