It almost brings me to tears to know at some point, we'll never hear that classic sound of those old drives again. So many fond memories of my childhood are attributed to this old tech. I still want to rebuild my old Zenith Data Systems PC/AT I grew up with.
If you record and keep saved the sound files of the recording you made of the sounds the old hard drives make you will hear it anytime you want for the rest of your life!
I’ve repaired and restored both my floppy drives and my hard drives but I do have modern SD card drives and Gotek virtual floppy drives etc. but I use them just to get images off the internet and back onto physical disks. I’d much rather save / load and run all my software off physical drives and media like it’s supposed to over the silence and speed of a virtual drive. Nothing like seeing and hearing the drive spin up to load the next level of Jumpman 😂 But I’m old… memories revived…
This is why I really want to modify my old non-functional drives to emulate seeking (interfacing either with the HDD light or intercepting IDE signals) so you get fully authentic sound while using far more reliable flash media. Currently I'm stuck figuring out how to make the spindle spin without the default board.
@@codahighland If you're talking about that thing that LGR showed off recently (might have been on his blurbs channel), it's a step in the right direction but its clicking emulation needs a lot of work to be more authentic, and then you still don't get the spinning sound. I personally am sceptical that a speaker can properly emulate the sounds and that the better solution is to create the sounds using the same physical movements, using either an old hard drive or something similar to produce the sounds. But then I'm a bit of a purist when it comes to such things.
@15:00, the magnetic stored tracks usually stay there. It is the mechanical track alignment of the stepper motor and head location which can go out over long aging of the drive mechanics. (IE: a new low level will generate new tracks whatever the alignment happens to now be. I have also recovered ancient drives by loosening and tightening the tracking stepper motor so long as I never touch the track 0 alignment optical sensor...) Newer drives which use a voice-coil to track the head don't have a stepper motor for positioning each track. It is the format on the disc itself which the drive electronics for these newer drives which adjusts the voice-coil positioning for an infinite track alignment. This is why low level formatting was removed from new drives, the disc needs the format to tell the voice-coil tracking electronics where to position the head. That is why the low-level is don at the factory on the discs with added special hardware.
I was going to post this. The old stepper motor HDD's were also sensitive to orientation. If you mounted the drive vertically, you needed to low level it in that orientation. It may work initially, but might go bad over time. A fresh low level would usually fix the issue, but with total data loss.
I remember being able to (finally) set a 1:1 interleave with my seagate st-238r RLL hard drive when I upgraded to a 386-25 mhz AND added a 16bit controller.
It is also possible to use DEBUG to perform a low level format on an MFM drive. You do have to KNOW your drive's geometry well enough to enter it by hand, and you have to take your time, but it can be done.
Steel shielded bearings (Z and 2Z types) are not sealed. The steel shield is just for some dust protection therefore they do dry out. One further problem is that these stepper motors have two bearings and from the outside only the outer one can be lubricated without taking the drive apart or saturating the motor with oil.
The stepper motor bearings (in the Miniscribe) are going to be problematic with old oil or grease, as you can only get to one bearing to lube it, the other being on the end that goes inside the drive, meaning you'd have to take the stepper off to do a full re-lube, and goodness knows how you're meant to do that without trashing the head alignment... :S
NEC used to market their MFM drives with a slogan "As sure as a sunrise". The D5126 was (and is) an excellent and very durable hard drive. I had two working 5126s among my possessions from 1980s to late 2000s - both of them survived a couple of house moves and being transported in cold, banged around a decade in cardboard boxes and so on. The last I used them was in 2009 - both worked without a hitch and with zero defects. Quality is a beautiful thing. Seagate drives were cheap, sad and unrealiable things compared to NECs. Yes, there are worse drives in the universe, but still.
The thing with Seagate drives is that the consumer models really are built down to a price. Their Enterprise drives are fine or they wouldn't get bought by the system integrators.
Gotta love old MFM drives... those were the days when computing was "fun"... none of this "automatic" stuff that makes it easy peasy these days... :) As for writing on drives -geez - many old drives have specific spaces to write bad sector information... and a note about magnetic fade - sure, it's going to happen, which is why data on any drive should be "refreshed" periodically.. (I'm sure there are specific magnetic properties of the disc coatings that would define their retention capabilities... not unlike floppies or old mag tapes).. As always - another fun video - thanks!
Low Level Format is the same stage of things as when a floppy disc is formatted but before the file system data is written. It writes out the track structure, all the sector gaps, the sector addressing, sectors filled with pad data (or what ever). Once that is done you can then go back and write out the logical structures for the file system using the OS's initialisation programs. The exact byte sequences written can be a bit controller specific so it's quite possible to LLF a drive with one MFM controller and not be able to read it with another. Some IDE drives were able to be truly LLF'd in the early days before drives went to variable sector counts per track (more on the outer tracks). Even then it was still possible to get a select number of drives to actually do a LLF and rewrite the whole track from scratch. Once drives moved to interleaved servo, ie did away with the dedicated servo surface, it became impossible to really do a LLF. All drives really only did a write/verify at that point and internal remapping using spare sectors to hide defects once that happened. Before I forget, it's not always possible to swap controllers on an IDE drive and get a fully sensible disk at the end of the process. It depends on how the controller records the sector defect remapping. If it writes it to some dedicated sectors on the disk, all good. If it stores it in the controller board in an eeprom or flash etc, not so good. The drive will work but any remappings will not be preserved with the potential for incorrect data being returned when sector X is read. I came across this years ago with a 2.5" HDD with a failed board. I had another of the same so I switched the board and it sprang back to life. Nice, I thought. Only catch was the defect map wasn't correct so occasional files had sectors of invalid data in them. Most of it was recovered so it wasn't a total waste of time. MFM / RLL drives with external cards are all constant bit rate with the cards all being fixed at the same data rate. IDE broke that relationships of course and SCSI never had it either, but there is another class of HDDs that use the same two flat cables, ESDI. ESDI disks had different bit rates. 5, 7.5 and 15 I think. It's been too many years to remember clearly. A good EDSI disc was quite fast.
I had a Conner specific device to do low level formatting on Conner IDE drives. It plugged into the extra little connector on the opposite end of the drive relative to the data cable. It made it fully possible to marry a different PCB to the HDA. I bought "bad" drives (if they spun up) for $1.00 each and sold the repaired drives for something like $0.50 per MB. There were quite a few drives "repaired" by putting the jumpers in the right place. I always did the full LLF regardless and put a 90 day warranty on them. Only ever had one drive returned out of I'm guessing around 250.
The CDC/Imprimis drive is a Wren II which was a very advanced drive for the time. It is also a rotary voice coil drive so it is definitely going to be self-parking.
There is the oil trick :D It may be possible to warm up the rancid oil quicker with an external heat source like a hair dryer or hot air station, I think
The first IDE drives were prototyped by Western Digital in 1986, and were first used in the Compaq Portable II (XT Interface). Tandy quickly adopted them too.
@@paulmichaelfreedman8334 no, actually standard ST-506 drives, with the western digital controller on a daughter board, that provided the IDE connector, much like the earliest SCSI disks with embedded controller.
When I started work in 1986, I thought the adjoining office next-door must have a coffee machine or something. Turned out it was their IBM PC's hard-drive clattering away! (I'd only used floppy micros and a VAX till that point). 😁👍
Based on your recent Amiga shenanigans my guess is you'll swap the boards and neither drive will work, then you'll swap them back and they'll both work...
Those old MFM/RLL drives from the 80s... they just have that nostalgic sound that makes the experience of using a retro computer just sound correct. The Seagate ST-225 (and its variants) is my most favorite retro hard drive of all time and I'm happy to have a working one in a Packard Bell PB88 XT clone system.
"When you have like the 5170 compatible controller - which this gold one is..." -- instant flashback to a box I moved a few weeks ago and sure enough I have that card and that same NEC drive lol. I'm going to have to play around with SpeedStor - that's one very useful utility.
I had good results with heating the steppers up to get them past the sticking point. I once put a really stubborn drive in the oven and left it @80°C for a while. afterwards i spun it up still hot. There was no longer any detectable data on the disk but it worked.Lubricating is a must as well. Using contact cleaner before using oil often made a difference as well.
Heating materials above their Curie point makes them completely lose their magnetic properties. Below that the stored field can become degraded. Maybe that's what happened when you heated your drive.
@@eDoc2020 The data was most likely damaged beforehand but the heat may have further degraded the magnetization. I had written the drive off anyways, so i had already tried different methods. In the end the drive only lasted for about another half year before it had a (fatal) head crash.
Low level formatting on MFM and RLL drives do indeed write a controller-defined track: This also means that a disk written on one type/brand of controller may not be accessible on another controller. And also if an MFM drive is really good, you can low level format it using an RLL controller and get 150% space and speed - and you will need to run spinrite's surface scan thoroughly. Debug G=C800:5 :-)
I'm always struggling with the voice coil terminology. At least Seagate in the case of ST4096 used the term "servo". This was a 5.25" full height beast with 5 platters, 9 surfaces for data, and the 10th surface solely used for servo control, as it had the patterns for keeping the heads over the tracks, regardless of thermal expansion. It was 78Megs IIRC. MFM, the only real thing.
Perstore made the wacky controllers that got you huge drives, >100 meg! Oddball device, for sure. Some WD controllers had a flaw in their BIOS that gave Type 17 drives 987 tracks instead of 977. The last ten cylinders in a low-level were off the media and always failed.
I'd have a go at bridging or replacing the ribbon cable in that last ST225. Fiddly but should be doable with patience and a very steady hand! (I had to do similar with one of my pocket TVs after I'd opened it up a few times to tweak things).
Wow, I didn't realise just how much I missed the sound of an old mechanical hdd. Great video, pure nostalgia with the MFM's. Thanks for opening one up, something I've never done or seen before on a drive this age.
I recall back in the mid 90's working for a placed who bought used computer to resell. About every issue you encountered, We had them as well. I agree with you 100% on the type of controller card. I got mad at the one you said was slower/ bad and threw it out the window! lol Fun times! Thank you for the video! :)
14:50 I actually had an issue like this with an m.2 SATA SSD recently. The computer would see it and its size, but would freeze trying to read it. Windows XP setup would blue screen trying to understand what was going on and a newer PC’s BIOS couldn’t secure erase it. But as a last resort I headed into Linux and, slowly, managed to get it to write a new MBR. I restarted the computer and it immediately started working. Secure Erase went through OK and it’s now in my X31 ThinkPad as its XP boot drive!
@@rpavlik1 Nah, not strange, poster wrote sata m. 2. Sata has a secure extension which if not unlocked at bios enum locks the whole drive and makes it only detectable but resonds as faulty device since the on drive controller refuses all commands but a few. Mostly unlockable with a proper sata host controller that supports hot plugging and the secure erase command set. Parted magic has some good implemented tools for exactly that.
@@rpavlik1 Funnily enough I had a similar, but less severe issue with another SSD last night. Froze my HDD dock, HP BIOS said "error with hard disk", booted into Linux and rewrote a new MBR and it's fine now lol It's very odd!
@@bloeckmoep The secure commands are also in more modern regular IDE drives. I haven't actually tested but XP shouldn't bluescreen if a drive is locked. It would just return read errors. I have seen some strange behavior with failing SSDs though. I had a drive which would lock up and then would _occasionally_ be readable after a power cycle and had speed going all over the place. And OP, I'm assuming you were using an m.2 SATA to 44 pin IDE adapter. X31 is fully PATA only. X41 is natively SATA but has an onboard adapter to PATA.
5-When it comes to recal or restores during the test program use, the drive itself can't tell where it is unless it has an encoder of some sort of the stepper motor so it can see exactly what has happened by counting pulses from the encoder. The test program is probably do a seek with verify command, which steps to track X, selects head Y and then looks at the addressing data flying by and sees if it matches the expected track. If not the controller card will signal an error or do an automatic recal / restore and try again depending on the option bits set. This is why you can format, or at least attempt to format, MFM drives with RLL and still have it work. A drive from the MFM only era has absolutely no chance in hell of decoding RLL itself (excluding of course MFM patterns which RLL encompasses) but will still happily power up, seek to track 0 and find it via the track 0 sensor and then do its step in / out testing and verifications by returning to track 0 and testing the track 0 sensor before declaring ready. There can be servo patterns written outside of the normal data area too which will also let the drive know something has gone wrong. 50:32 ... and there you go. I completely forgot that the ST-225 doesn't have a track 0 sensor. It is looking for the pattern on tracks before track 0 after smacking into the bump stop to tell that it is home. Other drives can have a sensor, its just that the 225 is a bargain basement drive.
Naphtha.. It does not hurt plastics. Your petroleum products/ oils/grease. Don't add more oil. Add Naphtha. It will restore the viscosity of the original oils. Try it. It will save you some time battling what you're doing. Also, it will remove things like sticker residue from your computer case us, without hurting the plastics. It'll even remove marker residue without hurting the plastics. Buy a quart. Do a series on it. You don't restore oils that have gotten hard by adding more oil. You add Naphtha solvent. You're welcome.
Sounds great. Where can one get it? I used "washing benzine" or is it called white spirit? Sorry, no native English speaker. Is that similar and ok, too?
The controller card that 'sucks'. I found that often you needed to low level format the drive using the controller you want to finally use. It seemed to make quite a difference.
My last MFM drive I used, was a 5MB drive that I put on its side to test (probably from a batch bought on some big fair, which was a lot of fun back in the days). I bumped it, and it fell flat.. I felt so bad. It didn't survive.. What I do still use on a daily basis, are floppy drives, and I craft Dual Floppy Drive sets (actually am crafting 6 as I am writing this), since people do still like them, like myself. MFM is much alike when it comes to interfacing, so it's quite cool and oldskool, relatively easy to understand, and it really gives me the nostalgic goosebumps! Just no more bumps to an MFM drive..
Adrian, I had an ST-225 that didn't initialize just like yours. I performed a low-level format through the ROM of the IBM XT hard drive card. It actually made it work.
With these drives it’s usually a good idea to Low level them when they are nice and warm, thermal expansion can be a thing, modern drives generally don’t have that problem as the servo track compensates and the material used doesn’t expand as much. As for what a LLF lays down is likely a high frequency pilot "baseband tone" with gaps between sectors and some rudimentary track/ sector information used by the controller and not visible by DOS etc
This brings back memories. I probably had every MFM, RLL, and early SCSI and IDE drive made cross my workbench back in the day. I kept a few dead drives on hand just to do PCB swaps and brought a few back to life that way.
I really like the ST-225 drives. I use to have two of them connected to my Tandy Radio Shack Color Computer with a B&B adapter that allowed me to use an 8-bit ISA MFM or RLL card with those drives. I personally never had any problems running the ST-225 as an MFM drive. As far as your floppy issues with CheckIt. I would probably use a Gotek drive emulator and use a write-protected disk image. Plus with a Gotek you could easily keep the M$-DOS 6.22 install disk images on a USB Flash Drive and other tools as well ready to go when you need them.
MFM is a recording technology, not the interface. The interface (the two ribbon cables) is ST-506. (My first HDD was an ST-238R which used RLL rather than MFM to record the data and thus got 30MB on essentially the same drive as the ST-255, but it used the same ST-506 interface.)
If I remember correctly, the old MFM drive had one head that had a servo track written to that platter surface, and the one ST225, I suspect had a bad servo head. Later drives utilized an embedded servo design and had servo info written at each sector. Servo engineers were very talented folks and I alway admired their capabilities when I worked at Micropolis.
ST-4096s were arranged like that, if I remember correctly, they had 5 platters and logically would have 10 heads, but only 9 were used for data, the 10th head was used for the servo track on one of the platters' surface.
Same. Epson Equity ii. Hercules green monochrome. Loved playing space quest on it. I think my dad paid it off for the next 4 years. I wouldn't have my career if he didn't decide to get one.
Same with IDE -- but when you do it, you're just making a request to the drive to do a low level format. Whether the controller honours the request is another matter! Modern drives will ignore the command, but I have heard tell there were a few drive models in the early IDE days where the controller would TRY to respond to the low level format request, but destroy the sector headers and make the drive unusable! So be very careful about low level formatting old IDE and SCSI drives (and don't bother with new ones), unless you have specific information that you're REALLY sure of, that your model is safe for it and it will actually do something useful.
It is supported with scsi drives of all ages, but it does not do as low level a format as the mfm controllers would. Basically it is a zero out/surface scan/error remap operation. I stopped the llf mid way with my very first scsi drive and it went into a dumb mode where it was visible on the bus, but I was not able to write anything to it. :-) Completing the llf snapped it out of it, but I had a good scare.
At least you could probably use the bad ST-225 as a parts donor (except for the heads), in the off chance you stumble upon a ST-225 riddled with a lot of bad sectors..
As a massive fanboy of old hard drives in general and the 225 in particular. I'm so please to see you revisiting this topic. :) I'll agree with you about the healing properties of a low level format... but I've got nothing in the way of explanation apart from nodding at yours. I also used to collect ST-157As and they were one of the first IDEs... you could still low level them... probably the last drive I had that you could low level. I always though a stethoscope would be a handy thing to have when mucking around with hard drives.
Yeah, they aren't the easiest thing to get working, but a true computer of the era needs a spinning disk... A flash drive just doesn't do it. I want to feel the case vibrate during drive access!
I'd try to solder those loose head wires to see if that fixes it. If not, make a clock out of it. I have an old IDE 3.5" drive with stepper actuator I made into a clock, using some of the head arms for the hands.
For freeing up motors where the bearings are gummed up from aging oil I have found that the best oil to use is automotive ATF (automatic transmission oil). Besides its lubricating properties it's loaded with detergents and as a result it cleans the bearing surfaces from the dried up oil and it keeps working like it was never seized. I have repaired several PC fans that were completely frozen and they keep on working for years after that.
Reminds me of the old days. Had a Miniscribe 6053 (very fast for that era) on a very souped-up 80386 with a Transputer as co-processor. Went through a whole stack of "fast" controllers to get maximum throughput, but in the end the standard WDC-controller was the best. Spent a lot of time optimizing interleave with Spinrite II on computers for friends, that viewed me as their computer-guru😂😂 man, those were the days!!!
Back in the day, our computer was constructed of mostly disused and broken hardware. All of our HDs were ones that failed at my dad's work. These MFM and RLL drives could almost always be brought back with a LLF (after un-sticktioning). They would often fail again months or years down the road, but then Spinrite became a thing and it would almost always recover the disk and our data. On those Seagate drives, it seemed like there was some wear and tear with the head positioning which could be fixed by Spinrite's trick of "jiggling" the heads to try to recover the track and then laying it back down, fresh. It was a little sad when IDE rolled in took away all the fun.
I don't understand why people would get upset about writing on the drive. I write on all mine, retro or not. Usually with the installation date and any notes related to the drive.
They are some kind of snobs, like audiophiles. The important thing in vintage computers is to fix them, use them, people complaining should open a museum and shut up
@@TheInsultInvestor The marker comes off with a little ISO. I've been a sysadmin for over 20 years and it's something I started doing early in my career. For old drives finding writing on them is usually helpful, especially if it is the drive geometry. It's not like MFM, SCSI, or IDE drives are rare or reliable.
I did something a few years ago with a 2tb drive that failed, it would definitely spin up but not work, no sound of the armature moving around, I tapped it is a few times, nothing. I had my Music and some other files on it, I had another 2TB drive that was pretty much identical but older, I took the bad drive apart and I swapped the armature from the working drive to the dead one and moved the PCB over just in case it had to be for calibration or some sort handshake thing with the armature, It worked, I moved all my stuff off it without the top on it and never used that drive again. I did all that in room that wasn't filtered of dust, but I mean the stuff I had on it I could replace if I needed to, but why no try it, it worked well, a lot less problematic than I thought it would have been.
Late reply but that sound is soothing/COOL . As a "old timer". Thank you Adrian for all your videos. As a enthusiast, I miss the days "listening" to HDD's knowing it running :P
MFM drives are so cool... when they work :D I have a couple of ST225s that are recognised by the controller and sound fine, but low level formatting fails on them, I should also try swapping out the boards from a working one some day.
Ahhh I sometimes miss the sound of those old drives spinning up. I think my first hard drive in an XT was a 225, or atleast one very similar. Good times. 😊
28:00 - You can disconnect temporarily the stepper motor from the PCB, then turn on the drive. With the stepper motor disconnected (hence no power applied to it) you'll be able to move it freely while the disk is spinning, without the need of doing that while switching the drive on and off repeatedly.
I used to use RLL controllers on MFM drives. There was a Norton disk utility that I can't remember the name of that would find bad sectors and I would run that over and over again to mark out the sectors that couldn't handle the 50% data density increase.
The very first step if any mechanism is stuck is to disassemble and clean all its parts! Sometimes up to three times in a row. Adding excessive lubricant can make it even more stuck.
To answer on the drives being low level formatted to work again; basically, yeah, magnetisation will naturally fade over time. The technical term is bit rot. It's why most floppy disks no longer work since they're more susceptible than hard disks. It's generally recommended, even with newer hard disks, that you should refresh your data(rewrite all data on the drive) every 3-5 years. Some drives can last much longer(an estimated 70 years), but that's with ideal temperatures/humidity and regular usage. A good quality of drive also makes a difference.
Based on my knowledge of bit rot and other electronic repair your theory about the loss of magnetism is holy believable. Especially with being able to fix it with a format. Newer drives that would take a long long long time to happen bit rot is rather slow but with older drives it's possible. Although that's probably more an issue because of miniaturization. Large drives like some of the very very early hard drives like a hawk drive tend not to have this problem simply because the sectors are so big like physically.
I used to have problems with these old disk drives, the problems came down to the spindle bearing. The spindle would seize and I found that the grease in the bearing had dried out and (presumably) created a hard lump that prevented rotation. Rotating the disk by hand freed it up and thankfully I never had to apply oil, I was concerned that excess oil would get onto the disk surface. Another thing that you could do in your testing procedure is use a CRO to examine read head signals (there are test points for this) and to check the read data signals on the Interface cables.
I had an HD fail as I was trying to back up files off of ZIP disks on a new Mac Pro G4 back in the day. Luckily apple lent me a new drive as was considering Drive Savers. PCB and firmware were idential. Swapped the PCB and saved everything. Saved a grand.
Don’t use a torch - use an electric heat gun. The torch has water vapor from the combustion process; and the temperature discontinuity from the flame edge tends to be rather abrupt - too far away and it won’t heat up, move it a little bit closer and you’re setting it on fire.
I really enjoyed this episode. My first one, or two PC's had MFM drives... I remember the dual ribbon cables... but parallel IDE replaced MFM pretty quickly.
yes, modern drives have lots of algorithms and programing to refresh tracks as well as move/restore damaged data. it does a lot of what spinrite does silently in the background. once a modern drive starts going bad, its usually been going bad for quite a while. i think you can do a refresh in spinrite.
You should low level format any drive when you change controllers, and you should enter the known bad tracks at that time, because sometimes they format and test good for a little bit, then fail. This is from experience of selling them for years.
Wondering if those steppers from old drives are standard unipolar/bipolar designs that could be used in all sorts of DIY projects instead of being E-wasted. Likely not the voice coil type in the Seagate (although I can still imagine some use cases for those), but the full 360° ones in the belt driven head assemblies, why not? Quite valuable and durable pieces of tech, unlikely to break anytime in a century, I'd probably buy used units at a discount instead of new. Also unusual form factors could be an advantage in certain cases.
Favorite MFM drive? I’ll start: Mitsubishi MR535, 40 meg MFM, 60 meg RLL, supported 1:1 interleave and weighed a ton. Never saw one come back and we sold hundreds.
The servo tracks on various drives act as a sanity check for the drive itself, if the drive hits cylinder 0 and bumps, then the servo track is damaged.
I recently discovered I had an external low profile case fitted with an AdSCSI Plus ST (for Atari) connected to an ST225 (which doesn't spin atm). I don't know yet if I'll try and get it working again, but if I do then I'm sure these videos you've made will be very useful. 16:15 - I don't know about IDE, but SCSI disks can be low level formatted. IRIX's fx allows one to do this, though it spits out a hefty warning that doing so can ruin a drive.
It's been a while since I used Steve Gibson's Gibson Research Corporation - Spinrite. Wow that takes me back to the days of Ray Gwynn's Double Aught Buckshot driver X00.sys which was needed to run more than a coupe of com ports.
MFM hard drives do need to have their format refreshed once in a while, because they will lose their magnetic formatting, which is what low-leveling does. i think you and Usagi Electric might be working on similar things, in that regard. IDE and later hard drives not only don't need this, you will damage the existing magnetic format and alignment, rendering the drive useless. this is why taking a platter from one hard drive and putting it in another will not only not work, it cannot be undone. if a hard drive only works when it heats up, it is heat damaged and will never work properly again without some heat. ps, writing on a hard drive's body won't hurt it at all
17:00 I don't have experience with hard drives, but I definitely did this with 3 1/2 floppy. Many floppies that would fail could be low-level formatted and serve reliably for many writes afterwards. This was definitely a time where you would write .rar files to floppies with heavy error correction, but I mean it worked, even if the error correction had to correct for some bad sectors.
If that CDC drive is a Wren, you might want to talk to Usagi Electric - he's trying to rebuild a Centurion, and might be able to make good use of it. As for the st-225: it's about 50/50 either way, but I lean toward a mechanical problem. (in before I view the startup...)
I am curious if you attempt to solder tiny wires to the head if you'd be able to read the drive. I also wonder how common this particular problem is. Perhaps a bit of delicate soldering may recover a number of such drives.
Yes. It would be only for experimenting, because once opened the drive is not reliable anymore. But I was expecting Adrian to attempt some tiny soldering before e-waste. Just for the video. It would be incredible if it just "freaking works" after the soldering... 🤯
@@GYTCommnts If he continues to fuss about with hard drives, it may be useful for him to get a little glove box / cleanroom. As a control systems engineer, I'd love to see what the firmware looks like for these drives. I've marveled at hard drives , and worked on similar technology professionally (piezo -servo micro-positioning)
Very interesting video. I have old Microscience HDD and I know that it worked ok last time around year 1990. I am collecting knowledge to turn it on again but for now I am not doing it because don't want to damage something.
Using that miniscribe drive as a main example I have a few questions about MFM drives .. - is the noise from that particular stepper motor normal or abnormal? - if abnormal, what would the problem be with it (ie. is some internal part of the motor assembly worn)? - is that stepper motor noise the normal type of noise that you hear when MFM drives are operating? (I'm used to hearing a similar noise which I always thought was the data being transferred) - if all MFM stepper motors make a similar sort of noise, why is that miniscribe motor so noisy? - is that miniscribe stepper motor, and the spindle noises too I guess, the product of bad or cheap parts being used, or an inferior type of hardware? - can you remove an external stepper motor from a hard drive in order to take it apart and service it or replace it? - if the spindle bearing assembly has a bad/worn/uneven roller can you replace the bearing assembly entirely? - If, as you mentioned, the original magnetic properties of the existing tracks/data on an MFM drive can weaken over time and the drive won't see/read them any more, or if the head assembly has moved out of it's original alignment and just doesn't see those tracks anymore, can you just somehow realign the heads back to where they were and have them read the data again? - If so, assuming those old tracks are there, how do you get the heads to scan for those original tracks? - Can the heads be moved in non-standard increments to try to find old tracks without special hardware or software? - Do you need some sort of special hardware/software or forensic type tools to scan for old tracks that the head is not aligned to? - If you were to find those old tracks can you get the hard drive to automatically re-write all the existing data so that the magnetic strength is increased and the drive can read normally again? I guesss (why did I type 3 s's there? I'm not in Slytherin) I'm curious as to how far you can strip down, diagnose and repair or replace the individual components of an MFM drive both mechanically and electronically. Can each and all individual components be replaced? I know that you can't open a sealed hard drive case to open air because of dust and particles but lets assume that you have a clean room to work with. I'm curious to see whether you can replace all the parts of an MFM drive that generally degrade and go bad, put in something newer or better/modern and have the drive be improved to a level of reliability that's good or maybe even better than the drive original design would allow it to be.
You can only get at one of two stepper bearings with your oil, and these use grease, not oil. The way to get these reliable is to replace the stepper motor. That said, if you just keep on keeping on excersizing those heads, extreme to extreme non-stop, the grease should soften and heat up. You should do this for 24 hours plus. After this, as long as you spin it up once a week or so, it should stay with you.
Ahhh, stiction... the bane of my 128 MB Connor IDE drives. Thankfully they lasted until I had upgraded to a new machine so I didn't lose anything. That's one phenomenon I'm happy to never encounter again.
Interesting video👍.I rememer when i saw one of old full drives in person (i think it was a scsi drive) and picked it up,It was heavier than a brick at the same size😂.they have thick layers of metal.
I would love to see a video on sticktion with some exploratory surgery on a drive to try to definitively determine what causes it. I saw a video no too long ago where someone disassembled a drive with sticktion, and discovered it was _NOT_ the bearings. After the head and everything was removed, the platters rotated freely. He thought it could be magnetic in nature.
Stiction is the _heads_ sticking to the platter. Since heads park at the edge of the platter it takes a lot of torque for the motor to start up if this happens.
@@olepigeon You'd probably need to replace the heads and/or the platters. I don't really know anything about what makes them stick so I can't say more.
When a hard disk makes 'clicking' noises while spinning up it may be due to damage on the disk, however most disks do have a landing zone, which is also a 'take off' zone. The heads won't move off that zone until the motor is up to speed, and then the heads will lift and fly over the good part of the disk. So unless the damage is great enough to destroy the heads, the drive will work even though it's making weird noises during spin up. I think you can low level format an IDE disk, there should be a disk init command in the IDE specification. However, the format is probably fixed in the drives firmware, and cannot be specified beyond a few 'canned' choices (drive specific).
![Adrian's Digital Basement ][](http://yt3.ggpht.com/ytc/AIdro_k7CrQRe71qN6c0Ua7A1P_I8KkxjqLa70u0zYeCvdvnKw=s48-c-k-c0x00ffffff-no-rj)
It almost brings me to tears to know at some point, we'll never hear that classic sound of those old drives again. So many fond memories of my childhood are attributed to this old tech. I still want to rebuild my old Zenith Data Systems PC/AT I grew up with.
If you record and keep saved the sound files of the recording you made of the sounds the old hard drives make you will hear it anytime you want for the rest of your life!
I’ve repaired and restored both my floppy drives and my hard drives but I do have modern SD card drives and Gotek virtual floppy drives etc. but I use them just to get images off the internet and back onto physical disks.
I’d much rather save / load and run all my software off physical drives and media like it’s supposed to over the silence and speed of a virtual drive.
Nothing like seeing and hearing the drive spin up to load the next level of Jumpman 😂
But I’m old… memories revived…
This is why I really want to modify my old non-functional drives to emulate seeking (interfacing either with the HDD light or intercepting IDE signals) so you get fully authentic sound while using far more reliable flash media. Currently I'm stuck figuring out how to make the spindle spin without the default board.
There are SSDs that make clicking noises with a speaker. :P
@@codahighland If you're talking about that thing that LGR showed off recently (might have been on his blurbs channel), it's a step in the right direction but its clicking emulation needs a lot of work to be more authentic, and then you still don't get the spinning sound. I personally am sceptical that a speaker can properly emulate the sounds and that the better solution is to create the sounds using the same physical movements, using either an old hard drive or something similar to produce the sounds. But then I'm a bit of a purist when it comes to such things.
@15:00, the magnetic stored tracks usually stay there. It is the mechanical track alignment of the stepper motor and head location which can go out over long aging of the drive mechanics. (IE: a new low level will generate new tracks whatever the alignment happens to now be. I have also recovered ancient drives by loosening and tightening the tracking stepper motor so long as I never touch the track 0 alignment optical sensor...) Newer drives which use a voice-coil to track the head don't have a stepper motor for positioning each track. It is the format on the disc itself which the drive electronics for these newer drives which adjusts the voice-coil positioning for an infinite track alignment. This is why low level formatting was removed from new drives, the disc needs the format to tell the voice-coil tracking electronics where to position the head. That is why the low-level is don at the factory on the discs with added special hardware.
I was going to post this. The old stepper motor HDD's were also sensitive to orientation. If you mounted the drive vertically, you needed to low level it in that orientation. It may work initially, but might go bad over time. A fresh low level would usually fix the issue, but with total data loss.
I remember being able to (finally) set a 1:1 interleave with my seagate st-238r RLL hard drive when I upgraded to a 386-25 mhz AND added a 16bit controller.
It's the HD controller that made the difference. WD1003 are 1:2 and WD1006 are 1:1 capable...
As I remember 30 years ago I fixed quiet a few MFM by low-level format with DNA (Diagnose & Analyse by Ontrack).
It is also possible to use DEBUG to perform a low level format on an MFM drive. You do have to KNOW your drive's geometry well enough to enter it by hand, and you have to take your time, but it can be done.
@@horusfalcon Yeah, still remember this method but a program such as DNA is handy since it can find the best interleave for me.
Steel shielded bearings (Z and 2Z types) are not sealed. The steel shield is just for some dust protection therefore they do dry out.
One further problem is that these stepper motors have two bearings and from the outside only the outer one can be lubricated without taking the drive apart or saturating the motor with oil.
The stepper motor bearings (in the Miniscribe) are going to be problematic with old oil or grease, as you can only get to one bearing to lube it, the other being on the end that goes inside the drive, meaning you'd have to take the stepper off to do a full re-lube, and goodness knows how you're meant to do that without trashing the head alignment... :S
Who needs head alignment if you can just low-level them again? 🙂
@@senilyDeluxe Anyone trying to recover data - eg. as data archaeology.
NEC used to market their MFM drives with a slogan "As sure as a sunrise". The D5126 was (and is) an excellent and very durable hard drive. I had two working 5126s among my possessions from 1980s to late 2000s - both of them survived a couple of house moves and being transported in cold, banged around a decade in cardboard boxes and so on. The last I used them was in 2009 - both worked without a hitch and with zero defects. Quality is a beautiful thing.
Seagate drives were cheap, sad and unrealiable things compared to NECs. Yes, there are worse drives in the universe, but still.
The thing with Seagate drives is that the consumer models really are built down to a price. Their Enterprise drives are fine or they wouldn't get bought by the system integrators.
Gotta love old MFM drives... those were the days when computing was "fun"... none of this "automatic" stuff that makes it easy peasy these days... :) As for writing on drives -geez - many old drives have specific spaces to write bad sector information... and a note about magnetic fade - sure, it's going to happen, which is why data on any drive should be "refreshed" periodically.. (I'm sure there are specific magnetic properties of the disc coatings that would define their retention capabilities... not unlike floppies or old mag tapes)..
As always - another fun video - thanks!
IMO the best automatic feature of newer drives is automatic parking. Head unloading is a bit nicer IMO but has its own issues.
Low Level Format is the same stage of things as when a floppy disc is formatted but before the file system data is written. It writes out the track structure, all the sector gaps, the sector addressing, sectors filled with pad data (or what ever). Once that is done you can then go back and write out the logical structures for the file system using the OS's initialisation programs. The exact byte sequences written can be a bit controller specific so it's quite possible to LLF a drive with one MFM controller and not be able to read it with another.
Some IDE drives were able to be truly LLF'd in the early days before drives went to variable sector counts per track (more on the outer tracks). Even then it was still possible to get a select number of drives to actually do a LLF and rewrite the whole track from scratch. Once drives moved to interleaved servo, ie did away with the dedicated servo surface, it became impossible to really do a LLF. All drives really only did a write/verify at that point and internal remapping using spare sectors to hide defects once that happened.
Before I forget, it's not always possible to swap controllers on an IDE drive and get a fully sensible disk at the end of the process. It depends on how the controller records the sector defect remapping. If it writes it to some dedicated sectors on the disk, all good. If it stores it in the controller board in an eeprom or flash etc, not so good. The drive will work but any remappings will not be preserved with the potential for incorrect data being returned when sector X is read. I came across this years ago with a 2.5" HDD with a failed board. I had another of the same so I switched the board and it sprang back to life. Nice, I thought. Only catch was the defect map wasn't correct so occasional files had sectors of invalid data in them. Most of it was recovered so it wasn't a total waste of time.
MFM / RLL drives with external cards are all constant bit rate with the cards all being fixed at the same data rate. IDE broke that relationships of course and SCSI never had it either, but there is another class of HDDs that use the same two flat cables, ESDI. ESDI disks had different bit rates. 5, 7.5 and 15 I think. It's been too many years to remember clearly. A good EDSI disc was quite fast.
I had a Conner specific device to do low level formatting on Conner IDE drives. It plugged into the extra little connector on the opposite end of the drive relative to the data cable. It made it fully possible to marry a different PCB to the HDA. I bought "bad" drives (if they spun up) for $1.00 each and sold the repaired drives for something like $0.50 per MB. There were quite a few drives "repaired" by putting the jumpers in the right place. I always did the full LLF regardless and put a 90 day warranty on them. Only ever had one drive returned out of I'm guessing around 250.
The CDC/Imprimis drive is a Wren II which was a very advanced drive for the time. It is also a rotary voice coil drive so it is definitely going to be self-parking.
There is the oil trick :D
It may be possible to warm up the rancid oil quicker with an external heat source like a hair dryer or hot air station, I think
The first IDE drives were prototyped by Western Digital in 1986, and were first used in the Compaq Portable II (XT Interface). Tandy quickly adopted them too.
yep, I looked it up too!
Even back in the 80s everybody loved WD drives. Still a top notch brand now.
Weren't those weird hybrids with esdi?
@@paulmichaelfreedman8334 no, actually standard ST-506 drives, with the western digital controller on a daughter board, that provided the IDE connector, much like the earliest SCSI disks with embedded controller.
@@tschak909 Did these use RLL or MFM encoding for the heads?
When I started work in 1986, I thought the adjoining office next-door must have a coffee machine or something. Turned out it was their IBM PC's hard-drive clattering away! (I'd only used floppy micros and a VAX till that point). 😁👍
Based on your recent Amiga shenanigans my guess is you'll swap the boards and neither drive will work, then you'll swap them back and they'll both work...
That's really funny. Yeah that would be just my luck!!
Those old MFM/RLL drives from the 80s... they just have that nostalgic sound that makes the experience of using a retro computer just sound correct. The Seagate ST-225 (and its variants) is my most favorite retro hard drive of all time and I'm happy to have a working one in a Packard Bell PB88 XT clone system.
"When you have like the 5170 compatible controller - which this gold one is..." -- instant flashback to a box I moved a few weeks ago and sure enough I have that card and that same NEC drive lol. I'm going to have to play around with SpeedStor - that's one very useful utility.
I had good results with heating the steppers up to get them past the sticking point. I once put a really stubborn drive in the oven and left it @80°C for a while. afterwards i spun it up still hot. There was no longer any detectable data on the disk but it worked.Lubricating is a must as well. Using contact cleaner before using oil often made a difference as well.
Heating materials above their Curie point makes them completely lose their magnetic properties. Below that the stored field can become degraded. Maybe that's what happened when you heated your drive.
@@eDoc2020 The data was most likely damaged beforehand but the heat may have further degraded the magnetization. I had written the drive off anyways, so i had already tried different methods. In the end the drive only lasted for about another half year before it had a (fatal) head crash.
love your nod to the hilarious Miniscribe „brick“-story 😅
Low level formatting on MFM and RLL drives do indeed write a controller-defined track: This also means that a disk written on one type/brand of controller may not be accessible on another controller. And also if an MFM drive is really good, you can low level format it using an RLL controller and get 150% space and speed - and you will need to run spinrite's surface scan thoroughly. Debug G=C800:5 :-)
I'm always struggling with the voice coil terminology. At least Seagate in the case of ST4096 used the term "servo". This was a 5.25" full height beast with 5 platters, 9 surfaces for data, and the 10th surface solely used for servo control, as it had the patterns for keeping the heads over the tracks, regardless of thermal expansion. It was 78Megs IIRC. MFM, the only real thing.
Perstore made the wacky controllers that got you huge drives, >100 meg! Oddball device, for sure.
Some WD controllers had a flaw in their BIOS that gave Type 17 drives 987 tracks instead of 977. The last ten cylinders in a low-level were off the media and always failed.
I'd have a go at bridging or replacing the ribbon cable in that last ST225. Fiddly but should be doable with patience and a very steady hand! (I had to do similar with one of my pocket TVs after I'd opened it up a few times to tweak things).
Absolutely. I would try the same. There is no "bye bye" 🙂
Wow, I didn't realise just how much I missed the sound of an old mechanical hdd. Great video, pure nostalgia with the MFM's. Thanks for opening one up, something I've never done or seen before on a drive this age.
I recall back in the mid 90's working for a placed who bought used computer to resell. About every issue you encountered, We had them as well. I agree with you 100% on the type of controller card. I got mad at the one you said was slower/ bad and threw it out the window! lol Fun times! Thank you for the video! :)
14:50 I actually had an issue like this with an m.2 SATA SSD recently. The computer would see it and its size, but would freeze trying to read it. Windows XP setup would blue screen trying to understand what was going on and a newer PC’s BIOS couldn’t secure erase it. But as a last resort I headed into Linux and, slowly, managed to get it to write a new MBR. I restarted the computer and it immediately started working. Secure Erase went through OK and it’s now in my X31 ThinkPad as its XP boot drive!
That is very strange. Glad you got it working!
@@rpavlik1 Nah, not strange, poster wrote sata m. 2. Sata has a secure extension which if not unlocked at bios enum locks the whole drive and makes it only detectable but resonds as faulty device since the on drive controller refuses all commands but a few. Mostly unlockable with a proper sata host controller that supports hot plugging and the secure erase command set. Parted magic has some good implemented tools for exactly that.
@@rpavlik1 Funnily enough I had a similar, but less severe issue with another SSD last night. Froze my HDD dock, HP BIOS said "error with hard disk", booted into Linux and rewrote a new MBR and it's fine now lol
It's very odd!
@@bloeckmoep The secure commands are also in more modern regular IDE drives. I haven't actually tested but XP shouldn't bluescreen if a drive is locked. It would just return read errors.
I have seen some strange behavior with failing SSDs though. I had a drive which would lock up and then would _occasionally_ be readable after a power cycle and had speed going all over the place.
And OP, I'm assuming you were using an m.2 SATA to 44 pin IDE adapter. X31 is fully PATA only. X41 is natively SATA but has an onboard adapter to PATA.
Apply some heat on the stepper motor to loosen up that old sticky oil, usually works. 👍
5-When it comes to recal or restores during the test program use, the drive itself can't tell where it is unless it has an encoder of some sort of the stepper motor so it can see exactly what has happened by counting pulses from the encoder. The test program is probably do a seek with verify command, which steps to track X, selects head Y and then looks at the addressing data flying by and sees if it matches the expected track. If not the controller card will signal an error or do an automatic recal / restore and try again depending on the option bits set. This is why you can format, or at least attempt to format, MFM drives with RLL and still have it work. A drive from the MFM only era has absolutely no chance in hell of decoding RLL itself (excluding of course MFM patterns which RLL encompasses) but will still happily power up, seek to track 0 and find it via the track 0 sensor and then do its step in / out testing and verifications by returning to track 0 and testing the track 0 sensor before declaring ready. There can be servo patterns written outside of the normal data area too which will also let the drive know something has gone wrong.
50:32 ... and there you go. I completely forgot that the ST-225 doesn't have a track 0 sensor. It is looking for the pattern on tracks before track 0 after smacking into the bump stop to tell that it is home. Other drives can have a sensor, its just that the 225 is a bargain basement drive.
I love the sound of the MFM stepper motors.
Naphtha..
It does not hurt plastics.
Your petroleum products/ oils/grease.
Don't add more oil. Add Naphtha.
It will restore the viscosity of the original oils.
Try it. It will save you some time battling what you're doing.
Also, it will remove things like sticker residue from your computer case us, without hurting the plastics. It'll even remove marker residue without hurting the plastics.
Buy a quart.
Do a series on it. You don't restore oils that have gotten hard by adding more oil. You add Naphtha solvent.
You're welcome.
Sounds great. Where can one get it?
I used "washing benzine" or is it called white spirit? Sorry, no native English speaker. Is that similar and ok, too?
The controller card that 'sucks'. I found that often you needed to low level format the drive using the controller you want to finally use. It seemed to make quite a difference.
My last MFM drive I used, was a 5MB drive that I put on its side to test (probably from a batch bought on some big fair, which was a lot of fun back in the days).
I bumped it, and it fell flat.. I felt so bad. It didn't survive..
What I do still use on a daily basis, are floppy drives, and I craft Dual Floppy Drive sets (actually am crafting 6 as I am writing this), since people do still like them, like myself.
MFM is much alike when it comes to interfacing, so it's quite cool and oldskool, relatively easy to understand, and it really gives me the nostalgic goosebumps!
Just no more bumps to an MFM drive..
I liked it before I even watched it cause it’s just that cool of an idea. Adrian you are the man!
Adrian, I had an ST-225 that didn't initialize just like yours. I performed a low-level format through the ROM of the IBM XT hard drive card. It actually made it work.
With these drives it’s usually a good idea to Low level them when they are nice and warm, thermal expansion can be a thing, modern drives generally don’t have that problem as the servo track compensates and the material used doesn’t expand as much. As for what a LLF lays down is likely a high frequency pilot "baseband tone" with gaps between sectors and some rudimentary track/ sector information used by the controller and not visible by DOS etc
This brings back memories. I probably had every MFM, RLL, and early SCSI and IDE drive made cross my workbench back in the day. I kept a few dead drives on hand just to do PCB swaps and brought a few back to life that way.
Fascinating! Thank you for taking the time to share this wit us!
I really like the ST-225 drives. I use to have two of them connected to my Tandy Radio Shack Color Computer with a B&B adapter that allowed me to use an 8-bit ISA MFM or RLL card with those drives.
I personally never had any problems running the ST-225 as an MFM drive.
As far as your floppy issues with CheckIt. I would probably use a Gotek drive emulator and use a write-protected disk image. Plus with a Gotek you could easily keep the M$-DOS 6.22 install disk images on a USB Flash Drive and other tools as well ready to go when you need them.
These drives hover around my age brackets & I love the sound they make. CompactFlash cards & SSDs just don't feel the same man!
MFM is a recording technology, not the interface. The interface (the two ribbon cables) is ST-506.
(My first HDD was an ST-238R which used RLL rather than MFM to record the data and thus got 30MB on essentially the same drive as the ST-255, but it used the same ST-506 interface.)
Thanx for the video. I haven't seen a MFM HD since about the early 1990s . Brings back memories.
If I remember correctly, the old MFM drive had one head that had a servo track written to that platter surface, and the one ST225, I suspect had a bad servo head. Later drives utilized an embedded servo design and had servo info written at each sector. Servo engineers were very talented folks and I alway admired their capabilities when I worked at Micropolis.
ST-4096s were arranged like that, if I remember correctly, they had 5 platters and logically would have 10 heads, but only 9 were used for data, the 10th head was used for the servo track on one of the platters' surface.
AFAIK the stepper-based MFM drives don't use servo tracks.
Nice, our 386 featured a 20MB MFM drive, was a real challenge to get Windows 3.11 installed.
Same. Epson Equity ii. Hercules green monochrome. Loved playing space quest on it. I think my dad paid it off for the next 4 years. I wouldn't have my career if he didn't decide to get one.
I sure love the seek sounds of them though... Desk vibrating goodness.
16:03 Low level formatting is possible to do on SCSI drives. I did many low level format on my Amiga, but it took many many hours to perform.
Same with IDE -- but when you do it, you're just making a request to the drive to do a low level format. Whether the controller honours the request is another matter!
Modern drives will ignore the command, but I have heard tell there were a few drive models in the early IDE days where the controller would TRY to respond to the low level format request, but destroy the sector headers and make the drive unusable! So be very careful about low level formatting old IDE and SCSI drives (and don't bother with new ones), unless you have specific information that you're REALLY sure of, that your model is safe for it and it will actually do something useful.
It is supported with scsi drives of all ages, but it does not do as low level a format as the mfm controllers would. Basically it is a zero out/surface scan/error remap operation.
I stopped the llf mid way with my very first scsi drive and it went into a dumb mode where it was visible on the bus, but I was not able to write anything to it. :-) Completing the llf snapped it out of it, but I had a good scare.
At least you could probably use the bad ST-225 as a parts donor (except for the heads), in the off chance you stumble upon a ST-225 riddled with a lot of bad sectors..
As a massive fanboy of old hard drives in general and the 225 in particular. I'm so please to see you revisiting this topic. :)
I'll agree with you about the healing properties of a low level format... but I've got nothing in the way of explanation apart from nodding at yours.
I also used to collect ST-157As and they were one of the first IDEs... you could still low level them... probably the last drive I had that you could low level.
I always though a stethoscope would be a handy thing to have when mucking around with hard drives.
yay love dem old hard drives! =) thank you so much, I always find them too much neglected in retro circles.
Yeah, they aren't the easiest thing to get working, but a true computer of the era needs a spinning disk... A flash drive just doesn't do it. I want to feel the case vibrate during drive access!
I'd try to solder those loose head wires to see if that fixes it. If not, make a clock out of it. I have an old IDE 3.5" drive with stepper actuator I made into a clock, using some of the head arms for the hands.
For freeing up motors where the bearings are gummed up from aging oil I have found that the best oil to use is automotive ATF (automatic transmission oil). Besides its lubricating properties it's loaded with detergents and as a result it cleans the bearing surfaces from the dried up oil and it keeps working like it was never seized. I have repaired several PC fans that were completely frozen and they keep on working for years after that.
Reminds me of the old days. Had a Miniscribe 6053 (very fast for that era) on a very souped-up 80386 with a Transputer as co-processor. Went through a whole stack of "fast" controllers to get maximum throughput, but in the end the standard WDC-controller was the best. Spent a lot of time optimizing interleave with Spinrite II on computers for friends, that viewed me as their computer-guru😂😂 man, those were the days!!!
Back in the day, our computer was constructed of mostly disused and broken hardware. All of our HDs were ones that failed at my dad's work. These MFM and RLL drives could almost always be brought back with a LLF (after un-sticktioning). They would often fail again months or years down the road, but then Spinrite became a thing and it would almost always recover the disk and our data. On those Seagate drives, it seemed like there was some wear and tear with the head positioning which could be fixed by Spinrite's trick of "jiggling" the heads to try to recover the track and then laying it back down, fresh. It was a little sad when IDE rolled in took away all the fun.
I don't understand why people would get upset about writing on the drive. I write on all mine, retro or not. Usually with the installation date and any notes related to the drive.
yeah, you're that guy
@@TheInsultInvestor and you're that person
They are some kind of snobs, like audiophiles. The important thing in vintage computers is to fix them, use them, people complaining should open a museum and shut up
@@TheInsultInvestor it's not like they're writing all over a rare box or game lol. Hard drives aren't special
@@TheInsultInvestor The marker comes off with a little ISO. I've been a sysadmin for over 20 years and it's something I started doing early in my career. For old drives finding writing on them is usually helpful, especially if it is the drive geometry. It's not like MFM, SCSI, or IDE drives are rare or reliable.
I did something a few years ago with a 2tb drive that failed, it would definitely spin up but not work, no sound of the armature moving around, I tapped it is a few times, nothing. I had my Music and some other files on it, I had another 2TB drive that was pretty much identical but older, I took the bad drive apart and I swapped the armature from the working drive to the dead one and moved the PCB over just in case it had to be for calibration or some sort handshake thing with the armature, It worked, I moved all my stuff off it without the top on it and never used that drive again.
I did all that in room that wasn't filtered of dust, but I mean the stuff I had on it I could replace if I needed to, but why no try it, it worked well, a lot less problematic than I thought it would have been.
Late reply but that sound is soothing/COOL . As a "old timer". Thank you Adrian for all your videos. As a enthusiast, I miss the days "listening" to HDD's knowing it running :P
Looks like we're entering the NO SPIN ZONE
MFM drives are so cool... when they work :D I have a couple of ST225s that are recognised by the controller and sound fine, but low level formatting fails on them, I should also try swapping out the boards from a working one some day.
Jurassic Park John Hammond :"Come on, Cone one little one" ... did you ever see that ?
Ahhh I sometimes miss the sound of those old drives spinning up. I think my first hard drive in an XT was a 225, or atleast one very similar. Good times. 😊
28:00 - You can disconnect temporarily the stepper motor from the PCB, then turn on the drive. With the stepper motor disconnected (hence no power applied to it) you'll be able to move it freely while the disk is spinning, without the need of doing that while switching the drive on and off repeatedly.
Drives of this era don't like being connected to other controller. Changing controller usually requires a new low-level format.
7:05 spinrite is the hhd king. used to listen to windows weekly and security now ehen i started my it job decades ago.. fun times. cool channel.
I used to use RLL controllers on MFM drives. There was a Norton disk utility that I can't remember the name of that would find bad sectors and I would run that over and over again to mark out the sectors that couldn't handle the 50% data density increase.
Retro computing concert: Drivival.
First band: 8-bit Blues
Second band: Head Crash
Closing act: Motherboard
I wish you used some wire wrap and tried to reconnect the read heads just to see if it fixed it.
The very first step if any mechanism is stuck is to disassemble and clean all its parts! Sometimes up to three times in a row. Adding excessive lubricant can make it even more stuck.
To answer on the drives being low level formatted to work again; basically, yeah, magnetisation will naturally fade over time. The technical term is bit rot. It's why most floppy disks no longer work since they're more susceptible than hard disks.
It's generally recommended, even with newer hard disks, that you should refresh your data(rewrite all data on the drive) every 3-5 years.
Some drives can last much longer(an estimated 70 years), but that's with ideal temperatures/humidity and regular usage. A good quality of drive also makes a difference.
I ALWAYS love your commentary on these, Adrian!!
Based on my knowledge of bit rot and other electronic repair your theory about the loss of magnetism is holy believable. Especially with being able to fix it with a format. Newer drives that would take a long long long time to happen bit rot is rather slow but with older drives it's possible. Although that's probably more an issue because of miniaturization. Large drives like some of the very very early hard drives like a hawk drive tend not to have this problem simply because the sectors are so big like physically.
I used to have problems with these old disk drives, the problems came down to the spindle bearing. The spindle would seize and I found that the grease in the bearing had dried out and (presumably) created a hard lump that prevented rotation. Rotating the disk by hand freed it up and thankfully I never had to apply oil, I was concerned that excess oil would get onto the disk surface.
Another thing that you could do in your testing procedure is use a CRO to examine read head signals (there are test points for this) and to check the read data signals on the Interface cables.
I had an HD fail as I was trying to back up files off of ZIP disks on a new Mac Pro G4 back in the day. Luckily apple lent me a new drive as was considering Drive Savers. PCB and firmware were idential. Swapped the PCB and saved everything. Saved a grand.
I remember changing the circuis from one 80 GB disk from other 40 GB both ide; same brand and they 80 GB drive si still running! 😊
I wonder if you take a mini torch and heat the bearing a little it'll loose old grease and also make the oil flow in just like solder.
Don’t use a torch - use an electric heat gun. The torch has water vapor from the combustion process; and the temperature discontinuity from the flame edge tends to be rather abrupt - too far away and it won’t heat up, move it a little bit closer and you’re setting it on fire.
I can remember MFM drives featuring in new systems back in 1989 FWIW.
I really enjoyed this episode. My first one, or two PC's had MFM drives... I remember the dual ribbon cables... but parallel IDE replaced MFM pretty quickly.
yes, modern drives have lots of algorithms and programing to refresh tracks as well as move/restore damaged data. it does a lot of what spinrite does silently in the background. once a modern drive starts going bad, its usually been going bad for quite a while.
i think you can do a refresh in spinrite.
You should low level format any drive when you change controllers, and you should enter the known bad tracks at that time, because sometimes they format and test good for a little bit, then fail. This is from experience of selling them for years.
Wondering if those steppers from old drives are standard unipolar/bipolar designs that could be used in all sorts of DIY projects instead of being E-wasted. Likely not the voice coil type in the Seagate (although I can still imagine some use cases for those), but the full 360° ones in the belt driven head assemblies, why not? Quite valuable and durable pieces of tech, unlikely to break anytime in a century, I'd probably buy used units at a discount instead of new. Also unusual form factors could be an advantage in certain cases.
Dont scrap potentially fixable ones, like I did way back when... Despite Adrien's huge stockpiles, they're getting hard to find!
Favorite MFM drive? I’ll start: Mitsubishi MR535, 40 meg MFM, 60 meg RLL, supported 1:1 interleave and weighed a ton. Never saw one come back and we sold hundreds.
That stepper motor on the bad ST-225 might be worth scavenging too...
The servo tracks on various drives act as a sanity check for the drive itself, if the drive hits cylinder 0 and bumps, then the servo track is damaged.
I recently discovered I had an external low profile case fitted with an AdSCSI Plus ST (for Atari) connected to an ST225 (which doesn't spin atm). I don't know yet if I'll try and get it working again, but if I do then I'm sure these videos you've made will be very useful.
16:15 - I don't know about IDE, but SCSI disks can be low level formatted. IRIX's fx allows one to do this, though it spits out a hefty warning that doing so can ruin a drive.
I've solved stiction with pulsing 12V into the drive. Sort of knocks it free, then running it for 5 or 10 minutes and the drives were good to go.
It's been a while since I used Steve Gibson's Gibson Research Corporation - Spinrite. Wow that takes me back to the days of Ray Gwynn's Double Aught Buckshot driver X00.sys which was needed to run more than a coupe of com ports.
SpinRite V6.1 is in prerelease development News Group for NG members.
MFM hard drives do need to have their format refreshed once in a while, because they will lose their magnetic formatting, which is what low-leveling does. i think you and Usagi Electric might be working on similar things, in that regard.
IDE and later hard drives not only don't need this, you will damage the existing magnetic format and alignment, rendering the drive useless. this is why taking a platter from one hard drive and putting it in another will not only not work, it cannot be undone.
if a hard drive only works when it heats up, it is heat damaged and will never work properly again without some heat.
ps, writing on a hard drive's body won't hurt it at all
17:00 I don't have experience with hard drives, but I definitely did this with 3 1/2 floppy. Many floppies that would fail could be low-level formatted and serve reliably for many writes afterwards.
This was definitely a time where you would write .rar files to floppies with heavy error correction, but I mean it worked, even if the error correction had to correct for some bad sectors.
Awesome stuff Adrian! as usual!
If that CDC drive is a Wren, you might want to talk to Usagi Electric - he's trying to rebuild a Centurion, and might be able to make good use of it.
As for the st-225: it's about 50/50 either way, but I lean toward a mechanical problem. (in before I view the startup...)
Upvote for this! :)
I am curious if you attempt to solder tiny wires to the head if you'd be able to read the drive. I also wonder how common this particular problem is. Perhaps a bit of delicate soldering may recover a number of such drives.
Yes. It would be only for experimenting, because once opened the drive is not reliable anymore. But I was expecting Adrian to attempt some tiny soldering before e-waste. Just for the video. It would be incredible if it just "freaking works" after the soldering... 🤯
@@GYTCommnts If he continues to fuss about with hard drives, it may be useful for him to get a little glove box / cleanroom.
As a control systems engineer, I'd love to see what the firmware looks like for these drives. I've marveled at hard drives , and worked on similar technology professionally (piezo -servo micro-positioning)
Very interesting video. I have old Microscience HDD and I know that it worked ok last time around year 1990. I am collecting knowledge to turn it on again but for now I am not doing it because don't want to damage something.
Lovely to hear all of those old noises again
Perhaps that corrosion is from organic type flux used during assembly all those years ago…
Those old ST225 disks look so simplistic. I wonder if we could produce some new ones for nostalgic purposes. :)
Using that miniscribe drive as a main example I have a few questions about MFM drives ..
- is the noise from that particular stepper motor normal or abnormal?
- if abnormal, what would the problem be with it (ie. is some internal part of the motor assembly worn)?
- is that stepper motor noise the normal type of noise that you hear when MFM drives are operating? (I'm used to hearing a similar noise which I always thought was the data being transferred)
- if all MFM stepper motors make a similar sort of noise, why is that miniscribe motor so noisy?
- is that miniscribe stepper motor, and the spindle noises too I guess, the product of bad or cheap parts being used, or an inferior type of hardware?
- can you remove an external stepper motor from a hard drive in order to take it apart and service it or replace it?
- if the spindle bearing assembly has a bad/worn/uneven roller can you replace the bearing assembly entirely?
- If, as you mentioned, the original magnetic properties of the existing tracks/data on an MFM drive can weaken over time and the drive won't see/read them any more, or if the head assembly has moved out of it's original alignment and just doesn't see those tracks anymore, can you just somehow realign the heads back to where they were and have them read the data again?
- If so, assuming those old tracks are there, how do you get the heads to scan for those original tracks?
- Can the heads be moved in non-standard increments to try to find old tracks without special hardware or software?
- Do you need some sort of special hardware/software or forensic type tools to scan for old tracks that the head is not aligned to?
- If you were to find those old tracks can you get the hard drive to automatically re-write all the existing data so that the magnetic strength is increased and the drive can read normally again?
I guesss (why did I type 3 s's there? I'm not in Slytherin) I'm curious as to how far you can strip down, diagnose and repair or replace the individual components of an MFM drive both mechanically and electronically. Can each and all individual components be replaced?
I know that you can't open a sealed hard drive case to open air because of dust and particles but lets assume that you have a clean room to work with. I'm curious to see whether you can replace all the parts of an MFM drive that generally degrade and go bad, put in something newer or better/modern and have the drive be improved to a level of reliability that's good or maybe even better than the drive original design would allow it to be.
You can only get at one of two stepper bearings with your oil, and these use grease, not oil.
The way to get these reliable is to replace the stepper motor. That said, if you just keep on keeping on excersizing those heads, extreme to extreme non-stop, the grease should soften and heat up. You should do this for 24 hours plus.
After this, as long as you spin it up once a week or so, it should stay with you.
Ahhh, stiction... the bane of my 128 MB Connor IDE drives. Thankfully they lasted until I had upgraded to a new machine so I didn't lose anything. That's one phenomenon I'm happy to never encounter again.
Interesting video👍.I rememer when i saw one of old full drives in person (i think it was a scsi drive) and picked it up,It was heavier than a brick at the same size😂.they have thick layers of metal.
I would love to see a video on sticktion with some exploratory surgery on a drive to try to definitively determine what causes it. I saw a video no too long ago where someone disassembled a drive with sticktion, and discovered it was _NOT_ the bearings. After the head and everything was removed, the platters rotated freely. He thought it could be magnetic in nature.
Stiction is the _heads_ sticking to the platter. Since heads park at the edge of the platter it takes a lot of torque for the motor to start up if this happens.
@@eDoc2020 Are you aware of any way to fix it? The drive works fine after it gets going.
@@olepigeon You'd probably need to replace the heads and/or the platters. I don't really know anything about what makes them stick so I can't say more.
Love the wd components. High quality
When a hard disk makes 'clicking' noises while spinning up it may be due to damage on the disk, however most disks do have a landing zone, which is also a 'take off' zone. The heads won't move off that zone until the motor is up to speed, and then the heads will lift and fly over the good part of the disk. So unless the damage is great enough to destroy the heads, the drive will work even though it's making weird noises during spin up.
I think you can low level format an IDE disk, there should be a disk init command in the IDE specification. However, the format is probably fixed in the drives firmware, and cannot be specified beyond a few 'canned' choices (drive specific).
Thanks Adrian that was a trip down memory lane :).
Hey, I have that same HP LCD monitor. It has so many inputs, it makes a great throw-down monitor.
That 225 might still be repairable if it's just a few disconnected wires. It might be hard to pull off, but I'd still give it a try.