Airlines around the world cancelled and delayed flights heading into the weekend after Airbus announced on Friday that it had ordered immediate repairs to 6,000 of its A320 family of jets in a recall affecting more than half of the global fleet.
The European Union Aviation Safety Agency (EASA), which is the main certifying authority for A320 aircraft, issued the instruction on Friday night as a “precautionary action”.
The fix mainly involves reverting to earlier software
As folk around here will – hopefully, for as all know I know nowt about tech – confirm this means that it was the recent update which was wrong…..
In the interim fly Ryanair as they use Boeings….
Yes.. It’s the latest update that causes the problems..How very…Microsoft, Boeing, SAP, ummm need I go on?…
from Airbus itself.
Although I can’t imagine how “intense solar radiation” in the cockpit can all of a sudden affect flight controls and drop a plane a couple 100 feet all of a sudden.
Unless they fed automatic instrument lighting/screen brightness adjustment to The Wrong Place…
Hell of a difference between an update to general purpose computing and an update to safety critical systems.
That’s either a rather an unexpected combination glitch or a software management process failure, possibly a design failure as well.
I’d expect full system separation and safe failure modes to be baked in. That’s the bit that surprises.
The ‘intense solar radiation corrupting data ‘ sounds far more like one of those really hard to detect and pin down hardware design flaws not software. You might have to fix it with additional software exception handling, error correction or rerouting methods in extremis but it sounds like unshielded hardware in a system that should not be unshielded (it does happen , its hard to get everything perfect in all conditions especially when weight is a factor).
They’re having to replace hardware in old aircraft.
Only if swamp gas from a weather balloon refracts the light off Venus first.
The A320 software I wrote lived on UV-erasable PROMs in plugin modules. Remember those EPROMS with the quartz windows? Ah, my.
So maybe the energetic solar radiation is doing the same erasure, or partial erasure trick?
But then, why would software version reversion cure the problem?
Unplug the bitrotted EPROMs and plug the originals back in? Normally you put a silvered sticker over the erase window to prevent exactly this problem.
Remember, yes… and Yikes!!
Best in that I ever encountered was during my Uni days, with spanking new Shiney chromatographs and D/RNA syntesis rigs.
Really lovely, but they were in a Hotlab where we worked with pathogens.Which meant that Outside Hours, the lab was flooded with Hard UV, Because.
This wasn’t a problem, until the quarterly maintenance, which called for every piece of equipment to be un-panelled and left open overnight.
Y’know.. just to be Sure nothing got inside.
This maintenance was done in relay, so after the fist lab got done…All the new kit stopped working…
Panic!!!
I was one of the peeps who knew about electronics and was asked to Have a Look, and saw, with the others, the nice clean windows on the EPROMS…
Not even stickered over “for ease of update”….
We did a collective Picard, and proceeded to organise a lecture in Maintenance, and the Importance of Nail Polish in Programmable Electronics…
That was the early ’90’s, and you’d *think* …..
Did you write L-104?
https://archive.ph/srgt4
There’s the answer.
There are some pretty long cable runs in an airplane. I wonder if perhaps the shielding (which adds to mass) is perhaps not quite adequate for current solar conditions. Which I understand are unusually energetic.
The reversion to an earlier version may be some extra checksums that got dropped because “they weren’t needed”, resulting in slightly better response times in good conditions?
Yeah. But it sounds like Chris knows way more about this particular area of software than I do.
Yup – this is the fundamental design difference between Airbus and Boeing, which might extend to allegories of the difference between EU and US systems.
Boeing: the pilot(s) fly the aircraft “with assistance from the computers”
Airbus: the computers fly the aircraft “with instructions (suggestions?) from the pilots”
The key here is the commands from the pilots in an airbus cockpit are fed to the computer, that then has a think about what to tell the aircraft to do.
A Boeing pilot is much much more directly flying the aircraft.
So a computer flaw is much more likely to have a catastrophic impact in Airbus than in Boeing.
Absolutely. There are infamous examples. Starting with their very first demo flight in 1988!
Worst was AF447 crash that killed 228.
From memory. Being pedantic, with the AF449, the computers did exactly what the pilots told it to do. It wasn’t a computer flaw.
It would probably be correct to say the design principle of the “unconnected” side joysticks was to blame – with mechanically linked yokes would immediately alert one pilot if the other was holding the yoke fully back, but with the side sticks, one pilot was unaware that the other was holding his stick fully back and keeping the aircraft in a stall.
Who said anything about AF449?
With AF447, weather conditions plugged the pitot tube used to measure airspeed. Plugged, it read double-ought zero. A problem with computer flight control is that is gets its information from instruments. It assumes readings are true. Zero airspeed at 40,000 feet. Computer went nuts! Jumped to FULL THROTTLE! Pilots were unable to wrest control from the computer, and it flew AF447 into the sea.
[Gamecock hopes Tim the Coder didn’t write that one.]
There’s a procedure for when airspeed readings go unreliable, called “pitch & power”. You fly the angle of attack at the power setting the chart tells you will result in straight and level flight at your altitude and weight. It’s what pilots are trained to do, if they’re paying attention and not having a Frenchy flounce.
As a bloke off the street, I’d have assumed its what flight control computers would also be programmed to do under those circumstances rather than freaking out (didn’t happen) or disconnecting, leaving the pilots to control the plane using exactly the same faulty instrument readings that tripped the computers.
There’s your design fault. Right there. And it’s not the first time.
That’s not what happened. The autopilot and autothrust disengaged and there were stall warnings. The PF reacted erroneously, as though they were close to landing (when 99% of stall warnings occur IRL) and he (not the computer) firewalled the throttles while pulling fully back on his (side)stick. As they were very near ‘coffin corner’ the plane entered a stall, which proved to be unrecoverable.
Dealing with ‘unreliable air speed’ is pilot 101, even on puddle jumpers.
Disputed.
Crossed wires, I was referring to AF447, not the Airbus lawnmower. No pilots left to dispute the findings on that one.
The demo crash WAS the computer doing what the pilot told it to do. In the demo, they flew the plane down near the runway to do a flyby. To get the plane to do that, they told it they were landing. Initiated a landing sequence. When the got down near the runway and “pulled up,” the computer INSISTED on landing . . . in the trees beyond.
Another incident I saw on youtube (can’t find it now), people, middle-east, as I recall, were testing new Airbus by running up the engines to full power. Only way they could get it to do that is to tell computer they were taking off. Once they got the engines up and saw what they wanted to see, they couldn’t get the computer to back down. It was taking off, dammit! It eventually broke its moorings an crashed into a hangar. No one killed in that one.
One result of this is that Airbuses don’t tend to dive into the ground when overshooting as a result of the pilots getting somatogravic illusion, as happens in 737s.
Despite all the flying I’ve done I’ve never heard of »somatogravic illusion ».
in some ways I’m glad I hadn’t.
I hadn’t heard the name, but I recall the “upset & recovery” training many years ago. Foggles on, instructor throws the airplane around for a while to disorientate you – then you have to recover on instruments.
The first time I put foggles on – I threw up 🙁
But given the problems at Boeing (and FAA) that have become public over the last few years, I’m not sure “fly Boeing instead” is particularly good advice.
The first time I put foggles on – I threw up
That’s a very strong reaction. But it certainly shows what happens when your senses conflict with reality.
I remember a programme by James Burke where they got a bloke to stand on a box, in what appeared to be a room with fixed walls.They then started moving the walls backwards and forwards to see what effect it would have on his sensory system – his eyes and balance. The eyes won – he fell off the box although standing perfectly still.
Boeing: the pilot(s) fly the aircraft “with assistance from the computers”
Except the 737 Nose Plant edition, where the aircraft dives into the ground despite the pilots’ best efforts to stop it.
Or the door falls off.
Maybe safer flying on MD11s then….oh wait!
There is that.
Though in both cases, there was also a fairly fundamental failure to “fly the aircraft” – the memory item was to disable the auto stab trim which was not followed in either case. Flawed design yes, but the pilots fundamentally failed in both cases.
The Ethiopian guys never throttled back, either. Tits out all the way down. The Lion Air crew on the previous flight had had the same problem and dealt with it.
All things considered, Ryanair is a remarkably safe airline.
Yeah, the 737 Nose Plant problem was due to multiple safety protocols being overridden.
It’s like reading about Chernobyl or Challenger. two other cases where safety measures were removed or overridden multiple times.
The door problem seems to be sheer incompetence though. Maybe because I haven’t seen the explanation.
Self-flying airplanes. Rhymes with self-driving Teslas.
“Thank you for flying Ryanair. In 10 minutes time, we will be landing in Malta Mogadishu airport.”
Time for an updated VC10, perhaps.