Air Asia Crash Report : Cracked Solder Joint on Rudder Limit Circuit Board

[EEVblog]

Solder is not generally regarded as a structural part of an electrical joint.

A common misconception.
By definition, a solder joint is used to mechanically fix the component (unless additional anchor measures are taken), whilst providing electrical connection too of course.

[coppice]

Solder is not generally regarded as a structural part of an electrical joint.

A common misconception.
By definition, a solder joint is used to mechanically fix the component (unless additional anchor measures are taken), whilst providing electrical connection too of course.

Its not a misconception at all. All good reliable soldering techniques are based on reducing the loading on the solder to the minimum possible, making it merely a filler between other most robust materials. Single sided through hole boards put a lot of loading on the surface solder, and the solder cracks with monotonous regularity. Through hole plated boards move the loading burden to the holes themselves, where the solder is well supported by the plating and the component leg, and reliability is much better.

[SL4P]

Solder is not generally regarded as a structural part of an electrical joint.

A common misconception.
By definition, a solder joint is used to mechanically fix the component (unless additional anchor measures are taken), whilst providing electrical connection too of course.

Dave,
I'm surprised at your response...!
I was taught to solder by a euro-mil-spec type of guy many years ago, and not to say he knows everything, I spent 40 years working on the highest quality professional equipment - and all sources indicate that good mechanical design MUST be in place to support good soldering - for long term reliability, and immunity from mechanical failure over long periods of vibration and environmental cycling.

Your experiences are respected, but on this one, I'd still opt for snug holes, correct (usually flush) component mounting and adhesive where needed - rather than none of the above!

Cheers!

[Psi]

If the hole was "snug" enough to provide all the structural support then the molten solder wouldn't flow nicely all around the joint.
Since solder flowing all around the joint is required the solder does provide structural support.

[crispy_tofu]

it gave me a feeling that their planes would run into all kinds of "fatigue" problems eventually

Not only AirAsia, fatigue problems are quite common on short-haul planes.
Control cables can snap (e.g. Air Moorea Flight 1121, this was also due to fatigue), parts of the plane can literally fly off (e.g. American Airlines Flight 587, the tail structure broke off), including part of the fuselage (Aloha Airlines Flight 243, which was an explosive decompression due to fatigue)...
But a solder joint is probably unheard of...

[EEVblog]

I'm surprised at your response...!
I was taught to solder by a euro-mil-spec type of guy many years ago, and not to say he knows everything, I spent 40 years working on the highest quality professional equipment - and all sources indicate that good mechanical design MUST be in place to support good soldering

In that case all through hole boards must have the every component physically secured to the board before soldering, and then stress taken off each lead by looping or whatever.
In that case I must have been blind for 35+ years, including more than a decade of which was spend designing military PCB's with extensive experience in vibration testing and component modal measurement because I've never seen such a thing, never been taught to do such a thing, and never seen any standard for such a thing.

Yes, mechanical securing of components is all part of good design practice when it's required, but the fact is that in the scheme of things almost every PCB ever made uses the solder joint as a mechanical support in some form for most of the components. To argue otherwise is to ignore reality of how boards are constructed.

- for long term reliability, and immunity from mechanical failure over long periods of vibration and environmental cycling.

Sure, and that's what potting and other methods are for when required. Most components do not get this treatment.

Come on, seriously, in your 40 years of working on the "highest quality professional equipment" have you ever seen PCB that de-stressed every solder joint?

Your experiences are respected, but on this one, I'd still opt for snug holes, correct (usually flush) component mounting and adhesive where needed - rather than none of the above!

Unless you glue or otherwise physically fix the body of every component to the board and de-stress every lead, by definition the solder joint is taking some mechanical stress. It must be, it's basic physics.
You said it yourself, use adhesive where needed. Flush mounting and snug holes does not remove mechanical support from the components. In fact, you can't have a snug hole and a good solder joint with solder flowing through the holes to the pad on the other side. And if you are using a single sided board, then that's bad design practice for mechanically critical boards to begin with.

[SL4P]

Obviously, we agree to disagree!
"Come on, seriously, in your 40 years of working on the "highest quality professional equipment" have you ever seen PCB that de-stressed every solder joint?" - yes.

"Yes, mechanical securing is all part of good design practice when it's required..."
That's all I was trying to say.
In many cases, this is all that separates consumer /prosumer, and mil-spec assembly standards.

[kaz911]

Mil spec is also country dependent..... Some countries specified curl mounts with 45 degree leg bend for high stress mill stuff when I trained.

So resistors legs was given a 270 deg circle (3/4 gentle twist around a screw driver when prototyping) before going in to the PCB and on the other side a 45 deg bend from pad to the side of the fattest trace - to offset stress away from PCB and towards the 270 deg bend plus offsetting stress at the PCB at where the copper trace was strongest. And then resistor was floating about 1/3 of an inch above the boards for all the Mill spec ARC net and early 2 MBit (yes 2 Mbit) Fibre network adaptors we designed.

That was around 1984 - and the Fibre 2Mbit was for " less option to snoop " - and a complete custom job for that particular country's military use). I had a roll of 2 km Fibre optics cable under my desk to  " stress- test " with *G*.. and at that time it was worth a small house......

[bitslice]

On that joint I can see corrosion of some sort, what caused that and why isn't the board protected from environmental effects?

[AF6LJ]

On that joint I can see corrosion of some sort, what caused that and why isn't the board protected from environmental effects?

To me it looks like dried rosin flux.

[Cubdriver]

On that joint I can see corrosion of some sort, what caused that and why isn't the board protected from environmental effects?

Well, it did go swimming in the ocean for a while.  It's likely weather resistant, but not intended for long term immersion since flying machines are typically creatures of the land and sky, but not the water.

-Pat

[SL4P]

I sometimes wonder if ultrasonics should be added to wave-soldering or reflow ovens, if only to break the surface tension of molten solder. Might have saved that particular joint.

An interesting idea...  I guess simple wave soldering and simple reflow is no longer used in high-spec boards, and part of the vapour-phase production process is serious chemical cleaning before... and a weird mix of vapours in the soldering cycle - to ensure removal of contaminants and other unexpected surprises.

I use ENIC gold passivated tracks - to help with better storage life - and soldering quality.

ADDED: The sneaky gotcha is supplier or inventory that allows oxidation of leads while on the shelf.  Airtight containers and component cleaning if available.

[CatalinaWOW]

While solder does supply some mechanical support for components, it is tough to make it do the whole job for things in tough environments.  Solder is an absolutely horrible structural material, with a different coefficient of expansion than most other materials used in PWBs, low fatigue life, low structural strength and worst of all poorly defined mechanical properties due to widely varying compositions in the as deposited case.  This means that additional support may be required unless several conditions are met:  Light components, small thermal stresses (small swings and limited number of cycles), low vibration (low amplitude and short durations) along with excellent initial quality and good joint design.  Good joint design may forbid additional mechanical support if there is no mechanical relief between the additional support and the solder joint.

Depending on the exact installation location of this board, it would have seen low to moderate levels of vibration for hundreds of hours, and moderate to extreme thermal cycling for hundreds or thousands of cycles.  The thermal cycling would what would worry me for this joint.  I have seen no front/back pictures of the joint, but the lead looks like it might be from a large component with pin leads.  This configuration can force the solder to carry large thermal expansion loads.

I worked on high reliability electronics my whole career, and failed solder joints were one of the most common problems.  A lot, probably the majority related to the initial quality of the joints, but many were mechanical failures that occurred during thermal cycling or vibration.  Gold embrittlement was often a problem - one of those things that is in the grey area between poor initial quality and the inadequacy of solder joints to deal with mechanical stresses. 

[bitslice]

Well, it did go swimming in the ocean for a while.  It's likely weather resistant, but not intended for long term immersion since flying machines are typically creatures of the land and sky, but not the water.

What I was getting at is that the electronics in planes must suffer from variable humidity, and every PCA I've seen that spends it's life outside of a computer case has some kind of conformal coating.
Something that is expected to last decades would surely be given as much passive protection as possible?

[SeanB]

Generally aircraft PCB will have a conformal coat on both sides, and larger components will also likely have extra support, either a tie down with some lacing twine, a clip riveted to the board, some epoxy applied after soldering and before conformal coating or a combination of these, followed by the conformal coat, generally a dip and brush coat or a mask for connectors then a spray application. Thus the coat can be thin and protective only, or so thick that the components will still work even if all the solder joints are broken.

To add to that some conformal coats are reworkable, some are solder through and some need removal before rework, often being different for each board in a unit ( don't you love separate manufacturers) or different for each batch ( same again) so you need to identify and do the rework appropriately. Recoat after rework also needs a matching coat, so you might have 3 different solvent removers, 3 different rework tool sets and 3 different coating pack sets ( all with a very fixed shelf life and an even shorter use before xx after opening) to do this. some are not compatible with the other, either they peel off, do not adhere or worse reach chemically and destroy the board. Plus the solder through ones are nice and generate hydrogen cyanide gas on heating, so you need good fume extraction on soldering.

[Rupunzell]

Crash this Scarebus (Airbus) A320, AirAisa flight QZ8501 goes far beyond cracked-failed solder joints.

Without knowing the specifics of this solder joint speculation and projection rules. Based on appearance only, it can be a cold solder joint, a solder joint that has been subjected to temperature cycling close to the solder's plasticity area and re-crystalized resulting in an intermittent joint, poor mechanical design of the pcb . Regardless, any aerospace-flight critical electronics and system must have redundancy to assure safety of flight.

So what if the rudder limiter uP failed, the system should simply keep on functioning on redundant systems.

As with most crashes of this nature it is not due to the failure of one single item, it is more often than not it is a chain of failures and actions that result in the disaster.

Significant part of the problem lies in the inherent design of the ScareBus flight control systems and pilot ergonomics. ScareBus flight systems are designed from day one to keep the pilots from abusing the aircraft and take over if the flight system decided the pilots actions violate the programmed flight laws built into these systems.

Commercial airlines were told and convinced by Airbus (ScareBus) that their passenger aircraft cannot get into stall situation due to the inherent design of their aircraft. This results in lesser pilot training that does not require extensive flight stall recovery maneuvers. This lowers pilot training cost and by projection pilot skills and competence .... a significant cost savings and operating cost reduction for the airlines operating Airbus (ScareBus) in their fleet.

Most every aspect of Airbus (ScareBus) designs are about cost reduction sliding on the razor's edge of safety -vs- cost.

Most reprehensible, irresponsible and a total Dereliction of engineering and management responsibly to humanity is the idea and belief that summing the two side stick controls without feedback is acceptable. Airbus ideology on this aspect of flight ergonomics is beyond defective and dangerous. It was for all the wrong reasons.

If one reads the NTSB crash report, one pilot was pushing the side stick down, while the other pilot was pulling the side stick up resulting in a near zero flight control input to the A320 flight control system. This made stall recovery impossible and doomed this aircraft.

What happened was near identical with the crash of Air France 447, same identical failure of the Airbus side stick flight controls total lack of feedback each pilot not knowing what they were doing.

The QZ8501 flight crash report can be found here:
http://kemhubri.dephub.go.id/knkt/ntsc_aviation/baru/Final%20Report%20PK-AXC.pdf

IMO, if possible avoid flying on ANY Airbus (ScareBus) aircraft. The ScarBus name given to Airbus is from those who work in commercial aviation and fly these aircraft daily, that is their professional response to Airbus aircraft.

As for a flight control comparison between Boeing -vs- Airbus:





Bernice

[MT]

If one reads the NTSB crash report, one pilot was pushing the side stick down, while the other pilot was pulling the side stick up resulting in a near zero flight control input to the A320 flight control system. This made stall recovery impossible and doomed this aircraft.

What happened was near identical with the crash of Air France 447, same identical failure of the Airbus side stick flight controls total lack of feedback each pilot not knowing what they were doing.

Long time since i read the report and besides the strange design with the sticks i recall that there was even more to it as the software was responding in wrong way when it was already in a stall position, as the co
pilot mowed the stick forward to increase speed to get out of stall the software set off the stall alarm,
he then mowed back the stick and could not understand the situation. This is the deadly combination of a untrained+arrogant/miss lead pilot and a faulty designed flight control system because he didn't follow
the ancient rules of flying:

Aviate, Navigate, Communicate. In short, fly the plane, everything else can wait.

[Cubdriver]

That does seem like a pretty awful design in that there's no coupling or tactile indication between the two side sticks (on opposite sides of the cockpit, and therefore not readily visible from the opposite seat) to let one pilot know what sort of input the other is applying to the flight controls.  I don't understand how anyone could think this is a good control system.

-Pat

[Howardlong]

This is the deadly combination of a untrained+arrogant/miss lead pilot and a faulty designed flight control system because he didn't follow
the ancient rules of flying:

Aviate, Navigate, Communicate. In short, fly the plane, everything else can wait.

Also in that order. So when the media ask why no Mayday was sent, you can be sure that either the pilot was dead or they were otherwise engaged, and as both a pilot and passenger that is the way it should be. Never underestimate the workload under such situations.

[MT]

Also in that order. So when the media ask why no Mayday was sent, you can be sure that either the pilot was dead or they were otherwise engaged, and as both a pilot and passenger that is the way it should be. Never underestimate the workload under such situations.

?! You mean that co pilot did the right thing when he was criticized in crash report for doing the wrong thing?

[Cubdriver]

Also in that order. So when the media ask why no Mayday was sent, you can be sure that either the pilot was dead or they were otherwise engaged, and as both a pilot and passenger that is the way it should be. Never underestimate the workload under such situations.

?! You mean that co pilot did the right thing when he was criticized in crash report for doing the wrong thing?

Well....  His priorities were correct, it's his implementation of them that left something (ok, a great deal) to be desired. 

-Pat

[Gixy]

That does seem like a pretty awful design in that there's no coupling or tactile indication between the two side sticks (on opposite sides of the cockpit, and therefore not readily visible from the opposite seat) to let one pilot know what sort of input the other is applying to the flight controls.  I don't understand how anyone could think this is a good control system.

-Pat

There is a light and an aural message saying "dual input". Moreover, a pushbutton on sidesticks allow the pilot to take priority. This has been used incorrectly by the pilot who didn'nt keep the button pressed.
As others, Airbus planes are certified and statistics are there to demonstrate they are well designed.

[MT]

Well....  His priorities were correct, it's his implementation of them that left something (ok, a great deal) to be desired.  -Pat

If you implement something incorrectly you are probably not aviating if it leads to a crash. Bonin did several
errors prior stall that caused them to get into a stall in first place. And when was in stall he was engaged in
"other things" while maintaining the stall by pulling the stick backwards while he should had been engaged
in the stall as such, apparently they didn't notice 54 seconds of constant stall warnings! Actually they did
lot's of things they should not have done.

[Cubdriver]

If you implement something incorrectly you are per definition not aviating. He did several errors prior stall that caused them to get into a stall in first place. And when he was in stall he was engaged in "other things"
while maintaining the stall by pulling the stick backwards while he should had been engaged in the stall as such, crash report details this. Actually they did lot's of things they should not have done.

If you have 3 minutes left to live what should you do? Start analyse the blinking light or fast as hell try to
fly again?

Hopefully not pull the nose up into an unsustainable climb attitude in the first place, but having done that a good first priority would have been, if nothing else, at least releasing the back pressure on the elevator to perhaps let the plane return to its trimmed airspeed and possibly recover from the stall of its own accord.  A more positive recovery could have been effected by lowering the nose and adding some power to completely alleviate the stalled condition.

This is not the first time a pilot causes a crash wile engaged into system issues one notorious one are the case of pilots trying to fix a lamp in cabin.

Ahh yes, the L-1011 descending into the Florida swamp after the altitude hold got bumped off while all three members of the cockpit crew were engaged in troubleshooting a landing gear position indicator lamp.  That was one of the first things I thought of when "aviate, navigate, communicate" was mentioned.  No matter what's going on, it's important that somebody flies the damned airplane!!

-Pat

[MT]

was one of the first things I thought of when "aviate, navigate, communicate" was mentioned.  No matter what's going on, it's important that somebody flies the damned airplane!!-Pat

Precisely , when all systems have gone bananas and all instruments are gibberish and all flight lamps are
doing the 1970 disco to the bumped up beat voice of stall warning in the cabin, wings on fire, smoke in
cabin, osmabinladin in cargo bay what do you do as a pilot? Make sure the damned airplane is flying and
land as quickly as possible!