You can use a hair dryer to heat the board up to failure. When the failure shows up, use a can of freeze mist, one chip or component at a time, until it works. That will identify your thermally intermittent part.
Thanks for making the video. I worked for NCR Corp from 1978 to 2003. I have a lot of memories repairing those Halk drives. Head crashes were very common and hearing the heads screeching on the platter you can never forget! Nothing like walking into a office building with a 20 lb tool briefcase and a 20 lb oscilloscope was a distinctive sign you were the geek of the times. And yes under the suit coat was a pocket protector.🤓 PS: most head crashes were due to the micron filter being clogged.
I am of the 486 vintage! And I can clearly recall that most of the people back then in most offices I have been to used to smoke, a lot 🤣 hence, most of the machines I scavenge from that era, have a certain yellow grimy patina that coated most of everything, and it was way worse on the ventilation side of things. Hence I can only imagine what these drives had to endure back then 😅
This takes me back. My first job in the late 1970's was as an engineer with a Wang computer dealer. For removable drive storage both Hawk and Phoenix drives could be attached. One of our jobs was to fix these drives when they went wrong - but back then you could order replacement parts! Our most common problem was when data errors were reported. This usually occurred either because the heads had drifted out of alignment or there had been a head crash. In either case a head alignment process had to be undertaken. This involved connecting the drive to a special diagnostic control unit, attaching a scope and carefully adjusting the head position until the scope showed the correct waveform whilst controlling the drive from the control unit. Over time you got a 'feel' for the alignment. Also I recall that the air filters had to be changed at regular intervals as they could get clogged (especially if they were in a smokey office environment which was common back then) and fail to remove dust particles properly which could cause a heads crash. One of the features of the Phoenix drive was that it had several fixed platters - but depending upon which model was purchased depended upon how many of these platters could be used and hence the capacity of the unit. But the number of fixed platters that could be used was determined by links on the control board - the drive being the same for all capacities! One of our jobs when installing these systems was to configure these links to correspond to what had been ordered. Or if the customer had initially ordered a small drive and then paid for an 'upgrade' to a larger model we just adjusted the links and hey-presto the customer had a larger capacity drive
@@markmuir7338 Back in the 1970's you could get disk drives which took a 200Mb removable disk pack which had 10 platters giving 20 sides of 10Mb each. These were big beasts and if there were several connected (you could have up to 8) you could feel the vibration in the room when they in use. Head crashes were a nightmare!
@@robertlewis4216 Possible with the 2200 MVP,, but I used these 200Mb types with PDP-11s. It was more usual to have the 200Mb drives with the Wang VS computers. I used Hawk or Phoenix drives with Wang 2200 VP/MVP computers. Then Wang brought in sealed hard drives (I think 8" Winchester?). These tended to go out of alignment with use and you had to adjust them using a scope. You could tell if they were out of alignment as they just kept seeking and seeking and seeking which you could hear.
It's important to understand that "erase" is a misnomer, the read/write head by itself is perfectly capable of obliterating the old data and writing over it. But magnetic gaps like in a read/write head have "fringing fields", like when you look at a magnetic field with iron filings, the magnetic field lines leak out beyond the gap in the head. This is still true for the latest hard drives today. The "erase" head simply blanks out a thin area on either side of the track so that it doesn't interfere with the next track over. Is that wasteful? Sure. That's why you have SMR recording on modern hard drives, they only erase one side so you can record closer together. Ideally the function should have been called "guard banding", but we're stuck calling them "erase head". So what's the difference? The timing of when the "erase" coil is powered up when writing. It's a very low-level function. There shouldn't be such a great difference in performance. Why they make different types of heads, I think boils down to what was easier to manufacture. Straddle erase heads seem to have been replaced very early on with tunnel erase heads on floppy drives, for example, but the only change was in timing in the FDC. Search for the Motorola AN917 "Reading and writing in floppy disk systems using Motorola integrated circuits", it's for floppy disks, but I think the ideas are the same. I won't put a link because the comment won't make it.
@@monad_tcp As tinkering in one's home lab gets better over the decades, I wouldn't be surprised if people start making their own equipment and platters. After all, we have access to xray scanning equipment to reverse engineer heads and such now.
You are one step closer in understanding the system. That’s valuable information to have. Imagine if someone reaches out and says - “I have a hawk drive that doesn’t work with my system” and you discover that it’s a PE head drive and they need an SE head drive…
Ex-Burroughs guy here. We used comparable drives on the very smallest machines, and my understanding is that straddle-erase is not actually to erase data (before writing) but to trim the somewhat marginal flux pattern laid down by the edge of the write head to improve reliability and potentially to allow more TPI (less inter-track crosstalk for a given separation). That might explain why the documents you were looking at showed the flux pattern differently: the illustration of the PE showed the written flux looking like a triangular wave while that of the SE trimmed the top and bottom (although the illustrator didn't quite understand what he was being asked to draw, so rounded it off). Don't know whether this will work, but /\/\/\ vs /-\_/-\_/-\ If the drive you've got similarly uses SE to increase density, there's also a possibility that the formulation of the oxide on the surface of the disc is different which might contribute to compatibility (or at least reliability) issues. As a footnote, we were trained during the early eighties to use iso on a lint-free (wrapped around a ruler borrowed from the customer) when doing an on-site repair, then to flush any remaining impurities off the surface with a liberal amount of Freon. Perhaps fortunately, they'd stopped issuing Freon as a cost-cutting exercise... Updated: I've got a very hazy recollection that one type of erase might have been for removable platters, and the other for fixed where (a) interchangeability wasn't an issue and (b) since the ring holding the platter was torqued down the overall dimensional accuracy was better and there was an incentive to try to increase the TPI.
Yes, either Kim-tech wipe replaced frequently or in a pinch coffee filters are relatively lint free. Always had a steel 6 inch scale in the pocket protector.
@@mikebarushok5361 We were still issued with packs of genuine lint-free cloths, together with the spring-pullers and link-benders needed by every electronics guy in the early 80s. And for those disks you needed more than a 6" ruler: clean while spinning and that edge is /aggressive/ :-)
@@button-puncher Presumably because it's easily-distilled to high purity but... heavens. Out of curiosity, roughly what year was that? It would be interesting to know when attitudes changed.
@@MarkMLl_uk My prof was talking about it in the late 90's. The US made it's sale/production illegal in 1995. It was supposedly contributing to the depletion of the ozone layer. Considering that the ozone hole in the atmosphere is the same size as it was 30 years ago, it sounds more like the environmental "scientists" don't know what they are talking about. Yet again.
I love that you have become apart of how I unwind on Sundays. Get home from church, have my artisan coffee, light up my cigar or pipe and enjoy some vintage computing. Thank You!
Yes, every single time I see one. I even started to wonder if one of the engineers thought about looping the wiring to put ears on either side of the head.
I can understand why the Phoenix drive scares you. I can still feel the cold brick in my stomach when i got a "system down" call and i knew thier primary drive was a Pheonix...
It's wild to me to think that 99% of the drive can be identical and it might as well be a completely different drive as far as the controller card is concerned!
@@UsagiElectric Terribly sorry but are you *sure* the straddle-erase needs a full revolution on CDC drives to overwrite data? Do you happen to remember from where do you know that? (No links please - youtube doesn't like those) I'm not knowledgeable in the subject and might be totally wrong but that's not how I understood the process. I thought in straddle-erase the write coil overwrites data while erase coils erase the guard-band, while with tunnel-erase heads the erasure happens a little ahead of time of writing That is what the jumper is supposed to change - it should introduce a delay between erase and write
@@jwhite5008 That's something I'm curious about, was the jumper on the drive PCB set correctly? Maybe you could also scope the output from the read amp to see if the signal is even written.
@@Stoney3K Actually I'm more interested in oscilogram of what is actually *read* from the platter. I think the issue might be unrelated to the head type. There is quite a lot that happens between issuing the format command and reading stuff back, and any of it can go wrong to produce such a result. Inconsistent head positioning or read/write attenuation for example See also my next comment
These drives used IBM 5440 disc packs, as did the first machine I ever programmed, an IBM, back in the 70s. The DEC RL drives also used these packs. They were known for their reliability and were usually rock solid. Nice seeing these things still spinning in Dave's garage
My First IBM PC had a whopping 15MB storage and i could only fill it up to 5MB with so much software available back then. Hard Drives were like Hard Currency.
I love these drive videos! As a data center guy and vintage fan I get a kick out of the fact that these names "Hawk", "Phoenix" are still in use for server storage.
Your detective work, and comprehension, is inspiring. Something that seamed overlooked, or perhaps forgotten, you didn't mention installing a StradelErase head controller card. Is that possible (two separate controller cards)?
Thank you so much! Unfortunately, Centurion never made a straddle erase controller card, so we'd have to build one from scratch. But, since I already have two working Hawk drives with pre-erase heads, we'll instead shift focus to something else, and reserve the straddle erase Hawk for a different project.
I was thinking on the same lines, while looking at the schematics and wondering how hard would it be to modify the controller to use both types of heads (selectable of course) Shugart drives used this types of head, have to look at the schematics but I would assume the mods will have to be related to some sort of delay between the erase pulse and the write one.
@@UsagiElectric Someone (CDC?) made a RWE card set up for straddle-erase heads, or the heads themselves wouldn't exist. Whether said card would work in a Centurion is another matter. 🙂 My guess is that somewhere there is (or was) a PE-equipped drive plugged into a SE RWE board that writes at half the speed it's capable of.
I've been thinking about that. Breaking taps has been doing silicon lithography, so I wonder if this is something that could be done today by "home" workshops.
I've been wondering where the original manufacturing of platters took place and what machinery was used. It would be awesome to see the original machinery refurbed and creating new platters.
Hawk Tuah Drive This is my first video I've encountered from this channel, and I am *heavily* intrigued by how this kind of tech almost makes enough sense to understand for someone like me who isn't an engineer, and the journey to keep it running is gonna keep me invested.
Hey there! It's been 40 years since i've worked on any of those old CDC drives and my memory isn't that great anymore so maybe i'm remembering this wrong but I seem to recall that the straddle erase head doesn't work like you think they do.. it is there to trim the sides of the written data to ensure the guard bands are preserved, you got a nice narrow track and this allowed you to get a slight increase in data density, they are enabled at the same time as the write head. I recally a low and high TPI mode on these drives too and many machines of the time used a hard sectoring card that was adjustable via jumpers or switches, this drive might have different sectors or tpi to your other drives. Head crashes were almost always caused by a clogged filter so make sure you stay on top of those, I doubt the originals are available anymore but you can probably fabricate something with a 99% .3 micron hepa style filter should you need to. Anyhow best of luck with it all and apologies in advance if everything i wrote it totally inaccurate :)
My mom and dad both worked for EDS for a good chunk of it's existence till it was bought out (funnily enough, they met when their two teams were having issues, and mom and dad were the two chosen to communicate between the teams). Our house is still is chock full of EDS whatsits; heck, I'm drinking out of an EDS mug right now! So hearing that name REALLY snapped my attention back to the video, lol!
@@PCFixerthat sounds like vintage (Tomb of The) Cybermen to me... none of your new-fangled "Delete Delete Delete".... although, if it's a stradle erase head, should that be "Erase Write Erase"?
RE: Fixed drive not writing. This reminds me of a customer of mine back in the 70's who had a removable pack Plessy drive. He was using this for archiving data. What he didn't realize was that the write protect indicator bulb was burnt out. The drive would accept a write command but would not indicate write fault. When he (we) figure this out, we examined his "most recent" archive, and the data on it was three years old! Lucky he didn't have to rely on an archive/backup. Check your write protect button and indicator bulb.
I haven't watched Pokémon since middle school, but I recognised that R immediately. Next episode I hope he'll put on that fetching cropped jacket they wore
When all heads fail: is there any documentation of how the head are designed and/or made? If you it should be doable to recrate them Otherwise once migt be able to disassamble an non-working head trying to reverse engeneer them Same spiel for the servo cards I know, i know it is a long and tedious project but we have come so far already
As seen at 27:44 the head mainly consists out of 2 transformers. And this is honestly how even a lot of modern hard drives are made these days with 20 TB drives as well. (except I think modern heads are MEMEs fabricated, since that is way easier...) But yes, one can sit under a microscope and hand wire a new head. Just need very steady hands and plenty of patience. How these heads were manufactured back in the day is a good question. I would not be surprised if they were manually wound by hand. Since that is how a lot of other things were made back in the 60's. (Magnetic core memory for an example were oftentimes manually threaded, a laborious task requiring a steady hand and plenty of patience.)
I think you are misunderstanding the straddle erase head. I think it means it straddles the full width of the track, a bit wider than the read write head. This is different to a tunnel erase head that has two thin strips that erase just the sides of the track. However, I don't understand why an erase head was needed for a digital drive as the write operation should overwrite any previous data. I think it was just to reduce noise from the edges of the track making head alignment less critical. I don't think any drives need two revolutions to do a write.
I found this info about floppy disk erase heads. From the Siemens manual, page 3-4: "The read/write head also contains a tunnel-erase or straddle-erase electromagnet, the function of which is to erase the edges of the recorded track as the data is being written. The width of the track is narrowed to approximately 0.013 inch by this technique, to minimize the effect of data previously written on the track and possible crosstalk between tracks." Discussion in the Siemens manual makes it clear that "tunnel erase" is one kind of head, and "straddle erase" is another kind of head; and that erase operation is concurrent with write. For instance pages 3-26 thru 3-29 describe the erase gating and logic during write: "The purpose of the auto erase feature is to provide the necessary turn-on delay between active WRITE and ERASE and the turn-off delay after WRITE goes inactive, but for tunnel-erase heads only. This is for option TE installed. Straddle-erase heads use option SE , which bypasses the delay circuits." A sketch of the tunnel-erased track shows that it measures .013 inches wide. The next document gives an explanation for why a delay is needed. A CDC (Control Data) application note "5.25 inch FDD format considerations and controller compatibilities" examines 8-inch drive technology and standards. It refers specifically to tunnel erase vs. straddle erase in the introduction, saying "The head style and drive tolerances determine the minimum gap of the 5.25-inch FDD formats". The two erase heads as used on 8-inch floppy drives are described on page 7 as follows: "Both head styles have as common parts a read/write gap and two erase gaps...The two erase gaps are used to erase guard bands on both sides of the data being written. These guard bands are necessary to to eliminate noise caused by old data that had been written slightly off track...A major difference between the two heads is the placement of the erase gaps....The erase gaps of the tunnel erase head are 36 mills behind the R/W gap whereas with the straddle erase head the gaps are on both sides of the R/W gap and extend approximately 11 mills behind it...Because the erase gaps of the tunnel erase head are so far from the write gap they must be turned on and off separately." Sketches of tracks for each head show that with the tunnel erase head, "used by IBM and most manufacturers", there is 6 mills of erase gap either side of the 13-mill track. With the straddle erase head "used by Shugart exclusively", each 6-mill erase gap is offset from the 13-mill track by 1-2 mills." So both types of head erase just the edges of the track and it is done while writing in the middle. Straddle heads do overlap the write head length wise so they are turned on at the same time. Tunnel heads are behind the write head so needed a delayed signal. On a floppy drive there isn't a separate erase signal so the delay for tunel heads must be in the drive logic, not the controller.
Really takes me back to my first few years at work. The MRI I maintained used a 11/750 for the scanner and a 11/730 for the standalone viewing console. Both had removable packs. Something was really satisfying handling those packs.
I was able to visit the LSSM in Pittsburgh last year and got the grand tour. It was one of the highlights of my visit to the states and can highly recommend this place. I even learned that I was wrong about a thing or two and since strive to do better.
I wonder if you will be able to find the appropriate hawk controller. A head rewind attempt would also worth the search of somebody which have the right tools ( and ideally knowledge). Amazing machines.
Every time you talk about the EDS PC I am reminded of the TV show "Halt & Catch Fire". The story you tell of Centurion getting into the PC market is so much like the fictitious company Cardiff Electric in the show.
Hot air rework station (for Surface Mount stuff) works well for precision heating of marginal components. Plus it gives you a reason to buy one if you don't already have one. LOL! I got a Quick 957DW a while back and it's really nice. I'm just a hobbyist. Didn't need anything more expensive.
Those drives greatly resemble the drives on Honeywell Level 6 Minicomputers from that time. I suspect they may be manufactured by a common OEM and then rebranded and sold by various other companies including Centurion. What that means is that there's a good change you can find parts like drive headswhich you seem to need from other stock and or drives if you can figure out who actually made the drives and who else used ones like yours for their equipment. Honeywell is still in business too so they may actually have some parts stored away in their parts stocks.
Yup! The SE heads were the original design, and then later, they came out with the PE heads. I think that timing on the SE heads vs PE heads could cause some problems on some particularly picky drive controller cards designed specifically for SE heads, so I imagine that CDC continued to manufacture them for backwards compatibility.
I LOVE watching these Videos, but man.... I REALLY, REALLY do NOT miss working on these massive old machines. We have it soooo good today with our modern computers. But it was all very new and exciting back then. 👍 I'm literally using an Intel NUC stuck under my Desk!
You should put a scope on the head signals to verify the write pulses get to the heads. The writes should be causing flux reversals no matter what. You could probably trigger on the sector pulses and get continuous view of a single sector, a bit like you did with the timing track on the Bendix drum. Then try some writes and see what happens. Also maybe try the same but disconnect the erase coil. Lastly check the pinouts? Maybe your SE heads are different.
@@UsagiElectric well if you decide to go with a bus card similar to the og I always thought it would be cool to use Sega Genesis 64 pin cartridge connector or a SNES cart connector for a computer like that because unless you want to make exact clones it's ok to be different if its still compatible
A Micro-ATX Centurion Motherboard! What a dream that would be! Similar to the Monotech NuXT except it's a full-Centurian Computer rather than an IBM PC!
As long as it has a thermal printer, I'm keen. Peripherals are the one thing people always forget about in vintage tech. They were the core part of what made them worth buying.
I remember learning about the AMD 2900 series, and how you could define your own instruction set in the microcode, this was back in the mid 80s, at Polytechnic (now converted to and called University). We had a great lecturer who was very excited about his subject. I hadn't realised it was used in so many systems.
every hard drive I've worked with going back to the very early 70s (and those were designed in the early/mid 1960s) did flying erase/write in one pass. the erase gap was right in front of the write gap in the same head assembly, and both worked together.
I used to work on some Honeywell Level 6 machines and they had Hawk drives. Fixing a head crash on the fixed drive was comical. They pulled out the bad platter, took out the heads. Sanded the heads with fine sandpaper and then washed them out with high percentage alcohol. Then the would put in a new platter, reinstall the heads and format the drive. That worked - most of the time.
I have found some discussion regarding the head configuration on the larger floppy drives. The straddle configuration just trims the written track left and right to create some security gap to the neighbour tracks and to erase potential remains from an old track that can be left over due to slight positioning errors. Trimming the track can happen simultaneously or after writing, so these additional heads can also be behind the write head. From what I understand erasing and writing is still done simultaneously, but the timing of the erase gate must be adjusted in relation to the write gate according to the head configuration which is irrelevant on floppy drives as this is handled internally. How is this implemented on the hawk drive? Are both gates exposed on the interface to the controller card?
Watching a voice coil actuator was always impressive 'back in the day'. But what what was really awesome was the hydraulic actuators on the IBM S360! It seemed to me they were faster than the voice coils.
I had a strange idea: perhaps the heads coils could electrically be protected by adding a inline fuse. A small board could be made with a surface-mount fuse, which could be put between the head and board that reads the head. It the board goes, the head may be protected. Perhaps a few Zener diodes could clamp the voltage, a bit higher than its working voltage, too.
David, since the badly crashed HAWK head was trash as far as you are/were concerned, why not take a no-risk chance at repairing or at least disassembling it under a microscope to better understand how it was built? You've got nothing to lose! I bought a stereo-microscope in order to attempt repairing a known-failed tape read (or write?) head for a DECtape drive. My friend had another head, so he swapped them out and sent me the bad one in the off-chance that I could get lucky trying to repair it. I need to re-start _trying_ to repair it for him!
I remember back in my minicomputer days working on Data General equipment exclusively. DG used Phoenix drives I believe. As memory serves they were 10 MB per platter with 20 MB total (10 fixed and 10 removeable). We also had some much older Diablo 33 drives that were 5MB removable only but they were being phased out in favor of the higher capacity drives. There also were some washing machine sized drives with stacks of removable disks that I think were 80 or 120 MB (can't remember). Those were also much older but still in use back in the 1970's. I was never aware of the head differences. Quite interesting!
Since you have an erase coil on either side of the write coil, with a straddle erase, you effectively erase before and after the location your are writing to. Which means that the data you are writing to the platter is erased after you write to it, no wonder you have corrupted data when you try to read it back. Since you have only four pins going to the head, presumably one pin for each coil with the fourth being a common, it would make sense to unpin the trailing erase coil, thereby converting the straddle erase into a pre-erase head. This might allow you to use this drive after all.
I think one easy way to describe the difference between pre-/straddle-erase heads is the coils in pre-erase heads are arranged in-line with/parallel to the data tracks and the coils in the straddle-erase heads are arranged perpendicular to the data tracks.
@UsagiElectric Thank you for sharing your adventures with us! Keep up the good work! Is it possible to find someone who can rewind the coils on the heads that have blown coils? I know the wire is extreamly tiny, but there has to be a way to repair those heads.
@@liquidsonly Yes,... no-a-days, programmers just add libraries, and code, no thought of how efficient you need to be... We definitely lost some skillz over the years... but we have also gained in amount of data you can process... I still love them old early 80's feelings..
That's 5 million ASCII characters without compression. That's a lot of text to type. Of course if you add images it will bloat up the document quickly.
Back in the day I seem to recall that main frame (ICL IBM etc) hard disks used significant air/vacuum filtration since the heads literally flew over the platter. Whilst a human hair would be pushed out of the way a smoke particle could cause a head crash. This may be an apocryphal story but I'm reminded of an issue, in Holland I believe, where for some reason a head crash occurred. Since work had to go on the removeable disk pack was tried in another reader. That crashed too. Now there are 2 readers destroyed. Nevertheless some bright spark tried another disk pack in the original reader with obvious results. Fortunately somebody with sense halted operations and called in the gurus to fix the fundamental issue. 5 1/4 inch disks held in the same hand as a cup of coffee can produce a similar effect. I know I watched it and had to sort it. The look on my boss' face when I took a pair of scissors to his precious disks will not be forgotten!
Need somebody out there with a watchmaker's precision to dissect some bad heads and come up with a way to make new ones or repair the old ones, there's certainly going to be a market for it!
Many years ago when I was a student, we had a VAX, and I recall a time when I was in the computer center, and we heard this horrible noise coming from the disk cabinet. But we were totally baffled in that the system was still up and running and had apparently suffered no ill effects. We shut the system down as a precaution. The DEC guy came out to repair the hard drive, and we found that the motor shaft had sheared off where it went into the brake, and the horrible noise was the two ends of the broken shaft grinding against each other.
As much as I love these videos on these machines, I’m kinda nostalgic for the early tube days of the channel. It’s what brought me in, what you were doing was unique and fascinating. I’d argue it’s time for a bit (lol) of tube computing videos.
We'll be getting back to that soon, I promise! The paper tape reader for the 1-bit tube computer has gone through about three different design revisions over the past few months. I'm waiting on some new parts to arrive and then we'll start building prototypes and filming that. After UE-1 is finished, we're already working towards a design for UE-2. The reason there hasn't been any videos on that stuff in so long is because I'm trying to do a lot of the initial design work first, then get to work constructing and building it.
@@UsagiElectric excellent! Given what’s going on with Intel, I’d love to see people go revisit the 4004, perhaps using discreet logic chip, or, woooooo, transistors built into gates. I’d bet it could fit on a single 4x8 sheet of plywood using transistors.
I'm fairly sure you're wrong about it using two passes to write a sector with the straddle erase head. The data itself doesn't need erasing -- the data head erases the old data when it writes the new data. The erase heads are only there in case there's a misalignment between the old and new data tracks, to remove any spillover. Erasing and writing is still a single operation. From what I've read about this, the only difference should be that the pre-erase method might need a short delay between turning on the erase and write heads, because of the separation between them, whereas the straddle erase method would not need this delay because the heads are in line. Since your controller board seems to be telling the drive to turn on data and erase together, it would seem that any such delay, if required, must be implemented somewhere in the drive itself, and the controller doesn't need to know about it. So I suspect your problem isn't caused by the pre-erase vs. straddle-erase issue, there must be something else wrong.
I wonder if you could get in touch with Lumafield to get the failed heads scanned. They did such collaboration with other channels so seems worth a try
Quite a blast from the past seeing that technology. I find it fascinating that you _like_ CDC drives. I used to operate a CDC Cyber 72-26... with an 808 drive attached. It was huge, and terrifying. It used a separate 1500psi hydraulic pump to drive the heads. Fortunately I got an offer to change jobs, and waved goodbye to that CDC forever.
@@Alexis_du_60 It was mainly terrifying Alexis. An 808 was tall - over 6 feet high. The platters were over 2 ft diameter, and there were a LOT of them. So that the heads did not droop as they were pushed towards the inner track, they rode a fixed steel bar that was about 3 inches diameter. So the huge mass of the head assembly did not cause the whole drive system to rock, they had two spindles, one on each side of the logic and head actuators. The two sides were driven in unison, in opposite directions. *Everything* about the 808 was massive... and despite the power applied through that hydraulic ram, they were slow to seek, with what had to be the longest latency in the rotating data storage game. So far I have not seen any reference to one on the web.
Mostly because it was solidified and hard as a rock. I would have had to use a chisel and a hammer to break it loose. It wasn't even sticky anymore, so leaving it there shouldn't cause any problems.
yeah i can see that, also having to chisel it out wont be good for the drive itself either i assume. as long as it doesnt become non solid gunk again it is fine, i get that reasoning. @@UsagiElectric
Do you have any idea how the heads fail? Maybe you could carefully take apart a broken one to see if you can find the broken connection. I worked on many old VCRs, I have never seen a open head coil! Maybe its not the coil itself, but the termination of the copper wire, so you could fix them
As a guy who has absolutely no idea what you are talking about while controlling the drive on the computer it kinda feels like you are playing battleship with it 😂
That servo control board is going to be interesting to fix. I'm betting those cans are op amps, and i counted 10 of them. So any one of those op amps (if that's indeed what they are) could cause a failure. I'd bet good money that you have a couple of "instrument amps" for the reference and feedback signals, an error amp. The error amp would feed into probably at least a proportional amp and an integral amp (may even have a differential amp, but hose are tricky to get right). Then a summing amp to sum up the control signals (from the PID amps), then output amps to actually drive the voice coil. And this is ALL analog, so good luck with that! Plus, there's no guarantee those op amps are even available any more. Sure, a new servo board could be engineered, but the cost would be eye watering. Unless someone just decided to do it as a fun project. ^-^
We nowadays take hard drives for granted, they're even being seen more and more like an obsolete technology with the advancements of solid-state storage, but it's astonishing if you think to all the R&D magnetic storage technology had to go thru to become the rather normal thing it is today.
Someone really needs to set up some virtual versions of these machines as teaching tools. It wasn’t until I started working with older more limited machines that I started truly understanding how computers work and how to write programs efficiently. Programming with extremely limited RAM and storage really forces you to think about cleaner more efficient ways of accomplishing tasks.
After seeing the differences between Pre-Erase and Straddle-Erase, it looks like the biggest problem is one of timing. The pre-erase head needs to have the erase head powered up for one "bit" time-length before the bit is actually written to the drive, while the straddle-erase can power up the erase head and the write head at the same time as writing the bit. There's no need to pre-erase the bit since writing the same bit to the same place won't change the magnetic field in that position, and writing the opposite bit to that location should just overwrite the bit which used to be there. It looks like the straddle erase head is more mechanically complicated than the pre-erase head, which is probably why straddle erase came later, but it should be easier to interface to it due to not having to juggle the timing so that the erase head wipes a bit followed by the write head writing to the same bit as you need to do in order to use pre-erase heads. Edit: Well, if they're powering both the erase head and the write head up at the same time, that means there must be some wasted space at the start and end of each sector when using pre-erase. That complicates things even more.
There might be circuitry inside the drive that does the delay according to the type of heads installed. In which case the same controller should work with both. I'm suspecting the problem is actually something completely different.
I had never heard of having to wait a full revolution of a disk in order to first erase and then write a sector. Is this really the case with many drives? Are there any downsides to pre-erase over straddle erase heads?
I think it's interesting that bunnies chew side to side. Mooo! 😊 Would it be possible to disable the trailing coil of the straddle erase heads? Is there enough power on the leading erase head to do the job? (my ignorance is bliss...) As always, fantastic video. Thanks so much.
I think you could make it work. But for that you have to change the write routine in the OS and add a flip fop two switch between erase and write. Every read would reset the flip flop. (how to set up exactly has to be discussed.) The Idea would be to change the OS that it would do the writing two times. One with only the erase head running and the other time only the write head running. It shouldn't be hard to find the writing routine and add a loop around it. (even in assembler) This would slow down write operations by 50% but would make the the hawk drive work.
Might consider a stethoscope to listen for any bad noises coming from the drives. Would amplify and pinpoint them beyond what the unaided ear could hear.
A Raspberry Pi Pico being spliced in between the card and the drive could logically translate the drive head logic between the two. If you want to keep it more period-correct you could probably cobble together some logic gate chips to do the same.
I watched your bunny do the read, write and erase of its lunch.
My favorite comment so far!
Wouldn't 'erase' require the camera pointed at the 'output' end? I much prefer this angle though.
@@jasonhaman4670 I think that would qualify as emptying the recycle bin 🙃
If you don't mind, I'll pass on the cpu dump...
Came here for the main video, stayed after for the bunny's lunch source code rendering.
You can use a hair dryer to heat the board up to failure. When the failure shows up, use a can of freeze mist, one chip or component at a time, until it works. That will identify your thermally intermittent part.
David needs to read this comment. So I'm commenting here to try to boost it for you.
Thanks for making the video. I worked for NCR Corp from 1978 to 2003. I have a lot of memories repairing those Halk drives. Head crashes were very common and hearing the heads screeching on the platter you can never forget! Nothing like walking into a office building with a 20 lb tool briefcase and a 20 lb oscilloscope was a distinctive sign you were the geek of the times. And yes under the suit coat was a pocket protector.🤓 PS: most head crashes were due to the micron filter being clogged.
You are so old you had a pocket protector, then realized it was uncool, without realizing they are cool again.
I am of the 486 vintage! And I can clearly recall that most of the people back then in most offices I have been to used to smoke, a lot 🤣 hence, most of the machines I scavenge from that era, have a certain yellow grimy patina that coated most of everything, and it was way worse on the ventilation side of things. Hence I can only imagine what these drives had to endure back then 😅
Only two 20lb items? Sheer luxury!! At DEC we also carried a 20lb briefcae full of microfiche and a reader!
*Hawk drives. They're all named after birds.
This takes me back. My first job in the late 1970's was as an engineer with a Wang computer dealer. For removable drive storage both Hawk and Phoenix drives could be attached. One of our jobs was to fix these drives when they went wrong - but back then you could order replacement parts! Our most common problem was when data errors were reported. This usually occurred either because the heads had drifted out of alignment or there had been a head crash. In either case a head alignment process had to be undertaken. This involved connecting the drive to a special diagnostic control unit, attaching a scope and carefully adjusting the head position until the scope showed the correct waveform whilst controlling the drive from the control unit. Over time you got a 'feel' for the alignment. Also I recall that the air filters had to be changed at regular intervals as they could get clogged (especially if they were in a smokey office environment which was common back then) and fail to remove dust particles properly which could cause a heads crash.
One of the features of the Phoenix drive was that it had several fixed platters - but depending upon which model was purchased depended upon how many of these platters could be used and hence the capacity of the unit. But the number of fixed platters that could be used was determined by links on the control board - the drive being the same for all capacities! One of our jobs when installing these systems was to configure these links to correspond to what had been ordered. Or if the customer had initially ordered a small drive and then paid for an 'upgrade' to a larger model we just adjusted the links and hey-presto the customer had a larger capacity drive
Wow - didn’t know you could download a larger hard drive that far back 😅
@@markmuir7338 Back in the 1970's you could get disk drives which took a 200Mb removable disk pack which had 10 platters giving 20 sides of 10Mb each. These were big beasts and if there were several connected (you could have up to 8) you could feel the vibration in the room when they in use. Head crashes were a nightmare!
Wang 2200?
@@robertlewis4216 Possible with the 2200 MVP,, but I used these 200Mb types with PDP-11s. It was more usual to have the 200Mb drives with the Wang VS computers. I used Hawk or Phoenix drives with Wang 2200 VP/MVP computers. Then Wang brought in sealed hard drives (I think 8" Winchester?). These tended to go out of alignment with use and you had to adjust them using a scope. You could tell if they were out of alignment as they just kept seeking and seeking and seeking which you could hear.
SLO-mo cronch cam at the end was awesome ❤
The real reason cameras were invented!
It's important to understand that "erase" is a misnomer, the read/write head by itself is perfectly capable of obliterating the old data and writing over it. But magnetic gaps like in a read/write head have "fringing fields", like when you look at a magnetic field with iron filings, the magnetic field lines leak out beyond the gap in the head. This is still true for the latest hard drives today.
The "erase" head simply blanks out a thin area on either side of the track so that it doesn't interfere with the next track over. Is that wasteful? Sure. That's why you have SMR recording on modern hard drives, they only erase one side so you can record closer together.
Ideally the function should have been called "guard banding", but we're stuck calling them "erase head".
So what's the difference? The timing of when the "erase" coil is powered up when writing. It's a very low-level function.
There shouldn't be such a great difference in performance. Why they make different types of heads, I think boils down to what was easier to manufacture. Straddle erase heads seem to have been replaced very early on with tunnel erase heads on floppy drives, for example, but the only change was in timing in the FDC.
Search for the Motorola AN917 "Reading and writing in floppy disk systems using Motorola integrated circuits", it's for floppy disks, but I think the ideas are the same. I won't put a link because the comment won't make it.
You've mentioned a few times that you're terrified of the Phoenix drive, but I forgot the reason why.
Looks like best gift we could possibly send to Usagi Labs is a nice big box of new/old stock pre-erase Hawk heads.
And some working controller cards.
I love old hard drives. Such beautiful machines.
Me too, that's my favorite part of retro computing!
@@UsagiElectric someone needs to figure out how to make more of those
It's so cool to see a hard drive operate on such a large scale. Not a tiny, sealed, monolithic block with just a connector on it.
@@monad_tcp As tinkering in one's home lab gets better over the decades, I wouldn't be surprised if people start making their own equipment and platters. After all, we have access to xray scanning equipment to reverse engineer heads and such now.
@@UsagiElectric Count me in too :-)
You are one step closer in understanding the system. That’s valuable information to have. Imagine if someone reaches out and says - “I have a hawk drive that doesn’t work with my system” and you discover that it’s a PE head drive and they need an SE head drive…
Ex-Burroughs guy here. We used comparable drives on the very smallest machines, and my understanding is that straddle-erase is not actually to erase data (before writing) but to trim the somewhat marginal flux pattern laid down by the edge of the write head to improve reliability and potentially to allow more TPI (less inter-track crosstalk for a given separation). That might explain why the documents you were looking at showed the flux pattern differently: the illustration of the PE showed the written flux looking like a triangular wave while that of the SE trimmed the top and bottom (although the illustrator didn't quite understand what he was being asked to draw, so rounded it off). Don't know whether this will work, but /\/\/\ vs /-\_/-\_/-\
If the drive you've got similarly uses SE to increase density, there's also a possibility that the formulation of the oxide on the surface of the disc is different which might contribute to compatibility (or at least reliability) issues.
As a footnote, we were trained during the early eighties to use iso on a lint-free (wrapped around a ruler borrowed from the customer) when doing an on-site repair, then to flush any remaining impurities off the surface with a liberal amount of Freon. Perhaps fortunately, they'd stopped issuing Freon as a cost-cutting exercise...
Updated: I've got a very hazy recollection that one type of erase might have been for removable platters, and the other for fixed where (a) interchangeability wasn't an issue and (b) since the ring holding the platter was torqued down the overall dimensional accuracy was better and there was an incentive to try to increase the TPI.
Yes, either Kim-tech wipe replaced frequently or in a pinch coffee filters are relatively lint free. Always had a steel 6 inch scale in the pocket protector.
@@mikebarushok5361 We were still issued with packs of genuine lint-free cloths, together with the spring-pullers and link-benders needed by every electronics guy in the early 80s. And for those disks you needed more than a 6" ruler: clean while spinning and that edge is /aggressive/ :-)
My electronics professor raved about how good of a cleaner that Freon was.
@@button-puncher Presumably because it's easily-distilled to high purity but... heavens.
Out of curiosity, roughly what year was that? It would be interesting to know when attitudes changed.
@@MarkMLl_uk My prof was talking about it in the late 90's. The US made it's sale/production illegal in 1995.
It was supposedly contributing to the depletion of the ozone layer.
Considering that the ozone hole in the atmosphere is the same size as it was 30 years ago, it sounds more like the environmental "scientists" don't know what they are talking about. Yet again.
I love that you have become apart of how I unwind on Sundays.
Get home from church, have my artisan coffee, light up my cigar or pipe and enjoy some vintage computing. Thank You!
Anyone else amused that the heads have little faces? Pretty cute for having been through such a traumatic crash.
Yes, every single time I see one. I even started to wonder if one of the engineers thought about looping the wiring to put ears on either side of the head.
I can understand why the Phoenix drive scares you. I can still feel the cold brick in my stomach when i got
a "system down" call and i knew thier primary drive was a Pheonix...
What makes the Phoenix drive so scary?
@@thegeforce6625 the emergency head retract, while staring intently at the scope at 2 in the morning with the customer needing system online at 5.
Thank you for taking us along! It is so much fun learning new things about old tech. I don't understand all of it, but I feel smart watching. 😁😁
Thanks for coming along on the ride!
The weirdness of head differences hits again
It's wild to me to think that 99% of the drive can be identical and it might as well be a completely different drive as far as the controller card is concerned!
My reaction every time he gets a haircut.
@@UsagiElectric Terribly sorry but are you *sure* the straddle-erase needs a full revolution on CDC drives to overwrite data?
Do you happen to remember from where do you know that? (No links please - youtube doesn't like those)
I'm not knowledgeable in the subject and might be totally wrong but that's not how I understood the process.
I thought in straddle-erase the write coil overwrites data while erase coils erase the guard-band,
while with tunnel-erase heads the erasure happens a little ahead of time of writing
That is what the jumper is supposed to change - it should introduce a delay between erase and write
@@jwhite5008 That's something I'm curious about, was the jumper on the drive PCB set correctly? Maybe you could also scope the output from the read amp to see if the signal is even written.
@@Stoney3K Actually I'm more interested in oscilogram of what is actually *read* from the platter. I think the issue might be unrelated to the head type. There is quite a lot that happens between issuing the format command and reading stuff back, and any of it can go wrong to produce such a result. Inconsistent head positioning or read/write attenuation for example
See also my next comment
These drives used IBM 5440 disc packs, as did the first machine I ever programmed, an IBM, back in the 70s. The DEC RL drives also used these packs. They were known for their reliability and were usually rock solid. Nice seeing these things still spinning in Dave's garage
wow that's both really cool and unfortunate!
also digital hi5 and yay for slo-mo bunnies!
High five!
Digital High Five back at ya!
Heck yeah!
Don McMillan Digital High 5, so 101 ?
My First IBM PC had a whopping 15MB storage and i could only fill it up to 5MB with so much software available back then. Hard Drives were like Hard Currency.
My first IBM PC had two 360k diskette drives, so cower before me Emperor.
I love these drive videos! As a data center guy and vintage fan I get a kick out of the fact that these names "Hawk", "Phoenix" are still in use for server storage.
Your detective work, and comprehension, is inspiring.
Something that seamed overlooked, or perhaps forgotten, you didn't mention installing a StradelErase head controller card. Is that possible (two separate controller cards)?
Thank you so much!
Unfortunately, Centurion never made a straddle erase controller card, so we'd have to build one from scratch. But, since I already have two working Hawk drives with pre-erase heads, we'll instead shift focus to something else, and reserve the straddle erase Hawk for a different project.
I was thinking on the same lines, while looking at the schematics and wondering how hard would it be to modify the controller to use both types of heads (selectable of course) Shugart drives used this types of head, have to look at the schematics but I would assume the mods will have to be related to some sort of delay between the erase pulse and the write one.
@@UsagiElectric Someone (CDC?) made a RWE card set up for straddle-erase heads, or the heads themselves wouldn't exist. Whether said card would work in a Centurion is another matter. 🙂
My guess is that somewhere there is (or was) a PE-equipped drive plugged into a SE RWE board that writes at half the speed it's capable of.
I know there was talk of recoating the platters, and possibly rewinding the heads, did anything ever come of that?
I've been thinking about that. Breaking taps has been doing silicon lithography, so I wonder if this is something that could be done today by "home" workshops.
There was an idea, but no direct plans, just wishes.
I've been wondering where the original manufacturing of platters took place and what machinery was used. It would be awesome to see the original machinery refurbed and creating new platters.
@@Nebulorum That was a great watch, I had not heard of Breaking Taps prior to your comment.
Looking at the design in the close-ups, unless they are soaked in some epoxy i can't get off, i should be able to rewind them just fine.
8:43 You absolutely can see a scratch on the top surface of the fixed platter. Bottom left corner of the screen.
I was lucky enough to be at VCFE when he was doing some of the work on the drive! Very happy to have gotten to watch part of this in person :3
Hawk Tuah Drive
This is my first video I've encountered from this channel, and I am *heavily* intrigued by how this kind of tech almost makes enough sense to understand for someone like me who isn't an engineer, and the journey to keep it running is gonna keep me invested.
Lights come on. Understaning emerges. Thank you. Very cool video.
Thanks! I learned so much about different head design in this one, it was quite fun!
How accessible are the coils on those heads? I have to wonder how feasible it would be to rewind the coils yourself.
That is definitely something we're going to tackle in the future!
Hey there! It's been 40 years since i've worked on any of those old CDC drives and my memory isn't that great anymore so maybe i'm remembering this wrong but I seem to recall that the straddle erase head doesn't work like you think they do.. it is there to trim the sides of the written data to ensure the guard bands are preserved, you got a nice narrow track and this allowed you to get a slight increase in data density, they are enabled at the same time as the write head.
I recally a low and high TPI mode on these drives too and many machines of the time used a hard sectoring card that was adjustable via jumpers or switches, this drive might have different sectors or tpi to your other drives.
Head crashes were almost always caused by a clogged filter so make sure you stay on top of those, I doubt the originals are available anymore but you can probably fabricate something with a 99% .3 micron hepa style filter should you need to.
Anyhow best of luck with it all and apologies in advance if everything i wrote it totally inaccurate :)
My mom and dad both worked for EDS for a good chunk of it's existence till it was bought out (funnily enough, they met when their two teams were having issues, and mom and dad were the two chosen to communicate between the teams). Our house is still is chock full of EDS whatsits; heck, I'm drinking out of an EDS mug right now!
So hearing that name REALLY snapped my attention back to the video, lol!
Good to see you;re using a TEk 465 scope.
Am I the only one who sees some smiley faces when looking at these heads?
Definitely not the only one!
I see the faces of CYBERMEN when looking at those heads! YOU WILL ALL BE MADE LIKE US.
@@PCFixerthat sounds like vintage (Tomb of The) Cybermen to me... none of your new-fangled "Delete Delete Delete".... although, if it's a stradle erase head, should that be "Erase Write Erase"?
•_•
ꍓ
●⍘●
Well, the cyberman does delete your platter if it gets too close to it...
RE: Fixed drive not writing. This reminds me of a customer of mine back in the 70's who had a removable pack Plessy drive. He was using this for archiving data. What he didn't realize was that the write protect indicator bulb was burnt out. The drive would accept a write command but would not indicate write fault. When he (we) figure this out, we examined his "most recent" archive, and the data on it was three years old! Lucky he didn't have to rely on an archive/backup. Check your write protect button and indicator bulb.
Ah! You are obviously a Pokemon fan! Go team Rocket!
Musashi and Kojiro - the real heroes!
@@UsagiElectric Team Rocket represent. They're always there for each other and they don't let defeat stop them trying again. Best role models
I haven't watched Pokémon since middle school, but I recognised that R immediately. Next episode I hope he'll put on that fetching cropped jacket they wore
I was wondering what "R" signified.
this lens is great, the depth of field lenses distort everything
I don't know how I got to your site and all I can offer you is a PDP, Trash 80 icy pole. Keeps up the fantastic work, my empirical data friend!
These are beautiful machines, well worthy of preservation
When all heads fail: is there any documentation of how the head are designed and/or made?
If you it should be doable to recrate them
Otherwise once migt be able to disassamble an non-working head trying to reverse engeneer them
Same spiel for the servo cards
I know, i know it is a long and tedious project but we have come so far already
As seen at 27:44 the head mainly consists out of 2 transformers. And this is honestly how even a lot of modern hard drives are made these days with 20 TB drives as well. (except I think modern heads are MEMEs fabricated, since that is way easier...)
But yes, one can sit under a microscope and hand wire a new head. Just need very steady hands and plenty of patience.
How these heads were manufactured back in the day is a good question. I would not be surprised if they were manually wound by hand. Since that is how a lot of other things were made back in the 60's. (Magnetic core memory for an example were oftentimes manually threaded, a laborious task requiring a steady hand and plenty of patience.)
I think you are misunderstanding the straddle erase head. I think it means it straddles the full width of the track, a bit wider than the read write head. This is different to a tunnel erase head that has two thin strips that erase just the sides of the track. However, I don't understand why an erase head was needed for a digital drive as the write operation should overwrite any previous data. I think it was just to reduce noise from the edges of the track making head alignment less critical. I don't think any drives need two revolutions to do a write.
I found this info about floppy disk erase heads.
From the Siemens manual, page 3-4: "The read/write head also contains a tunnel-erase or straddle-erase electromagnet, the function of which is to erase the edges of the recorded track as the data is being written. The width of the track is narrowed to approximately 0.013 inch by this technique, to minimize the effect of data previously written on the track and possible crosstalk between tracks."
Discussion in the Siemens manual makes it clear that "tunnel erase" is one kind of head, and "straddle erase" is another kind of head; and that erase operation is concurrent with write. For instance pages 3-26 thru 3-29 describe the erase gating and logic during write: "The purpose of the auto erase feature is to provide the necessary turn-on delay between active WRITE and ERASE and the turn-off delay after WRITE goes inactive, but for tunnel-erase heads only. This is for option TE installed. Straddle-erase heads use option SE , which bypasses the delay circuits." A sketch of the tunnel-erased track shows that it measures .013 inches wide. The next document gives an explanation for why a delay is needed.
A CDC (Control Data) application note "5.25 inch FDD format considerations and controller compatibilities" examines 8-inch drive technology and standards. It refers specifically to tunnel erase vs. straddle erase in the introduction, saying "The head style and drive tolerances determine the minimum gap of the 5.25-inch FDD formats". The two erase heads as used on 8-inch floppy drives are described on page 7 as follows:
"Both head styles have as common parts a read/write gap and two erase gaps...The two erase gaps are used to erase guard bands on both sides of the data being written. These guard bands are necessary to to eliminate noise caused by old data that had been written slightly off track...A major difference between the two heads is the placement of the erase gaps....The erase gaps of the tunnel erase head are 36 mills behind the R/W gap whereas with the straddle erase head the gaps are on both sides of the R/W gap and extend approximately 11 mills behind it...Because the erase gaps of the tunnel erase head are so far from the write gap they must be turned on and off separately." Sketches of tracks for each head show that with the tunnel erase head, "used by IBM and most manufacturers", there is 6 mills of erase gap either side of the 13-mill track. With the straddle erase head "used by Shugart exclusively", each 6-mill erase gap is offset from the 13-mill track by 1-2 mills."
So both types of head erase just the edges of the track and it is done while writing in the middle. Straddle heads do overlap the write head length wise so they are turned on at the same time. Tunnel heads are behind the write head so needed a delayed signal. On a floppy drive there isn't a separate erase signal so the delay for tunel heads must be in the drive logic, not the controller.
Really takes me back to my first few years at work. The MRI I maintained used a 11/750 for the scanner and a 11/730 for the standalone viewing console. Both had removable packs. Something was really satisfying handling those packs.
I was able to visit the LSSM in Pittsburgh last year and got the grand tour. It was one of the highlights of my visit to the states and can highly recommend this place. I even learned that I was wrong about a thing or two and since strive to do better.
I wonder if you will be able to find the appropriate hawk controller. A head rewind attempt would also worth the search of somebody which have the right tools ( and ideally knowledge). Amazing machines.
Thank you for the bunny photage. Best part about watching your videos the bunny tax is paid.
I don't know why exacly but I really enjoy these videos. Thank you for all your work to make them.
Every time you talk about the EDS PC I am reminded of the TV show "Halt & Catch Fire". The story you tell of Centurion getting into the PC market is so much like the fictitious company Cardiff Electric in the show.
Hot air rework station (for Surface Mount stuff) works well for precision heating of marginal components.
Plus it gives you a reason to buy one if you don't already have one. LOL!
I got a Quick 957DW a while back and it's really nice. I'm just a hobbyist. Didn't need anything more expensive.
Those drives greatly resemble the drives on Honeywell Level 6 Minicomputers from that time. I suspect they may be manufactured by a common OEM and then rebranded and sold by various other companies including Centurion. What that means is that there's a good change you can find parts like drive headswhich you seem to need from other stock and or drives if you can figure out who actually made the drives and who else used ones like yours for their equipment. Honeywell is still in business too so they may actually have some parts stored away in their parts stocks.
I guess PE is the newer tech (would explain why there's no CDC 1500rpm version of it)
Yup! The SE heads were the original design, and then later, they came out with the PE heads. I think that timing on the SE heads vs PE heads could cause some problems on some particularly picky drive controller cards designed specifically for SE heads, so I imagine that CDC continued to manufacture them for backwards compatibility.
I LOVE watching these Videos, but man.... I REALLY, REALLY do NOT miss working on these massive old machines. We have it soooo good today with our modern computers. But it was all very new and exciting back then. 👍
I'm literally using an Intel NUC stuck under my Desk!
You should put a scope on the head signals to verify the write pulses get to the heads. The writes should be causing flux reversals no matter what. You could probably trigger on the sector pulses and get continuous view of a single sector, a bit like you did with the timing track on the Bendix drum. Then try some writes and see what happens. Also maybe try the same but disconnect the erase coil. Lastly check the pinouts? Maybe your SE heads are different.
Ever thought of selling small centurion compatible kit computers using new chip kits?
I absolutely want to, that's just a new, uncharted avenue for me that I'm not quite sure how to run down. We'll figure it out eventually though.
@@UsagiElectric well if you decide to go with a bus card similar to the og I always thought it would be cool to use Sega Genesis 64 pin cartridge connector or a SNES cart connector for a computer like that because unless you want to make exact clones it's ok to be different if its still compatible
A Micro-ATX Centurion Motherboard! What a dream that would be! Similar to the Monotech NuXT except it's a full-Centurian Computer rather than an IBM PC!
As long as it has a thermal printer, I'm keen.
Peripherals are the one thing people always forget about in vintage tech.
They were the core part of what made them worth buying.
I remember learning about the AMD 2900 series, and how you could define your own instruction set in the microcode, this was back in the mid 80s, at Polytechnic (now converted to and called University). We had a great lecturer who was very excited about his subject. I hadn't realised it was used in so many systems.
every hard drive I've worked with going back to the very early 70s (and those were designed in the early/mid 1960s) did flying erase/write in one pass. the erase gap was right in front of the write gap in the same head assembly, and both worked together.
I used to work on some Honeywell Level 6 machines and they had Hawk drives. Fixing a head crash on the fixed drive was comical. They pulled out the bad platter, took out the heads. Sanded the heads with fine sandpaper and then washed them out with high percentage alcohol. Then the would put in a new platter, reinstall the heads and format the drive. That worked - most of the time.
I have found some discussion regarding the head configuration on the larger floppy drives. The straddle configuration just trims the written track left and right to create some security gap to the neighbour tracks and to erase potential remains from an old track that can be left over due to slight positioning errors. Trimming the track can happen simultaneously or after writing, so these additional heads can also be behind the write head. From what I understand erasing and writing is still done simultaneously, but the timing of the erase gate must be adjusted in relation to the write gate according to the head configuration which is irrelevant on floppy drives as this is handled internally.
How is this implemented on the hawk drive? Are both gates exposed on the interface to the controller card?
Thank you for showing us all these beauties!
Watching a voice coil actuator was always impressive 'back in the day'. But what what was really awesome was the hydraulic actuators on the IBM S360! It seemed to me they were faster than the voice coils.
I had a strange idea: perhaps the heads coils could electrically be protected by adding a inline fuse. A small board could be made with a surface-mount fuse, which could be put between the head and board that reads the head. It the board goes, the head may be protected. Perhaps a few Zener diodes could clamp the voltage, a bit higher than its working voltage, too.
19:53 Team Rocket, YEAH!
David, since the badly crashed HAWK head was trash as far as you are/were concerned, why not take a no-risk chance at repairing or at least disassembling it under a microscope to better understand how it was built? You've got nothing to lose! I bought a stereo-microscope in order to attempt repairing a known-failed tape read (or write?) head for a DECtape drive. My friend had another head, so he swapped them out and sent me the bad one in the off-chance that I could get lucky trying to repair it. I need to re-start _trying_ to repair it for him!
We used to use medical "tube gauze" on a tongue depressor with 99 % isoprophyl to clean disk heads.
David, Keep making videos please.. Lots to learn. Lots of engineering. Thanks.
Gosh, my college has an Eclipse system just like that one in the museum, what a blast
I remember back in my minicomputer days working on Data General equipment exclusively. DG used Phoenix drives I believe. As memory serves they were 10 MB per platter with 20 MB total (10 fixed and 10 removeable). We also had some much older Diablo 33 drives that were 5MB removable only but they were being phased out in favor of the higher capacity drives. There also were some washing machine sized drives with stacks of removable disks that I think were 80 or 120 MB (can't remember). Those were also much older but still in use back in the 1970's. I was never aware of the head differences. Quite interesting!
Since you have an erase coil on either side of the write coil, with a straddle erase, you effectively erase before and after the location your are writing to. Which means that the data you are writing to the platter is erased after you write to it, no wonder you have corrupted data when you try to read it back. Since you have only four pins going to the head, presumably one pin for each coil with the fourth being a common, it would make sense to unpin the trailing erase coil, thereby converting the straddle erase into a pre-erase head. This might allow you to use this drive after all.
I think one easy way to describe the difference between pre-/straddle-erase heads is the coils in pre-erase heads are arranged in-line with/parallel to the data tracks and the coils in the straddle-erase heads are arranged perpendicular to the data tracks.
@UsagiElectric
Thank you for sharing your adventures with us!
Keep up the good work!
Is it possible to find someone who can rewind the coils on the heads that have blown coils? I know the wire is extreamly tiny, but there has to be a way to repair those heads.
5 MBytes - wow... That's like my average .docx file.... My first computer was a TI994/A... And I loved it...
That's more of a condemnation of docx and the company that inventeded it than a shade on this drive.
@@liquidsonly Yes,... no-a-days, programmers just add libraries, and code, no thought of how efficient you need to be... We definitely lost some skillz over the years... but we have also gained in amount of data you can process... I still love them old early 80's feelings..
That's 5 million ASCII characters without compression. That's a lot of text to type. Of course if you add images it will bloat up the document quickly.
Back in the day I seem to recall that main frame (ICL IBM etc) hard disks used significant air/vacuum filtration since the heads literally flew over the platter. Whilst a human hair would be pushed out of the way a smoke particle could cause a head crash.
This may be an apocryphal story but I'm reminded of an issue, in Holland I believe, where for some reason a head crash occurred. Since work had to go on the removeable disk pack was tried in another reader. That crashed too. Now there are 2 readers destroyed. Nevertheless some bright spark tried another disk pack in the original reader with obvious results. Fortunately somebody with sense halted operations and called in the gurus to fix the fundamental issue.
5 1/4 inch disks held in the same hand as a cup of coffee can produce a similar effect. I know I watched it and had to sort it. The look on my boss' face when I took a pair of scissors to his precious disks will not be forgotten!
Need somebody out there with a watchmaker's precision to dissect some bad heads and come up with a way to make new ones or repair the old ones, there's certainly going to be a market for it!
Many years ago when I was a student, we had a VAX, and I recall a time when I was in the computer center, and we heard this horrible noise coming from the disk cabinet. But we were totally baffled in that the system was still up and running and had apparently suffered no ill effects. We shut the system down as a precaution. The DEC guy came out to repair the hard drive, and we found that the motor shaft had sheared off where it went into the brake, and the horrible noise was the two ends of the broken shaft grinding against each other.
As much as I love these videos on these machines, I’m kinda nostalgic for the early tube days of the channel. It’s what brought me in, what you were doing was unique and fascinating.
I’d argue it’s time for a bit (lol) of tube computing videos.
We'll be getting back to that soon, I promise!
The paper tape reader for the 1-bit tube computer has gone through about three different design revisions over the past few months. I'm waiting on some new parts to arrive and then we'll start building prototypes and filming that. After UE-1 is finished, we're already working towards a design for UE-2. The reason there hasn't been any videos on that stuff in so long is because I'm trying to do a lot of the initial design work first, then get to work constructing and building it.
@@UsagiElectric excellent! Given what’s going on with Intel, I’d love to see people go revisit the 4004, perhaps using discreet logic chip, or, woooooo, transistors built into gates. I’d bet it could fit on a single 4x8 sheet of plywood using transistors.
I'm fairly sure you're wrong about it using two passes to write a sector with the straddle erase head. The data itself doesn't need erasing -- the data head erases the old data when it writes the new data. The erase heads are only there in case there's a misalignment between the old and new data tracks, to remove any spillover. Erasing and writing is still a single operation.
From what I've read about this, the only difference should be that the pre-erase method might need a short delay between turning on the erase and write heads, because of the separation between them, whereas the straddle erase method would not need this delay because the heads are in line.
Since your controller board seems to be telling the drive to turn on data and erase together, it would seem that any such delay, if required, must be implemented somewhere in the drive itself, and the controller doesn't need to know about it.
So I suspect your problem isn't caused by the pre-erase vs. straddle-erase issue, there must be something else wrong.
I wonder if you could get in touch with Lumafield to get the failed heads scanned. They did such collaboration with other channels so seems worth a try
Quite a blast from the past seeing that technology.
I find it fascinating that you _like_ CDC drives. I used to operate a CDC Cyber 72-26... with an 808 drive attached. It was huge, and terrifying. It used a separate 1500psi hydraulic pump to drive the heads. Fortunately I got an offer to change jobs, and waved goodbye to that CDC forever.
Wait.. A *hydraulic* head stack servo?! That both sounds cool and oddly terrifying at the same time.
@@Alexis_du_60 It was mainly terrifying Alexis.
An 808 was tall - over 6 feet high. The platters were over 2 ft diameter, and there were a LOT of them. So that the heads did not droop as they were pushed towards the inner track, they rode a fixed steel bar that was about 3 inches diameter. So the huge mass of the head assembly did not cause the whole drive system to rock, they had two spindles, one on each side of the logic and head actuators. The two sides were driven in unison, in opposite directions.
*Everything* about the 808 was massive... and despite the power applied through that hydraulic ram, they were slow to seek, with what had to be the longest latency in the rotating data storage game.
So far I have not seen any reference to one on the web.
18:03 why did you not remove that solidified gunk at the bottom there? wouldnt that cause problems down the line?
Mostly because it was solidified and hard as a rock. I would have had to use a chisel and a hammer to break it loose. It wasn't even sticky anymore, so leaving it there shouldn't cause any problems.
yeah i can see that, also having to chisel it out wont be good for the drive itself either i assume. as long as it doesnt become non solid gunk again it is fine, i get that reasoning. @@UsagiElectric
High Five on the slow mo bunny 🙂
I had no idea there was such a big community dedicated to making working fallout terminals!!!
Biggest recommendation is a 45.5 mm lens if you want crispness and depth in one package
Why did you not clean the gunk from the bottom?
Do you have any idea how the heads fail? Maybe you could carefully take apart a broken one to see if you can find the broken connection. I worked on many old VCRs, I have never seen a open head coil! Maybe its not the coil itself, but the termination of the copper wire, so you could fix them
Put it on the 11/44. Then maybe you could read RL02 packs that are around here and there.
i want one of these in my house or man cave just because that would be one hell if a toy to mess around with
As a guy who has absolutely no idea what you are talking about while controlling the drive on the computer it kinda feels like you are playing battleship with it 😂
I suppose in a sense you're not entirely wrong, lol.
That servo control board is going to be interesting to fix. I'm betting those cans are op amps, and i counted 10 of them. So any one of those op amps (if that's indeed what they are) could cause a failure. I'd bet good money that you have a couple of "instrument amps" for the reference and feedback signals, an error amp. The error amp would feed into probably at least a proportional amp and an integral amp (may even have a differential amp, but hose are tricky to get right). Then a summing amp to sum up the control signals (from the PID amps), then output amps to actually drive the voice coil.
And this is ALL analog, so good luck with that! Plus, there's no guarantee those op amps are even available any more. Sure, a new servo board could be engineered, but the cost would be eye watering. Unless someone just decided to do it as a fun project. ^-^
We nowadays take hard drives for granted, they're even being seen more and more like an obsolete technology with the advancements of solid-state storage, but it's astonishing if you think to all the R&D magnetic storage technology had to go thru to become the rather normal thing it is today.
You carry a head to a Hawk 10mb hard drive head around in your POCKET? Neat!!
Someone really needs to set up some virtual versions of these machines as teaching tools. It wasn’t until I started working with older more limited machines that I started truly understanding how computers work and how to write programs efficiently. Programming with extremely limited RAM and storage really forces you to think about cleaner more efficient ways of accomplishing tasks.
After seeing the differences between Pre-Erase and Straddle-Erase, it looks like the biggest problem is one of timing. The pre-erase head needs to have the erase head powered up for one "bit" time-length before the bit is actually written to the drive, while the straddle-erase can power up the erase head and the write head at the same time as writing the bit. There's no need to pre-erase the bit since writing the same bit to the same place won't change the magnetic field in that position, and writing the opposite bit to that location should just overwrite the bit which used to be there.
It looks like the straddle erase head is more mechanically complicated than the pre-erase head, which is probably why straddle erase came later, but it should be easier to interface to it due to not having to juggle the timing so that the erase head wipes a bit followed by the write head writing to the same bit as you need to do in order to use pre-erase heads.
Edit: Well, if they're powering both the erase head and the write head up at the same time, that means there must be some wasted space at the start and end of each sector when using pre-erase. That complicates things even more.
There might be circuitry inside the drive that does the delay according to the type of heads installed. In which case the same controller should work with both. I'm suspecting the problem is actually something completely different.
I had never heard of having to wait a full revolution of a disk in order to first erase and then write a sector. Is this really the case with many drives? Are there any downsides to pre-erase over straddle erase heads?
"We'll plug the heads in and..." Plug and pray.
I think it's interesting that bunnies chew side to side. Mooo! 😊
Would it be possible to disable the trailing coil of the straddle erase heads? Is there enough power on the leading erase head to do the job? (my ignorance is bliss...)
As always, fantastic video. Thanks so much.
The straddling isn't in that direction, it's side to side. The erase heads trim off any old data left behind that wasn't quite aligned with the track.
I think you could make it work. But for that you have to change the write routine in the OS and add a flip fop two switch between erase and write. Every read would reset the flip flop. (how to set up exactly has to be discussed.)
The Idea would be to change the OS that it would do the writing two times. One with only the erase head running and the other time only the write head running. It shouldn't be hard to find the writing routine and add a loop around it. (even in assembler)
This would slow down write operations by 50% but would make the the hawk drive work.
Digital high five to you too!!!
Wait, it has been 3 years since you started fixing this system up...... I mean it has been worth the watch but where did the time go.
I love how the hawk heads have a smiley face on them 🙂. They’re definitely happy to see you.
Might consider a stethoscope to listen for any bad noises coming from the drives. Would amplify and pinpoint them beyond what the unaided ear could hear.
I forget how much stuff you've picked up along the way!
A Raspberry Pi Pico being spliced in between the card and the drive could logically translate the drive head logic between the two. If you want to keep it more period-correct you could probably cobble together some logic gate chips to do the same.
This video was quite interesting but as soon as he said that video isn't about the Phoenix drive I was immediately like "aww, maaaaan".