On Sun, Mar 13, 2016 at 3:34 PM, Jules Richardson <jules.richardso...@gmail.com> wrote: > Assuming for the moment that there were completely disabled surfaces > involved due to too many defects (i.e. that this was a 2-platter drive with > only one or maybe two active surfaces and the rest mapped out by the drive > firmware), do you know how common a practice that was in the early days of > ST506/412-type drives? > > I've heard various stories of it happening much later with more modern IDE > drives in the multi-GB range (e.g. a 20GB drive might be a 40GB > single-platter drive with one surface that had completely failed QC), but I > don't know if it was ever done in the ST506/412 era.
Often when components and assemblies are sold in a down-spec configuration, it isn't because anything failed test, but rather because they didn't take the time to do the testing on that portion. Testing a product is a significant part of the manufacturing cost, so there's actually significant cost savings associated with only testing a portion. This is often the case with memory chips and multicore processor chips; I don't know that it was done with hard drives, but manufacturing test time for hard drives was definitely expensive. On the other hand, I'm told that in mass production it is extremely uncommon for a side of a disk platter to have more than the allowable number of defects, so it seems more likely that a downspec hard drive would simply have fewer platters (and heads, etc.) installed, rather than having some disabled for excessive defects, or some untested.
