Sensors are highly reliable, especially the high-performance variety used in aviation/aerospace apps. However, human error sometimes interferes. It seems that failure to install angular velocity sensors correctly in the Russian Proton-M rocket resulted in a crash on July 2.
What happened? A source involved with the commission investigating the crash told the Interfax news service that angular velocity sensors were "connected the wrong way round" and delivered incorrect data to the guidance system.
The Russian space agency Roscosmos has not confirmed the cause. A source with links to the commission told Interfax that it is investigating a number of possibilities, including a problem with the navigation system, engine malfunction, and an error in the equipment at launch control.
A Proton-M booster rocket with a DM-03 upper stage and three Russian GLONASS-M satellites sits on the Baikonur Cosmodrome launch pad. (Source: RIA Novosti/Roskosmos press service)
Whether the cause was inverted sensors or something else, three navigation satellites and $1.3 billion of equipment was destroyed in the crash, according to Interfax. Just 17 seconds after takeoff, the rocket rapidly veered off course and tried to correct, but it plummeted towards Earth and exploded on contact only 2.5km from the launch site.
The Voice of Russia radio service said the rocket carried 172 tons of highly toxic heptyl fuel and oxidizer. Interfax said 100,000 residents were evacuated, and efforts are under way to detoxify the soil.
Wait a second--I hope this isn't true, because this is exactly what happened with the Genesis probe. It returned to earth and the drogue was supposed to come out after it had decelerated to a certain speed, but the accelerometers were mounted upside down so the deceleration was interpreted as acceleration and the drogue never came out and the probe cratered in the desert.
We can't necessarily be too smug about this. I seem to remember an American Mars probe that sailed right past that planet because someone put a period instead of a comma in a line of FORTRAN code. I have had friends that worked on the Saturn 5 that left because of the tension of building systems that could never be completely tested prior to a launch. We may be better off today with a combination of more experience and better test equipment, but we are also building more complex systems.
I'm not sure I'd put a lot of stock in an unidentified source and the Interfax news agency. I mean -- these are rocket scientists we're talking about, and they know how to install sensors. In any case, how would they know at this point?
Huge failure is right. One would think that there would be several tests and confirmations in place before actual launch. That's a huge loss of resources based on something that should have been easily caught.
The use of keyed connectors that can only be assembled one way is a good precaution to prevent reversed connections. Of course, if the connector is wired backwards that won't help. The second essential step is to gather baseline data and confirm that the device is operating properly. If indeed the sensor was installed upside-down, it should have been reported inverted data when standing on the launchpad. Exhaustive testing certainly proves its worth if it can catch a tiny error before it causes a huge failure.
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.