Debugging the 1959 IBM 729 Vacuum Column Tape Drive at the Computer History Museum

Debugging the 1959 IBM 729 Vacuum Column Tape Drive at the Computer History Museum


hello folks welcome back to our IBM 1401
exhibit our dinosaur computer from the 1960’s
and today I want to take you through one of the challenging repairs we had to
do on this tape unit. That will give you a chance to see how it is built inside, it is quite amazing actually. So this tape is called a vacuum column tape and the way it works is that instead of
going directly under the read heads, the tape is first is sucked in by vacuum in
this vacuum column right here, that Carl just opened for us, and then it allows the tape to coil into a reservoir here and the capstan can pull on it very fast. It can start and
stop so fast that it can just read one record at a time and of course that’s done so this can be
operated like disk drives which didn’t exist at the time In a second we are going to hit
the load button and you will see how the tape is sucked in the vacuum column.
You’ll hear and vacuum coming on first So Carl if you can load it. here we go and then you see the reels moving
here. It’s just the reel trying to keep the level of the tape at the right place
over here and behind this door is a little hole that senses vacuum and actuates a vacuum contact. So all the reels do is react to
replenish the supply of tape in the column, and the tape itself is driven by another motor right here at
a much faster rate So now let’s look at it from the back
to see how it’s constructed First there is a little bit of electronics
in it In this cabinet are the same,
rather simple SMS cards that we have in the main 1401 and lots of relay logic which
is almost impossible to follow on the diagrams. Quite a few lights to tell you
the state of the tape Then that big thing actually swings out and reveals all the motors
we have here There is no less than eight
motors. There are the two motors for the main hubs, two motors the capstans, the ones that actually move the tape, two motors for loading the tape one for the fast rewind, so that’s seven. And down there, I don’t know if you can see it, a super big one for the vacuum. So no less than eight motors to make this
thing work. I want to draw your attention to this
assembly here. This is a series of magnetic clutches and they are used to clutch
in and out the different motors that we just talked about. This one is used just for load, and these are the two main clutches for backward and forward The motors run continuously, the big ones there, and they are just clutched in and out which
allows the hubs to accelerate so fast and give it this jerky motion. So Carl is going to
load a tape and we’re going to try to see it from the back That’s the loading clutch and now you see the tape doing their
funny danse here and its just this clutch coming in and out on a regular basis until the tape is in position. Now we are reading and writing, you can hear the noise of the capstan starting and stopping You can see the hubs trying to catch up to replenish the vacuum column around this sensor point over here. and the other lower sensor point over there That’s what gives them this incredible random motion And our specific problem here has to do with the way the clutches are actuated Since they are rotating you need a slip ring to provide the current and those are these two copper rings and right on them you have a block of brushes that provides the current and of course with the time they have spent riding on traces, are failing and we need to replace them This is the original brush block with two brushes. This one is still there the other one is totally gone So here is a bunch of old brushes, they are all worn out and IBM won’t send us any new ones from 1959. So the only solution is to make new ones So here we go, our little machined brush with a hole in it good enough to put the wire in it. And voila, a carbon brush like in 1959. And another thing we need to reproduce the brushes is the very flexible wire at the back. We had a really hard time finding the right wire, but fortunately it turns out that the
company I work for actually makes wire in Oregon. We make it very high-end wire
for RF applications and one of the engineers over there Shashi heard my plight and I sent him a sample, and he made me the exact same wire, a whole spool of it, except he ugraded it to
silver-plated copper. So we have a the wire also faithful to the IBM original Here are the brushes with the wire epoxied OK, that’s the repaired brush block Two nice brushes. Now It should work perfectly fine. That’s the clutch – whao super heavy – and that’s the slip ring and that’s where our brushes go, they slide on those two copper rings. Putting the belts back together Here are the connectors Power is on. That’ clutch is working! That’s good, it means the new brushes work. There we go. No sparking? No sparking. Unfortunately. I liked the sparking, that was pretty spectacular. We don’t need that! [laughs] But I liked it! Alright, our brushes are doing their job Reading cards. OK, program is in. Tape not happy Oh yes [adjusting knob] Better when it’s in normal mode Iggy, I’m gonna use that drive! Just a second! Just do it on tape one that’s the one we want to know if it reads without errors Alright got the printer on There it goes The tape is working we are writing and reading on the tape
hopefully Is it supposed to print some errors in there? There we go. What does it say? Tape demo program, zero error counts. (Iggy) Wow. Zero error counts? So that did it! That did it. It was the sparking. So what happened is that we just wrote a whole bunch of stuff on the tape and it tells us that we didn’t get an error So the brush sparkng was causing the error and now we have good brushes
and I we’re good to go. We are going to try a high speed rewind. so it just unloads the tape to do the high-speed rewind and zip, off it goes And there is special motor to do that, this guy At one point it thinks it’s full enough, it reloads the tape and continues the rewind in low speed using the crutches and the regular motor It’s using that one and clutching in and out thanks to our brushes. There we go it’s unloading. Here we go. We’ve done it!

Leave a Response

Your email address will not be published. Required fields are marked *

100 thoughts on “Debugging the 1959 IBM 729 Vacuum Column Tape Drive at the Computer History Museum”

  • So the vacuum column is the "buffer" that allows the tape to be read faster than the reels can move… That actually makes sense. I was wondering how these things worked.

  • Wonder how often those carbon brushes need to be replaced with active use. I was just thinking the copper wire used for brush contacts looks an awful lot like de-soldering braid, but since you've got an entire spool now, no worries.

  • Wow. I'm blown away. I think there's as much mechanical work involved in this machine as electronic work in some modern systems. These are the days when finding bugs involved real bugs lol.

  • Very impressive, it's a pity that a lot of knowledge is disappearing. I have repaired a lot of old electromechanical organs, they do look a bit like these devices 🙂

  • Love those tape drives, they are as much mechanical as they are electronic. I remember seeing them as a kid in the early 80's and thinking how awesome they looked.

  • Thanks for a fantastic video.

    This takes me back to the beginning of my IT career when I worked for the San Diego Unified School District's Data Processing Department as an Applications Programmer. I was hired about the time the 1400 Series machines were being replaced by the IBM 360.

    I developed programs in COBOL for the Testing Services Department and I spent many weekends in the computer room and loaded/unloaded tapes from drives that were very similar to this one. During weekdays the machine was kept busy doing production jobs and the only time I could get my hands on the machine to test new software was on the weekends and that was a real treat.

    Thanks for keeping these mechanical and electrical masterpieces alive.

    Robert Montgomery

  • Such fantastic progress on such primitive machinery; I always wondered how those old school IBM reel-to-reel drives spun the data tape w/o snapping it…

  • Who came up with the idea 4 this machine? Then who made it? How? (Scratching head) what do blueprints look like?

  • hah! I own an old computer tape from the 80's, the vacuum tape drives were great, but the first manual moves from left to right reel until automatic tape loading by a finger was a little bit rubbish… 😀

  • I always wondered how those reels move so fast without shredding tape, now I know! Dampened by vacuum

  • MANY THANKS for putting this on. Ever since I was at school in the 1960's I've wanted to know how these tape drives work. You've certainly made an older man very happy.

  • 10:15 Was the current through the brushes 20mA, by any chance? I heard somewere that that exact level of current is enough to make the brush contacts self-cleaning.

  • 11:27 Funny, I don’t recall that sound effect being used in any SF movies or TV shows. It’s like nothing else I’ve heard on Earth. 🙂

  • These old tape drives brings back memories, the vacuum tape drives were far better than the modern drives that were attached the AS400 computer in the early 90’s. In that day there were ancient tape drives attached to a 4381 but boy were they fast. In those days IBM built quality and I was proud to work for IBM.

  • Anti slip ring I am. Inelegant, noisy, dirty, barbaric, and made to fail.
    1. Worked on large printers with precision registration requirements using solenoid spring clutches.
    2. Auto AC compressors for example, use hi torque magnetic clutches with a static field coil, ie no slip ring.
    Project?

    PS. Vacuum tape columns are an elegant solution.

  • It's absolutely amazing to me that these old machines actually worked, the engineers had so many different problems to solve beyond just the computation!

  • I had always wondered why the spools move independently and at such different speeds to each other. Now I know that the excess tape is being fed into a vacuum column.

  • There are still naval warships using tape drives. When I retired in 2014, the FFG-7 Class guided-missile frigates and some US Coast Guard gun-armed ships were transitioning over to CDs, but some, especially several foreign FFG-7s in nations with fewer funds for conversions and upgrades, were still using them. The tape drives are not IBM, but made by Sperry UNIVAC, all to "Mil-Spec" and MUCH more solid and heavy than the IBM drives you show here (half-inch steel frames and eighth-in steel cases, with heavy lock clamps around the doors to keep out water, etc.). The older drives, UNIVAC 1540s, were somewhat similar to the 1401 drives, but instead of a vacuum tape reserve buffer in two long tubes, they had a series of spring-loaded short arms with rollers at their tips. When you put tape through the machine from the source reel to the take-up reel, you threaded your tape through a narrow gap across all of those rollers, through the read/write head block, through another set of cushion arms, and onto the take-up reel, hand rotating everything to be tight. Then you hand-rotated the source reel forward a few turns (not quite to the LOAD POINT silver tape mark) to make sure it was secure on the take-up reel. This caused the two sets of arms to spread apart forming a pair of "W"s of tape, each arm of the W about a foot long, one on each side of the read/write block. You made sure the tape was not loose anywhere and then turned on the power. The arms suddenly spread apart even more and tightly held the tape and each arm could pivot under spring loading at its base, allowing more or less tape in the interval between the two reels, as needed, this being the cushion for fast tape motion that those two vacuum tubes do on your IBM drives. All loading was manual, including running the tape past the LOAD POINT silver tape mark and then manually pressing LOAD to cause the machine to move the tape forward and then backward to the LOAD POINT to be ready to run. Later, the UNIVAC 1840 drives appeared to slowly replace many, but not all, of the 1540s. These used a vacuum tube system like your IBM drives do, but not visible unless you opened up protective steel doors on the drive face, and were automatic: All you had to do was put the source tape on its reel shaft, lock it, and then feed the end of the tape into a small slot in the central read/write block, which was much bigger than used with a 1540. When you hit LOAD, the tape was sucked into that slot, the tape ran onto the take-up reel, the vacuum cushion inside turned on, and the tape was then positioned at the LOAD POINT silver tape mark ready to run. I assume that inside all of these machines, they were just as, if not more so, complex as yours are.

  • Beautiful systems. But l understand the hype when suddenly desktop stuff got possible… When all that room size machinery got squeezed together to one automatic box

  • Whatever application utilized round coals? I only know bow lamps and other furnace electrodes… Any other examples, comment section?
    Are this ones and their holders handmade?

  • I had no idea those IBM tape drives were so complicated. I learned a lot. I can't believe they used two separate vacuum chambers to accomplish the task of moving the tape. I am not sure if this was brilliant or the worst design ever. It seems like they faced a lot of obstacles and always seemed to find a solution. That explains why those IBM 729 tape drives were so tall.

  • Worked on similar units to these as late as the 80's in the Air Force. They used them to load programs onto a CENPAC unit and run automated test stations for maintenance on F-111s.

  • In the mid 70s this technology was shared with the other side of the iron curtain. They built a factory for tape in Dessau Elbe in the GDR, supposed to produce tape for all socialist countries, but its known the Russians built their own factory. The factory in Dessau also made tape for music cassettes, sold in the west under big brands like BASF too. In the factory, they used IT like in the video above for automation of the factory itself. Production ended just after the reunification of Germany, I have seen those machines standing in the rain ready for scrap.

  • THANKS for the video! the first computer on which I put my hand was an IBM ps2, late 80s. So for me the storage units start from hard drives with less than 100 MB and 3.5-inch floppies. I am surprised at the use of vacuum in these devices! p.s: the three guys with the red t-shirt are cool and the love for those machines makes them "young".

  • The enginerring complexity of that time to accomplish something so simple we take for granted today is amazing. IBM hired the best.

  • I want to see how the data was used in a real world situation. So what exactly did companies do with these and utilize the data after it was put on these reels?

  • THIS is what a computer should be. If there was a Platonic ideal of a computer, this with its chunky relay boards and blinkenlights and swooshy tape drives seeking and reading and reversing it is THIS

  • heh… the printout at the end.. it's kinda like what dd prints.. wonder if that's why it uses block size and count… would dd run on this?

  • Someone should write a simulation of this and expose it as a block device to the kernel then write an OpenGL visualisation app that can show tape reel position and how much is sucked into the virtual vacuum tubes and how fast it's spinning and stuff. Then you could make a filesystem on it and copy stuff around and watch the reels spinning as it did its thing 😀

  • WOW! What a mechanical nightmare. I used to be a tech at Control Data Corp and worked on the 60x and 67x tape drive made in the 70's. Light years ahead of these old drives, even for the 70's

  • There’s something deeply satisfying about actually fixing these old things instead of just letting them collect dust all broken down 🖖

  • crazy to see how much computers have evolved .. i bet its safe to say my phone has more processing power than this

  • Strange that the electromagnet of the clutch is inside the pulley, thus needing these burshes.
    It is the same principle as A/C compressor clutches in cars, and in these, the magnet is affixed to the compressor and just attract the clutch plate towards a continuously rotating pulley, thus not needing brushes.
    By the time this tape drive was made, A/C compressors already worked that way. Seems IBM just overengineered the thing a bit here.

  • What were you using for brush material? And where do you get conductive epoxy that's conductive enough to handle this clutch current? I was watching, hoping to learn how to metalize carbon/graphite so it can be soldered to.

  • Did you know that Mercedes Benz provides replacement parts for every Mercedes ever built? No matter how old? And they predate IBM computers by quite a bit.

    They might not be cheap by anyones standard, but they are available.

  • Oh dear, my computer isn't doing so well. Time to replace the clutch and throwout bearing, change some belts, and put in fresh dielectric oil.

  • Wow! Thank you so much for creating this video! I had never seen the inner working of machines like these. It is still amazing what people developed back in the days and it is funny to see all these young guys at work (including yourself at the lathe 😉 ) to get the machine working fine again. Thank you Marc.

  • Hi lads, Awesome job on that Tape Drive. These things always fascinated me but I was wondering how you conductive Glued those wires in the brushes????? I have a few of these to make as well.