Hacker News Reader

Comment by Animats | Hacker News Reader

A fix is being flight-tested.[1] The authority of the auto trim adjustment system is being limited. All planes that don't have it already will get the "AoA Disagree" detection, and a disagree will lock out automatic trim adjustment. Pilots will get more training on handling the aircraft with that system inoperative.

How bad is the handling with that system inoperative? The need for this came from adding bigger engines to a small airframe, which required putting them too far forward so they'd clear the ground.

[1] https://www.wsj.com/articles/boeing-plans-fixes-to-make-737-...

View on HN

> How bad is the handling with that system inoperative? The need for this came from adding bigger engines to a small airframe, which required putting them too far forward so they'd clear the ground.

My limited, layman understanding from the assorted coverage and posts by people who seem to understand this thing is that almost all the time the bigger, more forward engines do not cause a problem and MCAS doesn't have to do anything. It is just that on those very rare occasions when you get into a situation where it does make difference, MCAS helps you recover safely.

So, if the system can detect a problem with the AoA data and lock out MCAS, and tell the pilots that MCAS is disabled (and the pilots understand that this means some protection is gone), they should be able to finish the flight safely with at most a need to avoid pushing any limits that might get it into a situation where they do need MCAS.

Only those very very rare situations where the plane both gets into a situation where it needs MCAS and it simultaneously has an AoA sensor problem would be a problem.

I think this is why they can get away without triple redundancy and majority voting on the AoA sensors for MCAS. Losing MCAS is not an "Oh my god we are going to die!" situation. It's a "fly more conservatively for the rest of the flight" situation.

I don’t know specifically about the MAX, but based on the thrust vector if a pilot in a stall put in more power before nosing down the plane would pitch up and you’d be put into an even deeper stall. The MCAS is there to nose down automatically for you if a stall is detected.

My understanding is that in the 737-8MAX, pushing the control column fully forwards isn't necessarily enough to recover from stall (or potentially even a near-stall AOA).

MCAS was designed to trim the stabiliser down in a high AOA situation, to give the pilots a bit of extra downwards control authority and help them recover from a stall.

To compensate for flying with MCAS inoperative, pilots will need to manually trim down when recovering from a stall. I assume they will just train the pilots to always trim forwards when recovering.

I hope this results in both Boeing and Airbus rethinking their approach to balancing computer and pilot control.

The scales are too often tipped in favor of coddling pilots (handling too much for them), with the trade-off being that emergency checklists multiply and become longer - to handle situations where plane systems and sensors fail.

While I heard many stories of pilots who could get out of dangerous situations due to "extra training/knowledge acquired in their spare time", the big question is how many pilots did not get into dangerous situations because of the computer taking care for them. My gut feeling says we are better off with computers...

Just because I can handle a car, prefer manual shifting and electronic helpers rather irritate than help me, I would never dare to demand we shift control back to the drivers. In effect, I am /safer/ when oncoming traffic manages to stay in it's lane...

> how many pilots did not get into dangerous situations because of the computer taking care for them

Agreed, that kind of computer control is helpful and good. The problem comes when either (a) those secondary/contingent systems aren’t built to the same standards (redundancy, fault tolerance, etc.) as the primary systems; or (b) their reasons for existence are to artificially alter the flight characteristics of the plane, for no other reason than to make the plane fly “like another plane” or in a more familiar way — so the perceived aviation experience is effectively an abstraction, and a leaky one at that.

I’m not against abstracting away some of the flight characteristics, given that in doing so, proper due diligence is done to make the resulting system reliable and fault-tolerant.

The major problem I see here, is that no amount of software will make flying a shoebox a good idea. The primary systems were at fault too here.

> Pilots will get more training on handling the aircraft with that system inoperative

Sounds like the expense they were trying to avoid if the plane were reclassified. Is the plane now going to fall under a reclassification?

With two fatal crashes partly due to pilot error in the face of undpcumented features I think the classification is largely moot: every pilot is likely to refuse flying without extra training.

> All planes that don't have it already will get the "AoA Disagree" detection

It makes me wonder... if all planes had 2 Angle of Attack sensors, why was the disagree indication an extra option?

Was it almost pure price discrimination on the part of Boeing? A $4 light/buzzer for equipment, wiring and firmware already present, but only wired in for clients willing to pay an extra $10k per plane for the safe version?

Was it almost pure price discrimination on the part of Boeing?

Probably.

A $4 light/buzzer for equipment, wiring and firmware already present, but only wired in for clients willing to pay an extra $10k per plane for the safe version?

Nope, EICAS literally a warning message which is just text on the LCD.

> Was it almost pure price discrimination on the part of Boeing?

In at least one other case I vaguely recall, monitoring two sensors required bus bandwidth that was unavailable without a physical refit of cabling or electronics, so that could well have been the issue with AOA sensors and MCAS. I have no knowledge either way.

I imagine after the design of the 737 Max was completed, some customers requested that feature, and Boeing said "For just $100k extra, we'll design and build it for you". They then set an engineer to build and test the feature, probably costing nearly $100k. That airline then got the feature they'd paid extra for, and other airlines/groups of airlines didn't get it because it isn't officially part of the design.

Any proof for this interesting theory?

I imagine not.

That is indeed exactly what the nytimes article suggested

I guess this article? https://www.nytimes.com/2019/03/21/business/boeing-safety-fe...

But my question was more like this: If it cost, say, $10k to buy the option, how much did it cost Boeing to actually enable it? If it was $8k, ok, but if it was $8 or $80, then they're more culpable.

I worked for a company making small electronic devices with LCD screens. There were two models of the device, a cheap one with a smaller screen and an expensive one with a bigger screen.

After a careful look it turned out it was cheaper for the company to put a bigger screen in both models, but as a plastic cover on top to make it look small.

It didn't compromise the safety though.

I remember CD burners were like this. You could turn a 2x burner into a 4x with a firmware upgrade.

Overclocking brings up similar circumstances: The market may have been 50/50 for the low-speed and the high-speed versions of a chip, but the yield may have been 70, 80 or even 100% for the high-speed, but the manufacturers still wanted to create a low-end and a high-end market.

Overclocking brings up another example - a lot of AMD CPUs have extra cores that can be unlocked. They're manufactured as the higher-end chips, then just have cores disabled. Sometimes due to defects, but often they work just fine.

I recall some video card where you could change a resistor and fool the card into realizing all of the extra capacity it was built with and had available.

Much of that is related to binning rather than price gouging - https://www.tomshardware.com/reviews/glossary-binning-defini...

If that was 100% true, then Intel wouldn’t multiplier lock its chips.

TIL: binning

The Sennheiser HD 595 headphones from several years ago were exactly the same as the HD 555's except for being $150 more expensive and missing a piece of adhesive backed foam stuck to the inside of the ear cups behind the drivers that were put there to give the cheaper pair a worse frequency response.

Back in the late 80s, we ordered a $100.000 upgrade for a minicomputer system. The technician literally used a pair of pliers to cut a wire on the board, and the upgrade was complete.

Reply on news.ycombinator.com