When I was testing this back in May (discussed in separate thread), I eventually realized the 5GHz shutdown always (and only) occurred within minutes of a particular 80MHz-capable client (a Lenovo laptop) being powered up. While that isn't 100% conclusive, it did lead me to suspect the problem was triggered during "negotiation" of 80MHz operation by that client -- or perhaps the switching between 40MHz and 80MHz operation, or operation of that client in conjunction with other clients. In any case, once I found that setting the router to 40MHz eliminated the issue, I stopped testing further because I simply don't need 80MHz bandwidth.
My point here is that someone without any 80MHz-capable clients may never see the problem, regardless of the setting. I could also speculate it may be limited to specific brand clients or client chipsets/drivers, i.e., something the client driver does incorrectly in negotiation and the router or router's wireless driver code doesn't properly protect against that illegitimate client request. In that event, people without clients that use that particular client chipset/driver would never see a problem even with the 20/40/80 or 80 setting.
All I know for sure is that with my AC88U, across multiple versions of stock and merlin firmware, I never got more than 6 days without a complete 5GHz shutdown when set to 80 or 20/40/80 bandwidth. After setting it to fixed 40MHz, I haven't had a single 5GHz problem of any kind -- even when the router has been up, untouched, for periods exceeding 60 days.
I'm very happy with my AC88U despite having to turn off 80MHz to get stable operation. In all other respects, its great. But if I really needed/wanted 80MHz operation, I would have had to return it and get a different model -- because (at least with my client set) it was not stable at the 20/40/80 or 80 settings.