
Call me crazy
but I'd quite happily run one of these in the cellar, just for the fun of it!
That said, I guess it's probably worth a lot more money than I could justify spending on any level of nostalgia!
Features make the system. After all, what good is a gaming rig without superfluous neon-blue lights, an iMac without the almost-matronly lack of sharp edges, or a Verizon phone not forged by the souls of the damned upon an altar of bones?1 This old box logo This simple and self-evident truth carries over to storage systems …
This device should be resurrected as a giant jukebox. Any fool can have a $15000 of tubes and gold-plated transformers amplifying their music, but how many can claim that a gas tube reads and writes their digital media on archival film through the purity of a vacuum? MP3s would sound WAY better.
In 1969 I used to service a true-track radar system called a Decca Photoplot. It had a tiny PPI crt display (you stared down a microscope to focus it). The ppi was tracked across the tube by separate xy deflection plates, resolving the ship's log and compass. The moving image exposed a 35mm film. After 6 minutes the film moved on one frame, the ppi reset, and incredibly tiny metering pumps flooded the exposed frame with developer & fixer, then the image was projected upwards onto a huge plotting table.
Kelvin hughes had an even wierder system, with a temporily photosensitive glass plate servoed across an ultraviolet ppi crt . There was a TV camera pointing at the plate which was 'interrogated' with blue light and fluoresced where the crt had burned in the target tracks. A few analogue buffers & you could have as many displays as you wanted on raster TVs. The plate was periodically erased with light of a third wavelenght - a flash while the plate movement was re-centered.
I do hope someone kept some of them in a museum somewhere.
Lovely, the lengths we had to go to - reminds me of the original 7181 displays - text only and 2,000 characters on the screen.
The 2000 characters were stored on a spiral steel wire in the side of the case which was twisted at one end to represent the bits of each character, and took exactly the screen refresh time for the wiggles to get to the other end and be read out.
Of course if you slapped the side of the case, the whole display went bananas :)
...how about a 'this old box' item about the fearsome ICL Magnetic Card File (NCR also used this devilish technology) or the rather cool ICL Optical Mark Reader, which made direct offline human input not only possible but useful by reading marks off a sheet of A4?
IBM made a memory system in the same period that used wide, 8 inch long strips of magnetic tape stored in canisters shaped like wedges out of a cylinder. These strips were removed and inserted with mechanical fingers and wrapped around a spinning drum to be read/written by a magnetic head traveling along one axis.
Now the best part; The large vertical holder supporting the ( 8? ) canisters was rotated by a HYDRAULIC MOTOR to quickly bring the correct canister to the read/write station. The cabinet quivered each time the drum rotated. The canisters seemed to blur and then appear at the new position. It wasn't really apparent that they rotated, the movement was that fast.
The large, two section cabinet also contained a high pressure oil pump, accumulator to store the pressurized oil and a drip tray at the bottom to catch the oil leakage from the plumbing!
I don't remember the system name but the IBM techs called it the "noodle picker" for obvious reasons. Lots of storage with faster access time than a reel of magnetic tape. A set of 6 or so had their own room at a state run computer facility. I helped install one of them.
These were called IBM 3850 Mass Storage
The "tape" was read and then transfered to a real IBM 3330-11 (IIRC).
The machine I worked on (serial number low 50's IIRC) was a good machine.
The major issue with it people used it for data that was needed in near real time. I believe it was designed for slow access. Because it was slow the OS (MVS) kept getting tied up in long queues waiting for I/O to complete and since the OS was designed to keep the system extremely busy(100 percent) it didn't like long waits for various functions (to complicated to go into here) and tasks stalled out then the OS tried to drive more tasks busy and the queues got munged up because of the delays.
We never really had any interface issues between the applications and the device (it was really transparent). We did torture it a bit and as a result found a lot of OS/I/O interface problems. I think it would have been a solid hit if the box wasn't abused by over use.
...but not outstandingly weird technology for the time. Try looking up NCR's CRAM (Card Random Access Memory) storage or Univac's Rotating Head Drum (the heads moved the drum, with a magnetic card wrapped round it by pneumatics, was static).
The IBM unit pictured was something of a special project, but I think NCR delivered quite a few of their CRAM units. Hopefully there are one or two left somewhere in a basement.
Actually, in my opinion, the 1960's and 1970's decades were the high point of mechanical engineering. The engineering needed to make a 3250 BPI GCR tape drive (at that time the icon of high-speed computing) run reliably was significant. Now everything's done by digital simulations.
Thumbing through old Bell System Tech Journals I recal a system I think they called "Digital Microfilm" using an electron gun writing on metal tape with a 7 level greyscale. Not sure if it went into production and I think it was from the late 1970s (78?) but my memory could be playing tricks.
Yes it was a staggering achievement for the time. A triumph of imagination and solid engineering skills over an almost impossible challenge. If anything has changed for the worse since then I suspect it is the level of imagination shown. In terms of what is available to engineers, in terms of components and services its orders of magnitude above what was available to the average guy in the field back then.
And as others have pointed out it would still be readable in 100 yrs.
Today that storage would be roughly 32 DVDs (at 4GB capacity) or 7 140GB USB backup drives.
I seem to recall the early Bell Labs/Western Electric ESS telephone switch used CRT and video camera as dynamic storage.
I would love to see some of that.
Also, RANK (and others) made film to video converters that used a "Flying Spot Scanner" which was a CRT focused on the film as it was pulled through the gate. Since just a tiny spot on the film was illuminated at any one instant, a simple photocell was all that was needed to read out the image.
Fun!
Baird's orginal live TV broadcasts used flying spot scanners. The scene was filmed on a standard film camera, but instead of being reeled up for later use the film went straight into a machine which developed it, then (still wet) it was pulled though a (mechanical?) flying spot scanner. Output of the photocell was used for the almost-live TV transmission.
He didn't have electronic cameras, and mechanically scanning the whole scene directly required unworkably bright lights.
... the IBM 1360 was preceded by quite a few years (late 40s), by another memory device that used an electron cannon: the Williams Tube. This was a very early "DRAM" device that could hold about 1k bits on the face of a small(ish) CRT. Despite being a tad temperamental, it saw practical use in the first commercial British computer, the Ferranti MkI - and several other early machines as well.
"The 2000 characters were stored on a spiral steel wire in the side of the case which was twisted at one end to represent the bits of each character,"
For more details look up ultrasonic delay line. If this is the thing I think it is the Special Ssauce (c Lewis Page) of this device was that it used a "torsional" wave to store the binary digit (Torsional digIT?) as a twisting wave going down the wire. The difference being the wave speed. Lower speed = shorter wire = lower cost.
Today it's almost impossible to understand how *much* storage cost then. A The fact that so many different storage technologies were tried (and their R&D costs could be jusified) gives some idea of what the benefits of finding a cheaper storage methods would be.
It is ironic that the winnning system (DRAM) might be the one with the most expensive start up costs.
I believe the "Noodle Picker" was the "Data Cell":
http://www-03.ibm.com/ibm/history/exhibits/storage/storage_2321.html
The 3850 stored its data in "shotgun shell" reels of tape, picked by a mechanical spider and spooled to disk. The 2321 used strips, which it picked out of segments, ... as described.
I remember it well because I once made the mistake of mentioning it among "strange storage devices I have known" only to find I was dining with one of its designers. He forgave me. He also drove a Morgan, BTW, which may help explain...