This part describes the propulsion system of the lunar module.
Link to the official NASA handbook of the lunar module containing the description of the propulsion system of the lunar module
The lunar module's engine was working by mixing fuel with oxidizer.
So there were two different tanks to feed the engine.
In fact, unlike the ascent engine, the descent engine can be swiveled, which allows to realign the main engine's thrust with the center of gravity of the lunar module, at the end of the descent, when the fuel tanks have emptied and there starts to be a disalignement of the center of mass with the thrust of the main engine, but it also means that the horizontal thrusters of the RCS are no more horizontal, and that firing them either makes the lunar module go up or down, which must be compensated by adjusting the main engine's thrust.
There is a confirmation of the disalignment of the center of gravity with the thrust of the main engine of the ascent module along the ascent in a document "Apollo progress report" written in the sixties by a NASA engineer, David Hoagland.
David Hoagland says this:
"The LM ascent powered flight autopilot obtains control torque only by means of the reaction jets. The engine is fixed; it cannot swivel.
This control then operated very similarly to the free-fall coasting flight autopilots described above, but with the addition that the system estimates the torque arising from the offset of the main engine thrust from the center of gravity."
Because the center of gravity is shifted from the axis of the ascent module as the tanks are emptying, the combined forces of the thrust and the lunar gravity create a torque which makes the ascent module turn clockwise.
So that the ascent module keeps a steady direction, this torque needs to be corrected by applying a counter torque.
The only way of applying a counter torque is by firing the vertical lateral thrusters so that they produce torques which make the ascent module turn in the opposite direction.
If the torque created by the lateral thrusters could be exactly equivalent to the misalignment torque, the lunar module could keep going straight.
But the lateral thrusters could not be throttled, they were working in "all or nothing", and consequently it was not possible to adjust their thrust so that they would create a torque exactly equivalent to the misalignment torque.
The consequence is that they could not be permanently fired, but periodically instead; and as they could not be activated at a sufficient frequency (the guidance period was 2 seconds), the result is that it was not possible to avoid a consistent swaying move.
This animation explains what was happening: The misalignment torque was making the LM turn clockwise and go right; then the lateral thrusters were fired, and were making the LM turn counterclockwise, and go left; then the lateral thrusters were shut off (otherwise they would go on making the LM turn counterclockwise and go left), and the misalignment torque was making the LM turn clockwise and go right again, and so on.
The more the misalignment torque was strong, and the more the amplitude of the oscillating move was consistent, for the frequency of command of the lateral thrusters could not be increased to counter the misalignment torque faster.
This is absolutely not a supposition I am making, this is confirmed in the documentation by the NASA engineers.
Even Clavius is talking about it on his site, and saying this:
" As the off-axis thrust caused the ascent stage to rotate, the RCS jets fired to counter the rotation and return it to the correct attitude. This is why the films of the LM ascent seem to show a periodic sway or oscillation: the RCS "fought" the off-axis ascent engine."
The fact that the tank of the oxidizer is placed closer to the ascent engine is explained by the fact that the oxidizer has a greater density than the one of the one of the fuel (1.5 times greater).
If they were placed at the same distance, the difference of density of the propellants would make the LM turn on the side of the oxidizer tank.
At the beginning of the ascent, when the tanks are full, their contents are heavier than the containers, so the misalignment torque is limited.
But, as the tanks are emptying, the difference between the contents and the containers decreases, and consequently the center of mass of the LM shifts toward the fuel tank, with the consequence that the misalignment torque increases; as the lateral jets of the RCS cannot increase their frequency of correction of the misalignment torque, the consequence is that the amplitude of the oscillation of the LM progressively increases.
(The purpose of this animation is to make understand what happens and not to represent the exact reality).
Yet, it was absolutely not the interest of the LM to waste the fuel of its RCS, for it would need it at the crucial final phase when it makes its flipover maneuver to dock to the LM.
Was there a way to avoid using the RCS to correct the attitude of the LM?
Absolutely there was, and there even were several possible ways.
A first obvious way would have been to move a weight inside the lunar module to allow to shift the center of mass; the interest of this is that the ascent engine would remain in the axis of the lunar module.
Another way would be to allow the engine to be swiveled, which would also allow to move the center of gravity on the line of the thrust; but the disadvantage of this is that the ascent engine would not be in the axis of the lunar module any more.
Both ways would allow to avoid using the RCS to control the attitude of the LM which is only necessary because the center of gravity is not aligned with the thrust.
Not only it would give a more regular trajectory to the LM, but it would simplify the guidance, and, most important, it would allow to save the fuel of the RCS so vital for the final phase of the approach.
The NASA engineers made it illogical on purpose.
The engines of the RCS are placed at the four corners of the lunar module.
Each RCS module has four engines, two horizontal, and two vertical, which allow to turn the lunar module in all directions (by combining all the engines of the RCS).
Each engine is activated by an individual command.
A logic allows to individually activate the engines of the RCS.
The engines of the RCS which must be fired to apply the desired torque countering the misalignment torque in the ascent are those I have circled on this schema and are designated as A3U, B4U, A2D, and B1D, according to the terminology of the RCS.
They must all be fired in the same time in order to avoid a disbalance.
This schema shows the interconnection of the valves of the RCS.
When the RCS must be fed in fuel, the two fuel valves have to be open.
When the RCS must be fed in oxidizer, the two valves of the oxidizer must be open.
But, if any of the valves refuses to open, the RCS cannot be be fed either in fuel or in oxidizer, and then cannot work, because it needs both the fuel and the oxidizer to work.
So, why not have used this simpler schema?
The RCS will not work only if one of 2 valves does not work, and there is less chance that one in 2 valves does not work than 1 in 4 valves.
Now, with this schema, the RCS will not work correctly is a valve refuses either to open or to close; in the previous schema, it is only if a valve refuses to open that it won't work; if it refuses to close, the other valve which closes will allow to block the fuel of the oxidizer.
So, if the system must really be guarded against the failure of a valve, it is rather this system which should be used, with a duplication of the valves: If a valve refuses to open, the other valve which opens will still allow to let the fuel or oxidizer through, and, if a valve refuses to close, the valves of the other pair, which will be both closed, will allow to block the fuel or oxidizer.
There is a full security about the failure of a single valve, either refusing to open or to close.
This is the schema of the propellant measuring device.
It uses analog computers to work, that is electronic interfaces using operational amplifiers.
This is the schema of an analog computer of the propellant measuring device.
I see major faults in this schema.
The connections I have barred with red crosses should not exist.
And the adjustable resistors I have circled should not exist either.
Only the gain adjustment should exist.
This is what the analog computer should have looked like to be normal, without the abnormal connections, and unnecessary adjustable resistors.
This is the schema of the thrust chamber assembly, with the individual commands for each engine of the RCS.
The commands of the RCS are elaborated from the LM's attitude angles and the translation vector.
This is the the schema of the horizontal thruster selection which elaborates the commands for the horizontal engines of the RCS.
The devices I have framed are electromechanical relays; an electromechanical relay is composed with a coil I have circled in green, and a switch which is changed state by energizing the coil, and that I have circled in orange.
Here is how the electromechanical relay works:
- When the coil is not energized, the central point of the switch is connected to the upper contact.
- When the coil is energized, it attracts the switch down, and the central point of the switch becomes connected to the lower contact instead.
If the lower contact of the electromechanical relay is connected to the ground, the effect is that the output of the electromechanical relay is connected to the ground when its coil is energized.
The circuit I have circled in red is an OP amp with two outputs; these outputs are complementary to each other: When one is 1, the other one is 0, and vice versa.
The circuit I have circled in orange is an OR gate with inverted inputs.
In the symbology of Apollo, the AND gates and the OR gates are represented with the same symbol, but with a black point inside the OR gate to differentiate it from the AND gate.
An OR gate is a circuit which has several inputs (at least two), and which outputs a 1 if at least one of the inputs is 1, and 0 only if all the inputs are 0.
When there is a little circle before the inputs, this little circle means that the inputs are inverted before being applied to the OR gate (i.e. a 0 becomes 1 and vice versa).
The OR gate then outputs a 0 if all the inputs are 1, and a 1 otherwise (i.e. if at least one input is 0).
When there is only one input to that circuit, the output of the circuit is the inverted input.
Knowing this, we now can analyze this circuitry.
The circuits I have circled produce two outputs from the command inputs (I indicate with double arrows) which are complementary, which means that, when one is positive, the other one is negative and vice versa.
These outputs are inverted before being ored; as the outputs are complementary, so are the inverted outputs, which means that, when one is 1, the other one is 0, and vice versa; oring inputs means that the resulting output is 1 is one if one at least of the inputs is 1; here the little circles indicate that the inputs are inverted before being ored; and the inverted complementary inputs are also complementary, which means that one at least of the inverted inputs is 1, and consequently the output of the OR circuit 1 is always 1; this output is inverted by the next circuit, which means that the output of the circuit 2 is always 0; this output is inverted before being ored with another output, which means that the output of the circuit 3 is always 1; and the output of the circuit 3 goes into an Op Amp which will always energize the electromechanical relay that it is connected to, as the output of the circuit 3 is always 1.
As the electromechanical relay is always energized, no matter what, the switch that it is connected to will always be on the down position, that is connected to the ground.
That means that they could as well have connected the two inputs which are connected to the switch directly to the ground, the result would have been the same.
Now, the inputs of A4R/B1L and A3R/B2L are both Y-TRANSL and the ground, which means that A4R/B1L and A3R/B2L could have been connected directly to the same command.
Likewise, the inputs of B4F/B3A and A1F/A2A are both the Yaw and Z-TRANSL, which means that B4F/B3A and A1F/A2A could also have been connected to the same command.
And as the K3 relay is always energized, and thus connected to the ground, it means that the yaw command does not go into 4 horizontal thrusters; yet it should be able to go into all the horizontal thrusters, for all allow to turn the lunar module around the vertical axis (yaw) in both directions.
Now, the main point is that it makes no sense to compare rotation commands with translation commands to activate the lateral thrusters; indeed, the lunar module can be either rotated or translated, but not both in the same time; moreover there is no notion of the sign of commands; indeed, according to the direction that the lunar module is turned or translated, it is not the same thrusters which are used.
In fact, there should have been two inputs per rotation or translation command, one for each direction (for instance, Yaw+ to make the lunar module turn counterclockwise - the positive trigonometrical direction- and Yaw- to make the lunar module turn clockwise - the negative triginometrical direction).
And only one horizontal command would have been 1 at a time, and all the other ones 0.
The commands of the lateral thrusters would then have been a logical combination of the couples of horizontal commands.
- the lateral thruster A1F would have been activated if either Yaw+ or Z- is 1, and not activated in all other cases.
- the lateral thruster B1L would have been activated if either Yaw- or Y+ is 1, and not activated in all other cases.
- the lateral thruster B4F would have been activated if either Yaw- or Z- is 1, and not activated in all other cases.
- the lateral thruster A2A would have been activated if either Yaw- or Z+ is 1, and not activated in all other cases.
Each thruster is activated either by a rotation command or a translation command.
This is the the schema of the vertical thruster selection which elaborates the commands for the vertical engines of the RCS.
The circuits I have circled produce two outputs from the command inputs (I indicate with double arrows) which are complementary, which means that, when one is positive, the other one is negative and vice versa.
For the same reason as previously explained, the outputs of the circuits 1,2, and 3 are always 1.
The circult 4 receives two inputs which are always 1, but are inverted before being ored, which means that two 0 are ored, which gives a 0; the output of the circuit 4 is thus always 0.
For the same reason, the output of the circuit 5 is also always 0.
The output of the circuit 4 is inverted before being ored with other inputs; the inverted 0 becomes a 1 again, and the output of the circuit 6 is thus always 1; if feeds an OP amp which will always energize the electromechanical relay it is connected to.
For the same reason, the fact that the output of the circuit 5 is always 0 has for consequence that the output of the circuit 7 that it is connected to will always be 1; if feeds an OP amp which will always energize the electromechanical relay that it is connected to.
As the electromechanical relays are always energized, the switches will always be on the down position, that is grounded, which means that all the inputs which are connected to these switches could have directly been connected to the ground, which means that these inputs are useless, and two inputs would have been enough for the Op Amps.
Now, the first Op Amp, the one which is circled, has all its inputs to 0 (because of all relays being forced to the ground); that means that the vertical thrusters B4U and A4D it commands will never be activated.
The main point is the same as for the command of the horizontal thrusters: Instead of incoherently comparing the vertical commands, there should have been two inputs for each vertical command, one for each direction, and only one input set to 1 at a time, and the commands of the vertical thrusters would have been a logical combination of the couples of vertical commands (like I showed on the horizontal commands).
Once again, we here have a total fantasy.
Now we can see that the Roll command (which is normally the command which allows to apply the counter torque) is inputted on the + input of the command of the pair of engines A3U and B3D, and on the - input of the command of the pair of engines B2U and A2D; it means that it will activate the engines A3U and B3D when positive, and the engines B2U and A2D when negative, but never the four of them in the same time.
Now, it's where it becomes really tasty: The engines A3U and B3D, which are activated in case of positive roll, are two vertical engines directly opposed to each other; it means that they are going to mutually cancel their respective effects, and so the consequence is that the positive roll command...does NOTHING!
And the engines B2U and A2D, which are activated in case of negative roll, are two vertical engines which are also directly opposed to each other; it means that they are also going to mutually cancel their respective effects, and so the consequence is that the negative roll...also does NOTHING!
And we can see that the Pitch command is inputted on the + input of the command of the pair of engines B2U and A2D, and on the - input of the command of the pair of engines A1U and B1D; it means that is will activate the engines B2U and A2D when positive, and the engines A1U and B1D when negative.
And once again we find the same joke as for the roll command; The engines B2U and A2D, which are activated in case of positive pitch, are two vertical engines directly opposed to each other; it means that they are going to mutually cancel their respective effects, and so the consequence is that the positive pitch command...does NOTHING.
And the engines A1U and B1D, which are activated in case of negative pitch, are two vertical engines which are also directly opposed to each other; it means that they are also going to mutually cancel their respective effects, and so the consequence is that the negative pitch...also does NOTHING.
So, I think you have understood by now: The command interface of the lateral engines is a total absurdity which does absolutely nothing and does not allow to command the lateral jets of the RCS to change the attitude of the lunar module.
The lunar module is simply not maneuverable with this interface.
This is nothing but a big joke imagined by the engineers of NASA!
But they have even pushed the joke further as we are going to see.
...Anyway, even with a correct interface for the jet selection logic, this clown of AGC would still have been unable to control it correctly!
The normal way for the AGC to ccontrol the RCS is to acquire the attitude angles and update the LM's position from accelerometers, compare the measured angles and position with the ones which are desired by the guidance, and compute commands from the differences between measured and desired data that it sends to the RCS.
The AGC updates angles (and accelerations) by incrementing or decrementing counters according to the direction of angle's variation (with "hidden" instructions, which is a heresy, as I have already pointed out, for incrementing or decrementing these counters should be made by hardware); the guidance computes angles and position according to the trajectory which is to be followed; a measured angle is compared with the corresponding desired angle; the difference is compared with a threshold; if the difference is over the threshold, the thrusters which make this angle increase are activated if the difference is positive (i.e. desired angle greater than measured angle), and conversely the thrusters which make this angle decrease are activated if the difference is negative; if the difference is under the threshold, no thruster making this angle change is activated.
Now, what they do is a little different.
They compute the difference between these two values of the angle in a counter, and they convert this counter into an analog value with a digital to analog converter; the converted analog value modulates a high frequency signal, and the phase of this signal is set according to the sign of the difference.
Then this modulated signal is demodulated, and the demodulated signal is compared with a reference; if the demodulated signal is over the reference, the thrusters which make this angle increase are activated if the modulated signal is phased, and conversely the thrusters which make this angle decrease are activated if the modulated signal is dephased; if the demodulated signal is under the reference, no thruster making this angle change is activated.
This is uselessly complicated, it is much simpler to proceed like I previously showed, by just comparing the difference counter with a binary threshold, and activate thrusters if the difference is over this threshold according to the sign of the difference.
Converting the difference of angles to analog would only have made sense if the thrust of the lateral thrusters could have been adjusted; in that case the converted analog value could have been used to adjust the thrust of the lateral thrusters.
But, as their thrust cannot be adjusted, and they work in "all nothing", i.e. they can either be fully activated of not at all, but not not partially, the conversion of the angle difference into an analog value is an useless complication.
It cannot be said that this way of doing is completely impossible, but it is an useless complication, and engineers intending to make a serious project would never make useless complications.
But this is completely absurd!
When commands are to be transmitted from a controller to a drone, as there is no wire connection between the controller and the drone, the only way to transmit the commands of the controller to the drone is to modulate a high frequency carrier with the commands, so that these commands can travel through air up to the drone; in the electronic circuitry of the drone, the high frequency carrier is demodulated, and the controller's commands are extracted from it and applied to the drone.
But in the case of the RCS control, the commands are directly sent through wires, and not through air, and so this is a completely useless complication, uselessly complicating the circuitry and wasting power.
In other words, this completely useless complication is nothing but a new joke from the NASA engineers for an interface which anyway does not work and cannot make the RCS control the attitude of the lunar module.
So now we have a misalignment torque which comes from the shift of the center if gravity which cannot be corrected by swiveling the ascent engine since it is fixed; this misalignement torque cannot be countered by the lateral engines of the RCS since we have seen it can't work.
The consequence is that the lunar module is going to turn like a girouette, indefinitely, without this rotation having a mean of being countered.
The result is to be expected: The lunar module is going to crash on the lunar ground shortly after having lifted off.
Because, as the guidance task was unable to work regularly, very often the commands sent to the engines were not sent in time, which cas causing unwanted dangerous oscillations!
Since the engines are so important, and so vital for the safety of the lunar module, since their failure would fatally mean death for the astronauts, it is obvious that the NASA should have to be absolutely sure that they are in perfect working order before sending the LM to the moon, without the least doubt, even the smallest one.
Yet, on Apollo 10, during several of the lunar orbits, a critical fuel-cell temperature started to oscillate significantly, as shown on this figure.
An investigation led in a laboratory revealed that small, isolated disturbances in fuel-cell temperature were often present, as shown on this figure.
This investigation demonstrated that small, isolated disturbances could trigger an instability if the power loading ran sufficiently high and the temperature sufficiently low.
From this this information, they devised procedures to eliminate the oscillations, should they occur.
It is absolutely obvious that, this showing that the engine had a potential problem, even if they thought to have settled the problem, they should have sent another mission to check if the problem was really solved, and study in depth to get sure that the problem was no more occurring, since the engines are so vital for the safety of the mission, and that a failure of them is not allowed.
Not at all, they directly sent the next mission, Apollo 11, to the moon, with this "sword of Damocles" hanging over the safety of the ship; they left it to "chance".
Well, if I had been Neil Armstrong or Buzz Aldrin, I would have been less than reassured to fly in this death vehicle that the lunar engine was.
But perhaps that the NASA had the magical solution to prevent all these problems...
..and probably the red cross was ready to lend assistance to the astronauts on the moon in case of problem!