The Hits Keep Coming

This is another F-35 story. I know you’re probably thinking “when will this end?” and all I can say is: when the money-pump runs dry.

The F-35’s helmet is notoriously problematic. For one thing (probably the least thing) it costs about $400,000 and is an ongoing aesthetic disaster. It’s also, as WIRED calls it, “stupid cool.” [wired]

Although, I wonder if there’s maybe a word too many in that salad.

I recall reading that there were design problems (go figure) with the helmet and how it pushed into the back of the seat; the helmet is heavi(er) than a normal helmet and when you’re pulling G’s that’s a problem so the seat has to support it, just the right amount in the right places. It’s kind of interesting to see how to pilot is packed into the cockpit with all that crap surrounding them. Now, imagine that you’ve got to reach forward and fiddle with some touch-screen controls while simultaneously flying a really expensive killing machine. I don’t want to make this personal, but all I can think is that this arrangement of technology looks like someone made the mistake of asking Elon Musk to design an aircraft cockpit and then told him “money is no object.”

Why is the helmet so funky? Oh, that’s because it’s got augmented reality. Which, one imagines, might be cool or useful if done right. But, is it? Let’s first explore the capabilities of the helmet:

To go with what is possibly the most expensive and complicated weapon ever designed, a joint venture between defense contractor Rockwell Collins and Elbit Systems of America (RCEVS) designed a pilot helmet so advanced, it’s hard to say whether it’s an accessory for the plane, or the plane is an accessory for the helmet.

The head unit, made with help from Lockheed Martin (which designed the F-35) is way more than a protective shell. Built around a custom-fitted insert based on a 3-D scan of the pilot’s noggin, it combines noise-canceling headphones, night vision, a forehead-mounted computer, and a projector – not so different from the one in your office’s conference room – that displays live video on its clear visor.

Fortunately, pilots heads don’t change size. But having a $400,000 helmet that is custom-molded to the pilot is sure going to do a lot for making it easier to onboard new pilots. So, when the government of, say, Finland buys F-35s, do they have to ship the pilots over to someplace in the US to be scanned and have nano-tracking devices inserted under their skin by NSA? Joking aside, this seems to me to be a potential intelligence disaster for everyone except the US: we get intimate knowledge of your roster of pilots because they’ve got to come get custom-fitted, and then we can also give them a personalized head-mounted tracking device full of mysterious proprietary software. Given the US intelligence community’s fondness for putting malware in all the things, I can’t see how any defense department that thinks strategically would actually buy this stuff.

To make the helmet work as designed, plug it into a $100 million F-35 Lightning II joint strike fighter. There’s no Bluetooth or Wi-Fi here; a Kevlar-sheathed bundle of cords cascades down the side of the shell, providing input/output options for the plane’s communication system and computers.

That’s really amazing. Assuming it works. I have some experience with commercial virtual reality headsets (an HTC Vive) and augmented reality glasses (Q’s Magic Leap glasses) and “the software is finicky” is an understatement. In a jet fighter, it’s got to be flawless and that means every connector, every wire, and all the running computers and code have to not crash at an inconvenient time. I’m glad to hear there’s no bluetooth or Wi-Fi hauling the (presumably impressive amounts of) data back and forth, because those are not especially reliable systems in the face of hostile interference.

My first instinct, naturally, was to search for “F-35 helmet reliability” because I’m a cynic and a nay-sayer. It turns out that, like with everything else to do with the F-35, earlier versions didn’t work so good. Now, of course, everything is fine. Help me believe that. [airspace]

Ideally the display would follow instantly and seamlessly, but in early generations of the helmet, the display lagged behind a change in head orientation. According to an April 2017 Government Accountability Office report, those problems have been addressed.

“There are a lot of factors that need to work well together,” says Joe Ray, a manager at Rockwell Collins–ESA Vision Systems, which provides the helmet display system. If the developers conduct a test and see a performance they weren’t expecting, he says, “we bring the pilots back in and we duplicate that [finding] in the lab.”

It sounds to me as though some lucky pilots are flying around with a head surrounded by alpha-test code. But, seriously, there is so much money being spent on this project that they are going to get it right eventually. I’ll leave all the operation questions like patch management and regression testing unspoken-of, because I’m sure that they are not a problem.

Here’s where it gets interesting, to me: the aircraft has 6 cameras in it, providing visual coverage in all directions. So the pilot can look down at where their feet would be and “see” through the bottom of the plane. From the sound of it, the touch-panels that they are expected to be able to also use, must be overlaid in the correct location or something, so the pilot can still fumble with them. Or something. And apparently the F-35’s software can do the same thing that caused crashes in the Boeing 777 Max: it can take control of the aircraft away from the pilot and start doing what it thinks is best:


One of the marvels of this airplane is the digital flight control technology. You are telling the airplane to go up or down, speed up or slow down, go left or right. And the computers figure out what’s the best way to do that, and they’re going to move the flight controls to do it. And the interesting thing is, they may not do it the same way twice. So let’s say the airplane gets damaged, and one of the flight controls is no longer available. A legacy airplane would still try to use that surface because it doesn’t know any better. The F-35 digital flight control systems will say, “That surface isn’t doing much for me anymore, so I’m going to have to compensate by using some other things. Maybe I’ll have to move them a little bit more to get the same effect because the pilot still wants to turn left.”

Sounds like the pilot has two choices: sit back and enjoy the ride, or eject and enjoy the ride. To be fair, the pilots interviewed in that piece sound guardedly enthusiastic about the plane. Of course, several of them are Northrop Grumman test pilots. Slap me and call me “cynical” but I’m with Upton Sinclair (“It is difficult to get a man to understand something, when his salary depends on his not understanding it.“)

Here’s the thing I’m wondering: the F-35 was originally being touted as a stealth fighter/strike aircraft. Then, when it turned out to be too wing-loaded to be a decent knife-fighter – and nobody seems to engage air-to-air at knife range, anyhow – they began touting it as a mobile sensor platform. At that point, I wonder whether the air-to-air combat feature of being able to see in all directions is really worth it. Wouldn’t some kind of strategic situation management display be more useful? The F-35’s current “mission profile” (i.e: “purpose”) is what they call “beyond visual range engagement” – i.e.: you’re shooting long-range missiles at planes you can’t even see. Being able to see through your plane would be useful if you were in a classic dogfight, of the type that hasn’t happened since the Korean War (and even then, it was notable because it was rare). I realize that every little advantage in warfare can spell the difference between victory and defeat, but it worries me that the US’ strategy is to have a small number of very fancy gizmos, whereas every single component of that complex, fancy gizmo has to work right or its advantages are negated. Meanwhile, the Air Force’s procurement process appears to be optimized for producing insanely expensive stuff that needs to be tweaked and massaged endlessly until it works right, at which point it’s obsolete (e.g.: F-22).

All of this comes back to the Air Force’s love of flying. I understand that, I really do, but those days are as much in the past as knights in armor charging at each other in the lists. It had its moment, and that moment is gone. The Air Force is becoming an expensive version of the Society for Creative Anachronism, except they kill people. Most of what the Air Force continues to do with its expensive weapons platforms is drop high explosive on inexpensive insurgents.

------ divider ------

If you’re interested in Air Force procurement, you should read up on the “fighter mafia” – Pierre Sprey and John Boyd and a few other pentagon insiders [wik] who had the radical idea of creating a pure fighter aircraft. A knife-fighter’s knife, if you will. The result was the most successful aircraft the US ever built, the F-16. Which has since been turned into a general-purpose platform, mostly to hang bombs all over for dropping on inexpensive insurgents. The fighter mafia allied itself with the military reform movement [airforcemag]:

The Military Reformers were an obscure lot when they first emerged on the national stage around 1980. There were only about a dozen of them, mostly retired officers and midlevel systems analysts from the Pentagon and the defense industry. The outside world had never heard of them. They were not even called “Reformers” yet.

Their basic message was that the US armed forces were addicted to high technology and complex weapon systems. Such weapons were so costly that relatively few could be bought. Complexity made them hard to use and maintain, leading to readiness problems and reduced sortie rates. Even worse, the Reformers said, these complicated weapons were not as effective in combat as simpler, cheaper ones.

I assume you can see the profound influence that their thinking had on mine. One of the characters in the reform movement was Franklin “Chuck” Spinney, who is (I think he’s retired and cruising around on a boat somewhere now, he used to blog, or fulminate-blog [the blaster]) the author of the “defense department death spiral” paper. Perhaps I should write a piece about it, but if you’re seriously interested in this stuff you should go directly to the source. [time] Spinney’s view is godlike in its omniscience. Unfortunately, Spinney makes me extremely uncomfortable because at his core he’s still an imperialist: he can’t break away from the assumption that all these weapons are necessary and that the US needs to defend itself against those inexpensive insurgents.

Meanwhile, the procurement system’s history is repeating itself: the Air Force has decided it needs another cheap, pure, fighter. To maintain that air superiority over whom, exactly?


  1. Numenaster, whose eyes are up here says

    As a former federal procurement officer, I’m very much enjoying reading these. Thanks for the info (you’ve saved me a subscription to Aviation Leak & Space Technology) and the analysis.

    Disclaimer: I was with the Forest Service, so the most impressive thing I could buy would probably have been an helicopter.

  2. astringer says

    “One of the marvels of this airplane is… ” identical to Ardupilot + Pixhawk or any other 100 quid ($138.4 ) ‘copter with PIDs and extended kalman filters to make it fly. If one of your (say) six props snaps, the filter accommodates and it stays flying. Underwhelmed I am…

  3. JM says

    The general problem with the Air Force fighter and bomber procurement is that they are designing expensive systems for a war that never happened and probably never will. Their big long term planning and airplane construction is around fighting a big war with one of the biggest global powers. This hits all of the branches but the Air Force the worst. For whatever reason they don’t seem to be able to plan for what they actually do on a day to day basis. This isn’t entirely the Air Forces fault. Poor budgeting and oversight by Congress along with defense contractors exploiting the situation for their advantage also push thing along.

    The F35 does seem to take the cake though. It is the perfect example of a project so big that nobody will point out that it’s a failure because the whole house of cards would fall down. Instead just spend more money until it does something that can be pointed to as a success. Which isn’t hard because it’s normal combat mission is launching some missiles at undefended targets, as long as it doesn’t fall out of the sky and nobody asks any questions about it’s operating budget things look good.

  4. komarov says

    “””And the computers figure out what’s the best way to do that, and they’re going to move the flight controls to do it. And the interesting thing is, they may not do it the same way twice. “”””

    This would worry me as a pilot. Adapting to damage sounds fine, but it’s not clear how dynamic this system is supposed to be. “Not the same way twice” sounds a bit ominous in that the plane might decide, due to completely intransparent and minor factors, do things radically different despite broadly similar conditions. For the pilot that would mean they can never fully learn how their aircraft handles. Normally you’d practice some manoeuvre a couple of times to get a feel for it. But if all you’re really doing is ask the computer to do X and it keeps coming up with different solutions to “do X” that’s not something you can learn or rely on.

    Oh, it also occurs to me that most practice will be run in simulators. This raises the question if the simulator is smart enough to mimic an overly complex plane with its own internal decision-making processes based on its environment*. Maybe all those thousands of hours logged in the simulator will turn out to be useless because the thing used for training on wasn’t a good model of an F-35.

    *Where “environment” may equal a ton of data processed by a megaton of blackbox algorithms that seemed to work during software development.

    P.S.: Being the contractor to build the F-35 training simulator must be a dream job. “How do you know the sim is unrealisic if the real F-35 can’t even do [Lofty Design Objective] yet? Checkmate, DoD… and speaking of cheques….”

  5. jrkrideau says

  6. Pierce R. Butler says

    … the aircraft has 6 cameras in it, providing visual coverage in all directions. So the pilot can look down at where their feet would be and “see” through the bottom of the plane.

    Since the human head doesn’t have even 180° rotation capability, this must include some form of virtual rear-view mirrors, obligating the rest of the display to use a sort of fisheye lens effect to not create blind spots (same must apply to the instrument panel(s) as well). No doubt even sims of the actual display are so classified not even “Q” can see them [snarkity snark snark], but I imagine the net result resembles ’60s-psychedelic-movie visual-effects dizzification.

  7. jrkrideau says

    The more one reads about the user-interface (cockpit & helmet) for the F-35 the more one gets the idea that it was designed by a bunch of techie whizz-kids who grew up on computer games but who have no idea of pilot capabilities or needs.

    If they get the helmet to work, it will take amazing amounts of time for a pilot to learn how to use it. I am thinking probably in the 1,00’s of hours to really be able to make use of everything. Of course, perhaps 90% of the functions are unneeded for normal use and will be ignored.

    Given how carefully the even the seat seems to have to be tailored tho the pilot, does this imply that each F-35 is a bespoke edition that can only be flown by one pilot until cockpit alterations are done? Do pilots travel with their own seat as well as helmet?

  8. jrkrideau says

    @ 6 Pearce Butlar
    but I imagine the net result resembles ’60s-psychedelic-movie visual-effects dizzification.

    I had not taken the thought that far but I think you may well be right. Jack up my training hours estimate by a few more thousand and cut the prospective pilot pool by 50%.

  9. kurt1 says

    If I were a F-35 pilot I would try to get the deal with my government that Vladimir Komarov got with the soviets. Or an Osprey pilot. Why does the airforce like to build really expensive deathtraps?

  10. sonofrojblake says

    One of the marvels of this airplane is the digital flight control technology

    They missed a bit out there.

    One of the marvels of this airplane, and the F-22, and the Typhoon, and y’know most fly by wire aircraft. Planes have been thinking they know better than pilots for well over thirty years. Part of the reason Air France Flight 296 crashed was that the pilot spotted that they were approaching a hazard and yanked the stick back, which combined with the immediate application of full power would probably, in a plane where the stick was connected to the control surfaces, have seen them over the trees. As it was, the stick was connected to a computer, and the computer said “ah, yeah, nah” and didn’t alter the elevator settings, and then less than five seconds later (as the Wikipedia entry entertainingly puts it) “the turbines began ingesting leaves and branches”.

    Also echoing astringer@2 – there’ve been Youtube videos up for years of geeks flying drones, then deliberately literally snipping blades off the props with scissors and watching the software work out what’s happened and keep the thing flying.

    The central message is: if you’re in the seat of the F-35, you are not a pilot, you are a passenger making suggestions to the pilot, who is a computer program. And you’re going to have a hard time convincing me you need to be there at all…

  11. says

    The central message is: if you’re in the seat of the F-35, you are not a pilot, you are a passenger making suggestions to the pilot, who is a computer program. And you’re going to have a hard time convincing me you need to be there at all…

    Yet, the Air Force airplane-riders still feel and act like they’re Manfred Von Richtofen, come back to earth.Remember Top Gun featured F-14s versus some old MiGs – basically, Mike Tyson VS PeeWee Herman, but everyone was pumping testosterone like it was dragster fuel.

    The Air Force still insists in putting cannon/gatlings in aircraft because, you never know when you’ll be in a dogfight and be able to do a good ole gun-pass just like Eddie Rickenbacker. The F-35 had all kinds of design issues with the gun/gun pod, because of the problem of making a stealth gatling. Adding a gun brings in loads of interesting design constraints, like deflection from the recoil, damage from the muzzle blast/gases, and the casings (to say nothing of a huge ammunition feeding system) We’re talking some real dumb ass shit, here.

  12. says

    Its probably not worth a separate post, but one other thing all this complexity conceals: it is impossible to test. The combinations and permutations are too many and you might get subtle interdependencies. Like, what if a particular sensor starts failing intermittently – will the plane suppress the input or start gyrating wildly? You have to hypothesize all the potential errors, which is problematic if you think about a system like an F-35 as a flying data-center of interdependent servers. The plane is not a computer, it’s a network that has highly asynchronous processes.

  13. sonofrojblake says

    At least the F-14 requires a pilot.

    Adding a gun brings in loads of interesting design constraints

    I think I remember reading somewhere that if you fly an A-10 too slow and try to fire the gun the plane stalls – the recoil from the monstrous amount of lead flying out the front measurably slows the plane down to the point it drops out of the sky. Given that that plane’s entire design brief is to fire that gun while the plane is fairly low, that’s a hell of an edge to an operational envelope.

    Entirely off topic, if you’re not already aware of it, I imagine this is the sort of thing you might find funny.

  14. Dunc says

    Since the human head doesn’t have even 180° rotation capability, this must include some form of virtual rear-view mirrors, obligating the rest of the display to use a sort of fisheye lens effect to not create blind spots (same must apply to the instrument panel(s) as well).

    The way this works with headtrackers for recreational flight-simming (which is in some ways even worse, since your display is static while you turn your head) is that the movement is scaled – a small head movement results in a much larger change of the apparent viewpoint. I quit flight-simming around the time such things were becoming commonplace, but it’s supposedly not that hard to get used to.

    Also bear in that the military regards things like the monocular eyepiece used by Apache pilots – which is apparently an absolute nightmare to learn to use, and which a significant percentage of potential pilots simply cannot cope with – as perfectly normal.

  15. Reginald Selkirk says

    @13 sonofrojblake …think I remember reading somewhere that if you fly an A-10 too slow and try to fire the gun the plane stalls – the recoil from the monstrous amount of lead flying out the front measurably slows the plane down

    No, you didn’t read that. The 30 mm GAU-8/A Avenger famously uses depleted uranium ammo, not lead.

  16. says

    It is correct that the A-10’s flight control software puts the engines to maximum as soon as the gun starts spinning. Early versions of the plane fell from the sky when the gun fired.

    Rutan’s Project Ares solved this problem elegantly. I have posted about Ares before of you want to check the archives.

  17. says

    Reginald Selkirk@#15:
    NEWS BULLETIN: We interrupt this F-35 story to bring you a B-2 story.
    Air Force Seeks 3D Scanner to Reverse-Engineer Parts

    That’s nuts. For one thing, a competent machinist ought to be able to blueprint the part by careful measurement and CAD. Even with a scan, you’ve got to do that work, if it’s some aerospace component that has to fit correctly – you can’t just scan a part, 3D print it, and slap it into a plane. Maybe that would work with a WWII-era aircraft but anything worth doing is worth doing right.

    This is an interesting topic, though. Frequently I see people saying “why can’t modern engineers make anything as cool as an F-1 rocket engine anymore?” The answer is complicated – there are new things in modern machining techniques that make some parts a lot easier, and that means that the fundamental design changes. There’s no need to make the engine the way the original was, because the original had to be welded together by hand. Now, large parts can be produced with sintered metals in CNC’d molds, and they’re better and stronger. But making the molds isn’t exactly easy, either, and the CNC has to be driven by a complete 3D modeling chain of modelers who are also aerospace engineers/machinists. So, in the old days the problem was finding a bunch of insanely great welders and great machinists now it’s finding great machinists and modelers. It’s a whole different tool chain and that results in a whole different engine.

  18. dangerousbeans says

    I guess a blueprint wouldn’t be much use to the modern procurement chain?
    I saw a video a while back about someone 3d printing an intake manifold based off a laser scan of the engine bay ( This tech exists, and it doesn’t really seem that news worthy in 2021.

Leave a Reply