Old timer here with a real life application of this strange tech. . ... the IBM 5496 Data Entry machine used this technology for storing key strokes during the data entry. That was in days of yore when punched cards roamed the Earth. Each punch card stored only 96 columns of data so the need for storage was small with only 6 bits per column. So were talking only far less than 1K. The advantage for this tech was that older 'keypunch' machines punched holes in the card with each keystroke... so one typo and the card had to be junked. With the ability to store all the keystrokes in some form of memory one could key the whole card, back up and make corrections then when you were satisfied that the card was properly keyed, press the button and the card punched. Not so many wasted cards, not so much rekeying. I fixed these things for several years in the early 70s and found that delay lines were extremely reliable,. I only remember one delay ever failing. The memory didn't have to be fast, just reliable and cheap. Competing memory at the time was expensive.

@imaginerus

7 жыл бұрын

RCW Thank you very much for your report! I wouldn't have thought that these things were actually reliable, especially for being mechanical _and_ dynamical memory, where there is such much potential for things to go wrong... But I kind of like the technology of the early computer days, when people came up with such creative ideas; where today it's only about getting the same stuff smaller and faster!

@ohokcool

7 жыл бұрын

How does it actually store anything? I'm very confused

@ohokcool

7 жыл бұрын

My guess is that the wave is propagated from the magnet to the sensor at which it is thrown back to the magnet to restart the wave. I don't see how it could keep memory without this kind of feedback loop keeping the wires from reaching mechanical equilibrium

@rich1051414

7 жыл бұрын

Magnetic core memory was so reliable, it was still used until fairly modern times.

@fliptrontube

7 жыл бұрын

The twist does travel along the wire from a transmitter at one end to a receiver at the other end. It creates a faint signal at the receive end. It is then amplified and then sent back to the transmitter. For example, a bit (a twist pulse) might be 1 microsecond long. The twist travels along the wire at a speed determined by the wire material. Lets say the time for the twist to travel the full length is 1000 microseconds. We could send a new bit every microsecond, and the wire would have 1000 bits stored in it. Every microsecond, a new bit is transmitted, and a bit that was transmitted 1000 microseconds ago, will come out at the receive end. Unlike RAM in current technology computers (the R stands for Random, which means you can get at any bit at any time), these delay line memories are Sequential access memories, and if you want bit 378, you need to wait for it to come out, which will depend on where it is in the wire at the time you need it. Maybe just 1 microsecond, or maybe 999 microseconds.

It's called a magnetostrictive delay line. The transducer causes a tortion of the end of the wire, and that torsional wave runs down the length of the wire, and is eventually picked up by the transducer at the receiving end of the wire. There were earlier ones that used a compression (sound) wave, but the torsion wave was less affected by mechanical noise, for example from working on an attached keyboard. In 1974, I was working on a mainframe computer site that used VDU's (video display units) made by Conrack that used acoustic wire coil like this for the screen memory. They also used disk units that had four platters in them that turned about a horizontal axis, so the disks were vertical - and had a diamater of about 4'6" or possibly 5 ft. They were huge, and had a head for each track. On that site, we had three Boroughs mainframes, two with 4-wire core memories and one with a 2-wire memory. The sense amps sensed the current on the X- and Y- wires when a core changed state. It was all wonderful stuff. I never worked around magnetic drums (which preceded disks), but they were quite heavy, and I once read that "In the event of a bearing siezure, the kinetic energy in the drum would be released with consequent, extensive damage ..." - not to the unit, but - "... to the computer room". You can just imagine the cabinet starting to spin, tearing itself free of its cables and going walkies around the computer room. Wow!

@jimsteele9261

7 жыл бұрын

The first Burroughs mainframe, a B6700, I installed used that "head per track" disk. Zero seek time. ;-) Reliable stuff, the only failure I recall was one of the head modules. Someone told me Bell Telephone used a modified version of that disk to play recorded messages like the time, or "can't connect" things. The terminals I referred to earlier were also made by Conrac.

@jacksonlefteye

7 жыл бұрын

"going walkies" >. fascinating story though

@Roxor128

7 жыл бұрын

Australia's first computer, CSIRAC, had drum memory for secondary storage. The drum memory system used the motor out of a washing machine.

@DownhillAllTheWay

7 жыл бұрын

I don't recall your name, but it seems you worked on a very similar site to where I was. Mine was a bank centre in Brent. I can't remember the mainframe types now - it definitely had two B7600's with 4-wire memory, and the third mainframe - I think it was also a 7600, but 5200 also rings a bell - had a 2-wire core memory, which was bigger (a megabyte?), but also a bit slower than the 4-wire cores. There were a couple of sister sites - PNCU was one. The disks weren't all that reliable. You may remember that they had touch sensors, and if any head touched the disk surface, they would all retract. It got to be such a nuicance that the platters were sent away for teflon coating - then when the teflon started peeling off and getting under the heads, they were sent away for "decontamination". Who would ever have thought then, that we would ever have 256GB on a micro-SD card! I remember an article in Wireless World (an authoritative mag in its day), saying that with the 1Mbit chip, we had reached the limit due to stray particles from the sun being able to change the charge on the micro-capacitors that RAM used.

@jimsteele9261

7 жыл бұрын

I worked mainly at an auto parts plant in Michigan. The 6700 there had what we called "planar core". It was made up of modules of two pcbs attached with a lot of gold plated pins. Each was 32k x 20 bits. One board had the sense amps and drivers, the other the core. The next generation machines all had chip memory. I do remember some talk about those early ram chips being subject to a random error if a stray cosmic ray hit in the right spot. There was something about a batch made with ceramic packaging that was slightly radioactive causing errors. Fun stuff.

You know, we look at something like this with our modern eyes and think, that is ancient, but it really was brilliant stuff. It always amazes me to see how the people of the past used what they had available to solve the problems they faced.

Back when I was a field engineer at Burroughs, I worked on a terminal that used this kind of tech as the screen memory. It was in the mid 70's but they were obsolete then... Something about a shortage of the new models... There was a screw on there that adjusted the tension on the wire to adjust the memory timing. I always took my rubber mallet over there to bang on the circuit boards, and a "pot tweaker" to adjust the memory. One of the other pcbs in the terminal was a hoot, by today's standards... the character generator used a huge array of individual diodes to define the font. Fun stuff. ;-D

@e.s.oteric5068

7 жыл бұрын

Oh the days of diode logic :) Love the calibrated hammer!

@killerorca1

7 жыл бұрын

my calibrated hammer was named mem killer

@bshoke

7 жыл бұрын

Jim Steele my great-uncle worked for Burroughs, then unisys he actually had a set of cards from some of the old machines actually he would take the big cabinet cases and put counter tops on them to make work benches, but he had a memory module that consisted of wire mesh it seems like he said something about the wire could overheat and break fairly easy (it's been a while and I can't ask to double check) also had a bunch of 4 transistor flip flop cards, optical punch card reader.

@jimsteele9261

7 жыл бұрын

I heard about a guy who made a coffee table out of one of the platters from a fixed head disk unit.

The Olivetti coil memory is out of the Programma P101. A programmable calculator that could add, subtract, multiply and divide, calculate square root and store data. It had a read / write mag card that stored the "programme" sequence required for the calculations. NASA had a number of these for the Apollo 11 moon landing, evidently to help tell the antenna where to locate earth. Olivetti were quite leading edge with this device, HP "borrowed" some of the tech and wound up paying some royalties for the pleasure. Olivetti design was amazing, from their mechnical calcs right through their commercial machines such as the P603, A4, TES501 and Praxis 48 typewriters. Ah, good years fixing those beasts. Although a horror memory is dropping the twin single sided 8" floppy disk drive of a TES501 down a double flight of stairs when the bottom of the shipping cardboard box unglued.

@JeffreySJonas

3 жыл бұрын

I kinda used the Olivetti Programma 101 in jr high school, preferring the "newer" Compucorp 025 Educator programmable calculators. But the Olivetti allowed storing programs on magnetic card!

Hi xjet, what a cool video! Only half through it I realized the Olivetti board was from the Olivetti Programma 101, the first personal computer in the world! I'm from Italy and I was lucky to met it's creators two years ago, during the Trieste Mini Maker Faire. The were proudly explaining people how the machine worked and why was revolutionary at the time. The also had one of it still working! I recognized it because of the unique "forest" or "wafer" of resistor, on the left of the coil. They explained us how they initially used only nichel for the memory, and the twist would occupy something like 3mm of wire while travelling along it, and then they welded the nichel to the steel because the twist would occupy half the space (1.5mm) and allowing them to store more data! One of the most fascinating thing was how the Olivetti in Ivrea was one of the most advancef labs, they were great teamworkers and their main source of knowledge was a very big and constanty updated library that included books, magazines, research papers and so on. It was very fascinating talk with these people and grasp what was like the R&D back in the days.

@KB4QAA

6 жыл бұрын

en.wikipedia.org/wiki/Programma_101

I used to work for Olivetti back in the day. Those delay line memory modules were used in the Auditronic 770/730's and the TC380/350'. They were very reliable as far as I can remember. The modules on the circuit boards with the 8 or 6 hole cap were simple gate modules created using discreet compnents. the large gaps between the rows allowed the cards to be placed closely together with the rows interlacing. God I feel old LOL. I'm still repairing computers but I'm 63 now and will retire in 16 months.

@stevew8233

7 жыл бұрын

Looks to me that the Olivetti memory was from a Programma 101 desktop calculator and accounting machine. I was at college ca. 1968 and we had one of these as a cheapo computer substitute. It had a printer, magnetic card input and output for programs. If I remember correctly, calculator operations on the keyboard could be recorded on the magnetic cards and some form of simple branch logic and subroutine capability was possible. It was a nice modernistic looking unit for that time. Funnily enough, the company I went on to work for a year or two later based their first computer on delay line memory, but in that case the delay line was a mercury tube which regenerated sonic pulses along its length. That was the Univac I. Later: Wiki has a description of the Olivetti: en.wikipedia.org/wiki/Programma_101

Well that made me feel rather old, as I recognized much of those components! My father came to Canada in 1945, having sold up his electronics shop in Cardiff, Wales, with the money he made from turning obsolete radar crts into TVs. He went on to work, in Canada, on the DEW line radars in the arctic,mines, power plants, etc., and had me stripping old (mostly tube stuff) for components (got zapped by charged capacitors many times!). I also remember touring the IBM offices in Winnipeg, Manitoba, with my brotherinlaw , who showed me how they had just changed from large computer punch cards, to small punch cards, and the banks of twirling reel to reel tapes!

7:13 That construction technique is called "cordwood construction" Yo place axial leaded components in a tightly packed pattern between two PC board layers. Name comes from the visual similarity to a stack of cordwood. It allowed a high part density, but was harder to service, particularly for interior components. This one isn't too dense, but some are quite incredibly packed.

I worked at Olivetti in the early 1970's as a programmer, a skill I learned on the job(there was no other way at that time). My first machine was a very heavy business machine (the A-5) which had a lot of electromechanical printing capability and had 2K of programmable memory, programmed in assembler language. It was really a 32K but most of it was taken up by a burned in interpreter for the keyboard entered instructions. External memory was on mag striped cards. This machine could handle most accounting functions for a small business (accounting, payroll, etc). Anyway, our office ( in NYC) had lots of legacy equipment lying around including these wire memories and coil and core memories. This is the first time I have ever seen/heard a discussion about them. SO...thanks for the memories.

I learned more watching 1960's navy electrical training than any book I've found. They did so much with such basic tech it's crazy.

I believe the ICs on that last board that you showed were called Flat Packs (or maybe FlatPaks) back in the day. I once watched technicians soldering a lot of those early ICs to boards at a Texas Instruments facility. I once built a homemade little computer system that used a magnetic drum memory for its main storage. I found the drum at an electronic surplus store. Most of it tracks had been trashed but a few remained and I got it to work again. That project took me from 1964 to 1967.

When I see things like this I can't help but think of the genius and the insight of the engineers and dreamers that came up with this stuff. Nowadays we are streamlining and refining tech, but seldom do we re-invent the wheel. These people didn't re-invent the wheel... they invented the wheel it's self. That takes much more creativity.

@cup_and_cone

10 жыл бұрын

Couldn't agree more. All we do now is figure out how to put more into the same broken systems, not create new systems.

@myselfremade

9 жыл бұрын

What about DRAM? what about flash memory? What about fiber optic communication? Those are all amazing innovations JUST related to this one

@celluskh6009

7 жыл бұрын

Shame this is the top comment, as it takes a poor student of history to think 'old' inventions didn't come from countless iterations of even older ideas, just as they do now.

@DeGuerre

7 жыл бұрын

Cellus is correct. Mechanical delay line memory grew out of other delay-based memories, such as cathode ray tube memory and mercury tube memory. Most people don't know that DRAM actually dates to the 1930s! The Atanasoff-Berry computer used capacitors built into a rotating drum to implement the fast refresh mechanism required.

@erikdravn

7 жыл бұрын

He may be perfectly correct, I would not argue against his comment. However, he completely missed the point of mine. ;)

Dude, I am absolutely fascinated by this old technology, mainly by how the engineers of that time managed to do things. I'd really love it to see more of your content relative to this topic!

I really enjoyed this - geek fix for the day complete. Would be great to see more when you have time!

@xjet

9 жыл бұрын

I'm probably going to be doing some teardowns over the next few weeks -- since my primary activity (building and flying RC models) is off the agenda for a while.

I can imagine building a complete (but simple) computer out of relays. Simple logic gates. What I can not figure out is how somebody could figure, "Hey, know what we'll do? We'll store data by twisting a wire!" That could NEVER have occurred to me.

@allanrichardson1468

8 жыл бұрын

+BlackEpyon Some of the prototypes used a long, narrow pool of mercury. It's just like sonar: send out a pulse of sound and wait for it to come back. The delay lines, regardless of their physical form, just sent out sound pulses and waited a predetermined length of time to reach the other end, counting pulses so they would know how many bits had been sent. After a given "address" came back to the sensor at the other end, it was either refreshed and copied to the computer, or replaced with new data.

@BlackEpyon

8 жыл бұрын

Allan Richardson And I finally found out what that box with springs in old sound mixers was: the reverb.

@david203

8 жыл бұрын

+Allan Richardson Wasn't just prototypes. Some or all of the Univac computer systems were built around fairly large cylindrical mercury tanks with peizoelectric transducers to induce and detect sound waves.

@dunxy

8 жыл бұрын

+BlackEpyon I expected magnetic recording when i first spotted it.I know they recorded audio on wire at some stage, so data recording should be feasible.

@david203

8 жыл бұрын

+dunxy The first Univac tape drive storage units used steel tape. The motors had to be very strong to handle it, and rusting was a problem. I'm not kidding, I've seen samples of the tape.

All those old components, with their bent, solder-coated leads sticking into the circuit board, are like images from a travelogue through many hours, days and years of my past life.

The Sperry-Rand Univac computer system, which was based on a large mercury tank ringed by piezoelectric transducers to store memory bits as streams of sound pulses, had a console meter to indicate the accuracy of the mercury delay memory. You could see the meter wiggling randomly as the accuracy varied. This was a true mechanical delay memory, and it had the clumsiness of any mechanical component. The correct pulse sequence was constantly being regenerated using redundancy to eliminate the frequent errors in propagation and detection of the sound pulses.

I used to work on a radar that had vacuum tube analog computer in it. AN/MPS77 I think was the designation. It was used in the ArcLight Rolling Thunder program in Vietnam. Pain in the rear to keep working and calibrated but it did work.

Old technology always fascinates me. Really enjoyed it. I never saw those memory things- reminds me of a wire recorder!

When you look back at mechanical type computing devices from the mid 1900's its actually incredibly ingenious how they did these things. I really love your videos and appreciate you showing, most of these things would be all but forgotten in this digital age if we didnt have people like you.

You are certainly not wasting your time, I think these old tech videos are great, and nice to have a slight understanding of how there devices worked. TY for taking the time to make them.

Interesting, thanks for sharing, I never knew about this delay-line memory before. I've used/built computers since the '80s, still have a Tandy and a Timex around here somewhere. But that's as far back as i go. Used the 5 1/4" floppies as recently as the early 2000's in some 386/486s and such.

@kevinsocal59

7 жыл бұрын

I have a 'Tandy Sensation!' put away here. I think the keyboard weighs a couple of pounds, LoL

Delay line memory, wow! It's like saying a word against a mountain, then hearing the echo, amplifying that echo to have it sent out once a again as a form of storing information.

Great vid! In 1968 I was just out of tech school. There, we learned about time-delay multiplexing on a unit that incorporated a time-delay electron tube for the timing standard. I remember our instructor telling us how lucky we were to learn on such "modern" equipment, because the previous model used exactly the same-delay line mechanism you discuss here.

Pretty interesting to see such early SMT at 8:04.

@PierreaSweedieCat

9 жыл бұрын

Cool! Thanks!

@RalphDratman

8 жыл бұрын

+Maria Engstrom Surface mounted componenents did not become really popular until the 1980s, if I rememberer correctly

I came in during the early '80s. A friend of mine's dad worked for IBM in Cherry Hill and brought home one of the first production PCs they marketed in '82. I've used all the tech involved from then to now since. It is amazing how far things have come in the time they have. Thanks for the walk down memory lane and some info on memory I've heard about but never really seen up close. Cheers!

Fascinating. I still have a functional 8088 Computer with multiple 5 1/4" disks, and a Commodore Vic 20 complete with the cassette recorder and several programs on cassettes.

acousticl delay line memory, worked by depending on the binary length, sending a torsional wave round and round until extraction. your description was was very good. The olivetti part you are showing was from the very first printer calculator released in 1965 at a cost of over $3,000 which today would cost about $40,000, just for a calculator, quite shocking really. Just goes to show how expensive electronic devices were in the early days, now you can buy one for $2 in wallmart.

In the 1970s, I can remember equipment like that sending "read", "update" and "write" pulse to each row of a ferrite core store. The transducers arranged around the delay line produced the time interval between the pulses. Shift registers sent a set of three pulse to each row in turn. The duration for a set of pulses was 80 micro sec.

Many years ago, when I was a computer tech, I used to repair video terminals that used a delay line that looked similar to the first one in this video. The sensor transducer was the one connected to that long brass screw. That screw was used to range the delay line. Turning the screw moved the transducer along the wire. You'd turn the screw one way, until the text started breaking up, then back the other way, counting turns, until it started breaking up again and then turn back to half way between those two points. The recirculation was on the terminal logic board. When a character was inserted, the data was actually buffered in ICs and similarly when deleting a character, it would be deleted in those same ICs. I also used to work on core memory back then, and still have a core plane from a Collins computer, which is on the shelf behind me. Here is a picture of that core plane. It's 4K bits and there were 32 of them stacked in a module, with 4 modules in the computer for a total of 64 K bytes. drive.google.com/file/d/0B5LapMwk8iPrNkFZUXA4cmd5dTg/view?usp=sharing

What a lovely collection of very rare stuff. I used to work on calculators which used wire delay line memory. I didn't know the pulse was generated by twisting the wire, I thought it was piezo electric so thank you for that information. What did surprise me with those delay lines was how unaffected they were by shock. I remember trying to bang one with a screwdriver handle to get the memory to change and couldn't, very impressive.

Pretty cool stuff. When I first started with PC's, the 5 1/4 floppies had just come out which were so much more a joy to use than those cassette recorder storage method.

Just a few notes on this lovely old stuff... The twin sensors on the first wire thing is confusing. All I can think of, is a speed option. As in, the longer the line, the more storage, but the longer you wait for your data to come through. So maybe having one halfway along the wire lets you set a half-capacity twice-speed mode? As far as modern DRAM, it STILL needs refreshing! Just a bunch of capacitors, one per bit. DDR5 SDRAM is still DRAM! The construction method for the PCB with the vertical components hanging off, between the main PCB and the little PCBs is a method called "cordwood construction". At the time it was a way of fitting more components into a space. Must've been a bugger to repair though. For whatever reason, it fell out of use. The "delay lines" on the board with the microchips, I don't think are. I think they're inductors, for power smoothing. The fact they're next to all those capacitors, presumably for the same purpose, makes me think that. Lastly the 5 1/4 inch drives ended up with 1.2MB, after many long years! But I think 8" were long dead by then. As densities improved there was no need for the gigantic 8" disks. Maybe some terrible old machine somewhere still uses the 8 inchers, best left in it's dark corner if it does. Interesting hardware you got there mate!

@0MoTheG

7 жыл бұрын

No the delay lines are actually delay lines, just not mechanical but a bunch of coupled oscillators.

@greenaum

7 жыл бұрын

Coupled oscillators? How does that work? I noticed the capacitors seemed a bit weirdly wired up. The circuit with the two coils, one either side of a dozen or so caps. And there were 2 copies of that setup. That the one you mean? Far as I know no delay line would work like that. If it were some sort of bucket brigade, wouldn't it need switching transistors on each element?

@0MoTheG

7 жыл бұрын

greenaum "lumped electromagnetic delay line" looks like a LP filter and is used at it's resonant frequency. Useful if only little delay in a narrow frequency range is needed.

@0MoTheG

7 жыл бұрын

In general search: "electrical all-pass filter circuit". That is what these are from a filter standpoint. The information that they store per stage is a constant, therefore a distributed transmission line like the mechanical one shown here is more practical than chaining lumped ones.

@0MoTheG

7 жыл бұрын

greenaum "bucket brigade, wouldn't it need switching transistors" you might be thinking of a CCD best known for their use in image sensors, these were also used as memories but do not need a true switch, only an electrode that moves and creates the "buckets".

That is just weird. I love it. It's like somehow storing data on a clock spring. I've still got no idea how that works, but I appreciate the video. I don't feel quite so old now. I had a ZX81 when I was a kid.

How can you not love old tech?

The assembler /disassembler reference, on the floppy brings back memories from highschool. My dada comoany had an IBM system 32 i remember as a kid, i the 70s. i use to look at its tech manual back then.

Its pretty incredible to see how far we have come, I didn't even know the coil twist memory existed.

I wonder what's the read/write speed...

Eons ago, in college we were blessed with an Olivetti Programma 101. This in an age where slide rules were still king. It had programmable magnetic cards and provided many of us with our first computer experiences.

@xjet

10 жыл бұрын

Snap! I think that was the first computer I ever got a chance to use too!

Well, you're the first person I've found to actually pay attention to how your sound is captured with that lovely mic so I'll defo be watching more of your stuff

We used to use magneto strictive delay line memory like that to hold a complete 80x25 VDU screen of data back in the early 1970's - they were rather temperature sensitive. The board you showed with ferrite rod wound and discrete capacitor delay lines would be using them as timing elements where a series of events need repeatedly to be performed - possible in that read, and re-write in the dynamic chip array.

hey Bruce, I never used the 8 inch diskettes, when I started I used the 5.25 inch double sided disks, but I had seen the 8 inch, and much larger disks. It's really great to see how far technology has come. thanks for the look in to the past!!!

Hi Bruce, its good to look back an see this old tech and how far technology has leaped forward, One suggestion for you to teach us rc modelers is how new ignition systems work on all these new gas/petrol engines an the pros an cons of this reinvention back to petrol power... back to the bench...

Excellent historical closeup review. I've never heard of these delay line memories or simply glossed through them when reading about them. But this video really makes them come to life.

As other people commented, I would really like a demonstration, even if just with oscilloscope probes and manual signal injection

@cyberyogicowindler2448

Жыл бұрын

Here is one about the Friden EC-130 calculator hardware. kzread.info/dash/bejne/mJ2Zj7xmnc3Uqbg.html

wow, those boards should be in a museum. The National Smithsonian museum of American history has a very primitive walk though the evolution of computers. I was impressed to see a block of Core memory, what you just showed was astounding that it still exists in such a pristine state. Wow...

Had to ask my father about this stuff. He used to work for Honeywell and IBM as an engineer in the early 60's. He described the function of the delay lines on the second board from Olivetti. Somewhere he has a stack of early RAM boards. Probably down by the DEC Rainbow 100+ LOL!

Is it possible to see a demonstration of one working?

@rasvial

7 жыл бұрын

I mean there wouldn't be much to see.. electricity flowing through a wire?

@mxxc

7 жыл бұрын

But the mechanical twist of the wire..

@illustriouschin

7 жыл бұрын

we could see it store data and then display it.

@BisMaxx

7 жыл бұрын

I too am confused as to how this works and would love to better understand.

@Ciogos

4 жыл бұрын

kzread.info/dash/bejne/ZHZ92pSxXbSYlqg.html check it.

Thanks for posting this. I heard about the Olivetti desktop computer from the 1960s (really a desktop programmable calculator), and they mentioned that it used delay-line memory. I was curious how that worked. The only old-time memory I'd heard about before was the mercury chambers, which was another kind of mechanical memory (it used acoustic waves in mercury to store bits of information), and core memory.

I used 8” disks regularly with a CMX 360 and 3600 videotape editor. It was old tech, but it worked well. We would use those disks to save edit decision lists (EDLs) for clients, so if they had any changes, they could come back in the next month and re-do their stuff. Mind you, at home, I was using 1.44 Mb Sony diskettes as well as hard drives for my PC. But I was using the high-density double-sided soft sector 8.5” disks well into the 1990s.

Wow, this is absolutely incredible. Whoever came up with this was a genius.

What kind of wily frikkin' genius came up with this? My jealous brain is not even capable of imagining the bar napkin conversation that led to using pulsing wire twists to store data. With the incredible power of modern computers and all the amazing new things being developed almost daily, it's easy to look at how comparatively weak computers were back in the earliest years and then think of the engineers as primitive as well. The fact is these folks were absolutely brilliant to do so much with so little.

I must say, I am glad I stumbled across this channel :) As a long term subscriber to other such channels (including of course Dave)... I really enjoyed this vid. I am not personally into RC stuff but have subscribed to this channel in the hope I'll see more! You have a very easy going attitude, and a great presenting style. Will deffo be watching more!! So greetings and respect from the UK :)

I have several old calculators that use delay line memory. A Smith Corona-Marchant Cogito 240SR, a Friden EC-132, A Monroe 925, a Sony Sobax ICC-600W, and a Canon 163, off the top of my head. I SO wish I had one of those old Olivetti calculators. I've actually got over 100 calculators, but those are my oldest ones, and specifically ones featuring magnetostrictive delay line memory. I can tell you that the Cogito had a 480 bit capacity delay line.

Right on the money. IBM used them on some banking terminals which I used to teach in the IBM Customer Engineering school in the 1970s. They only held 512 words of memory, where extremely reliable and where integrated into an IBM 1130 processor.Gabriel Shaw

Didn't see in the comments, but the delay line in the unit that had the coils was probably an L-C delay line. Series inductors, shunt capacitors. We used to use custom built analog video delay lines for timing NTSC video signals in a television station. Measure with a scope how much delay is needed for a signal, then you have the option of using enough coaxial cable as the situation demands, or - as we'd usually do - order a custom delay line, with 75Ω impedance. When I was just out of high school in Minnesota, a couple of Air Force guys agreed to give me a tour of our local radar station (you never knew when those sneaky Canadians might attack!) The rotating dish received both reflections from airborne objects and ground clutter. A mercury delay line was used to cancel out the pattern received in the last transmission/reception and was effective in nearly eliminating the ground clutter and showing only those reflective objects that were moving. I asked how it worked, but they were techs and didn't know. I've always assumed there was some acoustic trick being used. Excellent video!

Enjoyed this very much. Keep old tech videos coming.

I had an old Sony Sobax nixie tube calculator about 30 yrs ago that had one of those in it, back then there was really no internet to speak of so took me a long time to research what the darned thing was!

Visuals like this are quite inspirational to this aspiring writer!

Fascinating! Thank you, xjet, for sharing this.

I'd love to see more of the old stuff. Thanks for this.

When describing the refresh function you said that it's a dynamic memory and that some of the early RAM chips were a bit like this... Well, they still work like this! The main memory in nearly all modern computers is DRAM based. Each memory cell must refreshed periodically. Also, the little ceramic chips that look like surface mount ICs (gull wing, flat pack) are probably transistor packs with two or three transistors per package. They may have had resistors in them to create RTL (resistor transistor logic) packages.

It's just crazy how that worked, so impressive.

delay lines is used in PAL TVs for storing a complete raster line full video details, for using it to merge chrome information with the next raster line. So chroma is half resolution, luminance is full resolution. But TVs delays are acoustic delays, so it works sound speed. The good side delay lines is, it is not just one device per bit, you can store kilobytes of information send to the delay line, and get it later in sequentian order, just to interpret and retransmit again, and if needed to change one bit or read one bit. So, it is really slow, you needs to wait for a whole memory cycle to just access the correct bit or bit stream

If you made a wire out of microscopic tin sheets embedded with fluorine atoms, it would actually become a superconductor (at up to 100 degrees celsius!). As such, the signal wouldn't get weaker as the length of the wire increased. I want to modify a DOS computer to have a whole bunch of these Delay-Line modules (in transparent casings); it would be awesome to actually watch it in action.

@zealotprime3796

7 жыл бұрын

It's an acoustic signal, not an electrical one, so the conductivity doesn't matter. Also, the wire isn't a complete loop, it has two ends, it's just coiled up to save space.

Excellent! I'm so glad i found this! very interesting as a 26 year old man who grew up with a cell phone in my pocket!

I love the hand drawn traces! I produced pcbs back in the day and while most of our designs were cad we had some artworks that were one of a kind, hand drawn with thruhole files created on a large granite digitizer floating table and (originally) transferred to tape reels ;-) We would need to color in the artwork from time to time and transfer them to new films whenever they got scratched from handling or whatnot. Good times!

Thanks - very interesting. Never seen delay line memory before. Heard about it from the old system progs in the early 80's when I was just starting my IT career.

Wow that was extremely interesting. I don't know about that things till now. Thanks for the show!

Fascinating. Not seen one of these before. I had an old ferrite core memory but it got borrowed and never returned.

Very interesting, didn't know how memory works. But still hard to comprehend how these little micro cards holds so much memory in such a small piece of plastic.

I am loving these videos! :) Thanks for making them.

More amazing than the technology itself is that people were more amazed and impressed with that ancient technology than people are today with 64000000000 bytes on a tf card. Technology seems to fail to impress the way it did back then.

@y2ksw1

7 жыл бұрын

Alasdair McC that's so true, but I'm one of these pioneers and I am really impressed every day. I believe we have different priorities. 😊

i thought this was a eevblog video at first

@RogerRHF

7 жыл бұрын

fox bodys and mustangs yeah theres alot of similarity

@DiegoCirilo

7 жыл бұрын

At first I thought they had lowered Dave's voice pitch, lol.

+1 internetz I'm not as "old" as you but even if i grew with the 3 1/4 diskettes i still like to see what was the world like before that. It's not much of a use but it can sometimes give you ideas that with nowadays' IC with millions of transistors all inside a black box you would have never thought of. Old PCBs and hardware in general are great learning tools. Being so big and with everything exposed to plain sight, you can poke around it with the oscilloscope and see how it actually works.

40 years ago The company I worked for had a metal punching machine with a Westinghouse (I think it was Westinghouse ) control that used the exact same delay line memory. By today's standards it was slow but it was "stone axe" reliable .

I support modern computers every day but I had no idea that's how memory started, thanks for showing us.

@sloperdude

10 жыл бұрын

I think Williams Tube memory started before delay line memory. There must have been an even older memory technology before that, when giants roamed the earth and massive vacuum tube computers were the high tech of the day.

This was great, I am entirely interested in seeing some more stuff like this

Assuming these floppy disks weigh about 30 g, that would make my actual 2 TB hard disk would be 468,750 tons heavy with that technology.

@UTUBESUCK666

8 жыл бұрын

Pierre Ripplinger lol

I love this, this sparks my intrest in learning more about this sort of stuff. Make more :)

Thanks for the historic information. I had a TRS80 radio shack computer, which used the 8" floppy disks. I think the trsdos operating system was on the disk and it was needed to start the computer. My disks were 64k memory.

That was so cool seeing the delay-line memory. I started back in the days of CPM with the single sided 5 1/4" Floppy's, that had `160 Mb (?). We would punch the case and flip them over getting another whole disk, They were something like $5.00 (US) a disk back then. We saved money where ever we could.

@LeRoy3rd

7 жыл бұрын

LicheLordofUndead Close, single sided 5.25" floppies were 360KB, by punching them you could make full use of them in a double headed drive (720KB). Definitely not MB though, there are 1024KB in a MB.

Awesome! Thank you for this, I grew up in the 80's mostly and read a lot about this kind of stuff in books but the oldest stuff I ever got to play with was a PDP 11

I just finished Cryptonomicon where an even more primitive RAM system is described in a pretty clever way, employing the same continuous feedback loop to keep data for as long as it is needed.

Never heard of this type of memory before, the only delay line I have came across are the ones that used a tank of mercury which I guess are even earlier than these, before that they used the Williams Kilburn tube were they wrote the storage to a crt and read it back via a metal grid over the display.

@0MoTheG

7 жыл бұрын

The most used delay lines were analog memories for the color lines in PAL and SECAM TV sets. That is what the M in SECAM means.

@Roxor128

7 жыл бұрын

+0MoTheG - Those used yet another delay-line technology. Basically a piece of glass (probably quartz) with a couple of transducers and cut to a shape that the path taken by a signal bouncing around from one to the other would take the 64 microseconds needed for a PAL or SECAM picture line.

What were these used for? I love looking at old tech. Something I saw at the space center (NASA) when I had an internship when I was much younger. Was and actual HDD with huge disk platters and had huge reader heads. It was the size of a V8 car engine! a When the platters started spinning it had a really cool winning noise and you could hear the reader heads accessing the data. It truly is amazing how far tech has come in such a short time!!

This is really advanced stuff. My first computer used mercury acoustic delay line memory. :)

Really fascinating Bruce. It's not so much the wonderment as to how far we have come from this old technology but more what kind of minds, and maybe pharmaceuticals might have even used recreationally, that dreamt up these devices especially calculating what strength of magnetic field was required to represent a 0 and how that was differentiated from a 1. Regards Nidge

I love these vids bruce, dont stop!

I just picked up a WWII era Aircraft radio receiver that was from the at the time state of the art ILS system that the Navy created to assist in aircraft landings on both land and carrier. This one was from a bomber as best as I can tell.

Props on the video! This was indeed very entertaining and quite interesting! I would very much like to see more of this stuff from you, sir! I'm 24 years old (as of yesterday) and am intrigued by things like this! I'd probably even go up a notch by attempting in getting one of these units to operate since electronic repair is my main hobby, to watch one of these memories operating would be cool!

For sure interested in some more old tech! Thanks

When I was at Caulfield Institute of Technology back in 1981, I volunteered to help clean CSIRAC, the 5th computer (maybe 6th, now that the UK declassified plans of Colossus). Anyway, CSIRAC had accoustic mercury-filled delay line memory, where an accoustic speaker generated waves in the mercury at one end of each tube and a speaker listened at the other end. Instead of torsion-waves like the video here, data was stored in a series of sound waves. See museumvictoria.com.au/learning-federation/csirac/hot-box---csirac-computer/

We also used these in old color tv's to slow the video signal so the audio could sync.

@Roxor128

7 жыл бұрын

Really? I knew about using 64 us delay lines in PAL and SECAM systems for the colour decoding, but I hadn't heard about it being used to synchronise the sound and the picture. PAL doesn't actually require a delay line to decode, but it does improve the picture quality. SECAM, on the other hand absolutely does require one.

@jirizlamal69

7 жыл бұрын

You have to slow the picture down as the speed of sound is much much slower! :-D))))))

@cyberyogicowindler2448

Жыл бұрын

@@Roxor128 PAL used some kind of piezo mechanism as acoustic delay line (later replaced with digital stuff) to compare 2 scanlines for colour difference.

@Roxor128

Жыл бұрын

@@cyberyogicowindler2448 Which is the "high quality" version of the decoding. The simple "low quality" version just subtracts the colour signal from the incoming signal. The simple decoder produces alternating light and dark lines of the decoded colour, with the hope that they'll average out when viewed from far enough away, whereas the complex one actually gives flat areas of the decoded colour. I can say that because I've actually implemented it in software and seen the results for myself.

I worked in the computer room of a power station in the early 70's. The mainframe was an English Electric 2140. Once a week, I had to check the alignment of the heads on the big mag' drums (mass storage) and set up the sense amps on the core store. I also got to build some great test equipment and diagnose faults down to individual component level. Tech' of the era was very rapid - (spin-off from the space race stuff maybe ?) I think the programming was mainly assembler /octal...

Amazing. Tech today is so much more advanced, yet in it's function, simpler than this old wizardry.

Absolutely fascinating , more please and perhaps greater depth as you seem to be an excellent explainer

I worked "around" this equipment. Glass delay lines made by RCA in--I think--Camden, NJ were used in equipment until the early 1990's. The product displayed market data and the machine was invented by Jack Scantlin. Scantlin Electronics marketed it as QuoteVue and Dow Jones marketed it as DowVue. Scantlin Electronics became Quotron and Quotron continued to support it until after it was purchased by Citicorp and destroyed (a little editorializing there). The QuoteVue was not a computer but a display device with several data inputs. The technician who installed it would get a mimeographed sheet from which he would program the machine on site with slide switches on the circuit boards. The QuoteVue could support up to four users. There were (from my memory) only twenty lines of data total. Typically, the top three lines would be the Broad Tape (Dow Jones News so called because Dow Jones was on Broad Street), the next two lines would be the NY Ticker, then a line of American Ticker. These would be programmed (with slide switches so that they were visible to all four users. Then there would be four "quote lines" one above the other. Using the slide switches again the technician would program the display so that the broker could see only his own quote line. That meant each screen had three blank lines so he could not see what quotes the other brokers were looking at. The delay lines held a line of information of sixty characters, I think. The Broad Tape was only 50 characters across. With the slide switches the tickers and DJ news could be "programmed" to slide across the screen or precess (scroll). This was described as the first "hardware split screen" display. Separate feeds were displayed in their own locations. South of Chambers street the tickers were current loop. North they were Western Union TLA's. Dow Jones was delivered via a Coherent demod. When this was first offered in the 60's Dow Jones did not have a 300 baud feed so the input was 75 baud. Dow Jones multiplexed 17 different news feeds across its network (including the Catholic New Service). The equipment was effectively indestructible and maintenance free. My interaction with it was first to deal with line outages and then to see if there was space in the user office to replace it with later more modern equipment which was about three times the size.

@richardhaas39

4 жыл бұрын

Oh, so if there were 60 characters a line and 20 lines (there might have been less) that would be 1200 characters of memory. All uppercase, but the font included fractions 1/8 through 7/8. Typically there would be blank lines between the ticker and news feeds. If there was sufficient screen real estate available the characters in one or more lines could be made double height (not double width though).

Used these in military equipment in the late 60s and early 70s. Most used old TTL logic (-6v = Logic 1). We eventually got shiny new 7400 series logic.