Any time a HD drops out of a raid0 configuration, the whole of the raid0 configurations is failed.
Any drive that fails, you lose almost all of the data. only data that fits under the stripe size of what is left is recoverable. That means small files like text files.
If a drive goes to sleep, and does not wake up in a timely manner, a raid card will see it as lost, and the whole RAID0 array is then inaccessable. How the raid device and software reacts varies, with the more Pro raid cards (expecting a raid drive) having the biggest problems.
If a drive has a few minor errors, and the errors require that the drive move data from bad sectors to good sectors or reserved sectors, it also takes time for the drive to complete this task. Because the drive is not responding in a timely manner the raid card can deem it to be failing completly, even if you can wait till it finishes, reboot (as needed) and have the array operational again.
Any drive reguardless of sleep or error, or tler, could potentially not respond in general, or be very slow to respond when it is simply failing, in the many ways that a drive can fail. A drive that is simply failing, will usually get way worse over a short time.
Drives that are designed for Raid (general) can be set so they never allow sleep. Also they have a reduced time allowed for responce when correcting an internal error or moving a bad sector around. See TLER for more info http://en.wikipedia.org/wiki/TLER
Good raid cards are way more sensitive to this delay in responce time, and really cannot be used with normal desktop drives at all (mostly). OS Software raid and cheaper raid systems that are more software based are more lenient, and do not have a fit as quickly when a disk does not respond immediatly.
If your using a good hardware raid card, you really need totally raid compatable drives to go with it. Even then any of the drives could have a problem, and knock a drive out leaving you with 3/4 of each of the files, which is None of it :-)
If a single drive in the array is having problems, you could check that drive with SMART testing and see if it is gaining a large ammount of bad sectors. IMO(that is my opinion) when you assemble a raid system the drives should match in age, type, speed, quality.
If I had the problem you are having and the drives are old and have been spun up for long, I would replace all 4 :-( . I also would go with less risk and make 2 teams of 2 disks, not 4. Always have backups that are not raid based, and do not require the raid hardware itself.