I worked on a tube computer that had a specific use, to encrypt and decrypt teletype communications. It was designed in 1956 and was used until the 1990s the KW26.
@@ComputerHistoryArchivesProject Thanks for the update. I worked on the KW-26 and the KW-7, one year training and 3 years on duty, 1969 - 1973. It was a fantastic design and the code was unbreakable, an electronic one time tape. The only flaw was if the same card was restarted - then the code would repeat. That's why the card reader had a knife in it to cut the card in half as it was inserted. if you have the time you might find this interesting www.jproc.ca/crypto/kw26.html
Hi Dave, that's an excellent web page, thanks very much! That's some great tech I was not fully aware of. Another double espresso, and I will finish reading the whole page. ~ Victor, CHAP
The LGP-30 only had 113 tubes due to its use of silicon diodes to make up most of its logic- just over a thousand of them. It also made the computer exceptionally small for the amount of computing power it had.
I don't recall the specific tube types, but according to the excellent reference book Tube Lore by Ludwell Sibley, some previously developed tube types were modified and manufactured as special-purpose variant versions specifically for computer use. Apparently, some commoner or standard issue tubes were prone to develop a sort of sleeping sickness, a chemical poisoning of the cathode, when they idled for long periods of time in a computer without having to conduct any significant plate to cathode current.
Sleeping sickness is usually associated with cold cathode tubes like nixies, and dekatrons. Vacuum tubes can have their cathodes damaged by being in cut-off (no current passing thru the anode-cathode path) too long. In that cases its called 'interface resistance'. Depending how the oxide emission layer is formed on the cathode, determines if the tube will be damaged. Tubes like the 5963 (special 12AU7) and 5964 (special 6J6) are an example of 'computer' tubes, immune to 'interface resistance'. However, that said, in later years of tube production, all vacuum tubes used the more robust method of forming the cathodes. Of course, by then, no one was still using vacuum tube computers. The Burroughs B205 came out ~1960. It still used tubes, but most of the competition were already shipping their first transistor based computers.
some of the sage components and a burroghs computer can be seen in the old tv show time tunnel and later scifi shows . a tutor of mine in college said that her first job as a young woman was to replace vaccuum tubes in a computer . when the computer ran, it dimmed the lights in philidipia .
Cool Vid. I did, however expect a voice with a British accent (even a generated one!) to use the term "valves" instead of "tubes". - And I'm an American! LOL.
The glow of the ANIAC at Naval Research Labs in the swampy Aberdeen Proving Grounds attracted moths that shorted the machine. The fix was to put screens on the windows.
On another machine, the Mark II computer, moths were also a problem. Google "Relay 70 Panel F" to see a famous computer history of something related to moths.
I didn't see the IBM 704 mentioned. I operated one beginning in October of 1964 at General Electric. The IBM 704 was the first mass produced computer with floating-point arithmetic. Hence some people now refer to it as the first super computer. Most of the time we would take card decks of scientific FORTRAN programs. Then read them and write to tape on an IBM 1401. The tape would then be taken over to the 704 and run creating an output tape which would then go back to the 1401 for printing. After about six months I moved from operating the 704 to programming mathematical and scientific applications in FORTRAN II on the 704. The IBM 704 was an excellent computer.
Hi artsietopology, Thanks for your note! The IBM 704 is a real classic. Good point. I think we have several other videos that mention the 704, and the 700 series family. Haven't met anyone who actually worked on one. That's quite an experience to have! Thanks again.
I don't know why they didn't use sub-miniature valves instead miniature valves as they're physically a LOT smaller and use less power (And lower operating voltages). If used in conjunction with PCBs it would resulted logic-modules that were smaller and lighter hence a smaller mainframe.
Hi Nicholas, that is a very good question. The smaller tubes took up less space, making the entire computer potentially smaller in size. Perhaps the sub-miniature tubes were most costly back in the early vacuum tube computer days. ~ something to look into. Thanks! ~ Victor, CHAP
@@ComputerHistoryArchivesProject "Perhaps the sub-miniature tubes were most costly back in the early vacuum tube computer days." Maybe, maybe not. I have some difficulty believing as sub-miniature valves were produced by the tens of millions in WWII for build radar proximity-fuses in AA-shells also during the late 40s, 50s and into the 60s sub-miniature valves were used extensively in avionics such radios, radars, missiles, navigation systems (Such as the AN/ARC-21 TACAN), etc. But still even with double-sided PCBs alone logic-units could've been shrunk. See this video on proximity fuses - ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-N0SgC78YFPc.html
Hi Johneygd, SAGE was an amazing computer system, networked defense computers of the mid 1950's. Here's a good video ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-06drBN8nlWg.html
Although it is called a “gun,” a light gun (like the Nintendo Duck Hunt game gun) doesn’t shoot anything; rather, it PICKS UP the initial flash of light as the CRT’s electron beam passes over the area of the screen where the gun is aimed. That time corresponds to what the display unit was “drawing” on the screen. Business and engineering systems often used the same technology in a small cylinder and called it a @light PEN.” I suppose the “gun” was meant to appeal to warriors.
@@allanrichardson1468 It propably seemed very fancy at the time but I find this SAGE "big boys missile shooter" a bit awkward and also, endlessly hilarious ':D
We should've stuck with vacuum tubes and made them smaller with advancements in carbon nanotube technology and molecular rigid bodies since electrons will start quantum tunneling their way in dielectrics in semiconductors. Mechanical computers, too. With such material engineering advancements, nanotube tech would've also yielded greater energy efficiency with nanotube vacuum flywheels that couldve been used to wind up the flywheel battery on our vacuum tube smartphone. Alas, maybe in the next life
Up until the late 1950s most car radios had vacuum tubes and a "vibrator" power supply -- I think the more expensive the car,the more tubes in the radio. Say,a Buick vs. a Chevrolet.
You mention the mercury delay memory and Williams tube memory, but you don't mention the much better magnetic core memory used on the MIT Whirlwind (and AN/FSQ-7). This was the biggest innovation of the Whirlwind and became the standard for 2nd Generation Computers in the 1960s. Still tiny by modern standards (typically 4K-16K tops).
Whirlwind replaced a Williams tube memory with a 32x32 array of ferrite cores in 1953. The innovation came from An Wang, based on an earlier idea by Frederick Viehe, I believe. www.computerhistory.org/storageengine/whirlwind-computer-debuts-core-memory/
Interestingly, they don't mention the series of Philbrick tube type op- amp modules labeled Gap / RR. It's my understanding those were used for some types of analog computing. I had about 50 of those some years ago, they came equipped with a pair of Telefunken 12ax7 tubes for each module! The other end of the module had standard octal plugs just like a 6L6 or other common output tube. Both the tubes and the Philbrick op amp modules sold pretty quickly on eBay.
good 'un, the computers covered here, are DIGITAL computers. That means, they run off a program which is a sequence of n-bit (digital On or Off) instructions. Analog computers operate in a completely different manner. I've programmed both ANALOG and DIGITAL computers. In an analog computer, you basically wire up a circuit, that often represents a differential equation, and all parts of that circuit run concurrently. Not sequentially, as step by step instruction on a digital computer. The output of an analog computer is often a graph drawn on a x-y plotter, that compares two voltages. One for the x-axis, and one for the y-axis. Can't use an analog computer to run a web browser. This is where language has a great failing ! The term "computer" has many meanings. The only reason it was used for both digital and analog computers, was that both computer types initially were used to solve math problems. A job originally done by a whole bunch of women, who's Job title was ... computer. See the movie 'Hidden Figures'. That's why, I KNOW that people who take things literally are idiots ! Language has a great failing at getting the meaning across accurately. that's why there are drawings, schematics, graphs, and other methods to get the message across.
The dekatron - a couple years older than I am and still working. Now that's amazing. btw, I used to check tubes in my dad's old repair shop. Took a couple minutes per tube flipping through the book for the right tube parameters and then dialing and calibrating. Might have to check six or or eight tubes before you found the defective one. How the heck did they check tubes on a machine with thousands of them? This might make an interesting video. I'd love to see something like that.
Hi d e austin and thanks for the comments. Makes me wonder too how they checked all the tubes so often. Notes on the early UNIVAC mentioned several interesting things. They tried to buy "computer grade" tubes, which had more reliability and consistently than general tubes for radio, etc. They must have had a daily procedure for checking tubes and having enough of the proper replacement types on hand. Sounds like a challenge.
Dekatrons (for the audience, in general) are cold cathode (no filament) counter tubes, and are both reliable and durable. The Harwell computer first used the British Ericsson GC10A, which glowed purple. I have a working one. When rebuilt (it was in storage for a long time) the GC10B was used, since a lot more of those were made, and they're still floating around, in decent quantity. Those glow orange. Orange is due to using neon gas. The purple gas for the GC10A, is not the same as the purple gas of the GC10D (later high speed), which also has a different pinout than the GC10A & GC10B. The GC10D has a very high dud rate, which I'm guessing is due to the gas possibly being hydrogen, which often bonds with metals, forming hydrides. You rarely find a dud GC10B.
It's pretty innovative to see how people made the most of the technology that they had available to them. Modern CPU's consist of trillions of transistors...... imagine what trillions of tubes would look like!
Hello, at 1.53 these are relays, in a calculator used to control 1 balance and a printer, there were a multitude of them, from GENERAL ELECTRIC brand, this calculator weighed about 80kg and contained 60grammes of 21 carat gold, it was the or contacts, I had noticed thanks to a flood at a supermarket customer where the butcher's laboratories were often in the basement, a blow of drying with hair dryer, replacement of the plate straighteners and the calculator started again, of there I said to myself that the contacts were made of noble metal better than copper. From then on I started to recover the calculators going to the destruction, because these old scales were replaced by a new generation, but how much gold was thrown, I dare not think about it. This scale had been patented in 1962 by Hobart, it was in 1975 that I joined it, there were quite a few left in France, the diagrams of the calculator alone measured 3 meters long and 70 cm wide, a "clamp "special to move away or approach the contacts (time delay). Excuse me, because online translation can be difficult to understand. Attached patent but the calculator is not there, because I think Hobart had been looking for a calculator already made in another company. patents.google.com/patent/US3045229A/en Bonne journée.
@@ComputerHistoryArchivesProject Hello Victor, I just realized that NCR is you too. So in the patent we can better see the Hobart printer and if we remove the iron-on label ejector (25) and all the rectangular bodywork, we end up with the good old NCR body in the flat position, it suffices to remove it. raise and we find the normal configuration with its numbers (20), and the well-known rounding where the buttons were replaced by Hobart by a block of "latch relay". the crank (also well known) being placed in the middle of the cover door located behind the ejector 25. For the calculator, it must have been used at one time at France Telecom for communication relays, I had seen the same ones about to be sent at the foot of a central site. Coming from the USA it must have been used in your home probably for the same use. I had kept everything of these machines from my youth but a stupid manager had tidied up without warning us and had all my documents thrown away, in France they are less sensitive to their industrial past and whoever is interested in it is laughed at, despite knowing a lot about the company. Hobart is Professional cooking, weighing closed in Europe (yet profitable in France) by the group that owns Hobart, then it was Kitchenaid in 1919 and sold in 1992 to Wirhpool by the Premarck group again but owned by ITW Illinois Tool Work. In 1975 the shareholders were private and only checked the good management, which is normal, the strategies nowadays are not always good, but it is so. Bonne continuation et Merci pour ces vidéos intéressantes.
How many BTUs were given off by each of the different types of tubes? These computers also had to have air handlers to supply cooling air. But was the air also filtered, cooled and reused?
Good question. I don't have a reference for BTU by tube, but am aware that early large vacuum tube computers were often cooled by "chillers", water-based machines with fans, etc. Tube based machines gave off enough heat to warm the room. Air conditioning came later. Even up to recently, some very large machines use water cooling, some use cryo technology. Hope that helps a bit. ~ CHAP
The heater current of a 6H6 twin diode is 0.3A at 6.3V, so you can calculate the BTUs from that and get a rough estimate of power required to run a vacuum tube computer.
@@ComputerHistoryArchivesProject Yes. Ranging from cathode ray tubes from the 1920s and 30s, wartime klystrons, acorn valves, thyratrons, electron multiplier valves (TSE4, EFP60), . In the 1970's whenever people were clearing out stocks of valves I added to my collection
Hi Mirek, that's an excellent question. Here is my take, and perhaps other people can give their opinions as well. Typically, the "texts" on computer history start with 1946 as the beginning era of First Generation computers, specifically, "electronic digital computers." Earlier machines, such as the Z3 by Konrad Zuse (1941), and the IBM ASCC (1944) were "electromechanical," not fully "electronic" and therefore not in the same class. Mechanical computers and relay computers that have no electronic components (i.e., no vacuum tubes or transistors) are arguably "pre-electronic." A distinction is made between "electrical, electro-mechanical, and electronic." Electronic machines control the movement of electrons in very specific ways. Electro-mechanical computers use electricity to power solenoids and relays, and such. -- {Maybe they should call the "pre-electronic computers" the "Zero Generation" computers} Hope this helps. Sorry for the long winded reply. ~ Mark (CHAP)
Hi Krish, I think if one tube goes bad, there will be problems in the computer. Some tubes go bad slowly, and had to be checked periodically. In those days, tube testing and tube replacement was a very big concern. Also, the best computer engineers would only buy tubes that were "computer rated" meaning they were designed to operate more efficiently for longer periods at specific voltages. In some cases, tubes were given initial burn in times to make sure they would last for the expected duration of the computer's need. Needless to say, the computers of the 1940's and 1950's drove the science to create better and more reliable tubes than the ones in radios. High quality "computer rated" tubes were in high demand, until the invention of transistors and integrated circuits, and eventually, microprocessor chips. Great question. Thanks! - Victor at CHAP
Apparently some computers designed some important logic in parallel, so that one vacuum tube heater burning out wouldn't necessarily stop the computer. Others designed it so the clock would stop if one of the tubes burned out, allowing online replacement and resumption of program execution.
The Swedish BESK computer, completed in 1953 and fast for its time, contained 2400 tubes. It ran in average for 5 minutes between HW failures, but this was improved in 1954. See en.wikipedia.org/wiki/BESK
in Denmark we had DASK, built in 1957 by the Danish Regnecentralen (state calculation department, today the department for digitalization). This is the only existing video of it in operation during the Danish parliament election in 1960 ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-z_8uV5Sawp4.html At this point it had 80,000 tubes and magnetic drum storage. And for those not so good at Danish, the guy talking about it in that video is the lead designer. He was mildly perturbed because the reporter calls it an electronic brain and then he explains how the machine works. It more or less did the same job as the American machines: statistics and calculations, and projectile calculations for the military. DASK remained in service until 1967 when it was scrapped and sent to the national Technical Museum where it remains in storage
That is a great historic video of the "DASK" machine! Thank you for pointing to it. At 80,000 tubes, it ranks at or near the top of vacuum tube computers(!) (RCA Bizmac and IBM SAGE come very close). I wonder if there is a full translation of that Danish film. It would be great to get ownership approval to re-host that film here with English subtitles. It was fun to watch, however, since they clearly show many parts of the machine. ~
Would perhaps be better titled "early digital tube computers", as it ignores the many analog machines created by, for example, GAP-R Philbrick - which continued on as a useful tool well into the solid state age (see - opamp). And of course, the inventor of the vacuum tube depends on which side of the pond you are on. On this side, it was Lee DeForest. In truth, like the light bulb, governments tried to grab the glory for first (see, flight for another example) regardless of who was really first, or if the inventions' time had simply come and the inventors independently came up with their invention around the same time (inside a news cycle latency). Often it was as much a matter of who was better at self-promotion than it was of inventing competence.
Doug, you make some good points. The "who was first?" question is often a very difficult one to pin down, and often depends on one's point of view. Good comments. ~ Victor, at CHAP
Hi MrCelroy, I found the original RCA video. I am uploading it to this link: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-vEPS1uYRCdE.html Thanks for asking! ~ Victor at CHAP