ApolloHoax.net
Apollo Discussions => The Reality of Apollo => Topic started by: Peter B on September 06, 2012, 03:04:30 AM
-
I was wondering if the Brains Trust could clarify something for me. "Eagle" landed long on the Moon, but why?
During the descent burn, Armstrong reported he was seeing features a few seconds early, and concluded that they'd be landing long.
Then, late in the descent, Armstrong saw the intended landing site was full of large boulders, and slowed his descent to overfly them, making them land long.
So when Armstrong looked as this intended landing site, was it the one at the centre of the original landing ellipse, or was it a landing site determined by the fact that the early part of the descent was already making them land long?
Does my question make sense?
-
Going from my memory of the various things I've read about it over the years, the original cause of the 'landing long' was incomplete purging of the air from the docking tunnel prior to separation of the LM from the CSM. The result of this was a little push that was not accounted for in the original descent plan.
The original comment about 'landing long' then referred to where the computer was taking them during powered descent, which would have been long due to that initial extra push. In other words, not in the centre of the landing ellipse. Armstrong's taking manual control to overfly the boulder field then left them landing even longer. So they would have laded long even if Armstrong had not taken manual control in the final phase.
-
Agree. The tunnel gave them a little more delta-V than expected and was not noticed. The effect of the mascons also contributed, IIRC.
-
Thank you for the information about the air in the docking tunnel.
I suppose what I'd like to know now was whether the LM computer recognised they were landing long as a result and was able to compensate, or was a long landing already locked in?
Or, to put it another way, if the boulder field hadn't been where it was, and Armstrong had landed "Eagle" where its trajectory had brought it, would the landing still have been long?
-
I've heard the explanation that the long landing was due to air trapped in the tunnel. Gene Kranz, among others, gave it in his book.
But I don't really buy it. The accelerometers in the LM's IMU were designed to sense and integrate every push on the LM -- not just the descent engine but the unbalanced impulses from the RCS engines and even miscellaneous forces like the pressure from the steam vent on the sublimator. There's no other way to do it; even if the actual propellant flow rates were measured extremely accurately that wouldn't give you the exact impulse because the Isp of every real rocket engine varies slightly from the norm.
The only accelerations that an IMU can't directly sense are those of gravity. Every IMU therefore needs an accurate model of the gravity field of whatever planet is nearby. The moon's gravity field was still not known very accurately during Apollo, especially on the far side. Even as recently as last year our understanding of the moon's gravity was fuzzy enough to justify a special mission just to measure it: that's what the two GRAIL spacecraft are doing. For a time the higher order terms of the Earth's gravity model (above 18th) were classified out of concern that they'd help other countries improve ICBM navigation. (All the numbers are now openly published, as any spacefaring nation can determine them.)
So I think errors in the moon's gravity model are the most likely reason for the navigational error.
One way that the tunnel pressure could still have been responsible is if Eagle's state vector had been accurately determined by ground tracking before separation but loaded into the PGNS after separation, thereby erasing the IMU's "memory" of the separation push. I don't know if this was done, but I tend to think not because the FIDO would have known of this problem and avoided it. And I'm pretty sure there was additional tracking and a state vector update after descent orbit insertion that would have removed all errors up to that point.
A contributory cause could be biases in Eagle's accelerometers. A static bias in an accelerometer will appear as a continuous small force that will be double-integrated, giving a position error that increases with the square of time. Normally you can eliminate these during known zero-g flight, but if there's a small continuous force you don't know about one could crop back in.
Scale errors also produce steadily increasing position errors. They're harder to calibrate because you need to know the "right" number. Even on the ground gravity varies from place to place.
There was no provision for Armstrong and Aldrin to compensate for this error even though they knew it existed. Fixing this was one of the major objectives for Apollo 12: to allow enough real-time compensation for position errors to permit a precision landing. That's why Surveyor III was chosen as the target.
-
Or, to put it another way, if the boulder field hadn't been where it was, and Armstrong had landed "Eagle" where its trajectory had brought it, would the landing still have been long?
Yes. Armstrong comments fairly early in powered descent that they're long. This is when he still had the windows facing the surface, which was specifically to time their passage of certain landmarks.
Eagle passed over a boulder field north of West Crater. It got its name because it was near the west end of the landing ellipse. Eagle landed beyond it so "West Crater" is actually east of the landing site.
-
Yep these guys got it right. I watched a great little documentary about all the things that did go wrong during the Apollo missions and this was touched on exclusively.
Armstrong seen that the planned landing site was unfit for doing do and dangerously close to a large crater. He announced that he was going to land long as to say he was going to overshoot the expected LZ and it even had Aldrin interviewed saying" he told me we were on course for a dangerous land and announced he was going to land long. That's when the 30 second alarm went off and I didn't want to add to his stress but I was feeling the moment and its tension."
I forget the exact number but it said that Armstrongs heart rate rose to around a 150 beats a minute during that decent. Ill check it out right now and repost if I made errors.
-
I stand corrected. They said that right off the bat Armstrong seen the auto pilot was going to take them far. He announced right away" were going long" which Aldrin says" I couldn't believe he could size that up so early and predict our course"
At which time the main computer read that it had crashed, the alarm code went off and one of the guys at Houston calculated that the computer was still flying the module and gave the go to continue the flight. That is when Armstrong seen the auto pilot was going to land them right into a football field size crater and that is when he took manual control of the Eagle. At which time it is said his heart rate spiked to 150 beats a minute. At the time of the 30 second alarm they were more than 4 miles off course...to which he landed the Eagle so softly that the shock sensors on the lander didn't even register a reading.!
-
To be clear, the computer didn't crash. It had a fixed maximum time in which to complete all of its assigned tasks before being required to return to the highest priority tasks at the start of the list. A failure to complete all the tasks set off the alarm. Houston had run simulations under which this occurred a short time before the landing, so the controllers knew what the alarm meant and how to tell if the LM computer was still performing the necessary tasks for a landing. One of the unsung heroes of the mission was the guy who programmed the simulation exercise that gave the controllers the confidence to make a call that the mission could proceed.
-
To be clear, the computer didn't crash. It had a fixed maximum time in which to complete all of its assigned tasks before being required to return to the highest priority tasks at the start of the list. A failure to complete all the tasks set off the alarm. Houston had run simulations under which this occurred a short time before the landing, so the controllers knew what the alarm meant and how to tell if the LM computer was still performing the necessary tasks for a landing. One of the unsung heroes of the mission was the guy who programmed the simulation exercise that gave the controllers the confidence to make a call that the mission could proceed.
Thank you sir for the clarification...that was the wording the doc used...it never really did explain it too well or better than I did I think...appreciate the added clarity and explanation!
-
The only real problem with the computer was that it unnecessarily distracted the crew (and ground) at a very critical time with a comparatively minor problem -- arguably a non-problem, even. The crew was so distracted that they did not notice that the autopilot was taking them into the boulder field north of West Crater, necessitating a rather aggressive manual fly-over by Armstrong at the last minute. Had he not been distracted by the alarms, he could have redesignated his landing point earlier in the approach phase and had a much less heart-stopping landing.
As already stated, the alarms resulted when the computer ran out of real time in executing all the tasks assigned to it. The root cause was misdesigned peripheral hardware on the rendezvous radar antenna that generated high rate spurious interrupts that "stole" so many cycles from the CPU that not enough were left to perform every scheduled task. Any spurious interrupt stream would have had the same effect.
Although this was not expected, the MIT designers anticipated and designed for it. They prioritized their software; the most important tasks are run first, and then the less important tasks until nothing is left to do or the computer runs out of time.
For example, the computer periodically reads the accelerometers and IMU gimbal angles and integrates their data into the spacecraft state vector, its estimate of current position and velocity. Missing an update introduces an error. During descent, the computer kept track of the spacecraft's decreasing mass and adjusted the throttle accordingly. With the IMU data it adjusted the engine gimbal and fired the RCS engines to maintain the proper attitude And had it not responded quickly and properly to Armstrong's hand controller, he would have ordered a manual abort.
But some activities were not nearly as critical. For example, the first alarm occurred when Aldrin entered VERB 16 NOUN 68 to display the difference between the landing radar altitude and that from the computer's estimated state vector. The extra load of this calculation was the "straw that broke the camel's back", as Aldrin immediately and correctly suspected. But the only effect of the 16-68 command was to show some numbers on the DSKY. Skipping the calculation merely updated the display less frequently. Not ideal, but certainly far less serious than skipping the more critical functions already listed. So when the computer ran out of time, it automatically "triaged" the workload and kept the critical functions going -- just as MIT had designed it.
In so doing, it turned on the PROGRAM ALARM light and Aldrin had to key in a sequence to determine its type. It meant nothing to the crew so they had to wait for the ground to tell them that it could be ignored. It distracted both of them unnecessarily at a critical time.
In retrospect, a better design would have implemented one or more "computer overload" lights that flash momentarily whenever the system runs out of real time and is forced to skip less important tasks. Yellow indicates a slight overload dropping only minor tasks; red indicates a more severe problem. The lights would be on only while the problem persists, making it easy to see the effect of changing the computer's workload.
An even more general purpose fix would be an analog meter showing CPU utilization with yellow and red zones. Such meters were actually standard equipment on many of the IBM mainframes of the day, and operators used them to adjust their job mix to more fully utilize the CPU, or just to entertain themselves during a boring shift.
-
Just to be clear, we are talking about the famous 1201 & 1202 alarms, right? If so,then here is a good article from one of the people involved in writing the LM software...
http://www.hq.nasa.gov/alsj/a11/a11.1201-pa.html
-
Yes, that's right. I've read those explanations and I think I understand them. The basic problem was a rapid stream of spurious interrupts from the hardware reporting the rendezvous radar position. The details of how these interrupts were generated are interesting in themselves but not really relevant to the computer's behavior. The bottom line was that the spurious interrupts occurred so rapidly (about 6 kHz) that the time spent servicing them left too little time to execute all the scheduled tasks in real time.
This forced the computer to prioritize, which it did by restarting and displaying a program alarm to the crew. The PROG ALARM light came on and the crew had to enter a sequence to determine the nature of the alarm. In retrospect it would have been better had the computer simply flashed an "overload" light and continued on its way without requiring any manual intervention at all.
Of course, this is 20-20 hindsight because there was no reason to expect the computer to run out of real time in the first place. It had all been tested extensively on the ground and at least 15% idle time was always available as a margin. If not for the design bug in the rendezvous radar position measurement hardware there would have been no problem. Fortunately the MIT programmers built in some protection in depth that allowed the computer to continue gracefully even in the event of an overload, something they didn't think could happen, but did.
-
There was another computer "non-problem" with far more serious consequences than the Apollo 11 program alarms.
The first Ariane 5, Ariane 501, blew up about half a minute after launch because of an incredibly simple but lethal software bug. The inertial reference platform (analogous to the Apollo IMU) encountered a program exception and halted. It even spit out a diagnostic block that the autopilot interpreted as bogus position data. Thinking it was suddenly way off course, the autopilot commanded a sharp turn and the resulting aerodynamic forces tore the rocket apart.
There were two inertial platforms but they ran identical code and crashed at the same time for the same reason.
The IMUs crashed because of an out-of-range overflow in converting a floating point number to a fixed point number. The programmers knew that overflows would cause a trap so they explicitly checked most of them first. But checking required time, their CPU was too slow to check them all, and they figured the conversion in question "couldn't possibly" be out of range, so...
The variable was an estimate of horizontal velocity maintained prior to launch to calibrate the accelerometers. (Since the earth-referenced velocity of the launcher on the pad was exactly zero, any non-zero value had to be an accelerometer bias). Although only needed before launch, the estimate was updated for some time after launch to aid in rapid recycling of an aborted launch.
This unit flew successfully on Ariane 1-4 dozens of times because horizontal velocity did not overflow before the routine was turned off. (The assumption that an overflow "couldn't happen" was based on the properties of Ariane 1-4.) But Ariane 5 built up horizontal velocity more quickly, so it did overflow.
There are many good lessons in software engineering from the Ariane 501 failure. Most of them were immediately obvious in hindsight, and the fact that they weren't obvious prior to the flight is one of those lessons.
-
Armstrong seen that the planned landing site was unfit for doing do and dangerously close to a large crater.
Yeah, this is the idea that was rattling around in my head that led to my original question. Was that boulder field in the centre of the original landing ellipse? If so, then it would appear that "Eagle" had got back onto course after Armstrong earlier reported they'd be landing long. If not, then two separate issues combined to push the landing point much further past the intended landing point.
-
Some years ago, an astronaut toured New Zealand on a number of speaking engagements. I can't remember for sure who he was, but IIRC he was a Space Shuttle crew member on one of the earlier shuttle missions (Gordon?)
I any case he spoke at an Astronomical Society meeting in Christchurch in the mid 1980's, where we had a number of computer enthusiasts, and the subject of the 1201 and 1202 alarms came up.
IIRC his explanation was that there were two radars involved in the LM descent; a landing radar which gave the LM pilot accurate height information above the lunar surface, and a rendezvous radar, the primary task of which was to give range indication during lunar orbital rendezvous with the CM after completing of the lunar surface part of the mission, but it also had a secondary role of keeping tabs on the CM in the case of a descent abort being required.
The issue, according to this astronaut, was that there was a procedural flaw in one of the LM checklists that left one of the rendezvous radar's switches in the wrong position during descent. As a result of this, the computer was trying to process too much information, so it kept restarting itself and reporting an error. The astronaut said that the labelling on this switch was ambiguous, and that it was changed on subsequent missions.
The question I have now is, did I understand what he said correctly or incorrectly?
-
The boulder field was immediately north of West Crater, so named because it was near the west end of the landing ellipse. The autopilot directed them there precisely because it had been running "long" throughout pretty much the entire powered descent. Armstrong knew it from his visual checks through the front windows before he yawed the LM 180 degrees to its face-up attitude.
West Crater is not to be confused with "little" West Crater, the smaller crater just east (sic) of the actual landing site to which Neil ran and snapped pictures just before the end of the EVA. His footprints there and back are easily seen in most of the LRO images of the site.
-
Armstrong seen that the planned landing site was unfit for doing do and dangerously close to a large crater. He announced that he was going to land long as to say he was going to overshoot the expected LZ...
No, Armstrong announced long before West Crater that they would land long:
To Aldrin:
102:36:11 Armstrong (on-board): Okay, we went by the three-minute point early. We're (going to land) long.
Then to Houston:
102:36:18 Armstrong: (To Houston) Our position checks down range show us to be a little long.
Following this in the ALSJ:
[Armstrong, from the 1969 Technical Debrief - "Our downrange position appeared to be good at the minus 3 and minus 1 minutes points (prior to ignition). I did not accurately catch the ignition point because I was watching the engine performance. But it appeared to be reasonable, certainly in the right ballpark. Our crossrange (north/south) position was difficult to tell accurately because of the skewed yaw attitude that we were obliged to maintain for comm. However, the downrange position marks after ignition indicated that we were long. Each one that was made indicated that we were 2 or 3 seconds long in range. (That is, they were reaching landmarks 2 or 3 seconds early. One second corresponds to about a mile of miss distance.) The fact that throttle down essentially came on time, rather than being delayed, indicated that the computer was a little bit confused at what our downrange position was. Had it known where it was, it would have throttled down later (to kill a little velocity). Landmark visibility was very good. We had no difficulty determining our position throughout all the face-down phase of powered descent. Correlating with known positions, based on the Apollo 10 pictures, was very easy and very useful."]
West Crater wouldn't have been visible at this stage and they didn't see it until after pitchover about six minutes later:
102:42:33 Armstrong: (With some urgency in his voice, possibly as he sees West Crater) Give me an LPD (angle).
[Armstrong, from the 1969 Technical Debrief - "In the early phases of P64, I did find time to go out of Auto control and check the manual control in both pitch and yaw and found its response to be satisfactory. I zeroed the error needles and went back into Auto. I continued the descent in Auto...We proceeded on the flashing 64 and obtained LPD availability, but we did not use it because we really weren't looking outside the cockpit during this phase. As we approached the 1500-foot point, the program alarm seemed to be settling down and we committed ourselves to continue. We could see the landing area and the point at which the LPD was pointing, which was indicating we were landing just short (and slightly north) of a large rocky crater surrounded with the large boulder field with very large rocks covering a high percentage of the surface. I initially felt that that might be a good landing area if we could stop short of that crater, because it would have more scientific value to be close to a large crater. (However), continuing to monitor the LPD, it became obvious that I could not stop short enough to find a safe landing area."]
[Armstrong and Aldrin, from the Crew Observations chapter of the Apollo 11 Preliminary Science Report - "This crater was later identified as one we had informally called West Crater during our prelaunch training."]
Once you know where to look, it's easy to find West Crater in the wonderful AS11-40-5882HR (http://www.hq.nasa.gov/office/pao/History/alsj/a11/AS11-40-5882HR.jpg).
This site helps:
http://www.boulder.swri.edu/~durda/Apollo/landing_sites.html
-
Once you know where to look, it's easy to find West Crater in the wonderful AS11-40-5882HR (http://www.hq.nasa.gov/office/pao/History/alsj/a11/AS11-40-5882HR.jpg).
That can't be the rim of West Crater because the view is down-sun, i.e., looking to the west, so West Crater was almost directly behind the photographer (Armstrong?) This is the rim of some other crater even farther downrange in the landing ellipse or beyond it, to the west of the actual landing site.
-
Some years ago, an astronaut toured New Zealand on a number of speaking engagements. I can't remember for sure who he was, but IIRC he was a Space Shuttle crew member on one of the earlier shuttle missions (Gordon?)
I any case he spoke at an Astronomical Society meeting in Christchurch in the mid 1980's, where we had a number of computer enthusiasts, and the subject of the 1201 and 1202 alarms came up.
IIRC his explanation was that there were two radars involved in the LM descent; a landing radar which gave the LM pilot accurate height information above the lunar surface, and a rendezvous radar, the primary task of which was to give range indication during lunar orbital rendezvous with the CM after completing of the lunar surface part of the mission, but it also had a secondary role of keeping tabs on the CM in the case of a descent abort being required.
The issue, according to this astronaut, was that there was a procedural flaw in one of the LM checklists that left one of the rendezvous radar's switches in the wrong position during descent. As a result of this, the computer was trying to process too much information, so it kept restarting itself and reporting an error. The astronaut said that the labelling on this switch was ambiguous, and that it was changed on subsequent missions.
The question I have now is, did I understand what he said correctly or incorrectly?
That is pretty close to my understanding although I have also read accounts that put more emphasis on Aldrin. Thankfully the ground crew was prepared.
-
There are many good lessons in software engineering from the Ariane 501 failure. Most of them were immediately obvious in hindsight, and the fact that they weren't obvious prior to the flight is one of those lessons.
I once read the following on a forum in which the discussion was about the STS51-L, and the systemic failures that lead to the disaster, although this particular post related more to the Colombia disaster.
This text was part of a post by a member of that forum, and I liked it so much that I saved it for future use. I do quote it from time to time in relation subjects other than Apollo / Spaceflight
The value of failure is that you almost always learn something new about what your system can or cannot tolerate, plus you are forced to look at all of the potential failure modes as part of a disciplined failure review process.
Every time you succeed, all you learn is that you were some combination of good and lucky, and you never really know the ratio of one to the other.
When you fail, you generally have a good idea of just how unlucky you were that day, and how to improve your odds tomorrow.
- Stranger-on-a-Train (straightdope.com) 7 December 2011
-
Once you know where to look, it's easy to find West Crater in the wonderful AS11-40-5882HR (http://www.hq.nasa.gov/office/pao/History/alsj/a11/AS11-40-5882HR.jpg).
That can't be the rim of West Crater because the view is down-sun, i.e., looking to the west, so West Crater was almost directly behind the photographer (Armstrong?) This is the rim of some other crater even farther downrange in the landing ellipse or beyond it, to the west of the actual landing site.
Oops, my apologies, I pasted the wrong photo number from the wrong film. The correct one is
AS11-37-5447HR (http://www.hq.nasa.gov/office/pao/History/alsj/a11/AS11-37-5447HR.jpg).
This site helps:
http://www.boulder.swri.edu/~durda/Apollo/landing_sites.html
-
Armstrong seen that the planned landing site was unfit for doing do and dangerously close to a large crater. He announced that he was going to land long as to say he was going to overshoot the expected LZ...
No, Armstrong announced long before West Crater that they would land long:
To Aldrin:
102:36:11 Armstrong (on-board): Okay, we went by the three-minute point early. We're (going to land) long.
Then to Houston:
102:36:18 Armstrong: (To Houston) Our position checks down range show us to be a little long.
Following this in the ALSJ:
[Armstrong, from the 1969 Technical Debrief - "Our downrange position appeared to be good at the minus 3 and minus 1 minutes points (prior to ignition). I did not accurately catch the ignition point because I was watching the engine performance. But it appeared to be reasonable, certainly in the right ballpark. Our crossrange (north/south) position was difficult to tell accurately because of the skewed yaw attitude that we were obliged to maintain for comm. However, the downrange position marks after ignition indicated that we were long. Each one that was made indicated that we were 2 or 3 seconds long in range. (That is, they were reaching landmarks 2 or 3 seconds early. One second corresponds to about a mile of miss distance.) The fact that throttle down essentially came on time, rather than being delayed, indicated that the computer was a little bit confused at what our downrange position was. Had it known where it was, it would have throttled down later (to kill a little velocity). Landmark visibility was very good. We had no difficulty determining our position throughout all the face-down phase of powered descent. Correlating with known positions, based on the Apollo 10 pictures, was very easy and very useful."]
West Crater wouldn't have been visible at this stage and they didn't see it until after pitchover about six minutes later:
102:42:33 Armstrong: (With some urgency in his voice, possibly as he sees West Crater) Give me an LPD (angle).
[Armstrong, from the 1969 Technical Debrief - "In the early phases of P64, I did find time to go out of Auto control and check the manual control in both pitch and yaw and found its response to be satisfactory. I zeroed the error needles and went back into Auto. I continued the descent in Auto...We proceeded on the flashing 64 and obtained LPD availability, but we did not use it because we really weren't looking outside the cockpit during this phase. As we approached the 1500-foot point, the program alarm seemed to be settling down and we committed ourselves to continue. We could see the landing area and the point at which the LPD was pointing, which was indicating we were landing just short (and slightly north) of a large rocky crater surrounded with the large boulder field with very large rocks covering a high percentage of the surface. I initially felt that that might be a good landing area if we could stop short of that crater, because it would have more scientific value to be close to a large crater. (However), continuing to monitor the LPD, it became obvious that I could not stop short enough to find a safe landing area."]
[Armstrong and Aldrin, from the Crew Observations chapter of the Apollo 11 Preliminary Science Report - "This crater was later identified as one we had informally called West Crater during our prelaunch training."]
Once you know where to look, it's easy to find West Crater in the wonderful AS11-40-5882HR (http://www.hq.nasa.gov/office/pao/History/alsj/a11/AS11-40-5882HR.jpg).
This site helps:
http://www.boulder.swri.edu/~durda/Apollo/landing_sites.html
Thank you sir