that's just a weak raid controller. it's very common to replace 1 drive at a time and have it rebuild then you can expand your raid. if different drives didn't work - this would be impossible. been doing this for years with hp's raid controllers (with BBWC/FBWC) and LSI controllers - works great. heck i had a old scsi drive fail and i popped in a 146gb because it was cheaper than trying to find a new 36gb and it's 36/36/146gb - no problems.
cheap soft-raid controllers - problems. but if you don't spend $50 on a LSI controller (like the br10i i have for sale with 8 ports using a fan-out cable included in deal) - then well you shouldn't be playing with raid.
and of course use raid-edition drives

you should know by now that only a few folks have written custom drivers to handle consumer drives (qnap/drobo) - the free raid solutions and hardware raid don't tolerate this. qnap abstracts the error correction to their own solution so if a AV drive (tler=0) and a raid edition drive (TLER=8) and a consumer drive (30-180 seconds) is handled well with the drivers then you don't have that problem - but you must tune your o/s to handle long timeouts. iscsi is pretty cool because you can withstand 10 minutes of timeout and continue in most operating systems - some apps might freak out (sql) but the system won't BSOD. slick.
so you get what you pay for still counts here.