[Editor’s note: One of the most intriguing clues in the MH370 mystery is the fact that the airplane’s satcom system logged back on to the Inmarsat network at 18:25. By understanding how such an event could take place, we can significantly narrow the range of possible narratives. In the interest of getting everyone on the same page in understanding this event, I’ve asked Mike Exner for permission to post the content of a detailed comment he recently provided. One piece of background: a lot of us have been referring to the satellite communications system aboard the aircraft as the “SDU,” but as Mike recently pointed out in another comment, it technically should be called the “AES.” — JW.]
Until we have more evidence to support the theory that the loss of AES communications was due to the loss of primary power to the AES, we must keep an open mind. Loss of power may be the most likely cause (simplest explanation), but the fact is we do not know why the sat link was down between 17:37 and 18:25. My reluctance to jump to the conclusion that it must have been due to the loss of primary AES power is based on decades of experience in the MSS (mobile satellite service) industry. It’s not just another opinion based on convenience to support a theory. Let me elaborate on a few possible alternative explanations.
The potential for loss of the pilot carrier, due to the orientation of the aircraft in relation to the satellite, was increased as soon as the airplane turned WNW. Between the time of this turn (circa 17:50) and the time of the FMT (final major turn circa 18:25-18:40), the aircraft was flying more or less toward the satellite where the antenna pattern was near a null. Don and I have both looked at the antenna pattern in some detail and concluded that the antenna pattern and coincidental direction of flight were unlikely to be so bad that the pilot carrier would be lost due to this geometry. Moreover, according to a MAS Press Conference on March 20, 2014, there should have been an ACARS message transmitted at 17:37, but none was received. ( bit.ly/QFbF6C ) At 17:37, the aircraft was still over Malaysia SW bound, so the HGA pattern would not have been an issue at that point. Taken together, loss of the pilot carrier due to antenna orientation appears to be a possible, but unlikely explanation for the outage.
Ionospheric scintillation has also been suggested as a possible explanation for the loss of service during this period, but there have been no reports of other aircraft in the vicinity suffering a loss of service, so this explanation is also unlikely. (Note: Ionospheric scintillation in the equatorial regions can be a big problem for VHF and UHF communications, but it does not affect communications in the L band as much.)
The MCS6000 AES, located in the back of the airplane, requires a continuous feed of INS data (position, speed, etc.) via an ARINC 429 link from the computers in the front of the plane. If the AES stopped receiving INS data for any reason, then it would not have been able to steer the HGA, or compute the required Doppler corrected transmitter frequency. Thus, it is very likely that the AES would be out of service if there was any loss of this 429 data link, or the information carried over the link. Given that there was no VHF voice communications after 17:19:24 and the Transponder Mode S data was lost after 17:21:13, it is certainly possible that the INS data flowing to the AES was disrupted due to a common failure in some piece of equipment in the E-Bay. This explanation for the loss of service cannot be dismissed as easily as the two previous theories.
However, there is one additional observation that tends to favor the loss of primary power theory over the loss of INS data theory (or the other two theories above). We note that when the AES logged on at 18:25:26, the BFO values for the first few minutes thereafter appear to have been drifting in a way that is more consistent with a restoration of primary power event than a restoration of INS data event. If the AES power had been on during the outage, the oven controlled reference oscillator would have maintained a stable frequency and there should not have been any significant BFO transients following the 18:25:26 logon.
In summary, there are multiple alternative explanations for the AES outage, but loss of primary power is the most likely explanation. Like so many other necessary assumptions, like the mode of navigation after the FMT, we have no choice. We must base the search on the most likely assumptions while maintaining an awareness that few of the assumptions have probabilities of .999.
Don,
T2 is one of the Thair radar sites Duncan had mapped on his blog using your supplied info.
As I was overlaying the radar sites onto the Lido image, it became fairly clear that you are correct in both of your assertions that T2 is the source of the Lido image, and also that the blind spot from the adjacent mountain is the cause of the gap.
Flitzer-Flyer
I was mainly querying the post reboot numbers. I don’t take the view that there was some malfunction, I see someone twisting knobs and doing things not exactly in the manual, maybe even plugging in foreign devices with who knows what on it, for whatever reason. Is it not a concern that the plane ostensibly changes direction at the reboot? I know some are absolutely determined to paint the reboot as incidental but I doubt it. Would anyone go as far as guarantee they(BTO’s) are not corrupted?
Orion,
Thanks for the clarification on your image: an ‘upvote’ for my conclusion that the RTADS-III network is the source of those RADAR targets.
We have some further detail with terrain profiled RADAR ranges from the head sites & environment data from the night 7-8 Mar ’14 which also supports the conclusion.
:Don
@Matty – Perth: I don’t think anybody can say with certainty that that either the BTO or BFO values are correct, especially after the login. However, if we are going to question the satellite data, there is a higher probability that the BFO values are not correct, either because they are corrupted or we are misinterpreting them. Remember: it is the BFO model that was used to eliminate the possibility of a northern route.
It is interesting to note that while MH370 remains missing after 11 months, and it took days to locate the crash site of QZ8501, GE235 was located “instantly”. GE235 ADS-B data was available within minutes after the crash, and showed the path down to 200 feet. Multiple dash cam videos were available on line and on the news channels within a few hours. And even though all 3 flights ended in the water violently, there were survivors in the case of GE235, but not the other two cases. (I realize that we don’t know 100% for certain the fate of MH370.) I think this is more evidence that real-time tracking and rapid response is key to maximizing the likelihood of survival. There may have been some survivors (initially) in the case of MH370 (unlikely), or QZ8501 (a little more likely), but we will never know because no first responders got to the crash sites quickly. Of course, that would not have happened in the case of MH370 anyway, assuming it went down in the SIO, but it could have happened in the case of QZ8501 if the ADS-B data had been transmitted in real-time all the way to the surface, and made available in real-time, like GE235. If QZ8501 had been tracked down to the surface, planes and helicopters could have been on the scene in a few hours instead of a week.
I think you’re right on Mike. If I recall correctly, there was also a slim possibility of survivors in AF447. The water temp was high enough for survival for several hours. You also have the hijacked 767 that ran out of fuel and crashed into water, also with survivors because it was witnessed from the beach. Locating time would seem to be a major distinguishing factor among the five accidents.
Matty: The BTO numbers after the reboot are still very consistent with independently calculated airspeeds and independently calculated fuel endurance. The method of determining the BTO and logging it at the Perth earth station hasn’t changed just because there was a reboot.
JS: Nope, it doesn’t make sense to me. X = known distance ? From where ?
The BTO is a function of the LOS distance from the Perth earth station to the satellite, then to the aircraft and back again to Perth. Even if the aircraft is stationary on the ground then the BTO will change as the satellite moves. Translate that into a radius from the sub satellite position [at a particular time] on a non-spherical earth and I think your simple linear Y=aX+b goes out the window.
new thread:
http://jeffwise.net/2015/02/05/guest-post-transasia-235-crash-timeline/
Flitzer-Flyer
I thought the only way to confirm the BTO’s would be to know the flight path? Victor summed it up in the end – we just can’t be certain, especially after the login – and this has stalked my thinking from the outset, albeit a gut feeling. The plane/pilot were in sneaky mode yet there is absolute trust in the signals it so strangely left behind. I hope it is found but it could be slightly shattering if the numbers just don’t add up in the end.
@Flitzer – it never goes out the window. There is a linear relationship between the distance between the satellite and the BTO, and you can easily get it from page 55 of the August ATSB report.
Using Y=aX+b:
Y represents the BTO
a is equal to 1 over 299791, which is the number of seconds it takes light to travel 1km.
X is the r/t distance between the satellite, Perth, and the plane
b is the “nominal” terminal offset of -495679us or .495679 seconds..
So, at 16:29:52, the formula looks like this:
BTO=153068km/299791-.495679s
Multiply BTO by 1million to get microseconds instead of seconds. The result is 14903, while the measured value was 14920. Sure, it’s close, except that the entire variation in BTOs for those 17 distances is only 200us, and there is no pattern as to which ones are higher than expected it lower than expected. At 16:29:17, the formula also predicts 14903, as expected since the plane didn’t move, but the recorded value is 14860. One is too high, one is too low.
That high/low discrepancy is what makes the correlation weak. By weak correlation, I’m saying that a formula, based on speed of light, that should hold true across all 17 doesn’t predict the correct BTOs. You could just as easily prove that BTO is 14840 no matter what, with a random variation of 100us.
There can’t really be any debate about whether they correlate or not. The debate would be over why they don’t, and some of the theories are pretty solid – for example the signal bouncing off a wall or simply variations in the circuitry timing. But, absent a BTO value for a known location far enough away, we can’t test the formula. We can use it, but we can’t prove that it’s right based on the numbers alone.
Out of respect for the thread, I’m going to let this go at this point and if we disagree, we can leave it at that.
Sorry to double post, but let me back up Matty’s point.
Part of the suspicion on the BTOs is that, to my knowledge, not a single BTO has ever been released for a known location other than KL. Not for any early flights, not for any other plane, nothing.
The sole released set of BTO data *in the universe* consists of 17 values from the same location, which range from 14740 to 14940. This, despite probably BILLIONS of BTOs being recorded in the last four years.
Just in case anyone is confused on this point: the ATSB and the IG both regard the BTO values, and the ping arcs derived from them, as among the most rock-solid data we have. JS feels otherwise. Let’s leave it at that.
JS:
Re “not a single BTO has ever been released for a known location other than KL”
This is not true. We have used all available observations out to the end of the available ADS-B positions for validation, and calibration. The BTO and BFO data after takeoff are spot on for all those known positions out to 1707.
In addition, there are hundreds of BFO and BTO values, not 17. (ATSB used 17 BFO observations for their calibration. The IG has used hundreds.) BTO observations vary from 11500usec at 1941 to 18400 usec at 0019, a change of 6900 usec. I suggest you check your assumptions, data and work. They make no sense to me.
Can anyone in the IG replicate Mike Chillit’s 3F1 position values? I’d thought possibilities were geodetic conversion (likeliest) and rounding (less likely).
If the IG can provide me with a spreadsheet that computes things his way AND your way, I’m happy to invest the time to work with him to sort it all out.
Thanks.
Where has Mike Chillit been getting his satellite data from ? Makes absolutely no sense. What has he been smoking?
Point of Information: This will be old news for most, but some may benefit from the following reminder.
BTO values (and the BTO Bias Calibration) are logged in the units of µsec. It is important to note that the BTO observations, ranging from 11500 to 18400 µsec, represent the round trip propagation time (less the bias) from Perth back to Perth via s/c and AES. Thus, the round trip path times must be divided by 2 to yield the single path times. The BTO measurement resolution is 20 µsec. That means the minimum change (like the LSB in a binary number) in the one way distance that the system can record is 3km or ~1.5NM. This also means the uncertainty in any given observation, due to the 20 µsec resolution, is ±1.5 km, or ±0.8NM. (There are other sources of uncertainty in the arcs, and geometric magnification of these values in terms of arc accuracy, but the BTO measurement system resolution limits the raw LOS BTO uncertainty to ±1.5 km, or ±0.8NM, assuming there are no other errors.)
For the same reasons, the sensitivity of the BTO Bias to measurement limitations is reduced by ½ once the round trip values are converted to single path values. Thus, an error of 20 µsec in the BTO Bias corresponds to a one way path error of only 3km, or ~1.5NM.
@Brock: I am emailing you a small spreadsheet that converts the ECEF coordinates presented in the ATSB report to LLA coordinates that are plotted against the subsatellite track shown in the Inmarsat paper (Ashton, et al.). The agreement is excellent.
I have no idea how @MikeChillit gets his results or the basis of his claims.
https://twitter.com/MikeChillit/status/563476064052465666
JS: Just 17 values – – Rubbish!
Check out the position accuracy of the BTOs at 16:43, 16:55, 17:07. Seems pretty good to me.
@Flitzer – the ATSB used only the 17 to calibrate, as far as I can tell. I concede that there are additional points out to 17:07 and those locations may be known with some accuracy. I decline to consider any points after 17:07 as known locations for the sake of validating the formula.
As Jeff indicated, we’re going to have to agree to disagree.
Don,
Thanks. You said “The aircraft has 5 separate means of voice comm: 3x VHF, 1x HF & SATCOM. The VHF radios are under his seat in the MEC with one antenna above his head & the other on the aircraft belly; the remaining systems rely on, at least, antennas aft of the wings.”
If you have this knowledge, could you explain a bit more about the design: where precisely antennas are located, what are sources of the power, where the 3 VHFs are located and who has access to them? It would be great if you could point out relevant documentation.
Btw, not sure if this was ever discussed: 3 VHFs + 1 HF were also silent. Assuming AES was disabled on purpose, what actions are needed to disable 3 VHFs + 1 HF?
Oleksandr
The VHF & HF radio LRUs are in the MEC (Main Eqpmt Ctr) also termed EE Bay. I previously posted a URL, or the components of one, to an image showing the VHF antenna locations. The HF antenna is embedded in the leading edge of the vertical stabiliser.
No need to disable radios, just don’t push the transmit button. However, the VHF & HF radio circuit breakers are on the flight deck overhead panel (the AES breakers most definitely aren’t) distributed across the L&R busses.
:Don
just be aware – battle for hearts and minds…
http://rt.com/op-edge/229851-uk-army-internet-culture/
Oleksander (& Niels),
A correction, the VHF antenna fitment for 9M-MRO is as shown in this excerpt from the AMM:
i.imgur.com/yWIaFWW.png
I wanted to show the aircraft maint manual’s diagram where the specific antenna type is called out, rather than one of many images available of 9M-MRO where the antennas are visible. Cross-ref to this image, bit.ly/BallPort . The AMM reference I posted for Niels is a early alternate fitment.
So, one VHF antenna located fwd upper centreline and other two VHF antenna top & bottom centreline of fuselage aft of wings. The convention, but by no means a limitation, is for L-VHF to be tuned for ATC comms; R-VHF for air-air & emergency chls; and C-VHF for Data.
:Don
Don,
Thank you for your response. As a follow-up on the radar coverage, I was interested to see if the terrain profile would give better insight into the altitude of the plane during the ‘Lido gap.’
As I’m sure you are well aware, I discovered that the gap is caused most likely by the radar range limitations, atmospheric conditions, and the altitude of the airplane- and not the mountains.
The following section diagram illustrates this concept using the range data provided for the T2 radar installation, overlaid on top of the terrain profile:
https://www.dropbox.com/s/hn45uhb89mhobun/T2.pdf?dl=0
If the plane had been flying low enough to be in the mountain’s “shadow” then it would have been flying too low for the radar to pick it up at all.
Interestingly, if you assume the radar coverage to be 100% of what is listed, and no atmospheric interference, then at 39,000 feet the plane appears to skirt along the radar profile’s boundary.
Do you think this “skirting” could be significant enough to cause the approximate 5 min gap in the 25 min radar trace?
Consequently, do you feel the range of the radar is actually higher than what is listed?
Wouldn’t the return signal of the radar theoretically be the strongest at the middle of the radar signature segment crossing through the profile- right where the gap is?
Did you find any atmospheric anomalies from that night which would cause the gap in the radar trace?
@Orion:
Given the location of T2 radar, did it pickup MH370 from IGARI as it supposedly turned around then flew back over Malaysia before seeing MH370 again in the Straight of Malacca?
Don,
Thanks a lot for the diagram. Now it’s getting clearer.
Well, at a glance L-VHF, C-VHF and HF antennas could be hit/smashed by a piece (or pieces) of debris originated from the cockpit area. The same debris could also result in a structural damage of the stabilizer.
A few more questions:
– What is the location of the satellite antenna?
– Is there any access to VHF & HF outside of the cockpit for the rest of the crew? I presume crew may have parallel access to VHF for safety reasons.
– You said R-VHF is primarily used for air-air emergency. This is the one, which has the antenna on the belly. What source of the power does it use: L, R, or something else?
Is there any association with the two power buses losing AC due to if an engine is restarted mid-flight? if the left engine is shutdown will power on the left side bus also be lost until the engine is restarted. Same for the right side, will the right side power bus will be off while the right side engine is restarted?
Orion,
Pls. check the recent map kindly shared by Don in this thread.
One of the most plausible explanations of the gap is that it is an artifact of data merging from two different radars. Previously I expressed opinion that one of them could be Indonesian Lhokseumawe; Don suggested Thai RTADS-Phuket and RTADS-Khoh-Muang (instead of T2) which is also very reasonable choice.
This does not, however, eliminate “low altitude” explanation, which still should be on the table in my opinion.
LouVilla,
Thanks for sharing your thoughts, but I think your explanation is highly unlikely. There are many reasons:
– Disabling ACARS would suffice.
– Not disabling anything would cause even a stronger effect. Even if interceptors are launched, then what?
– What about Twin Towers as a target?
– The 7th arc is huge; BFO only work in conjunction with some theory (=assumptions) with regard to flight mode, such as AP. There is a chance that nothing will be ever found. Why would not they leave more hints where to search if this was an idea?
– They did not leave any message whom to blame.
– I am not very clear how would such an act demonstrate the incompetence of Malay government? It could be any other country.
@Myron:
“@Orion:
Given the location of T2 radar, did it pickup MH370 from IGARI as it supposedly turned around then flew back over Malaysia before seeing MH370 again in the Straight of Malacca?”
GREAT QUESTION.
Oleksandr, myron, Nihonmama
It would appear that the Khoh-Muang is the most logical of the three sites. (Green ring)
https://www.dropbox.com/s/erk8dsrrgh3qmz2/3-radars.png?dl=0
The gap at Penang is because the radar sits directly below, and it’s range limits appear to correlate well to the end of Lido as well as almost to IGARI.
Both of the others should have revealed more about the segment up to and including FMT, imho.
But still, the question remains about the cause of the Lido gap.
Orion,
RADAR: my conclusion remains that the data came from the RTADS-III network in Thailand. There is radiosonde data from Phuket showing a temperature inversion over the sea that night. The three RADAR heads are integrated with a C2 system. Detection by Ko Samui (T2) would be at the edge of its range. The recorded pattern of the targets, first fix – last fix and the distribution east – west, certainly does not fit Western Hill nor the two TNI-AU sites.
What continues to intrique me is the obfuscation, delay, confusion, contradiction of the Malaysian authorities regarding RADAR sighting over the Str of Malacca culminating, two weeks after the loss of 9M-MRO with production of the ‘Beijing Lido’ image to confirm the track out over the Str of Malacca.
There was clear political interference from Najib’s office (PMO) emphatically contradicting the Chief of Air Force to state that nothing had been detected on military RADAR.
If the PMO had kept out of it I could accept a case of embarassed inattention.
Oleksandr,
The photograph I linked shows the SATCOM antennas. The ‘patch’ above door 3 is one of the two side mount HGA apertures, the other in same position on the other side. The SATCOM LGA is the third fwd of the vert stab. No access to VHF or HF radio service outside the flight deck. The SATCOM system can provide connectivity for “passenger amenity” with in seat handsets via the IFE system, I don’t know if that facility still existed on -MRO. Regardless, there were no AES initiated call attempts.
Power: if an IDG (an engine or APU gen) fails, the control systems automatically close the bus ties. The 777’s design evolved from the 747-400 & 767 automation which rendered the flight engineer redundant. One engine’s IDG can power the aircraft, the system manages the distribution, without pilot intervention and with minimal (10’s of milliseconds) interruption.
:Don
@don- power. What if both engines fail would that cut the power to both busses and satcom equip shutdown until one or another engine is restarted. The acars reported under performance issues. Then acars ended near IGARI.
Flitzer-Flyer
“The BTO numbers after the reboot are still very consistent with independently calculated airspeeds”
I thought we didn’t actually know the airspeeds? As Jeff says the BTO’s are the best data we have and the bedrock of the analysis, in fact it was impossible to move without a total buy in there. But if they can be faked then they can also be corrupted. To use Jeff’s words previously – a scaffold of assumptions – and taking them off the table as some would do is looks like a bias.
I am constantly sat back by the level of technical expertise on display here, but in equal proportion there is a tendency to discount inconvenient possibility and it goes back to another assumption: That MH370 was some kind of incident and not a crime.
Crunchers are saying trust us – but less and less do at this stage. Mike points out that there is an ignorance at it’s core and there is truth there, but there is also some arrogance on display. The public aren’t looking at a spreadsheet, they look at the panorama. Victor’s dichotomy is spot on – either it crashed in the SIO or someone made it look that way – and science dictates that you explore both, sooner or later, and stop assuming that everyone else is stupid when we probably have the knowledge right here on this blog to engineer a spoof on a minimal budget.
Thank you Don for the explanation of the integrated C2 network, the information regarding the temperature inversion, and the interesting observation on the target pattern and distribution. As the section diagram indicates, it would appear that the plane would have to be at significant altitude to even be at the range edge of Ko Samui. It is hard to imagine that nearly a year later the water is still as clear as mud.
“Matty: “independently calculated airspeeds”
Jef covered a number of these in his much earlier chapter on – – “How we can tell how fast MH370 was flying”. There are many pieces of evidence which build a picture on likely speeds. Independent approaches to this have produced speeds which are very consistent with the ping rings and with B777 flight characteristics.
@Flitzer
The IG and ATSB use of the word likely is getting a bit tiring. It is a vague and ambiguous term, and is used to disguise the fact there is a great deal of ambiguity in the SIO hypothesis – not only in terms of the data itself, but in terms of the assumptions using the data. Not to mention that it ignores both motive and lack of debris.
Recent news seems a bit slow. It is nauseating (sea conditions) in the SIO. I am sure the searchers wished you had guessed a more pleasant place.
Dennis – “likely air speeds”
It might take some discipline to eradicate that word?
The word “likely” has a clear meaning to mathematicians, engineers and scientists. It is not vague. Sorry if you don’t understand.
@ALSM
I am glad you are sorry for something. It is a good sign. I would suggest reading some Taleb, boring as it can sometimes be.
Did a lot of mathematicians, engineers and scientists forget they were dealing with a crime scene?
The word “likely” is EXTREMELY vague. It has two distinct common usages – colloquially as a statement of “absolute probability” (i.e. >50% chance), and technically as a statement of “relative probability” – often modified quantitatively (more/most likely, maximum likelihood) – in which >50% chance is (ironically, given the modifiers) NOT implied. Think of the statement: “If you roll two die, the most likely total is seven – but six and eight are also quite likely”.
Unfortunately, I’ve seen MANY technical writers use this word without sufficient context, leaving it to the reader to guess.
My rule of thumb is that if an intended audience infers the wrong meaning, it is the WRITER who is to blame.
***
But this is a quibble. More importantly: I will continue to berate anyone who berates the IG for their [s38, e89] estimate. Not only did they do their best, but their best was (IMO) extremely well-supported by the (dodgy?) data they were given to work with – though FWIW I’ve preferred Dr. Ulich’s approach since August, and note the ships seem to be almost SNEAKING ever closer to his spot (do they think they can add 4 degrees longitude to the priority zone without anyone noticing? Not…likely.)
@Matty
Those people generally have difficulty finding their way out of their houses in the morning. Don’t ask difficult questions unless it relates to BFO or BTO. Then you will get an ear full.
@Brock
You are being very kind. I have a long history of berating the IG, and for good reason. The assumptions relative to the flight mode after the FMT are outrageous IMO. I cut them no slack for very sloppy thinking.
@Brock
BTW, i have an SIO solution, put out there before the ATSB. Everyone who is someone has one (an SIO solution). I will post it if you want, It is North of the IG consensus.
Mea culpa. How about “plausible speeds” instead.
Matty: Where is the evidence for a crime scene?
Dennis: Has anyone found any debris anywhere ? If not, then where do you start looking ?
@Flitzer_Flyer
The lack of physical evidence does not prove the lack of an criminal act. And to not treat it as a crime scene an “UN-scene” crime scene would be irresponsible. An eight hour “accident” or “Hijacking” doesn’t hold water, be it the SIO or otherwise.
Flitzer-Flyer
I guess the last we heard from the investigation(bloody long time ago) they were treating it as a deliberate diversion and had put up the shutters? I think even when it was still on the ground, the likelihood that there was going to be a problem with the plane was low, and there is nothing to suggest even now there was anything wrong with the plane. There is plenty to suggest that whoever flew it performed a series of actions that were not quite out of the book. Even low rent crims use counter measures, sophisticated ones go a lot further, and misappropriating a 777 would require the some cover unless you had some immediate use for the plane. If you took signal data off the table what other options would you have to mislead an investigation once off radar? We know?? that someone did indeed misappropriate a plane.
You could throw a heap of life jackets out the door with a few beacons but that is quite an undertaking and would start to smell too. The perpetrator did not want to be found? Will the data lead us to the plane or will it just be a big boost for oceanography? I recognize there is a bit to go but it isn’t encouraging either.