Since we are talking about PSUs, for me the more expensive the PSU the higher the failure rate.
Most of the larger >1000 Watt PSU are all modular, and almost 25% of them have had a burnt +12V wire on the modular side of PSU, very difficult to fix properly.
My cheapest PSUs the Corsair CX750W the ones we all bought during our Litecoin days, never had any issues. No wires were burnt or fried PSUs.
for me, i have solved that modular problem from being burned with this..
- always use two lines per GPU instead of using both 8pin (8+8) from a single line.
- one sata or pata per riser
and make sure all connections are tight-standard in all electric related stuff..
if modular connection is ruined and if warranty is not an option, I solder a new connection...or if the board where it was soldered don't have a good soldering point due to burning...i trace the wire where that soldering point (board) is getting the 12v source and tap in the wire and solder it. normally the modular part is just a board which is getting it's volts and grounds form a bundle of wires behind that board.
EDIT:
I agree on expensive PSUs being crap but only up to the point where the manufacturer like corsair can inject a program to the PSU and make safety triggers like it turns off if a spike or the load is kinda heavy...in my case ax1500i is increasing sensitivity when it is going above 1000w..
that ax1500i really has miners in mind haha that prick PSU even wanted to be turned off and on by its switch...because when i traveled abroad i asked a person to just turn off the electric breaker and turn it on (that person is not a tech kinda guy)...and that PSU just stayed off like nothing happened and when i got home i switch that PSU off and on and then it resumed mining.....corsair ax1500i engineers fuck you...there you go