Have not messed with MFM/RLL drives in about 10 years but suggest you set the hd in 386 bios to none. Turn off and then re-start.
See if it stops clicking after a few seconds, it is common for some to click a bit as the controller gets the drive parms off the first tracks as written during the Low Level format.
Is possible that there is media issue in those tracks and controller can not read the drive parms to set up or it was setup different than actual native parms....
If it does stop clicking use DOS debug (-g=offset:5) and check the menu to see what controller thinks it was last LL formatted as, it should report it.. do not initiate a LL.
Sorry is all I can offer as am going from mostly long lost memories....will go dig out one my old hardware books and see if it gives any pointers...

douglar wrote on 2024-12-17, 01:17:

I've got an ST-138R Hard Drive

Is it a known good drive, or a find? Chances are it was retired due to damage and it's just toast at this point. Back in the day these Seagate drives were cheap and worked fine, but long term reliability was not their strong point it seems.
If you take the HDD in your hand and rotate it at various angles, does anything inside make a faint rattling noise? If so that would be a ripped off loose head. Althouth I do not hear any typical sound of the arm scraping the surface in your sample, but it's pretty short.

It might also be something more simple - wrong controller. Just because it's an R drive and RLL controller, and the cables were connected, doesn't mean it's actually formatted to RLL. Somebody might have just run out of time trying to set it up, or gave up with similar issues, and that's how you got it. Might be mis-configuration, these WD controllers have settings for different CRC/ECC modes (although these are usually factory pre-set). Some of these WD cards are also both MFM and RLL capable but set to MFM permanently for some reason. So try 17 sectors in BIOS as well to rule that out.

If you _really_ want to read the data, assuming the HDD is not broken, there might be other ways.

EDIT: And I'm going to ask the obvious question: Are you sure you connected the HDD 34-pin ribbon cable to the correct header?

Horun wrote on 2024-12-17, 02:56:

Have not messed with MFM/RLL drives in about 10 years but suggest you set the hd in 386 bios to none. Turn off and then re-start […]
Show full quote

Have not messed with MFM/RLL drives in about 10 years but suggest you set the hd in 386 bios to none. Turn off and then re-start.
See if it stops clicking after a few seconds, it is common for some to click a bit as the controller gets the drive parms off the first tracks as written during the Low Level format.
Is possible that there is media issue in those tracks and controller can not read the drive parms to set up or it was setup different than actual native parms....
If it does stop clicking use DOS debug (-g=offset:5) and check the menu to see what controller thinks it was last LL formatted as, it should report it.. do not initiate a LL.
Sorry is all I can offer as am going from mostly long lost memories....will go dig out one my old hardware books and see if it gives any pointers...

It is supposed to be as a good drive and it sounds like a good drive when it powers up, but I made the mistake of disconnecting the external battery before testing, so I lost the CMOS settings.

It doesn't click until the BIOS tries to read the drive, so if I don't configure it in the drive table, it sounds fine. I checked the cable orientation more times than I can count. Getting cable orientation reversed is a common mistake for me, but I was extra careful on this one since it was not IDE. I'm also using a floppy cable, and I didn't get the 34pin cables mixed up, yet.

I did try a different controller that was the same make and model for one boot and it did change the intensity and rhythm of the clicking permanently. Hopefully that didn't scramble the drive.

Moderately optimistic that the DOS debug (-g=offset:5) will help me out.

douglar wrote on 2024-12-17, 13:29:

Moderately optimistic that the DOS debug (-g=offset:5) will help me out.

Be very careful with that code, anything you don't like and it's better to hard reset or power cycle the machine than risk accidental reformat. There might be a verify option in addition to format but read the manual first to make sure it's non-destructive.

BTW did you try setting sectors to 17? Some BIOSes require it, the controller might be fixing it to 26 just before boot. Also figure out if you can bypass the HDD error and boot from floppy. If so try to dump the first sector of the HDD to floppy for analysis - the partition table would tell you a lot about how the disk was set in BIOS. Norton utilities should have a diskedit program, it can access HDD at physical sector level and is smart enough to show you the partition table (and in fact edit it too) and not just the binary dump of the data,

The small number of MFM and RLL I had back in the day, it was the controller that determined MFM or RLL, you could low level format either type of drive either way, but you'd gain 30% capacity of an MFM formatted to RLL or lose 30% capacity of an RLL formatted to MFM... but yeah, low level formatted to one type you couldn't use it with the other.

Not seen it specifically mentioned and it is something modern users ignore forget or don't know about, so you do have the skinny control cable hooked up as well as the data cable right? ... pin count has walked out of my head for a smoke break or something, it's 10 or 12 pin maybe.

As the wd-1006 is an AT controller in an AT-like system, that controller behaves like an IDE drive (actually, this is backwards: IDE drives behave like the WD1006). Thus, the controller is supported by the Mainboard BIOS and does not have its own BIOS that can respond to g=C800:5. If one would want to reformat the drive, check for an option like "format hard drive" or "HDD utility" in/next to the CMOS setup.

In contrast to XT RLL controllers like the WD1002 that had a custom microcontroller that translates Xebec commands into WD commands and could emulate a 17-sector geometry on RLL drives, the WD1006 can not perform a translation like that. You need to select a geometry with 25 or 26 sectors per track in the BIOS setup for RLL drives.

Periodic clicking sounds like error recovery attempts if reading a sector failed. If multiple read attempts fail, issuing a recalibration command is standard practice, and recalibration usually ticks. Assuming that both cables are properly connected, you should be able to read the MBR (sector 1, head 0, track 0), as long as the hardware is OK, no matter what drive type is set in the CMOS.

My next attempt would be to boot DOS from a floppy and check whether fdisk shows partitions. If not, the hardware is incapable of reading the MBR.

Hmm the WD manual for WD1006V-SR2 does say to use debug -g=ccOO :5 to enter board bios. Mueller's Upgrading and Repairing PCs 1st Edition also mentions the Western Digital 1006V-SR2 having a on board rom, page 404-405. The 3rd edition has some better hints as to ST-506/412 cabling including pinouts, etc....
a hint from someone with same controller: "(e.g. WD1006V-SR2). There you set the normal BIOS type to 1 and the card's BIOS takes over." Can not hurt to try it...

added: ok not all have the on board bios:
Feature F301R includes an optional ROM BIOS that allows the user to define the drive's parameters.
Feature 300R does not include the ROM BIOS and you must use the drive tables on your system's ROM BIOS that must contain the appropriate drive parameters.

Horun wrote on 2024-12-17, 18:04:

Hmm the WD manual for WD1006V-SR2 does say to use debug -g=ccOO :5 to enter board bios.

...

added: ok not all have the on board bios:

I stand corrected. Looking at the pictures uploaded by douglar to the Retro Web, douglar indeed has the version with the onboard ROM, which can be enabled/disabled using jumper W2. Generally, a BIOS is not required on WD 16-bit AT MFM/RLL controllers, but as certain early AT-type BIOSes only support 17 sector per track geometries, a BIOS extension on the RLL controller to add RLL geometries makes sense.

I was able to get into the WD 1006v-sr2 bios, but it only shows what’s in the regular drive bios table, except for #47, which just shows zeros.

I tried 17 sectors, no luck. Tried 25 sectors in case it started at 0, no luck.

Curiously I get different clicking patterns when I try different write pre-comp values.

douglar wrote on 2024-12-18, 04:24:

I was able to get into the WD 1006v-sr2 bios, but it only shows what’s in the regular drive bios table, except for #47, which just shows zeros.

There is no way the controller can identify the drive so it can only show the BIOS-set values. Though it's a bit odd it shows only zeros on the user-defined preset.

douglar wrote on 2024-12-18, 04:24:

I tried 17 sectors, no luck. Tried 25 sectors in case it started at 0, no luck.

Pity. Though sectors can't start at zero on PC-compatible machines (*) and the 25 setting is for cases where controller formats each track at 25+1 where the +1 is a spare sector. This however doesn't work automatically, any remapping would need software that detects and actually moves damaged sectors to spares. IMHO a waste of space, much better to just have a bad sector marked in FAT if/when it happens rather then waste so much space. Plus it's only one sector per track. Maybe useful for weird, old OSes that can't mark sectors as bad in software? Not sure about CPM and clones now.

(*) Well actually on the surface the sector headers can start the numbering from 0 instead of 1, but the chip will then add 1 when decoding the address and that's what will be reported to software via registers.

douglar wrote on 2024-12-18, 04:24:

Curiously I get different clicking patterns when I try different write pre-comp values.

That is even more odd. Precomp affects only writes. It has exactly zero effect on read because the bitstream is locked onto by PLL, no matter if precomp was used during write or not. In fact properly used precomp will make the flux reversal spacing more uniform, that is the whole point of it.

Some MFM/RLL controllers reserve one track for storing disk format parameters, it would usually be the last track. Not sure about WD but if it does that, and can't now read the data, it would explain the clicking and error (not the precomp though). Try with 26 sectors but different cylinder count. Try 616 as the last cylinder on Seagates is a parking space but is actually usable. Try 612 as this is what some disks had as well. Perhaps try all the values 610 up to 616. No need to test past 616, these particular drives will count the steps and won't let you go futher (in fact trying to go past 616 will cause the drive to recalibrate and reset to track 0).

It is odd that precomp effects reading the drive..
Just a question: you are using a real mfm/rll 34pin cable ? Floppy cables have diff end pins twisted and cannot be used in the same way as a real mfm control cable.
Is also possible the end part of cable is bad. As a test you could move the drive from the end and put in middle but change ID jumper on drive to 1 not 2.

Horun wrote on 2024-12-18, 16:58:

Just a question: you are using a real mfm/rll 34pin cable ? Floppy cables have diff end pins twisted and cannot be used in the same way as a real mfm control cable.

OK, that sounds like something I would do. I might have swapped the floppy cable and the MFM cable when reassembling the storage. First thing on the "to check" list when I get home.

douglar wrote on 2024-12-18, 18:14:

Horun wrote on 2024-12-18, 16:58:

Just a question: you are using a real mfm/rll 34pin cable ? Floppy cables have diff end pins twisted and cannot be used in the same way as a real mfm control cable.

OK, that sounds like something I would do. I might have swapped the floppy cable and the MFM cable when reassembling the storage. First thing on the "to check" list when I get home.

Have done it my self. Easily confused with the older 360k/1.2mb only floppy cables if cables get misplaced.
Just looking at the cables: Floppy has the 10-16th wires twisted starting at #1/red line. MFM/rll/esdi has the 25-29 wires from #1/red line twisted (or 6 to 10 from non-red line #34)....
The old rule is if twisted near #1 is floppy if near #34 then mfm/esdi...

Horun wrote on 2024-12-18, 21:29:

Have done it my self. Easily confused with the older 360k/1.2mb only floppy cables if cables get misplaced.
Just looking at the cables: Floppy has the 10-16th wires twisted starting at #1/red line. MFM/rll/esdi has the 25-29 wires from #1/red line twisted (or 6 to 10 from non-red line #34)....
The old rule is if twisted near #1 is floppy if near #34 then mfm/esdi...

In my defense, the twist was very flat on this cable!

Thanks for you assistance, the hard drive clicking has stopped, but it doesn't look like I'm there yet.

The controller card makes it past AMI post code 91, "Hard drive initialization" about 80% of the time now. It's gets to post code 00, which is "hand off to Int19h". It would be nice if it did that 100% of the time but whatever. I'll take progress when I can get it.

When I say "boot from C:" it doesn't do anything. Num lock still moves, but there's just the AMI rectangle at the top of the screen listing the settings, no action.
When I say "boot from A:" I've had a floppy read a couple tracks. With the gotek I can see what it's doing. I saw that it read tracks 14 - 17 on my Win98 boot disk, but it stopped and didin't do anything else. No messages, no errors.

The old system that this was in was using an IDE controller jumpered to just do floppy & serial & lpt. Odd, right? Maybe they knew something. Maybe I need to recreate that set up tomorrow.

Oh yah, earlier 386 and 486 boards had a HDD reset delay setting for old controllers/drives, then they stopped putting that in around 92ish. Might be you need to power on let it fail then hit reset button and being already spun up, might make it in time.

Ahh Yes as do some scsi stuff have a delay selection for older drives to wait for spin up.
As for cable: we would NOT be surprised how many times that was done or the opposite on the cable (trying a mfm control cable on a 360k IBM drive) long ago, probably lots. Had totally forgotten about it until checking some old manuals.
Got me thinking wrong or bad cable...Also the 1003v/1006v's cannot disable the onboard floppy part but can be moved to secondary floppy port 370 with JP4 jumped. If jumped be secondary so using another controller set as primary 3F0 might work ok.
For the AMI bios stuff am lost 🙁

douglar wrote on 2024-12-19, 02:31:

In my defense, the twist was very flat on this cable!

In my defense I never had a twisted MFM cable in my hands, wasn't even aware there are such. Every MFM HDD has a jumper to select which drive it is (a bit like master/slave on IDE but up to 4 options).
However looking into it the only thing the twist does on HDDs is swap the drive select signals. So depending on where you connect the HDD to the cable make sure your jumpers are in order. Also, I'm going to assume this was the only drive on the cable, ever? Thus it does have the terminating resistor pack installed?

douglar wrote on 2024-12-19, 02:31:

When I say "boot from C:" it doesn't do anything. Num lock still moves, but there's just the AMI rectangle at the top of the screen listing the settings, no action.
When I say "boot from A:" I've had a floppy read a couple tracks. With the gotek I can see what it's doing. I saw that it read tracks 14 - 17 on my Win98 boot disk, but it stopped and didin't do anything else. No messages, no errors.

DOS being stupid is one thing but Win9x boot floppy hanging as well might suggetst you are now not selecting the HDD at all, so the controller waits for the READY signal that never comes...

douglar wrote on 2024-12-19, 02:31:

The old system that this was in was using an IDE controller jumpered to just do floppy & serial & lpt. Odd, right? Maybe they knew something. Maybe I need to recreate that set up tomorrow.

The IDE controllers I've tried would all fail to work properly along my MFM/RLL cards. The suspicion is even disabled IDE channel was somehow grabbing the IRQ line and preventing the card from working. The only way I was able to get MFM/RLL to work with other HDDs was to use SCSI HDD - as the primary drive if possible, but if not (SCSI would detect the BIOS setting and add it's drive as 0x81, or D: in other words) I would boot from floppy and have the second HDD as SCSI and the first would be the MFM/RLL one. A bit annoying but at least this setup worked every time on every mobo (that didn't have integrated IDE).

douglar wrote on 2024-12-19, 02:31:

When I say "boot from C:" it doesn't do anything. Num lock still moves, but there's just the AMI rectangle at the top of the screen listing the settings, no action.
When I say "boot from A:" I've had a floppy read a couple tracks. With the gotek I can see what it's doing. I saw that it read tracks 14 - 17 on my Win98 boot disk, but it stopped and didin't do anything else. No messages, no errors.

Any luck ?

Horun wrote on 2024-12-20, 04:29:

douglar wrote on 2024-12-19, 02:31:

When I say "boot from C:" it doesn't do anything. Num lock still moves, but there's just the AMI rectangle at the top of the screen listing the settings, no action.
When I say "boot from A:" I've had a floppy read a couple tracks. With the gotek I can see what it's doing. I saw that it read tracks 14 - 17 on my Win98 boot disk, but it stopped and didin't do anything else. No messages, no errors.

Any luck ?

No luck so far. Tried swapping controllers but it didnt help. I'l get more time on the weekend.