I was a hard disk engineer for much of my early career... I worked on DEC DF32 drives which didn't even move the heads across the disk surface.. 4 Read/Write coils per head, and four heads per disk, gave a total of 16 tracks, storing 32k (12 bit, it was a PDP8 device) words... the early ones still used individual transistors, typically 8-12 on each board... these were later replaced with TTL, typically two ICs per board..... I later worked for a company called Priam, after some Greek hero or other, who made Winchester drives in 14", 8" and later 5.5"? We got up to about 760meg by about the time I left.... It's staggering how much more memory I have in my phone, than we could store on a few spinning plates only 30 years ago....
@@wimwiddershins Pretty much all modern drives use something called "embedded servo", where the servo carrier is embedded in wedges between the data sectors. Originally drive manufacturers didn't like it because it reduced drive capacity slightly, but it's been the standard for pretty much everything since the mid-1990s. What Quantum was dong with the optical sensor was a cost-saving measure; a standard "dedicated servo" (where all the servo information is recorded on one disc's surface and the others are used only for data) drive required a lot of bespoke analog hardware at the time, which was expensive, so Quantum split the difference and put an optical sensor on the head arm, so that they could treat it like a stepper motor. I'm pretty sure these drives also used an early, simplified version of embedded servo to keep the heads on track once the right track was selected; a few stepper motor drives did this, too.
To my knowledge only Quantum used the LPS (Light positioning system). All the later Quantum drives with "LPS" also used this system for servo feedback of the head position.
@@ultrametric9317 No, Steppers don't need feedback, as they are repeatable (usually) by design. Anything voice-coil needs some sort of feedback to close the positioning loop. Most modern drives use feedback derived from the magnetic pattern encoded on the platter itself, LPS used optical. Steppers fell out of favor very fast because they are slow as hell.
@@Ingineerix "Optically assisted Winchester"...around 1998-1999 they were talking about that as if it were a new technology. Quantum (and I believe Seagate as well) was doing that long before that.
Isn't it crazy that they put all that engineering into this device... and the failure points were just cheap parts. A piece of rubber and a bearing. Not the electronics, or the drive head. Not the platter or the drive mechanism. oh no no. A piece of rubber and a bearing. LOL
Pretty common really, a couple centuries ago a big steam engine investment would often break down due to a bearing or seal as well. These parts go through a lot of wear and tear!
I took apart a document feeder from a copier the other day that cost over $200 and sounded like a dirtbike due to a gear the size of a pea getting a crack in it. I could tell you more about printers that are perfectly fine and failing just cause a tiny foam pad is breaking down...
In some cases the issue is that at the time they were made, the technology for making the rubber or bearings wasn’t what it is todays so even the best just didn’t last that long especially with heavy use. Gradually over time they find new ways processing the rubber so it ho;DTS up better or new ways of designing bearings so they last longer. But this will only get you so far as their are issues that your not going to overcome with better rubber or better bearings such heat issues, wear and tear issues, exposure to the elements, etc. For example, all disk drives can only tolerate so much heavy use before the start to fail, all SSD drives can only be rewritten so many times before the individual cells all fail, Lithium-Ion batteries only can be recharged so many times before they don’t old as much of a charge, and so forth.. I was just reading about the invention of the lightbulb and how early light bulb designs (not invented by Thomas Edison BTW) burnt out too quickly due the wrong filament material being used or the fact they were not encased in a vacuum sealed bulb. What Edison invented was a filament technology that lasted long enough to make the lightbulbs useful for the home (along with technology getting electricity to the home and thus his lightbulbs). You can see the same improvement in reliability we’ve see with storage devices with cars. Early cars frequently broke down a lot and required constant maintenance but we progressed enough since then that is something modern cars don’t suffer from so much, at least with more reliable brands.
used to install the 40Mb ones into a number of IBM 5155 machines we had. Seem to remember that the heads auto parked when powered down for transport. Also was a TSR that flashed a '+' in the top right of the screen to show disk activity. The 5155 with dual floppy disks and a Plus Hard Card was a top machine. Think they were about a thousand dollars back in the day.
Those rubber bumpers in Quantum and Conner drives are notorious for turning to goo, but you can replace the bumpers or wrap them in tape and the drive then works fine after (if that's the only problem). You should put it back together!
23:57 that's probably an Alnico (Aluminium Nickel Cobalt) magnet, it's a pretty common magnet type until around the 90's when neodymium starts getting popular.
A lot of failures on these old drives were down to optical encoders, bits of plastic, the "rubber" parts, dirt in the bearings and so forth. They are not quite as susceptible to dust as people think, and you would usually get away with opening them up in an improvised "clean room". We used an old style dot matrix printer sound deadening hood as the "clean room", but I've seen a large plywood box with a perspex lid used as a clean area with a fan drawing the air out and a dust filter on the sides to keep the dust to a minimum. Think nuclear glove box, but without the hazardous contents and you get the idea. In the field I've opened them up in clear plastic bags and "operated" on them. They weren't particularly cheap, so fixing them could save you a lot of money, not to mention that often customers were lazy with their backups, so if you could recover the data you would dig them out of a hole.
This board actually has an internal ST506/412 disk interface. The D7810 is actually doing most of the work on the drive side of the board. It's an 8-bit microcontroller. The SMS OMTI20513 is the MFM disk controller and the DIP chip next to it labeled "Plus" is most likely a mask ROM containing its firmware.
Thanks for the analysis! Yet Adrian is also correct calling it "8-bit IDE" like. The first IDE drives were just the MDM disk controller and the hard drive electronics on a single PCB - with the only difference that for IDE this single PCB is drive-mounted and connected to the ISA bus using a cable and a "transceiver card", commonly called "IDE controller". In the hard-card, the transceiver card is eliminated by having the single PCB connect directly into the ISA bus.
@@tw11tube The drives were either MFM or RLL; given the capacity of this drive, it's likely to be MFM (the early drives tended to be 5, 10, 20 MB for MFM and 30MB, 60MB, etc. for RLL as they were typically 1.5 x the capacity of the same physical MFM device). IDE just meant Integrated Drive Electronics, which became known ATA when it had its own connector (which is a subset of the ISA bus).
I have just looked up the datasheet for the D7810 (pdf1.alldatasheet.com/datasheet-pdf/view/87657/NEC/D7810.html) and it looks to me like a hybrid of a Zilog Z80 and an Intel 8085 with some extras added on: 8-channel ADC, dual timers, dual zero-crossing detectors, USART, and hardware 16-bit multiply/divide instructions. Quite an interesting chip, actually.
There's a full wiring diagram and description of the chips for the original HardCard in US patent no. 4639863. That document says the ENDEC (the DIP to the right of the big OMTI chip) is an OMTI 20527, which is 2,7 RLL. This HardCard 20 is a later revision, and is different mainly in that it uses an OMTI 5055 combo chip (which combines the OMTI 5050 and 5060 mentioned in the patent), and a semi-custom version of the OMTI ENDEC. The chip below the big OMTI chip is the host bus adapter, and from what I can tell from the patent, it talks to the 5055 and the 7810 using SASI (like Adaptec's XT controllers do; I'm not sure about other brands). The 7810 firmware and the BIOS are in the ROMs in the lower left quadrant of the board.
I had a Connor Hardcard in a compaq luggable. I had no problems with it. I loved the + icon it put in the corner of the screen to show it was accessing.
In the 80's I used to work for xTech UK. We made the "Insider" hard card. A competitor to the card you have. These are made of basic parts. The chassis/case, the drive and the controller. The drive made by one company, the controller by an other and the chassis an other. The company on the sticker like us (xTech) would take these parts and put them together. In fact we designed our chassis and had them made but a metal work company and also patched the eproms on the controller cards to make them work on XTs as well as ATs. The big thing we did was have the disk powered connected to the PSU, where as the card you have gets it's power from the bus. We considered this a problem as the bus was never meant to carry the start up current that the disk pulls. I hope this info is of interest to you.
Those early drives used to be notoriously reliable, I once cleaned a 20MB drive with some Cleaning petroleum ( that's petroleum that leaves no residue) and then closed it and re-low-level formatted it. First on a MFM controller (as 20MB), and later even on a RLL controller as 30MB. And it worked without a hitch.
I agree. They got a lot of abuse with users and software guys carrying them round like USB sticks for software installs and they always worked. You could even pin them to be second cards in a 2 smart card machine.
My first job was the work study program at my high school. I got to work for the computer teacher, fixing computers around the school. One day he put me in the back room full of dead machines and taught me to take the hard cards out and sling them around really hard and see if that makes them work again. It worked far more often than it should have!
@@Walczyk yes. one data recovery trick involves using a toaster on heat-damaged platters. a very brief amount of time on top of a hot toaster, and with quick action, you can grab the misaligned data.
You can fix the next one. Lightly oil the bearings, remove that rubber, and I bet it will run fine. You can slip a tiny bit of capton tape over the sticky rubber once freed. May need a reformat and lock out bad sectors but there's no reason they can't be restored to working condition.
I approve! He can also, instead of removing the whole trim pieces, mark the points where there are holes for the screws, and remove the screws without damaging the whole trim uwu
over the bump stop? maybe use some solvent to remove the worst offending rubber, then add capton tape to make up the difference in diameter, and then add super glue on the side not facing the heads to prevent the tape from unraveling in the future
(15:49) That was a fairly standard hard disk internal setup for that period of drive architecture. That optical encoder was akin to the hard sectored 8-inch floppy disks of the mid-70s, and early 80s.
Agreed, there were very few 3.5" drives that were using the physical encoder and the MFM / RLL interface; most 3.5" drives had integrated drive electronics, AKA IDE and used tracking signals written into one of the platters.
I remember I used to have one of those same Hardcards but never thought to take it apart. Thanks for taking it apart and letting us see inside. That is a shame that they used rubber and not something more durable like a hard fiber or teflon but hey now you know why it was making noise.
Anything harder and the heads would vibrate too much when they hit the stop. I'd love to see a metal frame like this made for XT-IDE cards and an associated IDE hard drive, or maybe 3D-printed would be an option. I've got a Hyundai XT clone that only has two half-height 5.25" slots, and one of them is currently taken up by the hard drive. A hard card would be perfect for that chassis. Could always just use a CF card, but hearing the hard drives spin up and tick really touches my nostalgia.
Quatum drives also have that same failure. The guys at 68kMLA discussed on how to fix it (cut out/remove/replace the rubber bumper) and successfully brought back a dead drive. From this end it seems to be a simple fix: remove the cover, oil the bearing and remove/replace the rubber bumper without removing the heads or the platter. Care to attempt such a surgery on another dead hard card?
Hard cards were common in the late 80s and early 90s. A lot were kits. The bracket and 8 bit SCSI card, add the 3.5 SCSI drive of your choice. I bought one for a 40mb Connor drive i had.
Does it say anything about how many drive installations I did over the years, where to this day I can still remember the standard WD debug code you mentioned to low-level format the drives (and also the later 1006V SRx RLL boards, G=CC00) and still remember the drive settings for the Seagate ST-251 drives of 820x6x17, or Misubishi MR-535 (977x5x17 for MFM or x26 for RLL capacity.) 40MB or 65MB!
He didn't see or mention it in the video, but "ENCODER" is silkscreened on the main PCB, along with "R/W, MOTOR, & MOTOR." Pretty cool breakdown video.
@@darkwinter6028 Most printers use them these days, too, since stepper motors can't easily resolve the DPI that modern printers can produce. HP was using and optical servo tape on the print head carriage as early on as the first DeskJets, though a few other designs used raw stepper motors for a while.
@@lee4hmz I’ve taken apart printers that used a Mylar optical tape for position sensing and a regular DC motor to run the head back-and-forth; no stepper involved. Similar arrangement for the paper feed; the feed rollers just had an encoder wheel on it, and a DC motor. It’s possible that the motor was run in a servo loop; but it’s also as likely that the motor was just turned on and the print nozzles clocked off of the encoder pickup. 🤔
My first IBM compatible was an old Compaq luggable and it had one of these Hardcard 20 in it. That was back around '92 or '93. it worked fine as long as I had it. I wouldn't expect it to still be working now.
The Plus Hardcards were the Cadillac of harddisk-on-a-card form factor, back in the days. Most of the other manufacturers offered off-the-shelf hard disks with an interface card, that occupied the space of two expansion slots. The Plus Hardcards needed only one. I had, over the course of running my IBM PS/2 30-286, used two such cards, first a 40MB model, then a 105MB model (back in the days, it was plenty of storage). Never regretted the purchase. Wish I still have those cards. They are highly collectible now.
I have a fond memory of this device. I purchased one of these at the fleamarket at the Trenton Computer Fair and mounted it in my Amiga 2000 which was equipped with the Commedore Bridgecard. That was my first Amiga hard drive using a Janus drive with 19 megabytes of it's 20 dedicated to Amiga use. I became an expert at creating a slimmed down workbench to boot the works up and transfer control. I never had any problems with it. and I even ran the AMAX emulation from it as well.
8:03 I have tiny heatgun about 300 watts, to glue hobby things or so. (embossing don't know) bought it for stuff like this. It's not really hot that it burns everything, but helps soften glue.
Glass scale linear encoders are used widely in precision positioning applications, such as in CNC machine tools and coordinate measuring machines. They're less convenient and more expensive than sensors that can be interfaced directly electrically like Hall sensors, but typically can measure with better resolution. I believe most voice coil motors (VCMs) would use a few Hall sensors as feedback. I'm not totally familiar with all of this, though, The type of positioning measuring systems that I work with more commonly use optical interferometers, a whole other level of precision.
Newer hard drives use embedded servo - the positioning feedback uses signals from the drive heads themselves, reading special areas interleaved between the data sectors on the tracks. Complicated to design, but very cheap in terms of part count.
1986 was really too early to commercialize the 3.5" drives. They had more or less just figured out how to miniaturize the drives and had yet to figure out how to make them reliable when the hard cards appeared. I don't know of any of those first 3.5" drives that lasted very long. Back then the listed MTBF was typically 5-10,000 hours, compared to the 40,000 - 80,000 hours we have today, which means they knew the drives weren't going to hold up to extensive use 16:55 seen this parking brake release on much more modern drives as well. Once the discs are up to speed the plastic arm swings aside, presumably by the air, and the heads are released. It appears to be simply because the RPM sensor on the spindle motor isn't connected to the head logic at all, so the flap serves as extra protection to know when the disk is up to speed. It's probably much more critical on these old drives which relied on the air pressure of the platters spinning to maintain the correct distance between the head and the platters. You will also see the plastic arm swing back when the platters slow down, so presumably it's somewhat spring loaded and tell the heads to park. On this drive it looks like the park position is determined purely by the mechanical position
Quick tip as an someone who has a very similar iFixIt toolset; the inside of the top half of the case with the bits and drivers is divided into small square sections that are good for holding all the small screws and other parts. Not sure if it’s intentional design but it’s been helpful for me, saving me from having to search for and sort through the right parts to put things back together, some of which inevitably end up in hard to reach parts of my desk or on the floor.
the last time i saw one of these, it was about 1996, at a data recovery job. the owner had stacks of old hard drives with all of the problems old hard drives have. the hard card 20 did not work. i don't know if i'm misassociating a memory, but i seem to recall the failures having something to do with that tracking glass and losing tracking, so the heads could no longer read the data properly.
The white specs that where on the platter would never get off by themselves because they are electrostatically adhered to the platters. Those specs usually fall out of the filter after prolonged periods of disuse when they discharge, and upon spinning up the discs would attract them statically and that would make the hdd appear to be broken because the heads would not be able to read the tracks where the specs where. Those tracks usually contained the actual hdd geometry information.
I have seen this type of light positioning system on a huge harddrive linear motor. The same but very much larger. It actually uses a moire pattern, on one side of the glass strip is a light with a piece of film with lines on it with a slightly different distance between them as the lines on the glass strip on the head arm.
I actually have fixed a Conner drive by scraping off the goo and replacing it with electrical tape. It worked pretty well with no issues after that. Still use it in my 386.
The optical head positioning is typical for the time on Winchester drives. IIRC this was the source of the catastrophic failure of the Computer Memories 20Mb disks put into the original ATs in 1984. There was a design flaw, something like accumulated dust on the sensor grid lines would cause the sensor to lose track of the head position. The first full-sized disks (e.g. ST-506) used rails to position the head I think - that's the source of the particular whirring sound they make when working. The rail is the part that had the marking and the optical sensor was attached to the read/write head.
I bought a used 40MB hard card back in the early 90s pretty cheap. It was nothing like this though - mine was just a regular hard drive with an IDE connector that connected to the rest of the unit. It was basically an IDE controller and IDE hard drive all in one with a 16 bit ISA interface. I remember just removing the disk from the hardcard and just connected it to a regular IDE controller. None of the casing was glued on to the hard drive either. It was pretty easy to take apart.
I actually got one of those hardcard 20's working after replacing the rubber stop a couple of years back. That was actually a boxed drive though, plastic sealed, never opened. Every other one I've had is faulty. I just got a 40 recently and the rubber stop was fine. But the drive doesn't work. It scrapes a bit as it spins, so I think the platter is delaminated a bit. Haven't gotten it to work.
Can you believe it, after watching your video I decided to oil that bearing that you pointed out. I got oil all over the disk (I have a 40mb which has two metal platters). So I cleaned that off with alcohol. Then after giving the platter a little push to get it started spinning, the drive works! I can scarcely believe it! I think you might be onto something with oiling that bearing. Now if I could only figure out how to oil the bearing in the main motor. The cover on the back is not able to be unscrewed, and if I remove the platters it'll probably never work again. Hmmm.
The reverse anticlockwise thread is often used on drive shafts as if it were clockwise it would have the tendency to unwind, I would use a of key and some form to lock the screw like locktight. Reminds me of visiting IBM at Havent UK for Brishish Standards Institution BSI they even repaired the obsolete big drives think 10 inch multi disk that banks/milatery used. I think that the drive spins at x speed to shut down the read head or heads move to the inside of the disk, and the drive and parked, then the drive slows down and stops. The reverse happens when the drive has reached the correct speed the head is released. This is beacuse the function of the air between the disk and the head is reliant on the speed. Later disks were sealed in clean air with only the head movement inside all other bit outside the clean area. IBM drives were known to be noisy in operation and not only IBM but all other drives were prone to head crashes, due to mechanical failure or out side vibration or dropping. If the head hit the disk at speed it will score it thus usually fatally, cannot remember the disk speed but it was fast. All mechanical devices have their weakness thus ssd's. Thanks for the video
WOW! I have not seen one of those for many years! I had a 80MB one way back. It was Quantum that made them. Windows 3.1x needed a driver as well (that was referenced in config.sys). Ironically, it's interesting that currently we are seeing similar tech being used again in regards to M.2 slots and NVMe Solid State Drives (SSD) directly attached to the bus. Have a great week... great video.
RetroSpector78 opened a bad Quantum drive about 5 months back an although it was a different year drive and had a different head mechanism inside it had the same gooey bump stop in it as well.
That's very interesting that the bearing is gummy. Lisa Widget drives are *very* sensitive to gummy bearings... I'll have to try giving that a drop of oil on mine!
Based on the traces it looks like one ROM is for the ISA bus interface, and the other is the hard drive controller ROM. The vane is a part of the autopark mechanism. Moves in place from airflow pushing on the vane. Spinning disks move quite a bit of air. See Tesla turbines for an example. Locking the head until the drive is at speed and the heads are floating before allowing them to travel to the data populated portion of the drive.
I have a working Plus Hardcard 40. It wasn't working at first, but It would sometimes come alive when I gave it a shake. I opened it up and my failure was similar to what was going on in your unit. I removed the parking plastic things and I put a piece of tape on the sticky bumper. After that it would work every time. I suspect that the power it gets from the ISA bus is not really up to the task of getting things up and running after all these years. My Hardcard 40 has 2 platters and 4 heads btw even though it's MfD is late 1989. Interesting.
I have had my battles with certain rubber parts turning into goo. I have a friend who has a doctorate in polymer chemistry, so I once queried if there was any way to re-vulcanize the material. "Little to no chance!" was the answer. And I got some more info, which I quickly forgot. I think there were 3 different vulcanizing processes to choose in the first place. The oldest one was based on sulphur and produced quite permanent hard results on latex. Other raw chemistries like polyurethane use different vulcanizing processes, including one that uses platinum to catalyze it. The benefit of the newer processes is that you can obtain softer results. But then the Cross Linkin is less complete and some solvents or acids, maybe even sweat can start a de-vulcanizing and the elastomere (rubber) turns into the dreadful goo. No way reversing that decay. Aside from one tape recorder pressure roller hardening and cracking, all my rubber (and foam plastic) decay experiences seem to be towards de-vulcanization. As to the optical sensor, it would have been interesting to know whether the lines formed an incremental or an absolute encoder. I think it most likely was incremental and the electronics kept count of the position, based on some established known home.
Now that you know the failure mode, it might be interesting to try and revive one of the other broken hardcard drives by carefully opening them up to clean up the goo, and maybe lubricate that bearing.
The Hardcard 20 is actually rotary voice coil actuated. Some might argue this is the first "IDE" drive, it's certainly got the electronics integrated into it. Quantum actually owned Plus Development, so yes, this is technically speaking a Quantum mechanism although not designed by their usual engineers. LOTS of companies cloned the Hardcard very soon after it's release. In 1986, almost ALL drives were still half height 5.25". The big bad ST-251 doesn't show up until the year after! The very few 3.5" disks that existed were very much all full height, and this was considered a "Slimline" drive, not just half height at that. 11:35 This is a counterbalance weight for the swing arm. This allows the drive to operate more reliably in the vertical orientation that it is positioned in, and probably helps simplify PID values for the controller a bit too. 12:21 the false top platter is mainly used as a spacer for the hub to keep it very level, and then secondary allows them to use the same motor and power values. The hub does not have to be retuned for lower power with less weight - it weighs the same 20 or 40MB. Helps to simplify production and calibration. 12:45 The chip here is the head pre-amp. The primary failure on these drives is the rubber bumper turning to junk and either contaminating the drive or simply putting it out of alignment. 14:23 This is a cheapo "speed detector." It "detects" when the drive is up to speed and releases the head assembly. The brass weight on the end there moves out of the way of the counterweight on the head stack when the spindle achieves a set speed, determined by the weight of that arm and the spring underneath it. It is actuated by the forced air dragged by the platters during rotation. 15:13 This is an optical feedback sensor, which was a new thing at the time and allowed Plus to forgo having an odd number of heads for a voice coil tracking surface. Embedded servo had not been introduced yet, so this was a convenient way to get extra data from a voice coil drive with a lower cost AND increase it's reliability. Servofade kills many early voice coil drives and this totally bypasses that problem by moving the tracking somewhere else. It's actually pretty fast, too. This went on to become a Quantum technology called LPS, Light Positioning System I believe, which saw further use on the ProDrive lineup. 18:15 Almost no drives of this era could unload the heads, only commercial cartridge systems were capable of this and a select few very rare drives such as the Lapine Titan (aka Kyocera KC-30A) could actually do this. 19:07 Again for compensating against the extra non-existent platter, this balances against the extra 2 heads you don't have.
Ah the curse of the Quantum hard drive! I've fixed a couple of later 40mb SCSI ones by cleaning the gunk out and replacing with some heat shrink, ideally needs to be done in a clean room tho i improvised one. Never seen inside one that early but it does look like repair would be possible.
In the '80s my company was the Canadian distributor of Hardcards. Quantum owned Plus Development. They set it up in the hope of developing a high-margin consumer brand to offset their low OEM margins. At the time, these drives were very reliable with far few warranty claims and returns than other drives.
Great video!! I always wanted to own one of those HardDrive. But now I know enough that I better release this idea. They guess all of them are failing after a certain age. Thx, I enjoyed the video until the end. cheers, Peter
I'd love to see a metal frame, or maybe a 3D printed frame, to go with an XT-IDE card and an IDE hard drive. I've got an XT clone with only two half-height bays that could use something like this.
The placeholder disk is probably needed for the head unlocking vane to work -- if there is no disk riding on the top, the platter wouldn't move as much air and it would not be able to move the unpark vane out of the way, leaving the head stuck.
Reading the comments I was hoping this would have been a more careful tear down and turn into a repair. Ive oiled and brought back to life about 7 MFM and scsi drives even drives that I felt were broken. Gummed bearings, head park mechanisms and the goo bumpers can cause weird noises. I recommend always approaching the opening of a drive as if its repairable since you never know.
Very early voice coil drives did not have the servo tracking information on the drive platters. So you had to had to have some method for closed loop head position control. The earliest drives just used a stepper motor and an optical "zero track" photoeye to know where they were (or like the apple disk 2 which just stepped the heads far enough to know that they were at the position needed). This is a later design, and you needed position feedback in order for the voice coil closed loop control to work. Modern drives use servo info encoded onto the platters themselves, mainly due to the higher densities.
This video reminds of the many videos you see on YT from folks like Techmoan and others repairing vintage cassette players, turntables, and other vintage electronic devices that use rubber belts. They all suffer at some point from the rubber belt deteriorating and thus needing to be replaced. It also common to see old rubber pinch rollers and rubber coated wheels becoming sticky and requiring special chemical solution that removes the sticky surface of the rubber on these roller sand wheels so it practically as good as new. Its just seems that the rubber used in electronics only last for som long, and while the longevity is better today then it once was it never going to be forever.
If I had to guess the lines going to the motor, it is probably power, ground, enable, and tach. I’ve taken a fair number of hard drives apart myself (and have a cluster of rare earth voice coil magnets in the garage atm), but I’ve not seen one with an optical track sensor like that either, maybe why the arm had to be offset? At least we now know why these are unreliable.
If you look closely there is definite scratch in the middle of the disc,just right in place where i believe vital information for boot drive are located. I have saw tons of simillary scratched hard drives even quite recent ones.
Maybe on NTFS drives but not on MFM and RLL drives. Track 0, Head 0, Sector 0 had to be good but would wear out over time. There was a trick or two back in the day. You could move the home sensor making it think another area of the disk was T0 H0 S0 but you would have to low level format the drive again. The format routine was sometimes in ROM if you new how to jump to it in a memory editor utility like DEBUG. Also you could swap the head cables on some drives to use the other side as Head 0. Some scratched and worn drives would work fine and other immaculate drives wouldn't work at all. Th smell of burnt aluminium was the killer smell of a dead head crashed drive.
1. brushless motor 2. ceramic magnet 3. probably that optical setup with the flex cable (flex PCB, BTW, not an FFC -- flat flex are the white thin ribbons used on modern laptops and C= Plus/4 keyboards) is an LED + a photodiode or phototransistor -- four wires, not two ;) A pity it doesn't use a standard IDE or MFM style connector -- then you could change out the drive and have it work again. Quantum BTW was a good company at the time. My first system had an 85meg Quantum drive, built 1991 sold 92 still spins to this day. I need to check data integrity but it predates the era of drives that autoconfigure -- it doesn't know how to tell a more modern PC how to talk to it, so I have the C/H/S stuff written on an old flash card I keep with the drive -- a modern USB adapter can't do anything with it, though, because of that. If it won't autoconfigure, a USB adapter or external enclosure just can't make heads or tails of it :( I wish there were something out there for these old drives that would let you manually configure them, but still speak modern protocols, purely for testing purposes...
That's why I proceed very carefully with such autopsies. ...just in case I find a cure in the process and patient get chance for new life ;-) However, your video and drive sacrifice will help in many future repairs. Cheers! S
Well, thanks to the sacrifice of this drive, we now know how to fix the others. I wonder if that metal frame could be used to mount an IDE hard drive, coupled to an XT-IDE card. The Hyundai XT clone I've been playing with only has two half-height 5.25" slots, I've got one of them occupied with the hard drive, and I'd like to avoid doing destructive mods to the case if I can, so a hard-card type solution would be perfect for this chassis. I know, everybody's going to CF cards these days, but I just love hearing the hard drives spin up and tick.
I had a hard card of some no-name make about the time this was made, but it had a "standard" 20Mb 3.5" IDE hard drive on it. When it became too small for my needs I was able to remove the drive and fit a bigger 3.5" drive (I think it was 60 or 70Mb or similar) as it used the standard interface, and the ROM on the card supported different sizes of disk. Since the drive was an IDE drive the interface was considerably simpler that the PLUS card you show here, but only slightly younger. It was a time of sudden change! I remember replacing it with an IDE card and 170Mb drive when I upgraded my motherboard only a year later, because disk prices plummeted for some reason and I'd got a better job with more money to play with!
Nice bit of dissection, shame some parts got broken you could have tried reassembling it to see if it would work even for a short will to see what was on it.
That head lever design allows the heads to move in a more radial movement, keeping them nearly tangential to the disc tracks. Perhaps this was needed to minimise signal distortion and maximise signal strength?
Wait! What? IC on flex in 1986!?! I had no idea that tech went back that far. Our college was still doing through hole in 1990. SMD was a thing (but we really did not touch it) but I don't remember any mention of IC on flex! edit: the bad rubber stoppers is the same thing that happens on the WD40 and 80 MB drives. You can take them apart and put something (I use gaffers tape) around the rubber. Obviously the drives will be vulnerable to dust intrusion afterwards but it usually works to be able to read/image the drives.
back in the day we sold tons of hard cards. Very rarely had issues with them, but then again they were all new back then. I have also seen a very old one in an original Osborn portable, and that computer was the only one out of a half dozen that actually booted, because it was the only one with an HDD. Then i had to try and remember what command line commands worked with CPM.
I went back and checked where that sticky part was when it was assembled. And I remembered that I saw something which I first thought is a little button that pushes in when the head reaches the end so the drive knows where the head is. Turns out, the thing I saw was a part of that sticky rubber ring. It's at 15:09, the black stickiness is on the arm.
That would be an interesting project to possibly convert one of these to use CF cards or SD cards. Although, it is unlikely to be IDE or IDE compatible so that’s probably a bust. I did cringe but it was an impossible task given it was the first one you took apart. Lessons learned, the next one could likely be salvaged. I have a bunch of SCSI Conners for old Macs and I would like to refurbish them properly in clean conditions. Great video!
I have a Hard Card 40+ that doesn't work as well. I would bet if it has the same sticky issue, I could possibly fix it!! Maybe a thin strip of electrical or other tape around that sticky grommet and oil the bearings. What HDD Lubricant is recommended?? I think I could use a heat gun to carefully get the plastic covers off without mangling them!
Lots of early drives used a Moire grating to locate their position... it wasn't until the late 80's that heads could track the data track based on disk data alone....
oh please try to get it back to life, this will be heroic ;-) these old stuff might be repairable in opposite to the "nano specific" devices from nowadays. I know, half a year too late for asking, but better late than never.
Should have tried putting the drive/assembly in the freezer for a few hours first before tearing it apart. Once frozen, the gooey rubber might have released the heads long enough for you to save the drive contents. Even with newer drives you would be surprised how often freezing a non-working drive ressurects it long enough to save data that would otherwise be irretrievable.
I remember "back in the day" (around 1991 or earlier,) when a tech bulletin was sent out by the manufactuerer regarding problems with stiction on hardcards, and it specifically stated: "Hold the hardcard in front of you at a 45 degree angle, and shake like a ketchup bottle!"
Moving the pivot allows the arm to travel in a path more perpendicular to the radius of the drive. Modern drives compensate for the nonlinearity in the arm's travel with better electronics, more precise head movements, etc. This is a mechanical solution that simplified stabilizing the read/write rate saving some complexity and HW/SW performance needs.
Knowing what you now...know :/ IF you were to attempt a repair of a HardCard I'd suggest 2 things. Buy or print some head combs. Little plastic things that Wonder-Bra (lift and separate) the heads from the platters and prevent damage. The last bit is a trick from growing mushrooms, I think it was part of the PF Tech Method....Heat your oven up and leave it on during the process.. Make the oven rack your workspace and perform your surgery on the drive there. The continious rising heat will prevent particulate matter from falling and coming to rest on your patients delicate innards. In fungi farming this was to prevent contamination and destroying a harvest.
We had 100 and 200MB quantum branded hardcards when I was a kid. Worked great in our compay and victor 286's if you could honestly say they ever worked well.
While not as bad as David's "dremel the non-standard screws and stick a paperclip in the power connector", this was still an unfortunate way to destructively disassemble the drive. It would've been interesting to leave the flexes intact and attempt to spool up the drive to see what was causing the issue. Maybe (combined with unsticking the heads) you could've taken a look at the file structure.
![Adrian's Digital Basement ][](http://yt3.ggpht.com/ytc/AIdro_k7CrQRe71qN6c0Ua7A1P_I8KkxjqLa70u0zYeCvdvnKw=s48-c-k-c0x00ffffff-no-rj)
I was a hard disk engineer for much of my early career... I worked on DEC DF32 drives which didn't even move the heads across the disk surface.. 4 Read/Write coils per head, and four heads per disk, gave a total of 16 tracks, storing 32k (12 bit, it was a PDP8 device) words... the early ones still used individual transistors, typically 8-12 on each board... these were later replaced with TTL, typically two ICs per board..... I later worked for a company called Priam, after some Greek hero or other, who made Winchester drives in 14", 8" and later 5.5"? We got up to about 760meg by about the time I left.... It's staggering how much more memory I have in my phone, than we could store on a few spinning plates only 30 years ago....
This drive has an optical head sensor. How do newer drives do it? Signal strength feedback or something?
@@wimwiddershins They read track info from the disk, like Adrian said.
Multiple discrete heads - so, zero track seek time… just have to wait for the platter to rotate to the desired sector.
AMAZING!
@@wimwiddershins Pretty much all modern drives use something called "embedded servo", where the servo carrier is embedded in wedges between the data sectors. Originally drive manufacturers didn't like it because it reduced drive capacity slightly, but it's been the standard for pretty much everything since the mid-1990s.
What Quantum was dong with the optical sensor was a cost-saving measure; a standard "dedicated servo" (where all the servo information is recorded on one disc's surface and the others are used only for data) drive required a lot of bespoke analog hardware at the time, which was expensive, so Quantum split the difference and put an optical sensor on the head arm, so that they could treat it like a stepper motor. I'm pretty sure these drives also used an early, simplified version of embedded servo to keep the heads on track once the right track was selected; a few stepper motor drives did this, too.
To my knowledge only Quantum used the LPS (Light positioning system). All the later Quantum drives with "LPS" also used this system for servo feedback of the head position.
Gives "magneto-optical" disk a new meaning lol
Well I think all early Winchester drives used the same idea, because they used servo motors, not stepper motors. See above. I could be wrong.
@@ultrametric9317 No, Steppers don't need feedback, as they are repeatable (usually) by design. Anything voice-coil needs some sort of feedback to close the positioning loop. Most modern drives use feedback derived from the magnetic pattern encoded on the platter itself, LPS used optical. Steppers fell out of favor very fast because they are slow as hell.
@@Ingineerix Ok thanks
@@Ingineerix "Optically assisted Winchester"...around 1998-1999 they were talking about that as if it were a new technology. Quantum (and I believe Seagate as well) was doing that long before that.
Isn't it crazy that they put all that engineering into this device... and the failure points were just cheap parts. A piece of rubber and a bearing. Not the electronics, or the drive head. Not the platter or the drive mechanism. oh no no. A piece of rubber and a bearing. LOL
Pretty common really, a couple centuries ago a big steam engine investment would often break down due to a bearing or seal as well. These parts go through a lot of wear and tear!
I took apart a document feeder from a copier the other day that cost over $200 and sounded like a dirtbike due to a gear the size of a pea getting a crack in it. I could tell you more about printers that are perfectly fine and failing just cause a tiny foam pad is breaking down...
Probably the only thing they didn't test at temp or something like that.
Capitalism
In some cases the issue is that at the time they were made, the technology for making the rubber or bearings wasn’t what it is todays so even the best just didn’t last that long especially with heavy use. Gradually over time they find new ways processing the rubber so it ho;DTS up better or new ways of designing bearings so they last longer. But this will only get you so far as their are issues that your not going to overcome with better rubber or better bearings such heat issues, wear and tear issues, exposure to the elements, etc. For example, all disk drives can only tolerate so much heavy use before the start to fail, all SSD drives can only be rewritten so many times before the individual cells all fail, Lithium-Ion batteries only can be recharged so many times before they don’t old as much of a charge, and so forth..
I was just reading about the invention of the lightbulb and how early light bulb designs (not invented by Thomas Edison BTW) burnt out too quickly due the wrong filament material being used or the fact they were not encased in a vacuum sealed bulb. What Edison invented was a filament technology that lasted long enough to make the lightbulbs useful for the home (along with technology getting electricity to the home and thus his lightbulbs).
You can see the same improvement in reliability we’ve see with storage devices with cars. Early cars frequently broke down a lot and required constant maintenance but we progressed enough since then that is something modern cars don’t suffer from so much, at least with more reliable brands.
used to install the 40Mb ones into a number of IBM 5155 machines we had. Seem to remember that the heads auto parked when powered down for transport. Also was a TSR that flashed a '+' in the top right of the screen to show disk activity.
The 5155 with dual floppy disks and a Plus Hard Card was a top machine. Think they were about a thousand dollars back in the day.
Those rubber bumpers in Quantum and Conner drives are notorious for turning to goo, but you can replace the bumpers or wrap them in tape and the drive then works fine after (if that's the only problem). You should put it back together!
I think that bridge has been demolished
23:57 that's probably an Alnico (Aluminium Nickel Cobalt) magnet, it's a pretty common magnet type until around the 90's when neodymium starts getting popular.
A lot of failures on these old drives were down to optical encoders, bits of plastic, the "rubber" parts, dirt in the bearings and so forth. They are not quite as susceptible to dust as people think, and you would usually get away with opening them up in an improvised "clean room". We used an old style dot matrix printer sound deadening hood as the "clean room", but I've seen a large plywood box with a perspex lid used as a clean area with a fan drawing the air out and a dust filter on the sides to keep the dust to a minimum. Think nuclear glove box, but without the hazardous contents and you get the idea.
In the field I've opened them up in clear plastic bags and "operated" on them. They weren't particularly cheap, so fixing them could save you a lot of money, not to mention that often customers were lazy with their backups, so if you could recover the data you would dig them out of a hole.
This board actually has an internal ST506/412 disk interface. The D7810 is actually doing most of the work on the drive side of the board. It's an 8-bit microcontroller. The SMS OMTI20513 is the MFM disk controller and the DIP chip next to it labeled "Plus" is most likely a mask ROM containing its firmware.
Thanks for the analysis! Yet Adrian is also correct calling it "8-bit IDE" like. The first IDE drives were just the MDM disk controller and the hard drive electronics on a single PCB - with the only difference that for IDE this single PCB is drive-mounted and connected to the ISA bus using a cable and a "transceiver card", commonly called "IDE controller". In the hard-card, the transceiver card is eliminated by having the single PCB connect directly into the ISA bus.
@@tw11tube The drives were either MFM or RLL; given the capacity of this drive, it's likely to be MFM (the early drives tended to be 5, 10, 20 MB for MFM and 30MB, 60MB, etc. for RLL as they were typically 1.5 x the capacity of the same physical MFM device). IDE just meant Integrated Drive Electronics, which became known ATA when it had its own connector (which is a subset of the ISA bus).
I have just looked up the datasheet for the D7810 (pdf1.alldatasheet.com/datasheet-pdf/view/87657/NEC/D7810.html) and it looks to me like a hybrid of a Zilog Z80 and an Intel 8085 with some extras added on: 8-channel ADC, dual timers, dual zero-crossing detectors, USART, and hardware 16-bit multiply/divide instructions. Quite an interesting chip, actually.
There's a full wiring diagram and description of the chips for the original HardCard in US patent no. 4639863. That document says the ENDEC (the DIP to the right of the big OMTI chip) is an OMTI 20527, which is 2,7 RLL. This HardCard 20 is a later revision, and is different mainly in that it uses an OMTI 5055 combo chip (which combines the OMTI 5050 and 5060 mentioned in the patent), and a semi-custom version of the OMTI ENDEC. The chip below the big OMTI chip is the host bus adapter, and from what I can tell from the patent, it talks to the 5055 and the 7810 using SASI (like Adaptec's XT controllers do; I'm not sure about other brands). The 7810 firmware and the BIOS are in the ROMs in the lower left quadrant of the board.
You mean if the ROMs are alive and well (they tend to fail apparently), it's quite possible to rebuild this thing? 😃
I had a Connor Hardcard in a compaq luggable. I had no problems with it. I loved the + icon it put in the corner of the screen to show it was accessing.
In the 80's I used to work for xTech UK. We made the "Insider" hard card. A competitor to the card you have. These are made of basic parts. The chassis/case, the drive and the controller. The drive made by one company, the controller by an other and the chassis an other. The company on the sticker like us (xTech) would take these parts and put them together. In fact we designed our chassis and had them made but a metal work company and also patched the eproms on the controller cards to make them work on XTs as well as ATs. The big thing we did was have the disk powered connected to the PSU, where as the card you have gets it's power from the bus. We considered this a problem as the bus was never meant to carry the start up current that the disk pulls.
I hope this info is of interest to you.
Those early drives used to be notoriously reliable, I once cleaned a 20MB drive with some Cleaning petroleum ( that's petroleum that leaves no residue) and then closed it and re-low-level formatted it. First on a MFM controller (as 20MB), and later even on a RLL controller as 30MB. And it worked without a hitch.
I agree. They got a lot of abuse with users and software guys carrying them round like USB sticks for software installs and they always worked. You could even pin them to be second cards in a 2 smart card machine.
I think you meant to say 'unreliable'
My first job was the work study program at my high school. I got to work for the computer teacher, fixing computers around the school. One day he put me in the back room full of dead machines and taught me to take the hard cards out and sling them around really hard and see if that makes them work again. It worked far more often than it should have!
heat gun time for the next one! a repair of one of these would be epic
heat and hard drives do not mix, and can cause head misalignment.
@@iceowl but even just to soften the adhesive?
@@Walczyk yes. one data recovery trick involves using a toaster on heat-damaged platters. a very brief amount of time on top of a hot toaster, and with quick action, you can grab the misaligned data.
You can fix the next one. Lightly oil the bearings, remove that rubber, and I bet it will run fine. You can slip a tiny bit of capton tape over the sticky rubber once freed. May need a reformat and lock out bad sectors but there's no reason they can't be restored to working condition.
I've done this on old Macs. Just tape the rubber with a slim slice of electrical tape.
I approve! He can also, instead of removing the whole trim pieces, mark the points where there are holes for the screws, and remove the screws without damaging the whole trim uwu
p.s.
That rubber is the mark of the beast of those quantum HDDs, I had one of them branded HP from 1993 @106MB that had the same problem.
u should make a video on how to do it. please!
over the bump stop? maybe use some solvent to remove the worst offending rubber, then add capton tape to make up the difference in diameter, and then add super glue on the side not facing the heads to prevent the tape from unraveling in the future
(15:49) That was a fairly standard hard disk internal setup for that period of drive architecture.
That optical encoder was akin to the hard sectored 8-inch floppy disks of the mid-70s, and early 80s.
Agreed, there were very few 3.5" drives that were using the physical encoder and the MFM / RLL interface; most 3.5" drives had integrated drive electronics, AKA IDE and used tracking signals written into one of the platters.
I remember I used to have one of those same Hardcards but never thought to take it apart. Thanks for taking it apart and letting us see inside. That is a shame that they used rubber and not something more durable like a hard fiber or teflon but hey now you know why it was making noise.
Anything harder and the heads would vibrate too much when they hit the stop. I'd love to see a metal frame like this made for XT-IDE cards and an associated IDE hard drive, or maybe 3D-printed would be an option. I've got a Hyundai XT clone that only has two half-height 5.25" slots, and one of them is currently taken up by the hard drive. A hard card would be perfect for that chassis. Could always just use a CF card, but hearing the hard drives spin up and tick really touches my nostalgia.
Quatum drives also have that same failure. The guys at 68kMLA discussed on how to fix it (cut out/remove/replace the rubber bumper) and successfully brought back a dead drive.
From this end it seems to be a simple fix: remove the cover, oil the bearing and remove/replace the rubber bumper without removing the heads or the platter. Care to attempt such a surgery on another dead hard card?
pretty risky cracking open a hard drive to replace these parts if you don't have a cleanroom
@@StevenS757 Yeah, but it's also a 20MB drive. The bits are large enough to see at 4K. ;-)
these hard drives had some kind of tracking problem that would develop over time, iirc. no amount of hardware replacement can fix that. it's dead.
Hard cards were common in the late 80s and early 90s. A lot were kits. The bracket and 8 bit SCSI card, add the 3.5 SCSI drive of your choice. I bought one for a 40mb Connor drive i had.
I can recall these when they were new! I can well remember having to use debug G=C800 to access the drive controller BIOS!
Does it say anything about how many drive installations I did over the years, where to this day I can still remember the standard WD debug code you mentioned to low-level format the drives (and also the later 1006V SRx RLL boards, G=CC00) and still remember the drive settings for the Seagate ST-251 drives of 820x6x17, or Misubishi MR-535 (977x5x17 for MFM or x26 for RLL capacity.) 40MB or 65MB!
That piece of glass is called a glass scale. It a type of encoder, which is still used today for the linear axis' on CNC machines
We called it the graticule. Like on a servo feeding back it's position.
He didn't see or mention it in the video, but "ENCODER" is silkscreened on the main PCB, along with "R/W, MOTOR, & MOTOR." Pretty cool breakdown video.
And the full length glass encoders on CNCs are the best way to position the machine; better than a shaft encoder on the ballscrew.
@@darkwinter6028 Most printers use them these days, too, since stepper motors can't easily resolve the DPI that modern printers can produce. HP was using and optical servo tape on the print head carriage as early on as the first DeskJets, though a few other designs used raw stepper motors for a while.
@@lee4hmz I’ve taken apart printers that used a Mylar optical tape for position sensing and a regular DC motor to run the head back-and-forth; no stepper involved. Similar arrangement for the paper feed; the feed rollers just had an encoder wheel on it, and a DC motor. It’s possible that the motor was run in a servo loop; but it’s also as likely that the motor was just turned on and the print nozzles clocked off of the encoder pickup.
🤔
My first IBM compatible was an old Compaq luggable and it had one of these Hardcard 20 in it. That was back around '92 or '93. it worked fine as long as I had it. I wouldn't expect it to still be working now.
The Plus Hardcards were the Cadillac of harddisk-on-a-card form factor, back in the days. Most of the other manufacturers offered off-the-shelf hard disks with an interface card, that occupied the space of two expansion slots. The Plus Hardcards needed only one. I had, over the course of running my IBM PS/2 30-286, used two such cards, first a 40MB model, then a 105MB model (back in the days, it was plenty of storage). Never regretted the purchase. Wish I still have those cards. They are highly collectible now.
I have a fond memory of this device. I purchased one of these at the fleamarket at the Trenton Computer Fair and mounted it in my Amiga 2000 which was equipped with the Commedore Bridgecard. That was my first Amiga hard drive using a Janus drive with 19 megabytes of it's 20 dedicated to Amiga use. I became an expert at creating a slimmed down workbench to boot the works up and transfer control. I never had any problems with it. and I even ran the AMAX emulation from it as well.
8:03 I have tiny heatgun about 300 watts, to glue hobby things or so. (embossing don't know) bought it for stuff like this. It's not really hot that it burns everything, but helps soften glue.
14:30 - that's the brake for the r/w head. It unlatches once the drive is up to speed (from the airflow generated)
Glass scale linear encoders are used widely in precision positioning applications, such as in CNC machine tools and coordinate measuring machines. They're less convenient and more expensive than sensors that can be interfaced directly electrically like Hall sensors, but typically can measure with better resolution. I believe most voice coil motors (VCMs) would use a few Hall sensors as feedback. I'm not totally familiar with all of this, though, The type of positioning measuring systems that I work with more commonly use optical interferometers, a whole other level of precision.
Newer hard drives use embedded servo - the positioning feedback uses signals from the drive heads themselves, reading special areas interleaved between the data sectors on the tracks. Complicated to design, but very cheap in terms of part count.
1986 was really too early to commercialize the 3.5" drives. They had more or less just figured out how to miniaturize the drives and had yet to figure out how to make them reliable when the hard cards appeared. I don't know of any of those first 3.5" drives that lasted very long. Back then the listed MTBF was typically 5-10,000 hours, compared to the 40,000 - 80,000 hours we have today, which means they knew the drives weren't going to hold up to extensive use
16:55 seen this parking brake release on much more modern drives as well. Once the discs are up to speed the plastic arm swings aside, presumably by the air, and the heads are released. It appears to be simply because the RPM sensor on the spindle motor isn't connected to the head logic at all, so the flap serves as extra protection to know when the disk is up to speed. It's probably much more critical on these old drives which relied on the air pressure of the platters spinning to maintain the correct distance between the head and the platters. You will also see the plastic arm swing back when the platters slow down, so presumably it's somewhat spring loaded and tell the heads to park. On this drive it looks like the park position is determined purely by the mechanical position
Quick tip as an someone who has a very similar iFixIt toolset; the inside of the top half of the case with the bits and drivers is divided into small square sections that are good for holding all the small screws and other parts. Not sure if it’s intentional design but it’s been helpful for me, saving me from having to search for and sort through the right parts to put things back together, some of which inevitably end up in hard to reach parts of my desk or on the floor.
the whole pad where the bit set is in the toolkit is magnetic as well for holding onto screws and bits, but of course requires the bits to be magnetic
the last time i saw one of these, it was about 1996, at a data recovery job. the owner had stacks of old hard drives with all of the problems old hard drives have. the hard card 20 did not work.
i don't know if i'm misassociating a memory, but i seem to recall the failures having something to do with that tracking glass and losing tracking, so the heads could no longer read the data properly.
The white specs that where on the platter would never get off by themselves because they are electrostatically adhered to the platters. Those specs usually fall out of the filter after prolonged periods of disuse when they discharge, and upon spinning up the discs would attract them statically and that would make the hdd appear to be broken because the heads would not be able to read the tracks where the specs where. Those tracks usually contained the actual hdd geometry information.
thanks for fixing the splash logo, no more skipping at the last frame and the color is the right one :)
I have seen this type of light positioning system on a huge harddrive linear motor. The same but very much larger. It actually uses a moire pattern, on one side of the glass strip is a light with a piece of film with lines on it with a slightly different distance between them as the lines on the glass strip on the head arm.
Really enjoyed this one, AB. The quality and precision of the parts from that drive are amazing...and from 1982...wow. Cheers from Canada!
I actually have fixed a Conner drive by scraping off the goo and replacing it with electrical tape. It worked pretty well with no issues after that. Still use it in my 386.
gentlemen, welcome back to the shop!
The optical head positioning is typical for the time on Winchester drives. IIRC this was the source of the catastrophic failure of the Computer Memories 20Mb disks put into the original ATs in 1984. There was a design flaw, something like accumulated dust on the sensor grid lines would cause the sensor to lose track of the head position. The first full-sized disks (e.g. ST-506) used rails to position the head I think - that's the source of the particular whirring sound they make when working. The rail is the part that had the marking and the optical sensor was attached to the read/write head.
(06:55) That is a multiphase, printed circuit stator, brushless pancake motor.
I bought a used 40MB hard card back in the early 90s pretty cheap. It was nothing like this though - mine was just a regular hard drive with an IDE connector that connected to the rest of the unit. It was basically an IDE controller and IDE hard drive all in one with a 16 bit ISA interface. I remember just removing the disk from the hardcard and just connected it to a regular IDE controller. None of the casing was glued on to the hard drive either. It was pretty easy to take apart.
I actually got one of those hardcard 20's working after replacing the rubber stop a couple of years back. That was actually a boxed drive though, plastic sealed, never opened. Every other one I've had is faulty.
I just got a 40 recently and the rubber stop was fine. But the drive doesn't work. It scrapes a bit as it spins, so I think the platter is delaminated a bit. Haven't gotten it to work.
Can you believe it, after watching your video I decided to oil that bearing that you pointed out. I got oil all over the disk (I have a 40mb which has two metal platters). So I cleaned that off with alcohol. Then after giving the platter a little push to get it started spinning, the drive works! I can scarcely believe it! I think you might be onto something with oiling that bearing. Now if I could only figure out how to oil the bearing in the main motor. The cover on the back is not able to be unscrewed, and if I remove the platters it'll probably never work again. Hmmm.
The reverse anticlockwise thread is often used on drive shafts as if it were clockwise it would have the tendency to unwind, I would use a of key and some form to lock the screw like locktight.
Reminds me of visiting IBM at Havent UK for Brishish Standards Institution BSI they even repaired the obsolete big drives think 10 inch multi disk that banks/milatery used.
I think that the drive spins at x speed to shut down the read head or heads move to the inside of the disk, and the drive and parked, then the drive slows down and stops.
The reverse happens when the drive has reached the correct speed the head is released.
This is beacuse the function of the air between the disk and the head is reliant on the speed.
Later disks were sealed in clean air with only the head movement inside all other bit outside the clean area.
IBM drives were known to be noisy in operation and not only IBM but all other drives were prone to head crashes, due to mechanical failure or out side vibration or dropping.
If the head hit the disk at speed it will score it thus usually fatally, cannot remember the disk speed but it was fast.
All mechanical devices have their weakness thus ssd's.
Thanks for the video
WOW! I have not seen one of those for many years! I had a 80MB one way back. It was Quantum that made them. Windows 3.1x needed a driver as well (that was referenced in config.sys). Ironically, it's interesting that currently we are seeing similar tech being used again in regards to M.2 slots and NVMe Solid State Drives (SSD) directly attached to the bus. Have a great week... great video.
RetroSpector78 opened a bad Quantum drive about 5 months back an although it was a different year drive and had a different head mechanism inside it had the same gooey bump stop in it as well.
That's very interesting that the bearing is gummy. Lisa Widget drives are *very* sensitive to gummy bearings... I'll have to try giving that a drop of oil on mine!
Based on the traces it looks like one ROM is for the ISA bus interface, and the other is the hard drive controller ROM.
The vane is a part of the autopark mechanism. Moves in place from airflow pushing on the vane. Spinning disks move quite a bit of air. See Tesla turbines for an example. Locking the head until the drive is at speed and the heads are floating before allowing them to travel to the data populated portion of the drive.
Not to be a pain but that's basically what Adrian said.
I have a working Plus Hardcard 40. It wasn't working at first, but It would sometimes come alive when I gave it a shake. I opened it up and my failure was similar to what was going on in your unit. I removed the parking plastic things and I put a piece of tape on the sticky bumper. After that it would work every time. I suspect that the power it gets from the ISA bus is not really up to the task of getting things up and running after all these years. My Hardcard 40 has 2 platters and 4 heads btw even though it's MfD is late 1989. Interesting.
I have had my battles with certain rubber parts turning into goo. I have a friend who has a doctorate in polymer chemistry, so I once queried if there was any way to re-vulcanize the material. "Little to no chance!" was the answer. And I got some more info, which I quickly forgot. I think there were 3 different vulcanizing processes to choose in the first place. The oldest one was based on sulphur and produced quite permanent hard results on latex. Other raw chemistries like polyurethane use different vulcanizing processes, including one that uses platinum to catalyze it. The benefit of the newer processes is that you can obtain softer results. But then the Cross Linkin is less complete and some solvents or acids, maybe even sweat can start a de-vulcanizing and the elastomere (rubber) turns into the dreadful goo. No way reversing that decay. Aside from one tape recorder pressure roller hardening and cracking, all my rubber (and foam plastic) decay experiences seem to be towards de-vulcanization.
As to the optical sensor, it would have been interesting to know whether the lines formed an incremental or an absolute encoder. I think it most likely was incremental and the electronics kept count of the position, based on some established known home.
Now that you know the failure mode, it might be interesting to try and revive one of the other broken hardcard drives by carefully opening them up to clean up the goo, and maybe lubricate that bearing.
What's that line in the movie? Oh yes, "It didn't die, you killed it!"
The Hardcard 20 is actually rotary voice coil actuated. Some might argue this is the first "IDE" drive, it's certainly got the electronics integrated into it.
Quantum actually owned Plus Development, so yes, this is technically speaking a Quantum mechanism although not designed by their usual engineers.
LOTS of companies cloned the Hardcard very soon after it's release.
In 1986, almost ALL drives were still half height 5.25". The big bad ST-251 doesn't show up until the year after! The very few 3.5" disks that existed were very much all full height, and this was considered a "Slimline" drive, not just half height at that.
11:35 This is a counterbalance weight for the swing arm. This allows the drive to operate more reliably in the vertical orientation that it is positioned in, and probably helps simplify PID values for the controller a bit too.
12:21 the false top platter is mainly used as a spacer for the hub to keep it very level, and then secondary allows them to use the same motor and power values. The hub does not have to be retuned for lower power with less weight - it weighs the same 20 or 40MB. Helps to simplify production and calibration.
12:45 The chip here is the head pre-amp. The primary failure on these drives is the rubber bumper turning to junk and either contaminating the drive or simply putting it out of alignment.
14:23 This is a cheapo "speed detector." It "detects" when the drive is up to speed and releases the head assembly. The brass weight on the end there moves out of the way of the counterweight on the head stack when the spindle achieves a set speed, determined by the weight of that arm and the spring underneath it. It is actuated by the forced air dragged by the platters during rotation.
15:13 This is an optical feedback sensor, which was a new thing at the time and allowed Plus to forgo having an odd number of heads for a voice coil tracking surface. Embedded servo had not been introduced yet, so this was a convenient way to get extra data from a voice coil drive with a lower cost AND increase it's reliability. Servofade kills many early voice coil drives and this totally bypasses that problem by moving the tracking somewhere else. It's actually pretty fast, too. This went on to become a Quantum technology called LPS, Light Positioning System I believe, which saw further use on the ProDrive lineup.
18:15 Almost no drives of this era could unload the heads, only commercial cartridge systems were capable of this and a select few very rare drives such as the Lapine Titan (aka Kyocera KC-30A) could actually do this.
19:07 Again for compensating against the extra non-existent platter, this balances against the extra 2 heads you don't have.
I have a couple of these that don't have stuck bumpers, or bad bearings... but do need low level formatted which seems to be impossible :(
Ah the curse of the Quantum hard drive! I've fixed a couple of later 40mb SCSI ones by cleaning the gunk out and replacing with some heat shrink, ideally needs to be done in a clean room tho i improvised one. Never seen inside one that early but it does look like repair would be possible.
You can at least recover the data after a home made clean room.
In the '80s my company was the Canadian distributor of Hardcards. Quantum owned Plus Development. They set it up in the hope of developing a high-margin consumer brand to offset their low OEM margins. At the time, these drives were very reliable with far few warranty claims and returns than other drives.
Great video!! I always wanted to own one of those HardDrive. But now I know enough that I better release this idea. They guess all of them are failing after a certain age. Thx, I enjoyed the video until the end. cheers, Peter
I'd love to see a metal frame, or maybe a 3D printed frame, to go with an XT-IDE card and an IDE hard drive. I've got an XT clone with only two half-height bays that could use something like this.
Well at least that chassis bracket will come in handy for mounting a drive and a shorter board of your choice.
Was working on old take player and the belts turned to this oily solution it took a few days to get that from under my finger nails.
A Tuesday video? Adrian, you're playing with us...I like it!
The placeholder disk is probably needed for the head unlocking vane to work -- if there is no disk riding on the top, the platter wouldn't move as much air and it would not be able to move the unpark vane out of the way, leaving the head stuck.
Seen a few of those over the years. Tended to be fairly reliable. Cool piece of tech.
Reading the comments I was hoping this would have been a more careful tear down and turn into a repair. Ive oiled and brought back to life about 7 MFM and scsi drives even drives that I felt were broken. Gummed bearings, head park mechanisms and the goo bumpers can cause weird noises. I recommend always approaching the opening of a drive as if its repairable since you never know.
My AT had two similar type hardcards in it, both had errors etc. But worked at least.
Anxious to seen this one after the video I saw about the 5170. Great and quick follow up!
Very interesting! I’ve had a few Quantum 50-pin SCSI drives with heads stuck to that sticky stop.
Nice forensic work! The melty rubber is definitely the death knell for many components that have it sadly.
if i remember correctly, very early kind of "hard drive" do not have autopark, you need to issue command to park it before power off.
Very early voice coil drives did not have the servo tracking information on the drive platters. So you had to had to have some method for closed loop head position control. The earliest drives just used a stepper motor and an optical "zero track" photoeye to know where they were (or like the apple disk 2 which just stepped the heads far enough to know that they were at the position needed). This is a later design, and you needed position feedback in order for the voice coil closed loop control to work. Modern drives use servo info encoded onto the platters themselves, mainly due to the higher densities.
This video reminds of the many videos you see on YT from folks like Techmoan and others repairing vintage cassette players, turntables, and other vintage electronic devices that use rubber belts. They all suffer at some point from the rubber belt deteriorating and thus needing to be replaced. It also common to see old rubber pinch rollers and rubber coated wheels becoming sticky and requiring special chemical solution that removes the sticky surface of the rubber on these roller sand wheels so it practically as good as new. Its just seems that the rubber used in electronics only last for som long, and while the longevity is better today then it once was it never going to be forever.
If I had to guess the lines going to the motor, it is probably power, ground, enable, and tach. I’ve taken a fair number of hard drives apart myself (and have a cluster of rare earth voice coil magnets in the garage atm), but I’ve not seen one with an optical track sensor like that either, maybe why the arm had to be offset? At least we now know why these are unreliable.
If you look closely there is definite scratch in the middle of the disc,just right in place where i believe vital information for boot drive are located. I have saw tons of simillary scratched hard drives even quite recent ones.
Maybe on NTFS drives but not on MFM and RLL drives. Track 0, Head 0, Sector 0 had to be good but would wear out over time. There was a trick or two back in the day. You could move the home sensor making it think another area of the disk was T0 H0 S0 but you would have to low level format the drive again. The format routine was sometimes in ROM if you new how to jump to it in a memory editor utility like DEBUG. Also you could swap the head cables on some drives to use the other side as Head 0. Some scratched and worn drives would work fine and other immaculate drives wouldn't work at all. Th smell of burnt aluminium was the killer smell of a dead head crashed drive.
I haven't plugged in my Plus+ Hardcard in over 10 years... now I'm curious if it still works.
There was a video a few years ago where a guy slipped heat shrink around the rubber bumper and fixed the sticking rubber
I think this qualifies as an autopsy. :)
An autopsy where the body was still a bit alive in the beginning.
Adrian's Digital Butcher...
labels to each connector are printed on the pcb right beside them
ROTOR - MOTOR - ENCODER - R/W
1. brushless motor
2. ceramic magnet
3. probably that optical setup with the flex cable (flex PCB, BTW, not an FFC -- flat flex are the white thin ribbons used on modern laptops and C= Plus/4 keyboards) is an LED + a photodiode or phototransistor -- four wires, not two ;)
A pity it doesn't use a standard IDE or MFM style connector -- then you could change out the drive and have it work again. Quantum BTW was a good company at the time. My first system had an 85meg Quantum drive, built 1991 sold 92 still spins to this day. I need to check data integrity but it predates the era of drives that autoconfigure -- it doesn't know how to tell a more modern PC how to talk to it, so I have the C/H/S stuff written on an old flash card I keep with the drive -- a modern USB adapter can't do anything with it, though, because of that. If it won't autoconfigure, a USB adapter or external enclosure just can't make heads or tails of it :( I wish there were something out there for these old drives that would let you manually configure them, but still speak modern protocols, purely for testing purposes...
That's why I proceed very carefully with such autopsies. ...just in case I find a cure in the process and patient get chance for new life ;-) However, your video and drive sacrifice will help in many future repairs. Cheers! S
Well, thanks to the sacrifice of this drive, we now know how to fix the others. I wonder if that metal frame could be used to mount an IDE hard drive, coupled to an XT-IDE card. The Hyundai XT clone I've been playing with only has two half-height 5.25" slots, I've got one of them occupied with the hard drive, and I'd like to avoid doing destructive mods to the case if I can, so a hard-card type solution would be perfect for this chassis. I know, everybody's going to CF cards these days, but I just love hearing the hard drives spin up and tick.
I had a hard card of some no-name make about the time this was made, but it had a "standard" 20Mb 3.5" IDE hard drive on it. When it became too small for my needs I was able to remove the drive and fit a bigger 3.5" drive (I think it was 60 or 70Mb or similar) as it used the standard interface, and the ROM on the card supported different sizes of disk. Since the drive was an IDE drive the interface was considerably simpler that the PLUS card you show here, but only slightly younger. It was a time of sudden change! I remember replacing it with an IDE card and 170Mb drive when I upgraded my motherboard only a year later, because disk prices plummeted for some reason and I'd got a better job with more money to play with!
Nice bit of dissection, shame some parts got broken you could have tried reassembling it to see if it would work even for a short will to see what was on it.
That head lever design allows the heads to move in a more radial movement, keeping them nearly tangential to the disc tracks. Perhaps this was needed to minimise signal distortion and maximise signal strength?
Hey Adrian! Thanks for that video! I actually still have a Hardcard 20 in my computer stash!
Wait! What? IC on flex in 1986!?! I had no idea that tech went back that far. Our college was still doing through hole in 1990. SMD was a thing (but we really did not touch it) but I don't remember any mention of IC on flex!
edit: the bad rubber stoppers is the same thing that happens on the WD40 and 80 MB drives. You can take them apart and put something (I use gaffers tape) around the rubber. Obviously the drives will be vulnerable to dust intrusion afterwards but it usually works to be able to read/image the drives.
back in the day we sold tons of hard cards. Very rarely had issues with them, but then again they were all new back then.
I have also seen a very old one in an original Osborn portable, and that computer was the only one out of a half dozen that actually booted, because it was the only one with an HDD.
Then i had to try and remember what command line commands worked with CPM.
Actually don't remember if they were Plus hardcards, just that they were hardcards. Heck, that was 30+ years ago.
I went back and checked where that sticky part was when it was assembled. And I remembered that I saw something which I first thought is a little button that pushes in when the head reaches the end so the drive knows where the head is. Turns out, the thing I saw was a part of that sticky rubber ring. It's at 15:09, the black stickiness is on the arm.
was my first hard drive, hanging out the back of a schneider europc because i only had a half slot
A bit sad seeing not to be a rescue but I’ve learned how to do it. Thanks. Sometimes you have to do what you have to do to accomplish the mission.
That would be an interesting project to possibly convert one of these to use CF cards or SD cards. Although, it is unlikely to be IDE or IDE compatible so that’s probably a bust.
I did cringe but it was an impossible task given it was the first one you took apart. Lessons learned, the next one could likely be salvaged.
I have a bunch of SCSI Conners for old Macs and I would like to refurbish them properly in clean conditions. Great video!
We used a lot of these in the original Compaq Portables.
I have a Hard Card 40+ that doesn't work as well. I would bet if it has the same sticky issue, I could possibly fix it!! Maybe a thin strip of electrical or other tape around that sticky grommet and oil the bearings. What HDD Lubricant is recommended?? I think I could use a heat gun to carefully get the plastic covers off without mangling them!
Lots of early drives used a Moire grating to locate their position... it wasn't until the late 80's that heads could track the data track based on disk data alone....
wondering if the head bearing is made hard to rotate on purpose to keep the head in place after it reached the desired position
7:33 I can hear BigClive´s voice in my mind... "One moment please..." :)
.... and the spudger.
I had a hard card 42 in my packard bell 286 back in the day. And yes it was a quantum branded device
oh please try to get it back to life, this will be heroic ;-) these old stuff might be repairable in opposite to the "nano specific" devices from nowadays. I know, half a year too late for asking, but better late than never.
Should have tried putting the drive/assembly in the freezer for a few hours first before tearing it apart. Once frozen, the gooey rubber might have released the heads long enough for you to save the drive contents. Even with newer drives you would be surprised how often freezing a non-working drive ressurects it long enough to save data that would otherwise be irretrievable.
In retrospect, if you look closely you could see the goo of the rubber bumper 18:07
I remember "back in the day" (around 1991 or earlier,) when a tech bulletin was sent out by the manufactuerer regarding problems with stiction on hardcards, and it specifically stated: "Hold the hardcard in front of you at a 45 degree angle, and shake like a ketchup bottle!"
I've torn apart piles of hard drives and never seen an actuator arm like that, wild!
Yeah thought it was strange!!
Mitsubishi also had a series of 5.25" voicecoil based MFM and RLL hard drives. The 60MB RLL was the bomb back in the day.
Moving the pivot allows the arm to travel in a path more perpendicular to the radius of the drive. Modern drives compensate for the nonlinearity in the arm's travel with better electronics, more precise head movements, etc. This is a mechanical solution that simplified stabilizing the read/write rate saving some complexity and HW/SW performance needs.
I've seen stranger. A lot of Maxtor 5.25" drives had a 3.5-inch HDA inside of them sitting in its own metal "cage".
Knowing what you now...know :/ IF you were to attempt a repair of a HardCard I'd suggest 2 things. Buy or print some head combs. Little plastic things that Wonder-Bra (lift and separate) the heads from the platters and prevent damage. The last bit is a trick from growing mushrooms, I think it was part of the PF Tech Method....Heat your oven up and leave it on during the process.. Make the oven rack your workspace and perform your surgery on the drive there. The continious rising heat will prevent particulate matter from falling and coming to rest on your patients delicate innards. In fungi farming this was to prevent contamination and destroying a harvest.
The platter has very fine circular scratches on it. Platter should look like a high polished mirror.
We had 100 and 200MB quantum branded hardcards when I was a kid. Worked great in our compay and victor 286's if you could honestly say they ever worked well.
While not as bad as David's "dremel the non-standard screws and stick a paperclip in the power connector", this was still an unfortunate way to destructively disassemble the drive. It would've been interesting to leave the flexes intact and attempt to spool up the drive to see what was causing the issue. Maybe (combined with unsticking the heads) you could've taken a look at the file structure.