posted 07-14-2017 06:48 PM               

The Apollo 11 mission succeeded in landing on the moon despite two computer-related problems that affected the Lunar Module during the powered descent. An uncorrected problem in the rendezvous radar interface stole approximately 13% of the computer's duty cycle, resulting in five program alarms and software restarts. In a less well-known problem, caused by erroneous data, the thrust of the LM's descent engine fluctuated wildly because the throttle control algorithm was only marginally stable. The explanation of these problems provides an opportunity to describe the operating system of the Apollo flight computers and the lunar landing guidance software.

posted 07-16-2017 04:16 AM            

I can tell you exactly what happens if neither the engine or DECA would have any delay. The throttle oscillations become catastrophic. I know because I have simulated this. I am one of the developers of Project Apollo - NASSP (http://nassp.sourceforge.net), an open source project to simulate the vehicles of the Apollo program. And we are using the Virtual AGC emulator (https://www.ibiblio.org/apollo/) for our simulation. The Apollo 11 source code is available through this emulator, so we are using 100% identical software as compared to the flown Apollo 11 mission. Don Eyles himself actually supplied the source code for most Lunar Guidance Computer (LGC) software versions to the Virtual AGC project.

So, when we first tried the Apollo 11 lunar landing, we did get catastrophic throttle oscillations in Program 66 (Rate-of-Descent landing mode). The first delay I mentioned above is simulated in the emulator, so it is accounted for. As the next step we added the throttle lag in our simulated DECA and that fixed most of the oscillations. So even without simulating the actual delay of the engine response (which was 0.075 seconds), we can get a fairly stable behavior. As compared to the graphs on Don Eyles website we still get slightly higher oscillations, but it is not a problem for landing. We haven't yet added the throttle delay of the engine itself, since we get decent results without it. Here is a video of the simulated Apollo 11 landing I made a while back: https://www.youtube.com/watch?v=sHaS6sYJsMg The video shows an automatic landing, which uses a different throttle algorithm and does not have any throttle oscillation. The video was recorded before the DECA delay was implemented that made Program 66 with the semi-automatic landing mode work. P66 is what all flown missions used to land, including Apollo 11.

posted 07-17-2017 06:30 AM            

One of the important parts of the semi-automatic Rate-of-Descent (ROD) landing mode is that you need to be able to judge your altitude rate. Either from the computer display on the DSKY or direct Landing Radar measurements on the tape meters. With the knowledge of the altitude rate you can then adjust this rate with a switch, incrementing or decrementing the altitude by 1 feet per second with each switch actuation. This capability is severly degraded with oscillations this high. Due to the low gravity the cycling between 10% and 50% throttle doesn't immediately make you crash into the surface though. The altitude rate is cycling between about 1 and 5 feet per second. I levelled off at about 50 feet and then mostly had to judge my descent by looking outside, doing only a few adjustments with the ROD switch. I did not have to use manual throttle for the landing, 1-5 ft/s descent rate is in the acceptable margin. So I was able to land in the ROD mode despite the large oscillations, with a bit of trouble, but all in all still fairly safely.

If the Apollo 11 astronauts had encountered this situation then they might have opted to use manual throttle instead of using the computer controlled auto throttle mode. And as Don Eyles said in his "LM Tales", with the previous alarms they would have lost some confidence in the computer and might have aborted anyway.