Minuteman D-17b: The Desktop Computer Was Born in an ICBM
Вставка
- Опубліковано 22 тра 2024
- In the early 1960s when computers were typically the size of a car, the USAF took on the seemingly impossible task of cramming one into a missile. That missile was the Minuteman was an entirely new concept: a small solid fuelled rocket able to independently steer itself to the target. The resulting D-17b computer was so small for the time, I argue it may have been the first desktop computer.
The debt imposed on us all by nuclear weapons will never be paid off in full. But Minuteman became Minotaur, Polaris gave us GPS and Titan became Gemini. The guidance computers for all of these systems went on to evolve into the modern PC.
3D animation by Artem Tatarchenko:
/ artem.iskustvo
Sources:
www.dropbox.com/scl/fi/axthi6...
Simulator here (warning, it's not very good):
github.com/lambdaBoost/d-17b_...
Playlist for all public domain footage used in this video:
• minuteman-sources
Summary of why I no longer use Unity. For some reason a few people took offence to that when I mentioned it previously. I had no idea people were 'territorial' about gaming engines:
• Unity's Plan Won't Wor...
00:00 Intro
01:35 Background
05:40 PART 1: The First ICBMs
09:49 The Need for a Guidance Computer
18:07 Launch Sequence
24:32 D-17b Physical Configuration
36:00 D-17s given to Universities
41:45 PART 2: The D-17b Architecture
58:07 Simulated Programming Examples
1:06:10 Conclusions
At 26:14 I should have said 'dodecagon' not 'dodecahedron'. My script even said the correct term, I just read it wrong! Thanks @__a_4444
I recommend Usagi Electric's video on the Bendix G15 drum memory for anyone wanting to know more about exactly how the magnetic disc storage worked. The drum memory he describes is almost identical, just in a different form factor. He is able to communicate it far better than me and answers questions a few viewers here have asked in the comments.
1:01:19 should be the Gemini computer, I have absolutely no idea why I said ‘Mercury’ (which didnt have a computer). Thanks @AMcAFaves
ua-cam.com/video/ijRV_7sr4_k/v-deo.html
Thanks for the awesome video btw.
Are you British Perun? Your presentation reminds me of that
So technically the missile knew where it was because it knew where it was, and by subtracting where it was from where it ought to have been it can derive a difference, or delta velocity…
That is actually 100% correct
I don't get why people found that thing confusing. Its pretty straightforward.
@@AHHHHHHHH21 because the person in the video speaks in a very contrived fashion and spends a full minute on a concept that can be explained in two sentences
@@cockatoofan yeah fair
@@AHHHHHHHH21 The average person can barely use a smart phone.
"Jesus, your computer is slow. What are your specs?"
"One 800 kilotonne thermonuclear warhead."
"What?"
Some kids have RGB in their computer cases, this one had alpha, beta, gamma, x-ray, UV... whole spectrum, in lethal doses up to 20 miles.
I wouldn't exactly call a Mach 23 computer, slow.
@@totojejedinecnynick great to play Fallout! graphics on 50000ºk colors, and 800 kilotonne bass...
and a speed of 40000km/h
"Your first desktop will blow you away". (I'll show myself out).
That jokes pretty rad!
These are going ballistic
@@iitzfizz ICBM what you did there.
Cringe....
Dude! You’re not getting a Dell!
Real talk, "can it doom," is actually one of the best definitions of a computer I've ever heard. It's simultaneously very easy to understand and very accurate
In the late 90s that was one of the requirements we had to keep a system.. DX processor and Doom.
The only issue with that definition is you have to give a big preamble about ‘how its only hypothetical as youd need to address a relatively huge amount of memory…..and figure out a way to hook up the IO….and it’d probably run at like 1 frame per hour’.
i feel like 'can it doom' has an entirely different meaning when the computer in question is hooked up to a multiple-megaton thermonuclear warhead. and the answer is 'yeah dude, it sure can!'
"can it doom," (☞ﾟヮﾟ)☞ made my day XD
everything, quake, doom, fallout... even do light scribe (if you find a bunch of ruskys dancing YMCA close to a wall)
apparently, IBM was the abbreviation for intercontinental ballistic missile, its all coming together
Also, International Business Machines, producing hollerith machines that gave us widespread punchcards.
(electromechanical computers are half a century older than I believed for a long time)
How did they know before WWII that they’d be making ICBM’s
It’s crazy how forward thinking Big Blue is
Also, coincidentally, IBM is HAL shifted by one character place. Spooky, woo!
Imagine there was an imcoming ICBM but inside the re-entry vehicle it was just a computer.
An Inter-Continental Business Machine, if you will
@@scunnerdarkly4929that is a magnificent observation! 🙌🏼
21:28 The reason why thrust termination ports work is due to the St. Roberts Burn Rate Law. It states that the regression rate, or how fast the propellant is burning back, is equal to a constant a times chamber pressure raised to the constant n power: rb = a*Pc^n. The a and n coefficients are based on the propellant mixture and are determined through strand burner analysis. When the thrust termination ports open, the chamber pressure drops rapidly, and thus the regression rate goes down. This is known as 'quenching'. The ports are angled forwards because the regression rate will not reach zero immediately, and will instead taper off as gas exits both the nozzle and the open ports. The angle of these ports results in zero net thrust. Some ICBMs actually angle the ports further forwards and use them to 'back off' the spent stage from the warheads and guidance bus. If you'd like to learn more there are a few good NASA papers online on solid motor quenching.
That’s really interesting, thanks.
I also recall that one of the other stalwarts of space comms (Scott Manley, I think…) showed how another ICBM used bypass channels to steer the missile, by porting some of the chamber pressure sideways?
So they gave a serial number to every resistor in a computer and tracked it all using paper.
A logistical nightmare is an understatement.
Maybe Alexander only the ok, but man, that film was great
No, it wasn't. It was made with the solely purpose of collecting likes and views. There's no honor or glory in it's intended purpose.
@@barmalinithe guy built a working D-17 sim. Give him some credit
@@Jujukungfu I once built a working Sincler ZX Spectrum on pure logic gates, so from me certainly a pat on the back
Spinny magnetic things were apparently a pretty established memory technology back then. A fella on Usagi Electric youtube channel is restoring a 1956 vacuum tube computer that uses a large magnetic drum at its core. The drum that came in his machine was shot but he managed to find another one in better condition in a museum machine and swap it.
by the sounds of it they are quite reliable as long as you dont run it too cold, would be even more reliable if any of the maintenance tools still existed to deal with stuff like bearings and blitzed time tracks like it could originally do
@@nickolaswilcox425yeah Usagi Electric is actually putting this machine project on hold due to cold weather out of concern he'll just trigger another drum crash if he fires it up when it's too cold
The desk computer shown reminded me of a model of the Centurion mini computer, also being worked on at Usagi Electric.
For anyone wondering, the machine in question is the Bendix G-15. Not sure if it was this channel but there's a video about the guidance computer in a fighter jet that shows an earlier model sporting a Bendix computer at its heart, which given the size of those things when used as mainframes is mind boggling.
I need to catch up on Usagi Electronics - I love that channel. The way in which magnetic memory started off as drum format and moved to disc is weirdly identical to how vinyl records went from cylinders to discs.
Yup, core memory, used ferromagnetic toruses to capture state. This is where we get the phrases "core dump" and "... Dumped core..." Today.
JWST primary mirrors are made from beryllium with a thin gold coating.
Very minimal expansion coefficient , so the mirror alignment won't need massive corrections
Playing Doom on a nuclear missle sounds a bit too close to actually summoning an army of demons into the world.
lol, I can't believe a right-winger made me laugh without having to fear our future. Thanks!
@@ryanreedgibson I can do both 😁
Guess you wont like the idea that they play doom on the large colider now days 🤣
A CNC controlled by a Minuteman guidance computer is the most awesome thing I've heard in a while!
Amazing video as always
I wonder if the end command was the traditional G00 or "Moscow"...
Beryllium is exceptionally stiff. It is terrible to work with, the dust in particular though. Great video, thank very much for the time these take.
Yup it is really difficult to machine it. It is usually beryllium with a few percent of oxide as it is excessively brittle as pure metal. It machines similar to cast iron making crumbly dust rather than chips. Most of the structural bits of the MIRV platform are actually made of aluminum hardened with a few percent of beryllium and scandium. Super lightweight but a bit stronger than most common grades of steel.
…and I bet no one told the workers to not breathe it in to keep it a secret (like the poor folks in enrichment facilities who were told not to keep too many buckets of liquid close together…. but not why…)
Beryllium sounds absolutely awful to machine. I'm guessing the dust would be toxic...and probably explosive too.
@@Alexander-the-okExtremely toxic. As in "you were around someone who machined some without protections, your house needs decontaminating, your car will be taken away by authorities and disposed on, the workshop will be closed down, and you may never work again due to lung damage" toxic.
Not a hypothetical
@mumblbeebee6546 oh we were warned it was beryllium or an alloy thereof, the prints just had cryptic part numbers that gave no indication of what they were for but the general shape pretty much told the tale for the sa parts. The mirrors and stuff for satellites and such were far more interesting though. Yup the stuff is truly awful from a machining standpoint, got to get the speeds and feeds just right, too fast and it cracks, too slow and it takes chunks out feed too light and it eats the edge right off the insert, too heavy and it breaks the insert the part or both. Combine the work hardening of titanium with the brittleness of cast iron and you got beryllium. 💀💀💀
The Hard Drive architecture isn't as confusing if you think about it as a solid state version of delay line memory.
This is an excellent comparison. The fast access loops were dynamically erased and re-written just like delay line memory.
Yes, thanks for clearing that up very solid state version of delay line memoryish of you my good sir. I know these words, individually...
@@mrsquidly6395 So before we had things like high speed magnetic storage medium for computers we needed some way for computers to retain a lot of bits of memory punch cards weren't fast enough and couldn't be written and rewritten easily. Although memory cells could be made of tubes they were very energy and space inefficient for this purpose.
So the idea of delay line memory was made. Commonly filled with mercury for its density it was a long tube with a transducer at both ends. Bits were fed in as an audio waveform at one end travels through the mercury and is read out the other end. If the bits aren't read out and fed back in to the start they are lost.
@@Jackpkmn I'm gonna have to search this up because I'm still not getting the full picture even with how well you've articulated the concept to me. I can't quite tell if I'm dumb or if you're on a level I've yet to even begin grasping.
I saw mercury and all of a sudden everytime I try to comprehend it I just keep visualising one of those old lightbulbs and It keeps throwing me off. A.D.D brains there telling me "ah, we know mercury and it's in a tube so lightbulb" I'm starting to lean towards the idea that I'm probably just dumb haha
@@Jackpkmn does the wave propagates through the mercury bounce back and forward storing the information in diminishing waves and as a result acts like a very short memory?
Hmm, also if not, i thought sound propagates faster in denser mediums?
What does the delay serve?
That magnet hard drive sounds more like a parallel-track magnetic tape loop made out of rigid bodies instead of bands of elastic or celluloid.
Wait. It's planar drum memory.
Yea, the one used for the D-17 is a drum memory. The models produced for less-extreme environments (including the Hound Dog missile) were more like a single platter hard disks. See Appendix A of _MEM-BRAIN FILE, Four versions for DCA_ for details.
Early computers like this are fascinating, stuff like the core rope used in the AGC, really amazing tech.
The core rope modules were wild. When they reprogrammed the AGC to use as a fly by wire computer, they had to race to write the code before the core rope manufacturing facility was closed forever.
As a very young child I got to see a core memory computer being used in production. I found it really neat that one could see where every physical bit was stored. (I've tried to determine what precisely machine it was but to no avail.)
I'm a computer nerd and have no idea what core rope or an AGC is. 😂 Thank you for what I'm sure is about to be a very happy Google/wiki rabbit hole.
@@orangejjay Have fun!
@@Alexander-the-okcode being rushed? As old as code! 😂
26:13
That's not a dodecahedron. It's a hollowed out dodecagon prism. A dodecahedron is a polyhedron with 12 faces. The prism has 14 flat faces, not counting the extra ones from hollowing it out.
This is so much not a dodecahedron.
‘Extruded dodecahedron’ is what i called it because there's no way i was going to try and say ‘dodecagonal prism’.
...on further reflection, I'm still wrong. It should be 'extruded dodecagon'. The one chance I got to say that and I still got it wrong!
I was looking for this comment. Shout out to Mr Steeves my grade 9 math teacher for very effectively teaching us shapes. I think I made an icosahedron for that class.
27:00 - Huh! The pcb’s make me think of isolinear chips on Star Trek TNG. Very cool that this was possible back then!
Actually saw one when I was studying in St. John Fisher College in upstate NY. The construction and support structure on the PCBs was impressive, support for every single component: transistors, capacitors, resistors... very meticulous engineering.
Soviet R7 used measurement and correction ground stations. But the first generation of R7 was more of a technology demonstrator or stop-gap solution than weapon system
Later rockets used vacuum tube computers on board, with miniaturized vacuum tubes which had a size like a small part of the pencil and a lifetime from several minutes to several hours.
I need to try and find some good source material about the soviet rockets.
I certainly am not surprised later models used miniature vacuum tubes - a lot of soviet hardware did from the mid 1960s onwards.
I really appreciate the shoutout to Hal Laning. Besides doing the real-time OS for the Apollo Guidance Computer, he created the first compiler. People often credit Grace Hopper for that, but what she called a compiler was more like what we could call a linker today.
She did develop FLOW-MATIC which led directly to COBOL so she can be credited/blamed with creating the first high-level language for business applications, usable by people without any training in mathematics. (Guess whose work led to FORTRAN though.)
@@alexhajnal107Plankalkül?
@@peterfireflylund I was referring to Laning whose work directly influenced the design of FORTRAN.
It's kind of annoying that Hopper is remembered in popular parlance for the compiler....and then everyone has to point out it's a semantic quirk and she didn't technically invent it. Because she did SO SO much more than just invent the linker. She was an excellent presenter/orator too....I imagine she'll appear in one of my videos in the future.
Giving Hal his due does not reduce Grace Hopper's legacy by any noticeable degree. She and Margaret Hamilton were and are (respectively) towering figures casting long shadows, ironically given both stood only a little over 5 feet.
I love this series on early computing history! The intersection with the more advanced weapons of war is fascinating (although very much expected); I may abhor war but its still a very interesting topic (probably because of them having to solve novel, and difficult problems), which further complements my interest in learning about early computers and the like.
I appreciate you way of relaying information in videos quite a bit too!
Studying CS, i do get a more general, and vague ish, information on early computing history, but seeing those actual examples makes it truly impressive what they had to do in those early days.
War is just too damn interesting isn’t it? It never fails to amaze me how much early computer scientists were able to achieve with so little.
@@Alexander-the-ok It's a shame that humanity seems unwilling to devote similar resources to productive endeavours.
@@alexhajnal107 War can be more productive
WW2 had a lot of technological advances
although this happens more for total wars and cold wars,the modern insurgent style wars or small scale state wars dont produce much new technology.
44:02 APL(A Programming Language) is an Array oriented programming language meaning that it's main datatype is a multidimwnsional array, it uses special charracters to represent certain functions it is a combinatorics base language. So the code is structered as a chain of Monadic(operation with only one operand) and Dyadic(Operation with multiple operands) functions as well as other operators.
Ps:
If anything here is wrong feel free to correct me
Amazing video, thanks!
Storage time: 10 years
Operating time: 15 minutes
Operating environment: 10G's
Congrats you've made one of the best UA-cam videos of the year
Finally a channel to merge my both loves in life
Computer Engineering and Aerospace Engineering
Great to know the D-17 got put to good use. The Minuteman boosters are also getting a new life as part of today's satellite launch vehicles!
You have no idea how much I enjoy this video. I have always been fascinated by early computers. Back when I was in school and briefly after, I only had time to read *a little bit* into these computers. Here, you've gone *all the way!!!*
Thanks. Tbh I'm the same - always been fascinated by early computing, but only recently had the time to study these systems in more depth.
I am looking forward to the episode about the Navy's fully-networked, transistorized refrigerator-sized missile fire control computers, put into operation in ~1964. It is one thing to equip a computer to work control surfaces and thrusters, another to have it swing a robot arm into position and launch missiles. Many ship captains got into hot water trying to keep the computers off their ships, but the need for fast response left no possibility of them staying in the loop. Where more capacity was needed, they added more of them, and all talked via radio link to their counterparts in nearby ships to coordinate their activity. So a missile battery on a destroyer could be used to protect a nearby aircraft carrier, without crew help.
I gather those machines had a role in Apollo as part of the worldwide communication and tracking system, but the details escaped me.
you and Artem Tatarchenko are a match made in heaven lmao. he is so talented and really adds a lot to your really well written and interesting essays :)
Agreed, he's amazing at what he does.
While the animation is quite good, a few details do distract me quite a bit as a big fan of the Minuteman system.
The silo layout, in a forest and surrounded by buildings definitely seems more like a Soviet style missile silo installation. The US launch facilities are very spartan and remote, just a fence and dirt and gravel patch around the silo door, usually in the open plains of the upper Midwest.
Also the silo door actuator is much more violent. While a lot of footage of launches comes from the Vandenberg silos, where they slide open smoothly on rails, those silos were designed for repeated use. The actual deployment silos have a much more simple system. Hydraulic rams push with great force the concrete door on smooth metal rails with enoug energy to make sure that even if the silo is covered by dirt due to a near by detonation that it can get out of the way. The only thing stopping it is how much friction it has once it comes off the rails (and in reality could easily breach the perimeter fence).
Also the tail end renders show two exhaust nozzles. The Minuteman first stage had four nozzles fed by a single solid rocket engine.
Anyway, nitpicking but small details noticed by a pedantic nerd.
IIRC beryllium is a very stiff metal for its weight. It's also used in the JWST mirror segments for the same reason. Probably also has a low coefficient of thermal expansion. Nice modern IMUs still output an integrated delta_v over their sampling interval as well.
Man your videos get better and better. Thanks so much for sharing!
Back in the mid 70s we had one of the University donation units in our EE department. One of the advanced students built a seven segment numitron display (incandescent filaments) and switch setup that could be used to enter short programs.
The S/N on each component was no joke. For their time the components were top notch. Our unit consisted of the processor half of the round split apart and mounted on a rack.
I was very impressed by the reverse voltage protection. There was a huge stud mounted rectifier across the power input ready to short out the supply if it was connected up backwards.
Sadly no exciting projects were ever built around it.
Thanks for sharing your experience- it’s so interesting to hear from people that actually used these, even if they were just largely used as ‘nerd toys’.
Funnily enough, I made a video about numitrons before my channel took off. I have 6 of them on my desk displaying my channel subscriber count.
Assuming decent gyros and accelerometers, you can significantly minimise compounding error by using low-pass filters at your ADC inputs. This can drastically reduce the sampling frequency required to maintain a reasonably accurate positional and velocity vector. If implemented correctly, there is no quantisation error other than the bit depth of the ADCs, and you don't lose the contribution of any high frequency components, those get averaged into the signal by the filter.
Another great video as always! Appreciate you putting so much effort and attention into your work, keep at it!
I really appreciate your calm and clear narration!
I commanded a Missile Combat Crew in the early 80s. My system was Minuteman III. One morning when I was sitting in the Squadron waiting to go on alert, I happened to stumble upon an old Dash 1 Tech Order for the original Minuteman I system and started to read. I was shocked to learn that the first Launch Control system had no computer at all. The entire thing was based on mechanical relays. The Enable commands were simply electrical continuity circuits. Interrupt the circuit and the sortie enabled. If power at the capsule dropped, all the sorties in the flight would enable. Daily SCN tests dropped connected sorties into Standby mode so the crew could verify communication. The whole design was surreal to me given all the nuclear surety that was deliberately built into the software that I used. I couldn't conceive of that kind of system without a computer. But I guess it makes sense given the level of technology in the early 60s.
The computer in my capsule was a Rolm 1666B, especially built for the mission. Not much by today's standard but it did the job.
Sorry if this is a stupid question, but what was it like doing your job? I can't imagine the feeling of working next to such awesome (in the old sense of the word) machines.
@@GulmoharBloom Missile crew duty was a 24 hour alert consisting of alot of tedium interspersed with intense activity. It was cold in the capsule so I generally wore a sweatshirt. (The first thing you did on alert was take off your uniform - one of the many, Many, MANY things the movie "Wargames" gets wrong.) Taking off the sweatshirt was a good indication of how busy I was. We had daily tasks - making sure the capsule was OK, verifying the targeting info in our sorties, calibrating the missiles every now and then. Mostly we watched. "Status monitoring" was the parlance. Security, faults, maintenance, and comm.
Did I think about the possibility of war? No, not really. PBS did a documentary in 1983 called "Nuclear Outpost" at my base. One of the reporters asked me if I had nightmares about nuclear war. I said "No, I have nightmares about Two Officer Policy violations" (Being alone in the capsule was a serious offense.) I still had those nightmares 10 or 20 years after I got off crew.
That said. Just before I got on Crew, "something happened" (we will leave it at that) and the readiness of the missile crews went up to the point that they inserted launch keys. (That's a long way from actually launching missiles btw but it is still what you call a significant event.) That's only happened a handful of times in history. The command post on the base had a direct line to all the crews in the field, and the first direction was "Do not call home. Keep the outside lines clear." Everything was restored to normal in a few minutes as the "something" was resolved. Since the launch switches are sealed to prevent tampering, a lot of people were immediately dispatched to the field to replace the seals. They said the crews on alert were just white and still shaking. So, yes. We knew it was real. But it was mostly remote from consciousness.
Would I have turned that key if so directed? Yes. No doubt in my mind.
@@carljacobs1260 It just boggles my mind to meet (?) someone who would be/was involved in nuclear warfare. I'm not sure if I could have turned the key if it came down to it, but then I won't ever be in a position to do so.
Thank you for responding, it's given me a new perspective to consider as I read more about nuclear war. Have a nice day!
When I first realized, that I needed a computer, it was a Commodore +4 on display in a highstreet bookshop window, in 1984. -Now, if you know about the Commodore line of computer models, you might be going "Oh, no poor Jim. Not the Plus/4!", know it then, that at that time I was thirteen years old, and anything like a great looking computer would have to get funding, and to get funding the project would have to get approved by my father.
I did not get a computer before three years later, as my father; a geologist believed in "the human faculty for calculation far exceed that of such a toy".
When my father saw what I could do with my first computer he was not impressed, but when I upgraded to an Commodore Amiga 500 and showed him what I could paint in DeLuxe Paint 3, he bloody well went and bought me a RAM expansion, that cost the same as the computer.
I have told him, time and time again: You should have funded me, in 1984, and he figured so himself.
The US Department of Defense must be the cool dad; "I hear that you lads would really like to get to grips with programming? Here's a pick from the bargain bin"
Any computer is leagues ahead of calculating with pencil and paper.
Thanks for posting about that ground breaking machine :3
For anyone interested in programming, the plus 4 was a very nice little computer. Good for coding.
Yeah, I saw a couple of of preeecessors, probably for the Titan II, of these being thrown out from the local University. They were supposed to go to hazardous waste disposal, as they used a cooling system using liquid mercury. Really sturdy devices with inch-thick aluminum cases. You couldn’t see much inside the case as it was filled with plastic foam. Wish I had kept them.
I'm still kicking myself for not rescuing the DEC PDP that I saw in the trash outside a medical office in the early naughties.
@@alexhajnal107 I had the opposite experience-- I bid on two PDP-11/44 systems, won them, then found out they were just the cases, someone had pulled out all the cards.
Really liking the videos, you keep finding interesting subjects to cover.
Thank you for another awesome instalment to your growing long-form content library. I love the low level tech on here- please continue with the magic jazz :)
Alexander, you consistently impress me with your focus, dedication and contextual research - and yet you are brilliant at presenting, storytelling and not to forget, visual humour where possible. I wish I’d had professors and lecturers like you! Thank you for sharing your work with us!
Classic music and Minuteman missiles truly a blend for the end times.
Interesting video! Sat through the entire thing :) Amazing work!
Awesome videos bro, well done 👍
Alexander, this is a very interesting video. I used to work with a very early DCS, it was designed and built by the company that used it in its pharmaceutical manufacturing business as process control and batch reporting. I only had a bit of time to learn about it before we replaced it with a modern standard (S7-4xx).
I specifically recall an analog input board with hard-codeable limits (DIP switches and jumpers, 8-bit, low and high warning and alarms, it triggered an interrupt on the central and fed it to a card managing all the interrupts and their privileges.
It was a headache to implement the interrupts as the only documentation we had was a copied copy of someone’s handwritten notes that had been used during qualification and then copied again and put in a binder.
It worked for decades but there was a fatal flaw in the design and it hadn’t been picked up. It was quite the hassle to get the HAZOP reviewed by the chemist who just muttered „boy, we were very lucky“ and then update the documentation to reflect the corrections we had to make.
Fun.
How far is your SDS and the SMDS for this project? 😏
Keep it coming, love it!
The computer was under the platform that carried the RVs. It also had either an electrostatically levitated sphere gyroscope or a helium neon 6 axis ring laser gyro in the dead center. The main middle battery was a 28V NiCd battery but the ones in the actual nukes were magnesium based thermal batteries. What they got in the more modern ones is a lot smaller and more accurate.
I really want to do a segment in a future video on ring laser gyros.
The D-17B used discrete DRL (Diode Resistor Logic) circuitry with DTL (Diode Transistor Logic) where needed, but the MMII used the D-37C which was one of the first guidance computers to use SSI integrated circuits (The other being the Apollo Guidance Computer).
Can’t wait for that Polaris vid!
Excellent - was thrilled to see this out ❤❤❤
Anither excellent video, thank you for putting it together.
this is _such_ a good channel, good lord.
Beryllium is super lightweight, super strong, and has a very low thermal expansion coefficient, so parts made from beryllium alloys can be made with tighter tolerances than aluminum or titanium. There's also no risk of sparks or static buildup, which can reduce compounding errors in inertial guidance systems. But... beryllium poisoning is absolutely horrible, it makes asbestos seem like a mild nuisance in comparison
I love these videos you do :D keep it up!
very nice work!
Just like the real minuteman(men with muskets), it was never intended to be used, but only at the direst of times. The minuteman have made sure, we dont have to use them. That is a victory by itself.
21:30 That is trivial. The explosien caused a massive leak that bypassed the thrust nozzles completely. Thereby reducing the thrust not to nul but to a fraction. That influence sufficed to control the terminal velocity.
That’s what I initially thought but all the sources on this state it wasnt a thrust bypass but a near instantaneous termination as the reaction abruptly stopped.
A few other viewers have clarified its because the rapid drop in pressure blows off the top layer of fuel and stops the reaction.
That intro 👌
Goes hard
The Soviets used a ground-based guidance system in the R-7, the basis of which was copied from the Germans at the end of World War II. The R-7 ICBM was tested in the Northern Urals in 1959 and there were some casualties there during the testing phase, including the members of the Dyatlov group who were involved in the testing and evaluation process. They conducted these tests with active warheads set to a lower yield of 1.6 megatons (standard RDS-37 warhead). After completing the series of tests, the R-7 was officially handed over to the Soviet missile forces in 1960.
For those interested in more details about the Dyatlov Pass Incident, I recommend watching my videos about the entire case and its complete solution.
21:15 you say you don't understand how charges would blow out the reaction. I'm not going to explain how it worked in detail but one of the first commercial fire extinguishers was just a baseball sized glass container filled with gasoline. When you threw it down on the fire it would break and ignite, consuming all the oxygen and choking the fire out (hopefully or you'd have a slightly larger fire to deal with) I think the people in charge of the thrust termination design took inspiration from that lol
Great video!
Really amazing work. The entire video was interesting & enlightening. And the Godot work was the icing on the cake. Thank you!
The game Juno: New Origins would be a perfect visual aid for this channel
Fine video. Subscribed.
This is the most interesting video on youtube!
Afaik, modern computers actually just use NOR gates, and build everything else from there, since having a single kind of gate (that is functionally complete, like NOR) makes the silicon design significantly simpler
No way! They actually just use universal gates throughout!? I can’t believe I didn’t know that.
My understanding is that the abstraction is typically slightly broader with the _standard cells_ being not just NOR gates but also the other logical operators as well as flip-flops, 2-bit adders, etc. These are pre-optimized at the lowest level and then combined as you described. Same basic idea though. For mask-programmed devices though having a sea of NORs or NANDs could make sense (actual devices that I'm aware of use a sea of discrete N- and P-FETs). [Edited for clarity]
A lot of early computers were comprised entirely of NOR gates, notably the Apollo Guidance Computer which used 3-input NOR gate modules.
@@Alexander-the-ok You can make every other gate out of NAND or NOR gates, but on modern CMOS processes, the NAND gate ends up using less area, so it's the preferred gate. That being said, the other commenter is right, that ASICs are made up of "standard cells", which implement NANDs, NORs, XORs, multiplexers, etc., that the synthesis / place & route tools use. A noteworthy one is the AOI cell, which stands for And-Or-Invert. It turns out that a lot of boolean logic can be simplified to ANDing together groups of values, ORing those together, and then inverting the result. They're also particularly space-efficient on CMOS processes. Wild stuff.
@@MyloWhylo re: AOI cells Something similar is also made as a standalone IC (SOT26, SOT363, DFN1010-6). The 74LVC1G58 implements *Y = ~( (~A AND ~C) OR (B AND C) )* while the 74LVC1G98 implements *Y = ~( (A AND C) OR (B AND ~C) )* From that one can implement a 2-input NAND, a 2-input XOR, 2-input OR, etc. (for the '58), and a 1-of-2 data selector with inverted output (multiplexer), 2-input NOR, 2-input NAND, etc. (for the '98). Handy little chips to keep on hand.
Ur channel is so cool
Brilliant.
jeezo, UA-cam is hammering this with unskippable pairs of 20-second adverts every few minutes. I'm using a VPN so they're unintelligible ads for (I think) Danish supermarkets but I keep needing to mute
Sorry, youtube went crazy with automatic ad placement a couple of months ago. I've just deleted a bunch of them.
The form is not a dodecahedron that is a solid with 12 faces not sides, each face being a pentagon. The form of the D-17b is a hollow dodecagon base prism.
9:53 Thank you for this.
You say there is one in the Computer History Museum? A “nightmarishly difficult restoration project”? I have a funny feeling that it sounds like an open challenge for certain electric bunny themed channel or maybe a space hardware obsessed Marc’s channel in the future.
Great to know that there are still some extremely obscure, but historically significant computer artefacts left to dig into over the following years.
Cheers!
Chopin playing in the background just broke me :)
Drum computers were quite popular in the day. There are lots of examples of later systems.
Just for fun, I wanted to know how fast Doom would actually run on a D-17B as the CPU. There's no way it could store Doom on it's HDD. It probably can't drive a raster display either, so let's assume some IO upgrades to handle those parts. So here's my SWAG:
According to a post on the Doomworld forums, a user estimated that Doom needs 32 instructions per pixel, so we'll use that. He mentions that the 32 instructions are on a 486, so it's likely not the simple instructions the D-17B has, but I'm not going to dive into the Doom code to find out exactly what was used. Wikipedia lists the timing for the add, multiply and load instructions, so I'll arbitrarily use those. I picked an arbitrary mix of 10 adds, 12 multiplies and 10 loads. The wikipedia page says we can do two multiplies as the same time, so we'll do the timings that way. All told, we end up with a single pixel taking 4.8 milliseconds..
Assuming my guess about 4.8 milliseconds is right (which is probably wrong), we get the following:
Doom runs at 320x200, so rendering a single frame takes 310 seconds or just over 5 minutes. Doom runs as 35 fps (ideally) so 1 second of Doom takes about 3 hours to render..
The fastest listed speedrun of Doom is 4 minutes and 39 seconds. That speedrun would take 35 days on a D-17B.
Love it thanks for this!
There is one thing missing, and its the main reason i never coded the timings into my sim: read write to the disc.
Much consideration is given to this in the original programming guide as reading from a ‘sub optimal’ sector would need us to wait up to 10ms just for the drive to rotate to that sector.
I have no idea how much more time we’d need to add in for read/write. But every time a variable is read in incurs a minimum of about 80 microseconds.
The use of beryllium is because it's the lightest metal. It has excellent thermal and mechanical properties aside from it being fairly weak, it compensated that with being bendy. It is also used for solar panel arms on current spacecraft among (many) other applications in the space industry.
I have no knowledge of the d-17 or the minute man missile. Thank you for the data.
As for the R-7 rocket, i had the privilege of visiting a military training center for the r-7 on 1992. I and my group were given a tour but we did not recieve specific details.
The guidence control logic are based o. vacuum tubes. I got a good look at an example at point blank range. I do not know if it was digital or analog. My guess is analog.
The onboard tracking logic fills a large portion of an approximately 4 foot deep cabinet at the very top of the roughly 6 foot diameter main stage. There were 3 separate gyro instruments per coordinate but dont know what coordinate system was used.
I know that at least some of the trajectory correction data and decision making comes from the ground. I do not know how much, but some.
Learning to fly an r-7 was very prestigious, or so im to understand.
Unfortunately, thats all i know and I'd recommend taking even this "knowledge" with a grain of salt.
I think i might have pictures burried somewhere deep.
The fact that the missile could compute its location in 3d space on a globe with just gyros.. I mean it sounds right algebraically, but how tf did they manage to do it so accurately while strapped to a fucking rocket? All that rattling around..
We take GPS for granted now and I can see why the military was so interested in restricting civilian use of it in the early days. No one, short of a superpower, was going to be able to machine a guidance system so precisely. Without that hurdle though, it’s probably magnitude easier to make a functional icbm
This video has made me wanting a D17 replica!, (probably an arduino is overkill but it would be way cool having a 3d printed chasis, and wired ports to do electrically equivalent terminals), just to show off driving a 3d printer with an ICBM.
BLIMEY,! gonna have to rewatch, this one more than once, I reckon😅👍
really interesting stuff
If not mistaken, the first computer on russian lauch rockets was the s-530 on the 4th and last N1. As far as ICBM, someone knows?
Before that it was a program cam
Thanks!
You’re very welcome.
Beryllium has a couple of properties that could have made it a good material for the gyroscopes. High stiffness, high tensile strength, thermal stability, dimensional stability over a wide temperature range, diamagneticity, low density,...
Dow Chemical developed a process control system in house called the Mod5. I don't know if it used a D17, but I do remember the processors were repurchased missile controllers. They were programmed with a Dow specific version of Fortran jokingly called Dowtran.
Many early computers has magnetic drum (with multiple heads) (before magnetic core memory taken over ) as kind of cache, but making it into disc is truly another 3 levels at least not regarding environment wich it worked
5:22 The historic sequence of an operator at a console, I presume a first response operator simulating a launch initiation, is remarkably similar to the opening sequence of "Wargames". This is not surprising, as that was set in a first response bunker. But usually Hollywood gets a lot very wrong, in that case the actions shown in the historic video are almost point for point the same as the movie.
51:06 Actually, if we're allowed to clobber those two unused middle bits then split add/subtract is particularly simple:
split add: accumulator = (accumulator & ~0x1800) + (operand & ~0x1800)
split subtract: accumulator = (accumulator | 0x1800) - (operand & ~0x1800)
This is assuming 2's complement representation is used... I'm really not sure why you went with sign-magnitude considering 2's complement is the native format of both the D-17b and modern computers (for signed integer/fixed-point arithmetic), using sign-magnitude seems like it would just make arithmetic way more complicated to implement than it needs to be.
When you pointed at the need for high precision of guidance and possibility of corrections (22:07), you mentioned missing St. Petersburg and hitting Helsinki. Well, at the time the minuteman were developed and were operational it was Leningrad they were aiming at, since the name St. Petersburg was reissued only after the collapse of Soviet Union
I don’t think this correction is a must, but I figured it would be something to note
Great point. I never thought of that.
Thus Lt. Chekov in Star Trek had scotch "invented by a little old lady in Leningrad." How were the Star Trek writers to guess it would change back to Petrograd by then?
Amazing content, so many engineering channel videos just sound like Wikipedia article reads, but you go far more in depth and hold my attention the entire time!
17:30 - Only the Atlas D used this particular type of guidance system, the Atlas E and Atlas F used a fully internal inertial guidance system just like the Polaris, Titan II and Minuteman (Although its' guidance and control circuitry was all analogue).
OMG, why are you making me look up so much stuff in the middle of the night? People like you are the reason why my wife asks me why I'm still awake at night. 🤣🤣
I just listened to 4 different versions of the missile guidance copy pasta. Thanks
Those analog volt outputs may be usefull for analyzing the results with an oscilloscope? Seems to me quite handy to make a graph of a calculated trajectory?
(Or a simpel plotter to make a graph of it.)
Excellent video!
When you said Mercury Guidance Computer, I think you meant Gemini Guidance Computer, as it was the Gemini that was the first manned spacecraft to have a computer, not Mercury.
Did I say Mercury!!? Why did I say that?
That’ll be a pinned correction!
47:29 Ehh, a dedicated accumulator register is afaik really only a feature of ancient CPUs and their direct descendents (e.g. x86). More modern CPU architectures pretty much always just have a bunch of general-purpose registers that can be freely used as source/destination register operands for arithmetic instructions. The only recent architecture I'm familiar with that uses separate registers for various roles is a DSP, but it has not one but 16 accumulator registers.
I dont know if you take suggestions (or rather, requests), but I was looking into the A-5 Vigilante and found that it was one of the first (if not the first) bombers to utilize a digital computer. Apparently it was called the "Versatile Digital Analyzer" or VERDAN for short. If thats a bit too esoteric, the F-102 and F-106 had a flight control system that was incredibly advanced for the time called SAGE that worked with GCI. With SAGE the aircraft could practically fly and intercept a target from the ground, then fly back and land with minimal input from the pilot.
The VERDAN was a direct predecessor to the D-17! From what I understand, the D-17 was basically an up-specced and miniaturized version.
As for sage, oh boy, now that is a complex topic! I'll add it to the 'potential future videos' list.
23:01 I have a question. What does "maintaining separation" mean? From what I get, if the third stage is separated, the radar signature is decreased as well. However, "maintaining separation" would imply keeping the third stage? And I also don't get the retro thrusters fire at random times. Isn't the whole point of the guidance system to precisely fire on time different actuators? Or do the thrusters separate the third stage engines?
Sorry for all the confusion. I really like your videos, but sometimes the reach over my knowledge. Nevertheless I really like them, especially when I understand them. If you (or someone) could explain that to me, I would be very thankfull. Anyways, cheers!
P.S. I don't mind your accent and your dropped Ts. They give a nice touch
Correct, separating the 3rd stage from the payload makes the latter less visible to countermeasures. Firing the retro thrusters (on the disconnected booster) at random times may be to confuse tracking and/or make it difficult to guess the payload's position relative to the booster (assuming the it isn't visible on radar); this is just conjecture on my part though. Edit: Using a random interval would prevent this detail (booster/payload separation distance) from being ascertainable via espionage.
Damn! you have just added yet another item to my eBay wishlist...
the idea of using a multiple head HDD is pure genius, different sectors, useful for cache, RAM, etc at a given speed. but the lack of trust in transistors is disappointing.
To be fair, it wasn't so much a lack of trust - it was backed by statistical research. The early Germanium transistors failed frequently due to corrosion.