Air Asia Indonesia Lost Contact from Surabaya to Singapore

[Bergerie1]

You can add me to the list of non-believers too. As I recall, the max certificated altitude of a VC10 was 43,000ft. When I was a young pilot, the training captain I was flying with, took us up to a little over 46,000ft on a training flight at a reasonably light weight in smooth air and with a good horizon, to demonstrate the high and low speed buffets. There was a margin of only a few knots between the two and one had to be very delicate on the controls when edging towards each limit.
It was a non-standard exercise but taught me never to go there in rough weather. Limitations are there for a reason. Adding a margin in bad conditions is good airmanship.

[Volume]

This may be the case at present but I expect that the various tracks will be digitally filtered in due course. So that actual voices can be interpreted.

Fortunately there is already ready-made sophisticated audio spectral editing software for this, e.g, Izotope RX4. It easily allows a skilled forensic analyst to null out specific spectral patterns such as alarms which obscure speech.

Are these statements still true for modern, digital CVR? For the analogue recording I believe that you can filter a lot, because "it is all there" but buried, in a digital recording there is nothing between the bits. It is a bit like all the higher frequency events not recorded on the FDR with 1 second sampling rate. An oscillation at 1.1 Hz is recorded as an oscillation at .1 Hz as the real frequency is masked by the sampling rate.
What is the sampling rate for digital CVR? How does it match with computer generated sounds (alarms, chimes, cavallery fanfare...) which are also having their own sampling rate or rectangular waveforms. I never worked with digital audio, do you also have some frequency changes similar to my FRD example, so a sound of 42 kHz sampled at 41 kHz is recorded as a 1 kHz sound?

[marchino61]

Humans Can't hear ultrasound

What is the sampling rate for digital CVR? How does it match with computer generated sounds (alarms, chimes, cavallery fanfare...) which are also having their own sampling rate or rectangular waveforms. I never worked with digital audio, do you also have some frequency changes similar to my FRD example, so a sound of 42 kHz sampled at 41 kHz is recorded as a 1 kHz sound?

I think you'll find that even young human ears can't hear anything above 20 kHz. CDs are sampled at 44.1 kHz, over twice the highest perceptible frequency to humans. This ensures lossless recording. This is shown by the Nyqvist-Shannon sampling theorem.

I'd imagine CVRs would use the same principle.

[snowfalcon2]

Volume
The Nyquist theorem says that with a sampling rate of F Hz you can faithfully reproduce signals with a frequency up to F/2 Hz. So for example a CD is sampled at 44 kHz and it reproduces sounds up to 22 KHz. This is hi-fi sound quality.

I found a Honeywell CVR/FDR manual which states that the CVR inputs are sampled at 8 or 16 KHz and thus can record sounds up to 3.5 or 6 kHz. (This gives some margin to the "Nyquist limit"). This is "telephone line" sound quality. I was slightly surprised that the sampling rate is so low. Probably a legacy from times when solid state memory was much more expensive.

Edit: I might add that in order to avoid sampling errors like you describe (aliasing), almost all digital systems use techniques such as a low-pass anti-aliasing filter before the digital sampler, or use oversampling i.e. a sampling rate several times the signal frequency.

[Airbubba]

You don't need to be a lawyer to figure out there is no requirement to wear a headset at anytime.

You don't need to be a pilot to read the regs :

FAR 121.359 (g):

(g) For those aircraft equipped to record the uninterrupted audio signals received by a boom or a mask microphone, the flight crewmembers are required to use the boom microphone below 18,000 feet mean sea level. No person may operate a large turbine engine powered airplane or a large pressurized airplane with four reciprocating engines manufactured after October 11, 1991, or on which a cockpit voice recorder has been installed after October 11, 1991, unless it is equipped to record the uninterrupted audio signal received by a boom or mask microphone in accordance with § 25.1457(c)(5) of this chapter.

14 CFR 121.359 - Cockpit voice recorders. | LII / Legal Information Institute

Maybe your boom mike doesn't have a headset attached.

These are the U.S. requirements, they often filter down to other countries over time.

Now the selection of speaker and hand mic, above whatever, seems to be favoured by that group of F/Os who like to appear "cool" whilst reading the paper/Ipad or whatever THEY think is their function in the upper levels.

Yep, I think I commented earlier on this thread that many FO's try to look like they are doing everything but flying the plane while they are in the seat. I guess it's a generational thing.

[crippen]

Snow

This is "telephone line" sound quality


these are usually sampled at 8 k/c/s

[Mahatma Kote]

the CVR inputs are sampled at 8 or 16 KHz and thus can record sounds up to 3.5 or 6 kHz.

You get excellent quality sound at 8 kHz sampling (300 Hz to 3kHz, not 3.5 kHz). 8 kHz 8 bit uLaw or aLaw is the standard for landline telephones and as very reader knows that's more than acceptable (NB mobile calls are far lower quality). Going to 16 kHz gives a higher upper frequency ( 7 kHz ) but in general that's only useful for engineering analysis rather than voice clarity.

Where there is a big difference is the number of bits in the sample. This affects dynamic range and so say with 16 bits you can hear clearly the slightest whisper as well as the loudest bangs. 16 bit 44.1 kHz sampling is also well beyond the usual human dynamic range which is why it's used in CD recordings (fake golden-ear pedants not withstanding).

All of these figures are pretty irrelevant these days. 8 bit 8kHz sampling is 64 kilobits per second. CD quality is 705 kbps - a factor of about 10. This is compared to the storage increases over the past few years of factors of thousands to millions.

[FlightDream111]

May help

http://www.faa.gov/other_visit/aviat...Operations.pdf

[mcloaked]

@FlightDream111 thanks for that link - also useful is the FAA upset recovery link at FAA TV: Airplane Upset Recovery, Part 2

[8/8ths Blue]

HarryMann


At the time I read your post re equal opportunity I did think to myself that I may be entering a bear trap considering your usually erudite posts. I suppose the image I had in my mind of 8501 being in a tight high speed spiral at FL240 with perhaps a disoriented crew attempting to make sense of it all and the debris pics to date.

[Volume]

Unless this position means the actuator is fully extended (no idea which way it is installed) and the piston rod is bent and hence the rudder fixed in that position, I do not think that the current position of the rudder tells us anything else than the direction of gravity.

[captains_log]

THS

I see they recovered the tail section, no doubt the THS screw or assembly should give some indication last position.


Something like this:

[FullWings]

Someone said that would be nice to overlay the weather over the trajectory plot before crash.
See in link below an approximate, the time is 23:00 UTC and the flight path is the small yellow curve
http://www.mediafire.com/view/xxaocp...5e/weather.jpg

Nice overlay but that's a false-colour infra-red satellite picture rather than a radar plot. The satellite picture will tell you the temperature of the cloud tops, which corresponds roughly to their altitude but it won’t show areas of precipitation or strong air movement.

The area where the Air Asia went down, purely looking at that picture, could be an active cell going from 1,000’ up towards the tropopause or it could be the remains of an old CB anvil/blow-off at 45,000' with nothing underneath. Given the gyrations of the aircraft, it’s probably going to turn out to be the former but you couldn’t say from that picture.

[joema]

Originally Posted by joema

Fortunately there is already ready-made sophisticated audio spectral editing software for this, e.g, Izotope RX4. It easily allows a skilled forensic analyst to null out specific spectral patterns such as alarms which obscure speech.

Originally Posted by Volume

Are these statements still true for modern, digital CVR?

Yes, the audio bandpass doesn't greatly affect the ability to forensically extract speech from a cluttered audio environment. The technician visually edits audio spectrograms using software designed to identify and remove unwanted noises. It is vastly beyond tweaking an equalizer.

It is very sophisticated -- usually a cell phone ringing during speech can be removed so perfectly there is no significant artifact remaining.

The software already exists so it's not like they have to engineer something from scratch. In both music/video production and advanced audio forensics for law enforcement, audio technicians are highly trained using this software and accomplish amazing results.

[Nemrytter]

Nice overlay but that's a false-colour infra-red satellite picture rather than a radar plot. The satellite picture will tell you the temperature of the cloud tops, which corresponds roughly to their altitude but it won’t show areas of precipitation or strong air movement.

Satellite images can give reasonable estimates of precipitation. This one, however, can't
But looking at the raw data from that time there was quite a lot of heavy precipitation in the area - particularly to the Upper-Left of the crash site.

The area where the Air Asia went down, purely looking at that picture, could be an active cell going from 1,000’ up towards the tropopause or it could be the remains of an old CB anvil/blow-off at 45,000' with nothing underneath. Given the gyrations of the aircraft, it’s probably going to turn out to be the former but you couldn’t say from that picture.

It's the former - and in some places the cell had busted several thousand feet above the tropopause.

[McWho]

Is it correct on an A320 that you only have 2 ELACs and 3 SECs? Which set contains the normal law intelligence?

There is a comment on another forum that uncommanded rudder is suspected. But even if the rudder was at full deflection could this not be counteracted and control maintained through engine power and aileron input?

Just a speculation then from information I have read about the flight computers, if there were multiple sensor failures, say the pitots iced up, you might expect automatic computer failover and ELAC 1 would fail over to ELAC 2, if there was a fault with ELAC 2 as has been suggested may be a possibility with the maintenance log, would you then be left with 3 SEC computers which have no aileron authority?

[lomapaseo]

I don't have an argument with the tech talk about CVR audio freq analysis and filtering. It could be useful to decide if an alarm or other continuous noise has occurred.

However the real challenge is interpreting the words of the crew as spoken in their language and/or dialect. It's not just a fixed frequency question but a combination of sounds. Mess with that an you may never agree on the words in spite of the best guesses of "it sounds like"

[MrSnuggles]

There is a comment on another forum that uncommanded rudder is suspected.

Aaah, so that's why some posters ([3, 5[) have mumbled about rudder trim and stuff like that.

Well, I have been following this discussion here and no, only for those rudder trim guys noone have mentioned such gossip. But sure, this is for rumours too, as it says in the title.

Uncommanded rudder scenarios happened on a few Boeing 737's many many years ago. Hydraulic parts froze at altitude and when hot hydraulic fluid entered the part it jammed and caused a rudder hardover and even rudder reversal.

However all the scenarios where this happened were uneventful flights where the rudder all of a sudden slammed to one side and the plane nosedived straight into the ground. We are not looking at that kind of scenario here.

[toffeez]

McWho

A320 roll control is provided on each wing by one aileron, augmented by four roll spoilers.

[McWho]

There is a comment on the same forum that a loss of flight computers can result in the uncommanded rudder deflection on the Airbus.

Thanks for the comment about the spoilers, I had not realised they could be used for roll control which would imply that the SECs can provide this control in the event of ELAC failure? I did read however that the ELACs were the only source of aileron control on the Airbus.