PU 13: The Cosmic Ray That Broke the A320

This is Punching Upwards, episode 13 for the 7th of December, 2025. The Cosmic Ray That Broke the A320.

Broadcasting, as always, from the banks of the River Rhine, south of Echo Delta Delta Lima, which is Düsseldorf Airport, my name is Fab. Welcome to Punching Upwards, your podcast that looks behind the headlines and talks about the stuff the legacy media doesn’t talk about or gets wrong.

And we have a case of that today. And today I want to talk about an interesting software bug that caused the grounding of the majority of the Airbus A320 aircraft fleet recently. The A318, A319, A320 and A321, those are the four members of this aircraft family, is the best-selling jetliner worldwide. And obviously this outage caused a lot of grief at airports around the world last week.

The base airliner is the A320. And that’s probably what most of you have flown on at one time or another. The A319 is smaller. That is usually used for like shorter hops or routes that aren’t, you know, where not too many people fly the route. I’ve flown in one of those. I’ve never flown an A318, which is even smaller. That one is, if you look up a picture of that, that one’s crazy. That looks like a toy. Looks like somebody took an A320 and, you know, put it in the wash too hot or something. And the A321, I think I’ve never flown on either, is just like a bigger version of the A320. It’s longer. So they’re all like just different lengths. They all have basically the same kind of fuselage, kind of cockpit electronics, same engines, same wings. They’re just, you know, different lengths, basically.

Yeah, so the A320 had a little bit of a problem last week, and it caused a lot of trouble. Here’s a report from the 29th of November from France 24:

It’s a best-selling aircraft used by commercial airlines across the world. Airbus has ordered immediate repairs to 6,000 of its A320s, more than half of its global fleet, due to a flight control software issue. For safety reasons, the planes cannot fly until repaired. The flight system in question sends commands from the pilot’s side stick to elevators at the rear. It’s manufactured by the French company TELIS, which insists it’s not responsible for the problem. The incident that prompted the recall involved a flight from Cancun to Newark on the 30th of October. Several passengers were hurt after a sharp loss of altitude, forcing the pilots to make an emergency landing in Florida. After analysis, Airbus claims that intense solar radiation may have been the culprit.

We need to talk about solar storms, which are particles emitted by explosions on the sun’s surface.

This is an expert, and I’m using that term in air quotes here on France 24. We will be discussing … what that guy gets wrong immediately. The phrasing of solar flares being explosions on the sun’s surface, I guess they technically are, but, you know, when you think of an explosion, you think of something different. But, yeah, that’s … Let’s hear what else this expert has to say.

Project a cloud of particles and can reach Earth in about 20 hours. But if this happens with significant intensity, it can disrupt computers and therefore the flight control links to the computer give you bad information.

Repairs can be completed within a few hours for each plane, but disruptions will vary in length depending on the company.

We apologize for any inconvenience and will work closely with operators while maintaining safety as our top priority.

That was a statement from Airbus.

On Friday, Air France announced it was cancelling 35 flights, 5% of the airline’s daily total. Meanwhile, American Airlines said 209 of its planes were affected and that it expects some delays. Colombian carrier Abianca said the recall concerned more than 70% of its fleet. It said it would halt ticket sales for travel dates through the 8th of December.

So that is a relatively representative report of what was reported all over the country. In the media last week there was a lot of hysteria, you know all the Airbus planes need to be recalled and all of this and I’ve put this report in the show notes of course: fab.industries/podcast — there’s an episode list you can get to the show notes and as usual there’s all the material there’s all the other stuff I used to research this episode as well some good background if you want to read further into that.

So, let’s look at what happened. So on the 30th of October, 2025, a little while ago, an Airbus A320 operated by JetBlue was flying from Cancun in Mexico, and that’s Mike Uniform November, to Newark in New Jersey, that is Kilo Echo Whiskey Romeot when the plane suddenly dipped down and rapidly descended for about four seconds without inputs from the pilots.

In aviation, this is what is called an uncommanded pitch-down event. The aircraft fell from about 35,000 feet, so a cruising altitude, to around 20,000 feet. And when the pilots regained control, they further descended for an emergency landing in Tampa, Florida, Kilo Tango Papa Alfa.

The autopilot was engaged while this happened. So throughout this whole time, not the landing, but the event and even afterwards. The incident caused at least three head injuries, as well as another seven minor injuries on board when passengers and possibly cabin crew who were not wearing seatbelts collided with the roof of the cabin. I’ve actually seen this happen once. I think in that case, it was like a turbulence. It’s like when the plane … temporarily, you know, loses lift and just drops. And I’ve seen somebody hit their head, which is why I always wear a seatbelt when I’m on an aircraft and, you know, not walking to the toilet or whatever. And this is a good example of why you should always do that, always wear a seatbelt, because stuff like this can happen.

So the Airbus A320-200 in question, registration November 605 Juliet Bravo, was later found to have a faulty ELEC No. 2, the ELEC, E-L-A-C or Elevator Aileron computer, is one of two aboard the aircraft, part of a total of seven flight control computers aboard the A320. The two ELACs control elevators and ailerons, two different types of control surfaces on the aircraft. The elevators are located on the horizontal stabilizer on the tail of the plane and control the aircraft’s pitch, which means they move the nose up and down. The ailerons are located in the wings of the aircraft and control its roll, which means rotating the aircraft around its longitudinal axis, basically lifting the wings up and down. You’ve probably seen that if you’re sat in the window seat. It’s like this little … kind of like rudders people usually call them rudders although they’re not the rudder on the aircraft that’s on the tail but they’re on the wings you know and then one will go up on one side of the wing and at the same time the other one will go down on the other side and then it pitches the plane towards the side of the aileron that goes down.

Now the elevators in the back obviously in the control plane in the back of the plane which you will never see and you know those obviously pitch the nose up and down There are two ELACs because all systems in modern passenger aircraft need to be redundant. The Airbus A320 is a fly-by-wire aircraft. This means that unlike in older aircraft types where wires ran from the controls in the cockpit to the wings and tail to actuate the hydraulic systems that control the control surfaces. In a fly-by-wire aircraft, everything is operated by computers connected via digital links over electric cables.

The A320 actually was the first fly-by-wire jetliner and also introduced the iconic little side stick that replaces a traditional yoke on all modern Airbus planes. So if you fly simulator, you know this. This is how you can tell. you know, the big two manufacturers, Airbus and Boeing, apart. Airbus has these two little side sticks. There’s one on the left side of the cockpit for the pilot and one on the right side for the co-pilot, whereas Boeing has these traditional yokes that are in the middle, you know, between the legs of the two pilots. Those are also in modern Boeing, you know, modern Boeing are also fly-by-wire, but, you know, they still retain these yokes. It’s just a preference thing. You know, the side stick’s just been … It’s now just an Airbus thing. It’s almost a branding thing at this point.

On the 7th of November, the National Transportation Safety Board, the NTSB of the US, reported that the in-flight upset had occurred because of, quote, an ELEC switch change. On the 28th of November, the US’s Federal Aviation Administration, the FAA, issued Emergency Airworthiness Directive 2025-24-51 in conjunction with its European counterpart EASA and their EAD 2025-0268-E, grounding all A320s worldwide. But what had actually happened? And was this a solar flare?

So as it turns out, Airbus seems to have disabled some protective code when it went from software version L103+ to L104 for the ELEC-B system, which is the ELEC computers that are built into the A320, which is itself manufactured by Thales. Although the software running on these computers is developed by Airbus. And this is why Thales said they weren’t at fault because the software was at issue. The software update was meant to improve the flight envelope protection of the Airbus A320 and reduce the risk of a control loss with some software improvements. In fact, due to a mistake, it did quite the opposite.

So in computing, we call this sort of bug a regression, when you are trying to improve your software and in the process of doing so, you actually make things worse. And here they, I guess, forgot to patch some code in that used to be in this software. So that was actually a while ago, you know, just in this case, it just took a while for this regression to surface because the situation in which the code they left out matters is pretty rare. The code that inadvertently patched out in version L104 apparently protects against the effects of cosmic radiation on the flight control computers like the ELEC.

And because this version didn’t have those checks and the JetBlue A320 was then apparently hit by cosmic radiation, the system malfunctioned. To explain this, this is a little bit in-depth computer stuff. If you’re not so much a computer nerd, here’s a quick refresher. So the RAM in a computer, which stands for random access memory, stores information. And this is in every computer. This is your desktop computer, your iPad, your phone. So the RAM stores information in the form of an electric charge in a tiny electronical component. Basically, there’s some electrons that charge this component. Usually this is referred to a memory cell. When the component is charged, this represents a one. When it is not charged, it’s a zero. This is how all data is stored in RAM or generally on a computer in ones and zeros.

Cosmic rays, when they hit the Earth’s atmosphere, are absorbed, normally, but they generate, when they have enough energy, they generate a shower of secondary particles. Some of these are energetic neutrons. If those hit a chip inside a computer, they can create alpha particles and other forms of energy that have the ability to upset the charge of the electrons in a memory cell. And this can turn a one into a zero or vice versa. We’ve known this for ages. This is what is called a soft error.

I actually learned about this in high school, in computing class. Contrary to most of the reporting in the legacy press, this is not caused by solar flares. The sun generally doesn’t create cosmic ray particles that are capable of penetrating the Earth’s upper atmosphere to create particle showers that then in turn can cause soft errors. The rays that can do this come from deep space. And if you think about this, this is kind of logical because if the sun … If the Earth didn’t have a magnetic field and an atmosphere that would protect us from these particles and the sun would generally create particles that were energetic enough to penetrate the magnetic field and the atmosphere, then we would have a lot of problems. We would probably have a lot of cancer and life on Earth probably wouldn’t be that nice. So generally, this is not caused by the sun.\

But interestingly, intense sunspot and solar flare activity, what this guy on France 24 called explosions on the surface of the sun, that actually causes the Earth’s magnetic field to shift in a way that makes it harder for deep space particles to penetrate the planet’s atmosphere. Which means that the cosmic rays that can cause soft errors, like in the case of this JetBlue flight, are more likely during quiet phases with low activity from the sun. So it wasn’t a solar flare that caused this. It was a high-energy particle originating in deep space, outside of our solar system, probably. I actually found this quite interesting. I actually didn’t know that. I learned something while researching this.

So … this kind of thing happens actually more when the sun is not in like one of these sunspot cycles where it has lots of eruptions and solar flares that, you know, create this. It’s called solar wind. It creates these particles and then hit the Earth’s magnetic field and get, you know … get deflected and discharged and sometimes they hit the atmosphere. I guess this is what northern lights are. But actually, as it turns out, you know, that actually also protect us from cosmic rays.

So in 1996, IBM estimated, and I’ve put this study in the show notes, the IBM study that is, at fab.industries/podcast. If you look at sources at the bottom, there is a paper called Terrestrial Cosmic Rays by J.F. Ziegler in the IBM Journal of Research and Development from 1996. So this is what I’m referring to here.

So in 1996, IBM estimated a rate of one bit flip per month. And that’s a soft error, you know, flipping a bit from zero to one or the other way around. And they estimated a rate of one bit flip per month per 256 MiB, maybe byte of RAM for a desktop computer at C level. So, you know, roughly 256 megabytes. And maybe byte is actually a bit more than a megabyte. But, you know, that’s just … nitpicking, but I like to get things right. So one bit flip per month per 256 maybe byte of RAM for a desktop computer at sea level. This rate is about 300 times higher for a commercial airliner at cruising altitude, because the higher you get in the atmosphere, the more of these rays you actually get. Actually, the particle showers caused by these rays, the higher the likelihood is because the atmosphere is thinner. The further they go through the atmosphere, the less likely it is. They actually tested this in this paper.

You can read all about that if you’re interested. They actually tested this by putting computers deep under the earth or in a cave. You will eliminate these kind of soft errors. There are also other soft errors that used to be more common. They come from the packaging of the actual components used in a computer. When they have radioactive decay, that also causes energy that can cause this. But cosmic rays are actually a lot more common these days. And these days, this error rate is probably higher because increased chip density in modern chips means they are more susceptible to this. So we have this.

Actually, I learned about this in informatics class and computing class in school. Because this causes software errors. This can actually, you know, this corrupts your RAM. This happens to your desktop computer as well. And usually it doesn’t matter. Like, you know, if it’s just one bit somewhere in some random, you know, if it’s a JPEG in memory, it’s not going to cause a lot of trouble. But if it happens in the right place, it can actually cause, you know, crashes like back in the day, blue screens of death and stuff like that. So you know and so modern computers are more susceptible and the higher you are in the atmosphere the more susceptible you are so this is obviously a significant problem and one this one that is well known which is why the Airbus software had some error correcting built into it that checked for these random looking bit flips and would disregard them so you know if just the random bit flips in memory the computer would check that and go well this number doesn’t make any sense anymore.

The way just this one bit changed is not in the program. And you can do, I guess, error correction. You can do probably checksums for stuff in memory. I don’t know specifically how this works in this instance, but I guess that’s the kind of software check we’re talking about here. So when there’s a random-looking bit flip, it just disregards that. But then when that was patched out by accident, it was basically only a matter of time until a Cosmic Gray hit a RAM element in an ELAC on an A320 in just the right spot to make it do something unexpected. There were probably lots of these instances before. But this JetBlue flight was probably the first where the bit flip actually did something. Sometimes it just might not do anything.

And this is why ELEC number two on this plane, and ELEC number two normally is controlling the elevators, so they’re redundant … So if one… You know, they have these switches in the cockpit. If you ever flown, I mean, back in the day, you could actually go into cockpits. I remember this. I’m that old. But these days you can’t anymore. But, you know, you can fly a simulator. You can actually look at the panels and you can see you can switch all this stuff over. So in case your ELEC 2, you know, is malfunctioning or doesn’t work anymore, you can switch all this around.

So the other ELEC does what ELEC 2 is supposed to do. But generally in a normal situation, like on this flight, ELEC 2 is normally controlling the elevators. And because that has happened, it pitched the A320 downwards in this case. And it took a few seconds, that’s the four seconds in this incident, for the autopilot to correct this error automatically. So the plane was basically, the autopilot noticed that this was an uncommanded pitch down because the pilots weren’t in control. The autopilot was in control and the plane just went nose down. Which isn’t a bad thing.

This is not critical, obviously, because it happened at 35,000 feet. You’ve got a lot of way to go down before something goes wrong. The plane didn’t go too fast. The pilots, obviously, are always buckled up for this very reason but obviously some passengers got pretty badly hurt you know so some of them were probably in the cabin, uh, going going about their thing so this is why you should always wear a seat belt if you can. Again, this is this is good advice.

So to correct this issue um that appeared in this flight the FAA and EASA in Europe mandated a rollback of the ELEC software to version L103+, the one before the affected version, where these crucial software protections were patched out. A software update like this usually takes about 30 minutes, so I don’t know where the mainstream media gets the three hours from. I looked this up, everybody seems to think the update itself takes just a few minutes, but they obviously have to check that everything went according to plan.

And that should take about 30 minutes to an hour for each plane. But because many of these planes need to be flown to facilities where these updates can be taken care of, the FAA and EASA allowed airlines to perform limited ferry flights, that means without passengers on board, to get affected aircrafts to the necessary locations. And it was all done within a few days. So there were a lot of planes, thousands of planes that had to be patched, but obviously it’s just a software rollback. So it was kind of not as bad. It was just, you know, a big deal because this is the most flown aircraft in the world.

So lots of airlines were affected. Obviously, when they lose that many planes, people get stranded at airports and stuff like that. That’s always horrible. But at least, you know, nothing serious happened. This is not obviously nothing, something someone should panic about. I mean, this shows that something like this can happen and aircraft systems are still, you know, redundant enough so that something bad doesn’t happen. You know, if the autopilot hadn’t corrected it, the pilots could always have switched this ELEC off. And I don’t know for sure, but there’s probably a non-normal checklist that probably goes into that when your plane just uncommanded goes into an uncommanded dive. It’s probably disconnect autopilot, you know, and then some other steps and one of it is probably checking the ELAC. So, you know, it’s not that bad.

I just found this interesting because the story itself is interesting because it kind of shows what somebody who, you know, like me, who’s been reporting about software and security for many years now knows and programmers know that software is all the same. Programmers make mistakes and stuff like this happens. It does not only happen with your random GitHub project, your open source project or whatever you’re doing at work like your your web app that you’re that you’re coding in TypeScript or whatever this happens to everybody and this also happens at Airbus and this is why um there’s very very few software that can be considered almost bug free. And there’s some, you know, NASA is very famous for certifying software to like a very, very small percentile of possibilities that there’s bugs in there, which is very, very expensive. It costs like hundreds, I think hundreds of thousands of US dollars per line of code. It’s incredibly hard to do this kind of stuff. Yeah, this is just something you need to keep in mind and not only for planes, but this happens with everything, with payment systems, with cars, software in cars. It’s kind of the same thing.

And I think in cars, it’s probably even worse because we have very strict regulations in the aviation industry. There’s procedures for this kind of thing. Everything’s redundant. Yeah. And so usually this doesn’t happen. But sometimes, you know, combination of software or hardware errors actually cause a lot of fatalities, like in the Boeing case where they had a software that would also kind of pitch the plane down and it took two hull losses and many hundreds of people died before they figured out what the problem was. Actually, the second crash would have been preventable in that case because they actually knew what was up after the first crash. But, you know, something like this happens. So software is … When you have fly-by-wire on your aircraft or your car or whatever, this is what can happen.

And I also found this interesting, of course, because the media got it wrong. Everybody was reporting solar flares, and I think that might have even been in an Airbus press release but I’m not quite sure but I read about solar flares everywhere and people were talking about solar activity and I found this very interesting because I hadn’t I knew that this was cosmic rays as in they were from deep space, not from the sun. But I didn’t know that they’re actually more common or less common when there are solar flares. Obviously, this is a topic I’m very interested in. And ham radio operators know about this a lot because the sun weather as ham operators call it is very important when you try you know ham radios or amateur radio operators have very limited bands they can use and limited output power and to get across the planet to maybe from Europe talk to somebody in Australia you do like a technique where you bounce radio waves off the clouds and so the actual weather is important you know you have to kind of bounce the the radio wave off the cloud, and then it bounces down back off the surface of the Earth. Ocean works really good for this. And so you can actually go halfway around the planet. You can actually, you know, normally you’d only have, I don’t know, what is it, 13 kilometers to the horizon. So you can go over the horizon by bouncing radio signals off the clouds.

But if you go these long distances, it’s actually very important to know what the sun weather is like because the electromagnetic properties of the planet and of the magnetic field and of those particles that the sun emits are very important. In that case, you don’t worry about cosmic rays because they’re just very, very singular events. It’s just literally one particle. But it’s a particle that is like extremely highly energetic that just comes zooming in from deep space from somewhere I don’t even know. I would have to look this up. I can’t say offhand where they originate from. Quasars or whatever. No idea. But, you know, they come zooming in and then they go bing and they change a bit on your computer, which I was fond of. Already back in school, I found that fun, like that you have this kind of problem that there’s a ray from space that changes RAM in your computer, which normally you won’t realize. When you’re recording a podcast like me right now, if this happens, it’s not like on this audio recording you will notice one bit flip. And usually there’s also error correction and stuff like that. But there are critical applications, especially in cryptography, when this can actually cause problems. And there’s also another fun thing is actually there’s a way.

So in cryptography, obviously, you need random numbers and lots of Cryptographic algorithms use pseudorandom number generators, which are not really random. So it’s better when you actually use actual random numbers. And one way to do that is there’s actually chips that listen for cosmic radiation. Or general radiation. This is also caused by … if you have another radiation source, like if you have anything with radioactive decay, which happens, I mean, I think granite… Is it? Granite decays and actually causes radioactive gas. I think there’s gas in there. So stuff like that can happen. There’s actually chips that use this radiation in the environment. You know, those little particles pinging in basically random intervals of this chip, use that to generate, not pseudo, but actual random numbers. I always thought that was cool.

Yeah, anyway, I hope you learned something today. And if you learn something, and if you think this is worth supporting me for, you know, to keep this show going, you can do so like Michael Mullen-Jensen, Fadi Mansour, and Evgeny Kuznetsov, who are all subscribed to the podcast on Substack, which is great. I mean, you can subscribe there for free, which means you get an email when the show releases. But, you know, you don’t really have to do that. It’s mostly you can subscribe and then have a paid subscription, which helps me out financially, obviously. And yeah, if you want to do that, go to fab.industries/podcast. It’s all explained there. And I would also like to thank Sir Galteran, who supports the podcast via Fountain.fm, which is a podcatcher that you can use online. That’s a great podcatcher, but it also has the ability to put some Bitcoins in and then you can either give all the shows that you listen to a certain amount. You have a certain amount that you put in there and it gets divided by all the shows you listen to. Actually, I think by the seconds you listen to the podcast or you can boost them. You can shoot them some extra money. And yeah, Sir Galteran does that for this show, both of it. So thanks for that. Thanks to all of you. I appreciate it. You make this at least a little bit worthwhile. It’s a lot of research during the show, so make it worthwhile. I would need some more supporters, but this is a great start and I appreciate it.

And before I say goodbye at this point, I would also like to mention that this podcast is now part of my No AI Content initiative. If you want to know more about that, including why I’m doing it, you can also go to fab.industries. And on all the pages at the top navigation, there’s a little owl at the top. Click at the owl and it’ll tell you. But in short, this podcast is produced without the help of AI. I don’t use AI tools when I research the topics for the episodes, prepare my notes for the show, or in other parts of the production and or the recording. And, you know, click on the owl on fab.industries and it explains and links you to a column that I wrote why I think this is a very important thing to do.

So, thanks for listening to this episode of Punching Upwards. The theme music for the podcast is a track called Fight or Fall by Dev Lev, which is also not produced with AI and used under license. I will be back with more detailed coverage of interesting news stories, little facts dug out from behind the headlines, and stuff that you won’t get from the corporate news media next Sunday.

Until then, goodbye, and I wish you good luck. This has been Punching Upwards, podcast by FAB INDUSTRIES. New media, new rules.