If the Apollo computers were less powerful than my phone...

[Superfluous J]

I can make an adapter for that out of duct tape, a piece of plastic, a piece of cardboard, and an old sock!

(Gold star for whoever can name whom I'm quoting! )

Sounds like Red Green to me.

[Javster]

Sounds like Red Green to me.

So we can connect Mission Control to an iPad with duct tape, plastic, cardboard and a sock?

[DerekL1963]

One of the difficulties in trying to compare Apollo computers to modern ones and say "oh how did they do so much with so little?" is that there's a lot of apples and oranges comparison going on given that the Apollo computers were NOT general purpose do-anything processors accessing general-purpose do-anything RAM. When you make a generic "computer" that's not hardwired to specific tasks the burden is on the software to do all the heavy lifting. When you make a very specific one-purpose computer that DOES have a lot of things hardwired to specific tasks because it was designed in an era a little bit prior to the prevalence of 'general' do-anything sorts of computers, a large portion of that burden is moved from the software out to the "peripherals" so to speak.

This. The 'CPU' of the Apollo guidance system was really just one small component of a much larger and more complex system, pretty much nothing at all like an iPhone's CPU. While it was a Von Neumann machine, and theoretically Turing complete, it just wasn't very powerful... It made up for this lack by being embedded within a number of specialized support and interface circuits (IIRC some of the coordinate transformation were analog), and it's these that make it essentially impossible to compare then-to-now.

[Brethern]

Nobody has brought up the most crucial part of this: Computers were important to the Apollo program, but not even NEARLY the most important part. Columbus didn't even have a computer! I could use my smart phone to sail a ship across the Atlantic!

Apollo was full of the best computers on Earth in the brains of the people running the program and working those (actual) computers. They weren't saying, "Siri, give me directions to the Moon." They were calculating the things KSP does for us, by hand on paper, and entering numbers into the computers that were running software that they (or someone as smart as they) wrote themselves, and then writing that result down on another piece of paper and going to tell Neil and Co that they needed to thrust so far in such a direction in about this amount of time.

So what were they calculating anyways? Whenever I try and read it my eyes glaze over.

[Godot]

Nope you couldn´t I think ...

the Apollo computers my have been primitive by todays standard,

but this also made them rather invulnerable to the radiation in space.

I´d guess that todays smartphones would burn-out within short time (or give faulty readouts), if they would have to perform in the same environment as the Apollo computers

[Brotoro]

Nope you couldn´t I think ...

the Apollo computers my have been primitive by todays standard,

but this also made them rather invulnerable to the radiation in space.

I´d guess that todays smartphones would burn-out within short time (or give faulty readouts), if they would have to perform in the same environment as the Apollo computers

Why couldn't you just shield the phone? You could include a lot of radiation shielding and still come in MUCH lighter than the originalApollo computer system.

[Tex]

I'm so glad I could run a real landing with my phone. Anybody got a spare $50,000,000,000 for rocket expenses?

[ttb]

So what were they calculating anyways? Whenever I try and read it my eyes glaze over.

They were calculating such things as burn start times and duration, vectors for burns, re-entry angles, etc. Also most of the Moon descent was calculated and flown by the autopilot. Also, the AGC displayed info on the DSKY for the crew, such as altitude, MET, state vectors, and other info that the crew needed to know, mostly to verify that the computer was doing its job properly, and so the crew could call an abort if the computer didnt, if the need arose.

In some stages of the flight, including Earth ascent, the crew would compare the display from the computer to cards velcro'ed onto their panel, to make sure things were happening as per The Plan.

ttb

Edited January 7, 2014 by ttb

[Nuke]

Everyone is ignoring one crucial problem: You can't just plug an iPad into the Apollo LM, because one was made in 1969 and the other was made in 2010. They would not be able to communicate with each other. It's like trying to stick a cube into a round hole.

If you built a new ship to run off the iPad, though, that's another story.

not really. the agc was one of the first cpus to use semiconductor logic. logic was discrete (made with a bunch of chips, each performing a basic logic operation), mostly built out of 3 input nor gate ics. but the same rules of digital logic apply. at most you would need to change logic levels. agc ran at 3v high, 0v low, which would probibly work with modern 3.3v ttl (3v is enough to register as a logical 1). the way the pcbs were made for the agc is they same way they are made now. they used surface mount packages and multi-layered pcbs. the agc really is the forefather of modern technology. you could make them talk to eachother, question is why?

the only thing that really changed was how much stuff you could get on a semiconductor die. eventually it became possible to put more and more logic on a single chip, which eventually made the modern cpu possible.

[vetrox]

Hmmmm.

If only i had some money to burn. Would be quite a cool science project to launch a rocket into space and land back down using a phone. I dont quite think your run of the mill model rockets would be able to make it though

[Stargate525]

Yeah you can, and really... I don't get why people think this is so surprising.

It's basic, basic physics and geometry. Everything in space is parabolae and ellipses. You could, theoretically, do it with a slide rule. I believe they were actually prepared for that, come to think of it.

[ttb]

Hmmmm.

If only i had some money to burn. Would be quite a cool science project to launch a rocket into space and land back down using a phone. I dont quite think your run of the mill model rockets would be able to make it though

You don't even need the power of a phone to do this. There are sattelutes with arduino's running the show: https://ardusat.org/

Thanks

ttb

[OdinYggd]

Everyone is ignoring one crucial problem: You can't just plug an iPad into the Apollo LM, because one was made in 1969 and the other was made in 2010. They would not be able to communicate with each other. It's like trying to stick a cube into a round hole.

If you built a new ship to run off the iPad, though, that's another story.

Or if you know the electrical layout of the original computer, there should be a point in the system where the computer itself can be separated from the manual overrides in the signal harness.

Cut the harness at this location, and in its place insert a modern PLC brick with IO boards to match the voltage and current. That PLC board can then interact with your phone by protocol of choice, allowing the phone to operate all of the controls as though the original computer was still working- except now it is a lot faster. This is how retrofitting of industrial equipment is performed, the harness is separated at a point where the bare IO to the machine is exposed and a new computer or data interface is fitted in its place.

Of course the PLC itself would probably have more power than the original LM computer. Depending on its capabilities the entire LM computer could be reduced to a plastic box no larger than a shoebox, with a few thousand wires coming out of it. The power consumption would be reduced by orders of magnitude in the process.

[Nuke]

you just need to be able to look at the address/data buses. throw a comparator on the address bus, if the value on the bus matches another arbitrary value, intercept all read and write requests to that address with your own hardware. usually you copy the data into one register and out of another, that way you turn that address (or range of addresses) into your communications pipe. memory mapped i/o for the win.

[DerekL1963]

Or if you know the electrical layout of the original computer, there should be a point in the system where the computer itself can be separated from the manual overrides in the signal harness.

Cut the harness at this location, and in its place insert a modern PLC brick with IO boards to match the voltage and current. That PLC board can then interact with your phone by protocol of choice, allowing the phone to operate all of the controls as though the original computer was still working- except now it is a lot faster. This is how retrofitting of industrial equipment is performed, the harness is separated at a point where the bare IO to the machine is exposed and a new computer or data interface is fitted in its place.

If the LEM guidance computer was a piece of industrial equipment, that would be a relevant piece of information. But the LM's guidance computer doesn't really resemble what we would think of as a computer today. Yes, there's a CPU, but there's a whole raftload of other circuits (analog, digital, and discrete) as well. Making things even more complex, some of the analog signals rely on phase difference rather than voltage or current to convey information. Analog circuits in general can be dicey to splice into, but these are particularly difficult. (I've worked on these types of circuits, and I've seen a bad wire wrap take them down.)

Of course the PLC itself would probably have more power than the original LM computer. Depending on its capabilities the entire LM computer could be reduced to a plastic box no larger than a shoebox, with a few thousand wires coming out of it. The power consumption would be reduced by orders of magnitude in the process.

Exceeding the power of the digital portion is trivial. Replacing the full capabilities of the guidance computer... is not. Doing so requires additional support logic, emulators, and complexity - introducing a whole bunch of new ways for things to go all pear shaped. You'd be far better off to go all in and replace the entire guidance and navigation system, especially considering how tightly integrated the system is and the various advances in the related technologies over the past forty-fifty odd years.

[magnemoe]

Yeah you can, and really... I don't get why people think this is so surprising.

It's basic, basic physics and geometry. Everything in space is parabolae and ellipses. You could, theoretically, do it with a slide rule. I believe they were actually prepared for that, come to think of it.

Yes, orbital science is easy for an computer.

Main issue with an real world mechjeb is that you don't have all the data mechjeb has, you don't have your accurate position and speed for once but need help from earth to find it.

I guess engine and even more environment control require more computer power, the user interface will probably use most code.

The user interface on the apollo computers was not user friendly

[DerekL1963]

Main issue with an real world mechjeb is that you don't have all the data mechjeb has, you don't have your accurate position and speed for once but need help from earth to find it.

Only if your spacecraft lacks an accurate guidance system. (Which both the LM and CSM had.)

The user interface on the apollo computers was not user friendly

It was designed to be used by trained professionals, not Joe Sixpack, therefore 'normal' guidelines of user friendliness do not apply. The systems I operated for the Navy were anything but user friendly by 'normal' standards, but none of us had any particular problems operating it.

[Stargate525]

Yes, orbital science is easy for an computer.

Main issue with an real world mechjeb is that you don't have all the data mechjeb has, you don't have your accurate position and speed for once but need help from earth to find it.

I guess engine and even more environment control require more computer power, the user interface will probably use most code.

The user interface on the apollo computers was not user friendly

Yeah, but with a fairly basic lidar system, you should be able to get relative velocities pretty easily. It all boils down to trigonometry and physics, which are super easy for a computer to plug through.

Again, Environment control is an oxygen sensor, thermometer, and barometer. Rip open your air conditioner; that's the kind of processing you're looking at. Engine... It's a rocket. Less complex than your modern car engine; far fewer moving parts.

Actually... Top Gear should have used a modern car for their rocket program; the on-board computer probably could have handled it all.

[K^2]

Yeah, but with a fairly basic lidar system, you should be able to get relative velocities pretty easily.

For docking, sure, but when you're buzzing by something at something on the order of 1km/s, and so you need to measure velocities from hundreds of km away, this gets really complicated really fast. So if you are hoping to do rendezvous or venturing beyond LEO, you are going to do something way more complicated. You're looking at multiple ways of gauging position, velocity, and acceleration, as well as some smart filters to combine this information to make the best possible guess on the running values. The coding for this wouldn't be terribly complicated, but it's actually more computationally expensive than you might think.

On the other hand, if all you want is to be able to enter LEO, maintain roughly the correct orbit, and then re-enter somewhere over the ocean, you can do all of that with a GPS chip and an equivalent of the $1 micro controller.

[Stargate525]

For docking, sure, but when you're buzzing by something at something on the order of 1km/s, and so you need to measure velocities from hundreds of km away, this gets really complicated really fast. So if you are hoping to do rendezvous or venturing beyond LEO, you are going to do something way more complicated. You're looking at multiple ways of gauging position, velocity, and acceleration, as well as some smart filters to combine this information to make the best possible guess on the running values. The coding for this wouldn't be terribly complicated, but it's actually more computationally expensive than you might think.

Waitaminute... you need two... three if you want 100% accuracy... reference points to determine position. For almost anywhere in Earth's SOI, that's the sun, an arbitrary position on Earth, and a similar one on the Moon. That combined with known values of their separation could get you a fairly accurate position reckoning. It's, unless I'm missing something, three equations. repeating on a regular schedule gives you velocity, as well as acceleration.

But yea, if you want full auto-pilot, you'll need to be sampling that very, very fast. But it's still not overly expensive. I think.

[TheCardinal]

Waitaminute... you need two... three if you want 100% accuracy... reference points to determine position.

{snip}

Unless i'm mistaken, you'll need four reference points for an accurate determination of position. The angles with three reference points lead to two possible positions. The fourth reference point determines which of the two possible positions you're really at.

Edited January 9, 2014 by TheCardinal

[DerekL1963]

For docking, sure, but when you're buzzing by something at something on the order of 1km/s, and so you need to measure velocities from hundreds of km away, this gets really complicated really fast. So if you are hoping to do rendezvous or venturing beyond LEO, you are going to do something way more complicated. You're looking at multiple ways of gauging position, velocity, and acceleration, as well as some smart filters to combine this information to make the best possible guess on the running values. The coding for this wouldn't be terribly complicated, but it's actually more computationally expensive than you might think.

And with modern computers, that computational expense is all but meaningless. The Deep Space 1 and Deep Impact missions demonstrated autonavigation years ago. Deep Space 1 tracked asteroids as it's navigation reference, and automated stellar navigation is old technology. (As in "used operationally as far back as the 50's and 60's".)

You seriously overestimate the difficulty and underestimate the state-of-the-art.

[Nuke]

keep in mind that state of the art in radhard terms is 10 year old technology compared to what we are used to. i have a 10 year old computer that wouldnt run a modern version of windows very well, let alone any applications. though most of that is cruft that you dont need on a spacecraft os/applications. i think the biggest hunk of code in my game engine is actually the ui code. the most computationally expensive thing your computer does for you is to draw whats on your screen, even if its just ui graphics. this problem would be as hard as ksp's physics if a modern video card wasnt a friggin super computer. my graphics card (a geforce 560) can do around 840+ gflops, my cpu, an i7 3770k, can do around 85gflops. it gets that ~10x performance gain because its a special purpose parallel processing monstrosity designed for the sole purpose of letting you see whats going on with your system. the thing that makes computers slow is the need to interface with our dumb meatbag asses.

when it comes to actual maths for space flight, like coming up with reliable positional data from the output of an array of sensors, a 486dx could do that job (this is the cpu the hubble uses btw, unless they upgraded it). the algorithms are complex, yes, but the time it takes the cpu to figure it out is imperceptible to us humans. thats an old cpu. on a modern radhard cpu it takes less time to figure out where you are than the time it takes for the signal representing the sensor readings to get from the sensor to the cpu (obviously with imperceptible latency, but thats manageable).