Apollo Guidance Computer History ProjectSecond conferenceSeptember 14, 2001
Relations with GrummanJIM MILLER: The mission that I was involved with, which was subsequently called
Apollo 5, though we called it 206, was the unmanned first flight of the Lunar Module. But in the event of the flight, things began to go badly at the first scheduled burn. Although we had simulated it lots of times, when the command went to the engine, unknown to me, as the mission-program guy and as the former simulation guy, there was preliminary pressurization of tanks that took place between the fire command and when the engine actually lit up. I had been told by the person that wrote the descent burn software that it had to have a very narrow window in the startup for some reason or other - I don't remember what. We had a fairly narrow window in there. And, because of the timing of the interrupt that governed the accelerometer monitoring for ignition, it just missed reaching the delta-V threshold by milliseconds, if I remember. So the software shut it down. And, utter chaos took over in the Mission Control center. Everybody was climbing all
over everybody to find out what happened - totally preventing knowledgeable people from
finding out what happened. It was a bad scene. I don't know if that continued on other
flights. But it certainly screwed up this one. But, there was no telemetry coverage if the mission didn't go exactly perfectly. Of course, there was no need to reprogram it if that's what happened. So here we were out of telemetry coverage. Then they decided to use the LM abort guidance system to do that burn, and subsequently to do the 'abort stage' burn, which jettisoned the descent stage. completely without involving our primary guidance system. Control was taken away from the guidance software. Then, NASA abruptly handed control back to the primary guidance system. Well, nobody had ever planned for or thought that that would happen. Our guidance on the LM thought that it was in the descent configuration because it had not commanded separation. But by then, the descent stage was gone. So the digital auto-pilot had this sporty little vehicle when it thought it was controlling a much more massive one. So it was doing horrible limit cycles and burning fuel like nobody ever saw. NASA said, "What's going on?" I knew right away what was going on. Nobody, of course, had asked MIT anything when all this was happening. They just 'knew better' and took over. Of course, as I feared, they didn't know better. But I, again, sent information up the chain of command to the flight director: If they would activate the heat-soak program that we had been asked to put in there, the first thing the program would do would be to check agreement between the computer's input bit that indicated whether the descent stage was present and what the guidance system thought. And if it found a mismatch, it would correct the guidance system. So this was a way to get the mission back in sync with just one uplink command. When this was made known to the flight director, he said "We can't do that program. We've never run it in our practice simulations here at Mission Control." And here was this second thing in the flight software that we had put in that couldn't or wouldn't be used by the Mission Control people. Well, the result was not good. And, we took a lot of flak for that, a lot of it undeserved. But I think it really shook up the Mission Control people who realized that their ability to handle things was much lower than they thought. And their ability to understand and deal with the possibilities of what would go wrong in the flight computer and in the system other than that was rather incompletely developed. And I think they went through a lot of learning and made a lot of changes that led to success later, in situations like the first lunar landing, where something really unexpected did come up. And there were people there in Mission Control who were confident enough, or foolish enough, to say GO - and it worked. Without some of those blunders that happened early, there would have been a much worse
outcome. Well, that was fine. But for the simulations, we had to have a lot of values preloaded into erasable memory for every run. For convenience, there was a piece of code called "Mr. Clean" which would preload values we thought would be fine for erasable to start with. And Houston could override those before launch. But, it turned out that Houston never touched the erasable load, at least for 206. So the values we had in there for the simulator, and which always worked in the simulator, were what flew. And Grumman ran their simulations, and we never heard, or at least I never heard anything from Grumman about the descent stage pressurization delay. So something is funny there. Here was another one of these mismatches - we never saw what Houston might have done to the erasable load. It turns out they didn't do anything. But there were these interface gaps that weren't local but between geographic locations. But the fact that it went wrong caused people to shore up the defenses a little bit on missions that came later. But, without some of those mishaps, I don't know where we would have come out. But that was a tremendous disappointment to all of us that busted our tails on 206. And
it was never possible to point the finger. I could never point a finger at anybody. I
remember exactly where I was standing when the person who was responsible for the descent
stage burn -- I'm not going to mention his name but everybody here knows who I mean --
told me there were restrictions on the startup time window. So the number came directly
from him. I didn't make it up. So it wasn't something just crazy that we did. But, mishaps
helped us a lot. The accident that took three lives had nothing to do with the AGC
software, but it had the biggest effect. Because I think without that accident, we never
would have made the schedule and maybe never would have made the landing. site last updated 12-08-2002 by Alexander Brown |
|