I am currently singing "It's a miracle! a mirrrracle!" because I was finally able to boot my A2000 and run stuff for the first time in months.
During the last episode of this saga, after putting much work into getting the system to power on without a blinking power LED, the GVP A3001 decided it was going to stop recognizing the IDE drive. So my next plan of attack was to grab a cheap PCI 50-pin SCSI interface and try to figure out why the Amiga was also not booting from its old Barracuda SCSI drive attached to the GVP HC+8.
I got such a card, attached the 'cuda to a PC and dumped the whole disk. Turns out this was actually a 4.3GB, but only 2GB was allocated. And the disk geometry in the RDB was awfully strange, it was setup with 424 cylinders, 1 head, and over 9000 sectors. I found this puzzling but when I mounted the disk image in WinUAE it booted fine. Whereas the A2000 would show a bunch of disk activity and then just freeze at a black screen.
I tried mucking about with jumpers on the A3001, taking RAM out and putting it in. No change.
I found the manual for the 'cuda online and skimmed through it. I saw something about the drive "remembering" the geometry it's formatted with. So I was thinking, maybe if I reformat I can change the geometry? And then write a new disk image from MakeHDF? So I was fiddling around in the adaptec BIOS utility and selected the format option. Bad move, as that appears to have bricked the drive.
Tried the IDE again just for the heck of it. It decided to recognize the drive again, booted once, and then froze. After that it would only start booting and then say "DH0 is not a DOS disk." I was about ready to kill it all with fire by that point and gave up.
Later I had another idea. I read somewhere that the poor compatibility of the GVP A3001 IDE interface is because of timing problems. So I thought, maybe I can underclock the card and that will affect the IDE timing? I popped out the 50MHz crystal and put in a 24MHz one. That didn't fix the problem. I was checking to see if my ROM programmer supports 16-bit ROMs which I had never messed with before, because I was thinking about burning an Amiga diagnostic ROM I saw online. Then while I was putting some junk away I saw the Quantum LPS270. This was the first HDD I used in the Amiga 2000, before the 'cuda. I believed it to be dead but figured I'd connect it to the Adaptec PCI card and try to dump some sectors. Well, I dumped the entire thing, without any clunking. Wasn't this dead? I think I may have mixed it up in my brain with another SCSI Quantum drive from an old Mac which died.
I put the LPS270 in the Amiga and lo and behold, a workbench screen! What do I do now?!?!? I found the "memcheck" utility which was already on there and ran it. No errors. OK, so next I tried enabling the RAM on the HC+8, but that still doesn't work. Just leads to a blinking power LED. I disabled that again, and reinstalled the 32-bit RAM for the A3001. Booted and ran memcheck again, no errors. Then I ran AIBB. Turns out that by playing with jumpers earlier I had gotten the card into some half speed mode. The '030 was at 12MHz and the '882 at 24MHz. After that I reinstalled the 50MHz crystal, ran memcheck, ran AIBB, now at 25/50MHz. The documentation I found online for the A3001 jumper settings isn't entirely accurate, aside from the half clock mode, JP5 on the RAM32 board apparently determines whether the memory is mapped at 2MB or 32MB.
TLDR, it works (for now...)