Digital Fly-By-Wire: The Apollo Guidance Computer's final gift to the world 

Some corrections: Around 56:30 I briefly mention the 737 max incidents. What I didn’t make clear is these occurred because the aircraft DOESN’T have fly-by-wire. If they had been equipped with it there would be no need for MCAS in the first place (thanks @tonymcflattie2450) Thanks to @Ticklestein for pointing out that I mistakenly state at 21:35 that the photo of Margaret Hamilton includes the '36Kb of Apollo Source Code'. That photo is actually a collection of various revisions of the code they had around the office. The source code fits into a single (very large) binder and a copy recently sold at auction. We will be revisiting this point in a future video. Supercritical airfoils aren't for supersonic aircraft: they are designed for flow in the transonic region (ie large airliners). Thanks @LegateMalpais

"Then crammed it into an obsolete fighter jet..." *F8 Crusader* "I took that personally."

Virtual machines in the 1960's, programmers back then were of a different kind.

Westminster isn't actually machine readable, it was only inspired by the machine-readable numbers on checks, a MICR font called E-13B which only has numbers and banking symbols, no letters at all. It's kind of wild that anyone thought to make fake-machine-readable alphabet characters at the time, but sci-fi authors etc went wild for it!

The military is sooooo about fault tolerance. I remember Air Force guys telling me "You need to be able to launch a missile off the left wing even if the right wing is blown off and your plane's on fire." The 1553 spec is very clear that any device on the bus needs to be able to handle every kind of failure including bad data coming from every other device. I've read about all these planes and seen most of them because I used to live next to the USAF museum at Wright-Patterson AFB. I would always go to the museum annex which had disassembled SR-71 engines and research planes hanging from the ceiling. That's also Hangar 18 ... so, there's a YF-16 in there, but they also have some plush aliens in the old Norden bombsight vault because they have a sense of humor about it. I've never seen many details about how we actually got from the fly-by-wire F-8 to the F-16. That is, how we got from using an Apollo computer to creating the MIL-STD-1553 bus in 1973, which was put on the F-16 and then on the AH-64 in 1975. The AH-64 actually made good use of 1553. It had a really cool helmet-based targeting system for the gunner. The helmet also weighed 4 lbs. Today everything still basically uses 1553. Even the JWST uses 1553. My first software job was an internship at a military avionics contractor in 2007. Basic 1553 is a 1 mbit bus. That's pretty fast compared to MIDI which is a 31.25 kbit bus! But the F-35 needs a lot more bandwidth to send video to the fancy helmet that lets you see through the floor of the plane. The F-35 helmet actually weighs 5 lbs - and a lot of that's carbon fiber. It's ridiculous. I think NASA first adapted 1553 to optical fibre with MIL-STD-1773 in the early '90s. But it wasn't for speed reasons. It was optical because that keeps you from having interference problems. So we had to adapt 1553 to an actually fast bus for the F-35, and that eventually became FC-AE-1553 == 1 gbit 1553 over fibre. I wrote several avionics databus test suites for a cart that plugs into the F-35. And if I had a question about any specific details, I literally needed to look in manuals from 1973-1978. You know what's a pretty tricky job for an intern? Dealing with 8b10b encoded data that's too fast for regular software, so you need to write code for the PowerPCs embedded in the FPGAs on the analyzer, and you need to manually convert from big-endian to little-endian and between 8-bit, 16-bit, 32-bit, and 64-bit word sizes because you're on an Intel machine with a 64-bit PCI-X card but the Xilinx chip has PowerPCs (big-endian) on it, but it also has an Intel ethernet controller going back the other way. Plus the packets/frames of everything is scrambled. If you're looking at an ethernet frame that came across the 64-bit bus you'll see a few bytes of the payload before you even see the header, because you're flipping the endian-ness while also rearranging the words in terms of least-significant vs. most-significant word being on different sides. We used those analyzers for a lot of things. I remember my boss demo-ing 8 gbit communications over a single wire while the wire was also powering a 100W light bulb. Because NASA wanted fewer wires on something. Sorta like POE but it was Power Over Fibre Channel On Copper. 🤣 It seems like everything uses 1553, but a lot of civilian stuff is ARINC 429. Anyway this is actually a nice overview of the evolution of embedded systems - which is funny since it never mentions the term. Today every car has computers that can reboot a few hundred times per second. And that's all goes back to the Apollo computer which managed to keep Buzz Aldrin from killing everybody when he overloaded the computer by not turning off the CSM rendezvous radar when he turned on the landing radar. You're more likely to have a hydraulic failure than a fly-by-wire communications failure. I remember hearing about a plane (I think it was a 767 evacuating people from Baghdad?) losing hydraulics because it was hit with a shoulder-fired rocket. I think those guys had enough adrenaline to move the control surfaces a little, but they mostly managed to land the plane by modulating the throttles.

A detail: The reason the Sidewinder missile (excluding X model) has those gyro fins is because the way the guidance worked, it required there to be no roll at all to work. It's not really for compensating for adverse effects of controls in of itself. After all, other contemporary missiles didn't have those because their guidance was different and could handle (or even require) roll.

Don't use a Raspberry Pi as a flight computer, use three!

I have to say as a retired engineer this is the most fascinating and informative youtube video I have ever watched. A huge thanks for making this.

Great video! I worked on the follow-on F-16 production system as a junior software engineer late 70's to early 80's at Lear Siegler Astronics. I used to estimate and time subroutine performance. I helped a little with the voter routines. I also burned our program versions onto very expensive nuclear blast resistant ROM chips. Best engineering team I ever worked with. Our system was a big success. Your kind comments about the significance of digital fly-by-wire made me feel good about my years on the project. Thanks for your efforts!

I can't wait to see where you're going next with this channel. These lengthily researched perspectives from an engineer who has so obviously been steeped in the culture of rigorous analysis for so long, but lost none of their excitement for the variety of amazing technological accomplishments of human beings over the past lifetime or so, make for some of the most compelling viewing on YT. Don't stop! 😊

Be careful mixing up "analog electronic" with "analog mechanical". The early F-16 for instance had a fly by wire system, but it wasn't digital (ones and zeroes), instead it was analog electronics (continually varying electrical signals to electronic components). Later versions replaced the analog electronic flight control computer with a digital one because it was more compact, lighter and more reliable.

45:28 its no digression it fits right in with the meaning of analog digital interfaces, love it

Fact: NASA Invented the non-stick Velcro frying pan.

I used to fix the analogue F-111 navigation computer in the Australian air force. Half the work was servicing miniature gearboxes and was really interesting work.

the rope memory of the AGC is really such an archaic creation. You program your code, debug it, test it, then hand it off to another guy who converts it into a schematic matrix of iron cores who then hands that off to an army of old women who weave the physical code out of cores and wire

The shot starting at 35:29 is amazing! To me it's the first time that a full-CGI shot feels real, and passes the uncanny valley. The other day I saw a video of a Norwegian church rendered in Unreal 5 where they boasted about the realism of the scene but it just felt too perfect; for some reason this shot hits the spot and looks "real"

The book “The Apollo Guidance Computer, Architecture and Operation” is truly an excellent read. I was able to build most of an emulator with the info in it, it’s quite thorough.

Most supersonic planes are ugly?! F-104? SR-71? F-14? I think the F-22 is up near the top as well, but many seem to disagree.

The L-1011-500 was the FIRST wide-body commercial aircraft to have a digital autopilot (Collins FCS-240) instead of an analog system. It was produced in 1978 and was the first to use a supercritical wing and flew at M.90 but usually flew m.84 for lower fuel burn

I'm looking forward to more in-depth essays like this. This is a story of how rigorous engineering practices led to a quiet, but unqualified success that shaped our era. Engineering stories tend to follow a pattern where we often only learn the details of how systems work after they fail catastrophically, while engineering success stories are often sparse in technical detail. This video fills that gap well.

Definitely your best video so far! Man, this channel is awesome. Would love to see more aviation related content from you

wow that gyroscopic stabilization system is indeed really elegant ! Also how cool is it to own a sidewinder fin ???

I have to say that I literally stumbled across your channel. I appreciate your writing, opinions with fact based details in an organized manner. Your channel seems more baked in than I would have thought. Definitely shared. Good job and Thank You

As I am a college-graduated software engineer and longtime aviation enthusiast, I wrote several Quora entries on the non-FBW digital foul-up in the 737-Max MCAS system. I did not know, however, the story of AGC adaptation onto the F8 airframe, though I felt some nerdy embarrassment to consider the author's comment late in the video anticipating that he had lost everyone already. I was still there following along just fine. It was like sitting alone in a movie theater. Did I miss a social cue? Had to answer quick. No, I was only one who was in on the lingo. Ah, well then. Thanks for informative and relatively thorough coverage up to and including mentions of the F-16 and Space Shuttle systems.

What an epic production! A rare gem in the steaming pile that is normal RU-vid content. Chalk up a new subscriber ❤

these videos are so good. love your channel. your semi-informal presentation style works perfectly, and it’s obvious how interested you are in this; your enthusiasm about the topic is in itself very engaging. keep it up, and i really hope more people find your channel.

This is the first time I have come across your channel, and it is the most fun I have had watching a RU-vid video, EPIC stuff !

How the hell do you produce something great like this? It's like an hour long very high quality documentary.

What a fantastic piece of historical research. I was a teenager during the Apollo landings and I'm an aviation enthusiast, I knew the AGS computer was an impressive machine; but I had never heard about all this before. Wonderful and thank you.

15:30 There are direct coils to fire RCS jets, and the SPS engine. This is what the "RHC Direct", "Direct SPS", and "Direct Ullage" switches do. But they are separate redundant contacts.

You deserve like 20x the subscriber count. Love your vids.Keep it up please.

I came here knowing very little about flight controls but your ability to explain in simple terms made this really enjoyable. Thank you for sharing!

Just for the record, and not to be that guy, but the choice of TLI as an example of the AGC handling guidance on the Apollo missions is technically one of the few times the AGC was absolutely *not* providing guidance. The Saturn V guidance was handled by the IBM-designed LVDC (Launch Vehicle Digital Computer) which was located in the Instrument Unit in the 3rd stage. The LVDC handled all primary guidance for the rocket, including the TLI burn which was executed by the S-IVB stage. The AGC had monitoring capability using its own gyros and IMUs during launch and TLI, but was not actively calculating control signals nor was it in the loop under normal circumstances. There was a basic capability of the AGC to takeover control of the LV in an emergency, but the guidance provided by the AGC was far inferior to that of the LVDC, consisting of basic polynomial steering calculations during the S-IC first stage burn, and only attitude hold capability for the S-II and S-IVB stages. Incidentally, there was also a capability to interpret pilot commands via the RHC (essentially a sidestick) as well, and in the event of an LVDC failure during TLI, this would have been the primary source of backup steering input, as the AGC was not capable of generating detailed steering information for TLI, only monitoring of dV and orbital parameters. For me, the most truly impressive display of the AGC capabilities during Apollo was the terminal phase of the lunar landing. The AGC actively handed not only steering of the LEM, but also throttling of the DPS engine as well as actively calculating a predicted landing location on the surface. This was displayed to the pilots in the form of an angle, which they could use in conjunction with marks scribed on the window to see the point where the AGC predicted landing would take place. Using the RHC the pilot could re-designate this point left and right, or closer and farther and the AGC would adjust accordingly. Additionally the rate of decent was managed by the computer, with the pilot having the option to increase or decrease the ROD as desired. None of the lunar landings actually occured under AGC control, with all commanders opting to switch to manual control a few hundred feet before touchdown (pilots being pilots and all), but by all rights the AGC was absolutely capable of controlling the LEM right up until the point of lunar contact.

I watch all CuriousMarc videos for several years and I wish that I could first see your synthetic intro about the AGC. It provides an excellent overview !

Terrific video. Right up my street as an avid curiousmarc watcher & as a 30-yr fly-by-wire maintainer. Excellently written, narrated & choice of video (some clips I've never seen before). Superb effort & very watchable.

RU-vid has been recommending me your content for a while now and I feel ashamed for not having subscribed sooner. Really in depth and fascinating whilst being thoughfully presented videos about things I've always wanted to know more about - and delivered in a way that I can actually relax while watching. Really good, thanks so much.

That sidewinder fin is truly amazing. I am in awe of the people in our past and the solutions they invented. Your opinion on what makes a good looking super sonic plane is... let's say... unique. I think a lot of people find the Concorde pretty derpy looking... while the SR-71 is considered to be one of the most beautiful and coolest looking objects ever made by man. But... you be you!

Once again an amazing video Alexander. Keep up the good work! I find it awesome that we have people like you on RU-vid producing well thought out and informative content.

If you ever flew through severe turbulence in an airliner like I did, you experienced the wonders of Fly-by-Wire. Awesome technical video. Learned something new I didn't know about NASA's AGC.

Great video! I had an uncle who helped build the avionics in the Saturn V (Specifically the “Black Ring” Instrument Unit). I still have an old module of the erasable memory he gave me as a gift from his time at the program, along with original Mission Patches for Apollo 8-17. Learning more about the work him and so many others put in really shows just how impressive their designs were. The things that could be done with a drafting table and a slide rule are insane! Thanks for such an interesting look into a part of my family’s history, -J

Truely outstanding! The work that must have gone into this! Yet another reason why I have not bothered with a TV for years if such quality 'programming' such as this can be found here!

I have been watching YT videos for over a decade, and this is literally THE FIRST video that made me, Like, Subscribe, comment, AND SHARE. Congratulations! I'm impressed.

Todays modern avionics would not have been possible without Apollo project and the research carried out at MIT...

Nice work - lovingly researched and produced. Thanks for the list of sources - a bit of reading!

Another amazingly good video, thank you! As soon as I can, I will be supporting you channel for sure.

I'm a computer engineer, and while you're correct to say that your phone is not comparable to the Apollo Guidance computer, you do quite a poor job of describing the technical reasons for this. This is a great video, and I don't think many people will be misinformed by this, but there were many critical errors so I'm going to be that guy and nitpick you. It's mostly false to imply, as you do at 13:51, that it would be easier to build a chip from scratch specialized to go to the moon than copying an existing design. Nowadays, Designing a custom CPU from scratch is absurdly expensive, and there's rarely a good reason to do so. Such reasons are relevant to the case of sending a chip to the moon, but almost none of them are mentioned in this section. Some of the reasons you mention, such as writing an operating system dealing with interrupts and subroutines, are technical challenges that must be dealt with regardless of whether you use an off-the-shelf phone chip or a custom CPU. Others such as needing to rewire the CPU to spaceship peripherals instead of phone peripherals are only relevant if you are using an off-the-shelf smartphone rather than the CPU of such a phone. Since you explicitly say "engine of a car" rather than "car" itself I find this a bit disingenuous. Writing a custom interpreter for an assembly language is a foolish & unnecessary thing to do. A human-rated spacecraft should NOT be programmed in assembly today. Realistically a custom JIT interpreter will add a layer of abstraction between the software and the hardware--a completely unnecessary point of failure. Sorry if I've been critical. Your point is absolutely correct and this is a great video but I felt I had to address these technical inaccuracies. I thought I would also include some legitimate reasons why you would not want to use an off-the-shelf CPU design - Spare Hardening: This is the most important problem. You'll want to use more robust materials to make the chip than what a cell phone is made of, for sure. Insulating substrates probably don't want any of the cutting-edge

An excellent description. Well researched, engagingly presented and generally spot-on.

I had zero clue fly-by-wire came from the Saturn V despite knowing it came from NASA. Thanks man!

Thank you for all of your time and effort that went into this video. It was brilliant. Kudos to you.

Dude, this is an absolutely fantastic video, I thoroughly enjoyed the history and thought put into it. From the titan tragedy to the inspiration to make amazing content, I sincerely hope it pays off multiple times over!

I think this is the first time anyone's really explained the AGC to me in a way that really got me to go.. oh.. wow. Keep up the good work ❤

I just stumbled across your channel and I love it, thank you for such cool videos!

Finally! A fellow F8 appreciator!

Got be best video on the AGC fly by wire tests and that info on the Digital text font was interesting , keep up the good work

Just wanted to say that that title is amazing. Your videos tend to have really good titles but that title is genuinely inspiring.

What an insanely good video! I'm subscribing immediately. Thank you. As a software developer that loves aviation and the space race, this video hit all the points.

I'm now going to watch all of your other clips. You've been added to my "favorite channels" status 👍💪

Thank you Alexander, this is a wonderful video! I learned a ton and was truly fascinated! Awesome job!

Awesome video, I really enjoy this long form content! Keep it up

What a beautiful and well done video. Than you very much for sharing this work

Incredibly well done, thank you for putting this together.

Doood. The F8 was a MONSTER. It was primarily limited as to speed by the composition of the wind-screen. That happened around mach 2.5. The latter iteration was ridiculous. With the afterburner lit, it could climb to orbit-(ish).

As a pilot and aviation enthusiasts this was an amazing video and I learned and relearned things!

Thank you so much. I never realised the direct lineage of the AGC to modern fly by wire aircraft. I’m sure you might be aware but this was a revolution not just for aviation but for everything in the modern world.

Control Allocation is one of the most sophisticated and cool things I've come across yet in aerospace engineering

One of the best RU-vid videos I've ever watched

I remember listening to an interview with one of the engineers on the AGC who talked about how he was using many layer pcbs, something not common until the 1990s

Just found your channel, i sub'd within 5 mins of this video. Thank you for this and im sure the other vids will be epic too

Just found your channel, good to see another collector of strange display technologies.

An absolutely fantastic presentation! Thank you! Subscribed.

This was a great suggestion from RU-vid. Instant subscribe. Thank you.

What an interesting topic. Thanks for ur hard work

What phenomenal content man, congratulations.

Amazing video! As a utility pilot, I'm not a fan of Airbus school model, but you made me saw it from another very interesting prospective.

Can’t tell you how excited I am to see this uploaded.

amazing content! great presentation of what many think would be a very dry subject. subbed directly

fabulous insightful description of a massive subject well done

Fantastic video! I’ve just discovered your channel! I to have an obsession with the AGC and curiousmarcs channel haha

Fantastic video! You clearly have a wealth of experience to draw on and have done extensive research. Everything is explained clearly and in an entertaining fashion. It's also worth pointing out that the original stealth aircraft, the Have Blue prototype and the F117A that followed on from it, would have been impossible to get into the air without fly-by-wire.

Two Lessons I take from this: 1. Never fly a new type (or pretend old-type) in the first year of operation -- that chart was striking. Not as striking as the 737MAX plowing into the ground but still clearly important. 2. Heroes of modernization and great advances rarely get acknowledged. The team that came together for this was clearly incredibly capable but you couldn't politically capitalize on the effort and advancement. In fact, without the military adopting it, it may have been another decade or two before the economics forced the issue. Truly astounding work that I would never have known about without this video and our host here. Thanks for a great video and well-constructed story. As good as the Apollo Guidance Computer? No... but I think the folks who created it would be proud of this.

Thank you for making this video.

Outstanding presentation! Well done. One small point , if I may ... At 42:10, there's a very good shot of the wing in the raised (takeoff & landing) position. In this position, the leading edges droop and the trailing edge 'flap' and flaperon droop. You can see the small 'flap' next to the fuselage (not moving). Next to it moving outboard on the trailing edge is the flaperon. This device acts as the flap extension AND the entire aileron. Above the wing, and in front of the flaperon, are spoilers that you can see popping up when the flaperon moves to the 'up' deflection position. When the wing is in the down and locked position, the flaperon acts as the aileron for that wing. Respectfully, there are no flight control surfaces on the trailing edge, outboard of the wing fold hinges like on most other aircraft (as you show in the animated graphics). Those massive flaperons, so close to the longitudinal axis of the aircraft, were responsible for the exceptional roll rate of the Crusader,. A prized asset, essential in Air Combat Maneuvering. F-8's Forever! ^v^

What a great video! Thank you so much!

Absolutely love your videos!

thank you very much for posting

Actually the A5 vigilante had one of the first fly by wire systems. The electronics of the Vigilante were relatively advanced and complex at the time of its entry to service. It incorporated one of the first "fly-by-wire" systems on an operational aircraft, along with mechanical/hydraulic backup

This is one of the coolest videos I've ever seen

I learned a few things there. The use of virtual machines is one. I have heard (read / been told) that mainframes in the 1970s used them, but I had no idea they were used in the '60s by the space program, and the very early fly by wire systems. I am struck hard, by the confidence of the Air Force and General Dynamics, to design the F-16, not just to use FBW, but to be dynamically unstable and to depend on it, completely, at a time when it was still in testing and development and revision. I knew they built it with that concept as central, from the get-go, but I didn't know how new that technology was when they started using it. That's real bravery there. You've got to have some courage to bet the farm on something that's still in fairly preliminary testing ...

A really great and valuable presentation!

Fantastic content. Great stuff!

Startling to see your own model (the AGC) appear in a random video! I think i share your same fascination with that piece of kit.

16:27 TLI burns were performed by the LVDC (made by IBM), in the third stage of the saturn 5. The AGC is only used to monitor the burn and update the state vector.

Never knew that the AGC was the mother of fly by wire, thanks for doing what you do.

Quite excellent, very enjoyable!

Very well done!

took me 3 day to watch really great and detailed video

Liking the video! For added information/context, the Apollo Command Module also included an all-analog flight computing system, the SCS or Stabilization and Control System. It naturally wasn't as advanced as the AGC, but it could still provide orientation information from body-mounted gyros, attitude rate dampening, attitude hold, and semi-automatic control of engine burns. In combination with ground communications, it could be an effective backup for the AGC in a fair number of cases.

Incredibly informative video. Thank you for the effort. Wish you the best (and more video like that). Cheers from France

@27:36 -- another important improvement illustrated in this graph, is that you get flight-data-telemetry recording for-free compared to prior control mechanisms. Imagine the complexity of recording all flight control inputs from push-rods and control tension-wires!

Thanks for a great piece of content!