Yup, right in the controllers themselves...
See, I read recently that SCSI stuff is designed for WAY longer "continuous operations" in that page as a matter of fact!
Now, that seemed to contradict what I read in the past the essentially the drives platters are the same, bearings & all, as well as the actuator arms & read/write heads...
That sort of had me wondering... things CAN & DO change alot in this field over time, & what I read about them being pretty much mechanically the same for the most part (like drive bearings on the spindle etc.) could be wrong nowadays.
There's a trickle down effect from SCSI to IDE. Things like the fluid bearing are in that category. You must remember that first line SCSIs are spinning at twice the rate of a good IDE, so, in that respect the SCSIs are built 'better' than their IDE counterparts, but, a lot of it is just using older technology for the IDE disks. Eventually we see the same things go to IDE. I can't believe that their is much difference in the QC or any difference in the clean rooms. Still, all of us will agree that SCSI is
the good stuff.
Yes, I agree... I had one in "those old days":
A TekRam DC-600 ISA slot IDE Caching Controller with 4mb of 30-pin FastPage RAM I had added to it... & it was awesome!
(With it my old WD 212mb Caviar IDE disks outran even Vesa Local Bus controlled drives with their wider TRUE local bus (moreso than PCI, VESA was supposedly a truer implementation of a real "local bus", another factoid hanging around here in memory)).
I too, wish such things were still made for IDE/EIDE to be honest, it did make a diff. in the DOS/Win3.x 486 & early Pentium I days for disk performance!
*

APK
Actually I have one of the promise ones (8mb) in a 486-50 that I'm setting up as a pure dos box ... fooling with DR DOS and such

.