Hard Disk Confusion

[NeXT]

I'm assembling a late revision 5150 from parts and for the last 12 years or so I've had an IBM hard disk and controller hiding in storage for this day.
The controller is a 61-031099-00 floppy + hard disk controller. This that I can tell is the 16-bit board for a 570 and not a 5150, so that's not gonna work because it's 16-bit ISA.

The drive on the other hand I'm a bit confused about. MinusZeroDegrees says it's a Type 20 30mb 0665-38 30mb (MFM?) drive.
Google however says that 102485-ESP125-0665-38 is a 70mb ESDI disk. That makes no sense since I know these both came out of the same machine. This is a 30mb drive, right?

 

[krebizfan]

I don't recall IBM ever building an ESDI 16-bit ISA controller so if the hard drive was paired with it, the hard drive can't be an ESDI drive. Could you provide a picture of the label?

 

[modem7]

The controller is a 61-031099-00 floppy + hard disk controller. This that I can tell is the 16-bit board for a 570 and not a 5150, so that's not gonna work because it's 16-bit ISA.

Correct. The first generation, as shown at [here]. MFM.

This is a 30mb drive, right?

Yes, the -38 is the unformatted capacity. MFM.

IBM type 20. Per [here], expect to see a large "20" printed on the defect track label.

 

[bear]

MinusZeroDegrees says it's a Type 20 30mb 0665-38 30mb (MFM?) drive. Google however says that 102485-ESP125-0665-38 is a 70mb ESDI disk.

MinusZeroDegrees is correct. Google can kick rocks. The ESDI disks are 0667. IBM's 16-bit ISA ESDI controllers are for the RT.

 

[NeXT]

Right, sticking with that then.
Now to go and dig through my stack of hard disk controllers for an 8-bit card that will support it in a PC. Never actually spent any time looking this up but I am assuming the same requirements for adding a hard drive to a 5160 are the same (The card needs a BIOS as the computer does not provide one and the controller itself may have its own limitations on what drives it will work with, such as my WDXT-GEN with the F300 firmware will only support an ST-225) for a 5150 but 30mb is a strange one. I'm used to that capacity for RLL disks.

 

[Eudimorphodon]

A number of later XT controllers supported a “custom drive type” option in addition to whatever was hardcoded into the ROM, so supporting any arbitrary geometry is possible if you can scrape one up. (Looks like the WD1002A and S supported it?) So you just need to figure out what “Type 20“ means in terms of heads and cylinders.

One thing I vaguely recall about those auto configure controllers is they did something like waste track 0 to hold the config info and faked out to DOS that track 1 was track 0? Supposedly that could cause issues/disaster with some low-level hard disk utilities.

 

[NeXT]

So you just need to figure out what “Type 20“ means in terms of heads and cylinders.

The 0665 used in the IBM AT was 30 MB is size and a type 20 (733 cyl, 5 heads, WPC of 300, 17 SPT --> 30 MB).
IBM put a white sticker on them that contained the defect track list and the number "20".
Therefore, the presence of that "20" on your 0665 will indicate the 30 MB version.

Source on that: https://forum.vcfed.org/index.php?threads/ibm-hard-drive-type-0665.16425/post-223751

I know I have both of the controllers you specified but I need to double check if they have ROMs installed. I'm so used to using the LLF routine in the ROM.

 

[Eudimorphodon]

I'd think those controllers would be pretty worthless without the ROM, unless it was some OEM'ed version that was shipped with a proprietary XT clone that had the ROM support for it on its motherboard.

Anyway, the manual for the 1002S is on the usual suspect, you can hop into it with the usual DEBUG command and do the LLF as usual, it just has a simple menu that lets you input the geometry if you have the jumper set for autoconfig. (Autoconfig is of course kind of a misnomer here, it's not like IDE autoconfig where it sends the IDENTIFY command to the drive and gets the geometry automagically. It's "auto" in that it "automatically" on power-up uses the geometry you told it to use during format as opposed to, I dunno, asking you every time?*)

Edit: I guess it's also "auto" in that if you had two different drives formatted with these controllers and swapped them between machines they'd "automatically" know about the swap and configure themselves accordingly, since the config data *is* stored on the drive post-format, vs. living in NVRAM on the controller.

 

[modem7]

(Autoconfig is of course kind of a misnomer here, it's not like IDE autoconfig where it sends the IDENTIFY command to the drive and gets the geometry automagically.

Definitely misleading, which is why I always use the other term for it, 'dynamic', as I describe at [here].

One thing I vaguely recall about those auto configure controllers is they did something like waste track 0 to hold the config info and faked out to DOS that track 1 was track 0? Supposedly that could cause issues/disaster with some low-level hard disk utilities.

I encountered that with SpeedStor's media analysis functionality, recording it all (including workaround), at [here].

... such as my WDXT-GEN with the F300 firmware will only support an ST-225

Western Digital's F300 BIOS, a.k.a. Super BIOS, has dynamic formatting ability.
The following is from the user's guide for the WDXT-GEN, available at [here].

 

[Eudimorphodon]

Source on that:

Interesting that "type 20" is listed as having 5 heads. That implies it's a voice coil drive with three platters and one surface providing the "dedicated servo" information. It'll be interesting to see if that positioning data is still good 40 years later; if it's faded out the drive's a doorstop even if the mechanical parts are perfectly fine.

(* Fading/corrupted servo information is what's going to eventually kill all modern hard disks, of course, since it can't be rewritten without a special machine.)

 

[Svenska]

One thing I vaguely recall about those auto configure controllers is they did something like waste track 0 to hold the config info and faked out to DOS that track 1 was track 0? Supposedly that could cause issues/disaster with some low-level hard disk utilities.

Storing configuration on the disk itself is somewhat similar to modern SD cards and SSDs, they often store their own firmware on the flash they are exposing to the system. Which is the reason why some SD cards tend to fail rather violently when surprised (such as turning off power without physical ejecting). Many Raspberry Pi owners learned the hard way that quality differs wildly.

 

[NeXT]

Western Digital's F300 BIOS, a.k.a. Super BIOS, has dynamic formatting ability.
The following is from the user's guide for the WDXT-GEN, available at [here].

View attachment 1266717

That suddenly changes things. That PDF lists the F300 BIOS has support for the Microscience HH-725A 20mb disk. I bought one of those at Midwest and it's been sitting here with a shorted rail (a tantalum). I can save the IBM drive and keep it with its original cables and controller then for if I ever get a 5170 and pair the controller with the Microscience instead.