@Gorgonops Does the article just I backed up here make anything clearer to you, so you can translate it for us?
So, I can break down a few segments of it:
You can use third-party IDE drives in Apple Products as long as the IDE drive supports the identify command,
The "Identify" command is an IDE command set feature that when triggered makes the drive spit out a data structure that contains (in pretty gruesome detail compressed into a massive bitfield) a summary of what the attached drive is (vendor, capacity, etc) what data transfer modes it supports, and what extended command sets has available. Very, very old IDE drives don't support it; in essence their controllers just emulate a Western Digital WD-1003 MFM controller (the one used in the original AT) with a hard drive permanently attached and it's up to the user to set the BIOS to the right capacity.
My editorial take: It would not surprise me at all if a SATA drive connected to one of those bridges produces an "Identify" bitfield that contains data the driver in the 6500's ROM doesn't understand and freaks it out.
works at least at PIO mode 2 performance level
Here's a page about IDE PIO modes. "PIO" means "Programmed I/O", which roughly speaking describes the situation in which the CPU has to be actively involved in conveying each individual byte/word through the bus connector. (Verses DMA, in which the controller on the hard drive and a DMA controller on the computer's side automatically handle stuffing a discrete packet of data to/from the bus directly to RAM.) The number in PIO0-4 is an indicator of the bus cycle time. Mode 2 is the fastest that's supported by an 8mhz (that's the standard) AT bus; Mode 0 is IBM XT compatible and Mode 1 covers some other edge cases. Only *very* old IDE drives can't handle PIO mode 2; CF should support at a minimum Mode 2 because it was originally targeted at PCMCIA performance levels which correspond with the 8mhz AT bus.
has write caching turned on so that auto-reallocation of new spares is invoked.
Again, this is something that would only be a problem on *very* old drives.
The highest performance setting PIO mode the Macintosh Performa and Power Macintosh 5400 and 6400 series are capable of supporting is PIO Mode 4.
According to the developer note the 5/6(4/5)00 also support the
Multiword DMA modes defined in the ATA-2 standard along with the PIO modes.
Again, and this is just me speculating, I wonder if the fact that those machines try to enable DMA is why the SATA bridges are reported to work with 6(2/3)00s; perhaps the bridges work fine falling back to slower PIO modes but something goes wrong trying to enable these very old pre-UDMA settings. (IE, perhaps the bridges don't actually support them properly but the data in the IDENTIFY device structure misleads the 6500 into thinking they do.)
Most importantly, any IDE hard drive that is going to be used on these systems must support Logical Block Addressing (LBA). LBA is the method that drive manufacturers have adopted in order to go beyond the 520 MB DOS-BIOS limit
Extremely old hard drives (IE, the aforementioned ones that strictly emulated a WD1003) also only emulated the WD1003's method of accessing a location on the disk, IE, the location of a given chunk of data was described in three dimensional terms of Cylinder, Head, and Sector. (Often on these early drives the CHS values the drive accepted actually had no relation whatsover to the actual geometry of the hard drive attached to the controller, they just faked things up to the maximum capacity of the WD1003's registers to cram as much space as possible into the limited addressing possibilites.)
LBA is when you simply describe the data space on the drive as one long string of blocks going from 0 to the end of the hard drive. SCSI used LBA from day one (although technically it *does* allow CHS terminology to be used for backwards compatibility to SASI) and most IDE drives made after 1990 or so support it as well. So, again, this is something that would only be a problem on a *very* old drive.
(And almost certainly not a problem with a SATA bridge; if anything I suspect the SATA bridge would be a problem if you plugged it into a machine so old it tried to use CHS. CompactFlash cards are LBA but *some* also support fallback to CHS; often that's limited to devices smaller than 512 megabytes.)
Anyway, in short, I don't think there's anything here that tells you why the SATA bridges *don't* work. If the bridges supported perfect fallback to ATA2-spec modes then they should work, but I don't know how perfectly they do that. My guess is they *don't* do it perfectly because they're mostly aimed at machines that support ATA4 and therefore there's *something* about them (again, my guess is on the DMA modes) that gives the 6500 fits.
not a single problem! Heaven!
