Have ya noticed all the Pro NASA cheerleaders have NOT answered your post...
..
I have ALOT of Respect for Ron W and MAKG but they keep going on and on about testing, and more testing and even MORE testing......
My question is... How the hell did a simple math conversion get missed in what sounds like years amid hundreds of tests ????.
Well, there's some things you can test...and some things you can't. You test what you can, individually, and try to develop some integrated tests to test as many components together.
Some things CAN'T be tested, because the scale is too damn big. Imagine a landing radar, for instance, that's supposed to detect the surface at a 200 mile range and do something automatically based on the decreasing range. Hard to test that, on the ground. Sure, you develop simulators...but how to you verify that the simulator is accurate? What if it's NOT and you modify your hardware and software to go with it?
"Rocket Science" is a euphemism for something that's extremely hard, for a very, very good reason. It
IS extremely hard, and little tiny things have a tendency bite you in the butt. That's the big advantage manned aircraft have, in that there's an extremely intelligent computer in charge who can use his brain and knowledge to save the day. Not available in space, for the most part.
And stuff happens.
When I was a second looie, I was the on-shift technical expert for a US Missile Launch Detection satellite system. The thermal testing of a new satellite hadn't apparently been sufficiently instrumented, and a portion of a fuel line got too cold and froze. Blooie, there went the propellant and the mission.
Another time, a strong IR star refracted through the atmosphere and the system categorized it as a massive missile launch (I hate it when that happens). Cause? Wrong value of Pi in the computer.
A while back the launch of a new French booster went awry, and they had to push the boom button. They re-ran the telemetry from the attitude control system, and found the guidance commands had suddenly started turned to gibberish. Some bright spark happened to run it through an ASCII converter, and the gibberish spelled, "DIVISION BY ZERO ERROR." Turns out the rate system had been used from an earlier model rocket with less capabilities of the new one.
The fact is, it's impossible to detect everything. But we try. And THAT'S why we test the hell out of everything. Not to verify that the system is perfect, but to check everything that we CAN check.
I've been working in Space System Rapid Prototyping for about the last 20 years. We offer to build faster and cheaper, if the customer doesn't mind taking the risk that the system may not work. Here's our primary guide:
http://www.wanttaja.com/rapid.html
We commonly eliminate steps and tests that conventional systems feel are must-haves. The customers are aware that we're taking risks and things may fail.
Ron Wanttaja