Nice work figuring out it's the RAID at fault! Definitely not the first thing I would think would be the issue. In my experience SMART as a pass/fail is pretty poor. Usually I know a drive has serious issues before SMART claims it's failing. However, individual SMART attributes are really good to look at especially when watching trends! I've had some experience with drives developing bad blocks which get marked as pending before actually failing. Performance suffers when accessing them but the dumb drive won't remap the sectors until they are written to. The worst is when the pending blocks are part of the filesystem and not an actual file with performance taking minutes to access files... Bleck.
Check out Hard Disk Sentinel. There are other SMART monitoring tools including a number of free ones, but that's the only program I know that handles USB external drives. Useful to know if your external backup drive is having issues
Anyway, grab a trial or something and see if any of the raw smart DATA varies between the drives that look good and those that don't. Should be very similar data since they are RAIDed drives.
I usually use WD Green drives as internals for bulk storage even though their performance isn't the greatest since they run cool. However WD intentionally makes those drives have huge bad block timeouts. They do that so people buying lots of drives for RAIDs will buy WD Red's (same drive, more money, small timeout). Seems like that should just be a jumper option to me or something. I'm not sure how your RAID handles a single drive not responding because of bad blocks. Some drop the drive from the array, other work around it for a while, etc. I'm sort of surprised you won't see warning messages from the RAID controller somewhere though.