F-16 divide by zero error

[snopes]

Comment: http://www.rinkworks.com/stupid/cs_world.shtml

This story was told by people from Motorola and is supposedly included in
every microcontroller training course Motorola gives.

Test flights of F-16's were being conducted in Israel. The F-16's were
doing low height rounds. On approach to the Dead Sea, the whole navigation
system suddenly reset itself. The daring pilot landed the bird. HQ called
up Motorola and ordered a team on the spot ASAP. The ground tests went
perfectly, but every time the bird went airborn, it rebooted.

The pilots were getting restless. Flying on the border of hostile
territory without navcom, with the Arabs pointing their earth-to-air
missiles at anything that moves, wasn't that pleasant. Neither was
debugging the whole navcom in-flight. Then someone figured it out.

The height of the Dead Sea relative to world sea level is -400 meters. As
soon as the F-16 reached sea level, the navcom did a divide by zero,
crashed, and rebooted.

[Richard W]

That doesn't sound very likely. The software for military applications is meant to be really highly specced and proven using specialised languages and tools before you even start to write the code itself, and divide-by-zero errors are one of the most basic things to check for in critical area code.

I assume altitude gauges (if based on sea level, and hence air pressure, rather than height above the ground) aren't 100% accurate all the time, so you'd get system crashes even when taking off or landing near sea level if the air pressure changed - you wouldn't need to fly to the Dead Sea to find out what happened. And Death Valley is also below sea level, and close to areas where lots of US military test flying is done. Somebody would have noticed long before this.

Are altitude gauges like that even based on sea level / air pressure any more? The height above the ground seems a more important thing to know about in military flying, and that can be measured directly with radar.

[GenYus234]

I'd imagine that military aircraft would have altimeters in both AGL and MSL. The former is important for not hitting the ground, the latter for not hitting other aircraft. Above a certain altititude (12,500 ft?) civilian aircraft use MSL at 29.92 in Hg as their cruise altitude. This way, every aircraft at that level knows that every other aircraft is using the same altititude reading so an aircraft heading east is at the proper altitude to not collide with an aircraft heading west. Military jets heading to a different base would probably fly the same system so as to avoid other aircraft.

[UEL]

Originally Posted by Richard W

Are altitude gauges like that even based on sea level / air pressure any more? The height above the ground seems a more important thing to know about in military flying, and that can be measured directly with radar.

I don't fly a F16.

But I do fly other aircraft. And pressure is still the dominant manner in determining altitude.

Every once in a while, especially after a long time of flying and with variable weather conditions, I'll ask for a radar altitude. A ground based radar will tell me my true altitude, and I'll adjust my reading to my pressure. Then changes in pressure will change my altitude.

In certain areas, pressure altitude is the exclusive altitude measurement due to limited facilities to keep everyone informed of their true altitude.

[overyonder]

I'm not convinced that a F16 would have a divide by zero error, but I was confirmed first -hand by a F18 pilot that some computer systems sometimes do crash and need to be rebooted mid-flight. The pilot can of course continue to fly normally ("by hand") while the computer reboots.

I've also seen Airbus A319/320/321 have their computers rebooted at the gate (for what reason? that, I don't know).

OY

[crocoduck_hunter]

I believe I heard this story in the nonfiction book Tom Clancy wrote about the Air Force back in the mid-90s as an example of the many problems that plagued the F-16's development.

[UEL]

It would not surprise me that some completely basic error fit into the system during development. Expecially when you have different computers from different companies talking to each other. They are not necessarily talking the exact same language.

I have experience with on aircraft whose computer accepted mapping from Canada, but had difficulty recognising it as being from the Western Hemisphere. That is a fundamental error, right there.

Eventually, the errors will all get found out.

[fitz1980]

Originally Posted by UEL

It would not surprise me that some completely basic error fit into the system during development. Expecially when you have different computers from different companies talking to each other.

Sometimes the simplest errors are the ones that you don't catch until you actually field test something. It's possible that such a thing happened during the F-16's field testing. That it happened in Israel might have been an embellishment to make the story more interesting.

Having your navcom constantly reboot over Southern California (Death Valley) doesn't sound nearly as dangerous as it happening over Israel. I have no idea where the F-16 was tested but I'd imagine that it was pretty thoroughly tested here in the US before sending a top of the line experimental aircraft to Israel, a US ally that's (epically back in those days) surrounded by nations hostile towards the US. Not a place I'd want to field test my latest and greatest technology.

[Troberg]

I've seen a variant on that story in an old MSDN magazine (which I, sadly, don't have anymore, as it's probably about 15 years old). In that variant, it was the autopilot system in the F16 that worked perfectly in tests. Then, someone flew across the equator, and the autopilot promptly turned the aircraft upside down.

Now, computer are smart idiots. They do exactly what you tell them, no matter how stupid, so I can see how something like that might happen, but I don't find it likely. Could it slip through the safety routines? Sure, but probably not. Heck, NASA has lost two probes by mixing up . and ,.

Another part that makes this a bit suspicious is that you don't see aircraft flying in the Jordan valley and Wadi Araba. Not even civilian aircraft (except, of course, at more normal altitudes). It's a tense area, and that area is pretty much considered an unofficial no-fly zone. You can sometimes see an occasional helicopter there, but they stick to the sides, keeping well away from the border. Besides, who would like to fly where the AA units can fire down upon you, effectively removing most of your room to make evasive maneuvers? Especially in an area where even an unintentional border crossing might lead to an armed response. Once again, I'm sure it could have happened, but it's not likely.

One thing I know, though, is that many GPS units does not handle negative altitudes well. When I worked in Jordan, among the things we did was to do GPS based road feature inventory, and when the GPS got negative height coordinates, it threw them out as impossible and instead told us that we were halfway out of the atmosphere (about 5 metric mil, 50 km iirc). So, it's not unbelievable that technology will eff up under such circumstances.

[Troberg]

Oh, one side note I forgot: The Royal Jordanian Airforce also flies, among other aircraft, the F16.

I also found this image, of several aircraft of mixed nationalities, among them a Jordanian F16 and a US F16, taken at a mixed nationality exercise. It looks like the photo is shot from the Jordanian side. They are flying about sea level, though.
http://upload.wikimedia.org/wikipedi..._side_view.jpg

[dewey]

Why would you ever need to divide by the altitude? As the altitude decreased your result is going to get larger and larger to point where I would thing it would be meaningless before it reached zero.

dewey

[Troberg]

It could, theoretically, be some internal step in converting what the sensors provide to real altitude.

Personally, though, I would guess that, if this did happen (which i doubt), it would be some internal sanity check on the values that triggered the watchdog, not a divide by zero.

[UrbanLegends101]

Originally Posted by Troberg

One thing I know, though, is that many GPS units does not handle negative altitudes well. When I worked in Jordan, among the things we did was to do GPS based road feature inventory, and when the GPS got negative height coordinates, it threw them out as impossible and instead told us that we were halfway out of the atmosphere (about 5 metric mil, 50 km iirc). So, it's not unbelievable that technology will eff up under such circumstances.

Interesting on the negative altitudes.

I took my Garmin map76 to Lac Assal, Djibouti, a few years back and it had no problem with negative altitudes.

With that said, I didn't have it on when we went below sea level, so I can't say it would not have had problems crossing sea level.

[UrbanLegends101]

Originally Posted by Richard W

Are altitude gauges like that even based on sea level / air pressure any more? The height above the ground seems a more important thing to know about in military flying, and that can be measured directly with radar.

Altimeters in aircraft are pressure reading instruments. And to be accurate, they have to be adjusted to the current local barometric readings. If the local barometric reading is not available, known field elevations, are a useful means of setting the altimeter to adjust for the effects of varying barometric pressure.

At some point in very high altitude operations, such as the space shuttle, pressure reading altimeters stop working.

All pilots should be concerned about height above ground level, not just military, but most pilots, other than crop dusters, spotters and such, are not flying that close to the ground except while landing and take off operations.

Radar altimeters have some use, generally for landing, as they are normally calibrated for zero feet when the main landing gear touches the ground in the landing attitude.

In more normal flight operations, terrain changes make most radar altimeter information more a novelty rather than a real flight application. most very low level military flight operations also have forward looking radar systems which analyze the terrain in front of the aircraft to help maintain adequate ground clearance or even alert the pilot the aircraft cannot maintain low level flight due to large changes in terrain levels.

[UrbanLegends101]

Originally Posted by UEL

Every once in a while, especially after a long time of flying and with variable weather conditions, I'll ask for a radar altitude. A ground based radar will tell me my true altitude, and I'll adjust my reading to my pressure. Then changes in pressure will change my altitude.

In certain areas, pressure altitude is the exclusive altitude measurement due to limited facilities to keep everyone informed of their true altitude.

Tell me how this radar altitude information from ATC works?

I was never under the impression ATC radar was equipped to determine aircraft altitude. Now, if the aircraft has an encoding altimeter system, the transponder will report that altitude corrected to 29.92 inches Hg, and that information is displayed on ATC radar, but if the aircraft has no transponder, there won't be any altitude information.

[E. Q. Taft]

Well, I presume he's talking military ground stations. They may have good ol' "skin-paint" radar that could determine altitude.

[UrbanLegends101]

Originally Posted by E. Q. Taft

Well, I presume he's talking military ground stations. They may have good ol' "skin-paint" radar that could determine altitude.

And that may very well be the case.

Simple geometry, distance to the aircraft determined by radar, up angle of the radar system, do the math and determine height.

But I'd figure even the typical altimeter in a small general aviation aircraft is far more accurate than the height determination by radar, simply because the vast majority of ATC radar systems don't have the resolution in both distance and look angles to determine altitude to a greater precision.

With that said, there is radar which could do it, but those systems generally aren't available to ATC on a routine basis.

I've never heard of ATC giving an altitude check via radar. I have asked for and been given the altitude readout on the radar and compared that with the altimeter in the aircraft. All that does is to show that the encoded altitude readings are close enough to the pilot's altimeter.

In the unusual event the encoded altimeter readings are too far off from the pilot's altimeter, it may give altitude alarms to ATC, showing altitude deviations. I've been asked to turn off the altitude encoding because the difference between the two systems was large enough for continuous alarms with ATC.

[UEL]

Originally Posted by UrbanLegends101

Tell me how this radar altitude information from ATC works?

That's what I get for typing in too quickly.

When I fly, an Air Defence radar will give me my altitude. These are typically ones that will direct weapons into aircraft, so their 3D placement is quite precise.

Apologies.

[UrbanLegends101]

Originally Posted by UEL

That's what I get for typing in too quickly.

When I fly, an Air Defence radar will give me my altitude. These are typically ones that will direct weapons into aircraft, so their 3D placement is quite precise.

Apologies.

OK, well that changes the picture. I've been in US ATC facilities and never got the impression they really could do altitude measurements.

I'll have to do some research into how one might be able to do the same in the US. Are these ADR radio outlets published in general pilot information handbooks?

[UEL]

I don't know about fixed systems. Typically, it tends to be a "mobile" system in the areas I fly**. For example, in Afghanistan, it was furnished by the Australians, and was on a huge trailer.

There may be permanent ones, but I don't know.

**Mobile is a relative term. It may be on wheels, but it isn't moving anywhere fast. Much like a mobile home. A 5 tonne radar is not mobile, in my books.