Space Kadet: Dark Side of the Mun

Memoirs of a doomed Kerbal

Posted by Wisq on January 4, 2015

This is the third post of my highly-modded Kerbal Space Program playthrough. The first is here and the second is here.

[Final entry in the logbook of Bob Kerman, Scientist, aboard Korona 29]

The Mun is going to kill me. I’m thousands of kilometres away from it by now, on my way home, but it’s going to kill me anyway.

The ship will keep going, of course. It’s on a stable orbit, and besides, they’ve got all kinds of automation in this thing. They can fly it without me. Heck, that’s the whole reason I’m here instead of Jebediah. “You won’t need to fly it at all,” they said. “Everything’s automated,” they said. “Piece of cake,” they said.

I’ve got almost three hundred days of food left. All the cake on Kerbin couldn’t save me now.

It started an hour ago

No, scratch that. It would’ve started a few hours before that, but nobody knew anything was wrong until an hour ago. I was going over all my Mun reports, taking advantage of the travel time to get a head start on sorting the data. Next thing I know, all kinds of alarms are going off. The lights go out, Mission Control radios something garbled about an “electrical failure”, and then … silence. Not just the radio, but the entire ship.

Bob in the dark

I may not be an engineer, but even I know what that means. No life support. No heaters, no CO₂ scrubbers. I helped crunch the numbers back when we were designing the life support systems, so I also know what happens when it fails: Within two hours, I’ll either freeze to death or die of CO₂ poisoning. With any luck, I’ll pass out before then.

All because the Mun is blocking the Sun and I’m right in the middle of its shadow.


It began about a week ago. We needed a ship with enough delta-v to get to the Mun, but we still only have these little 1.25m tanks. We have some plans for 2.5m tanks and engines, but we need stronger materials — the ones we have can’t handle the fuel and exhaust pressures.

Instead, we came up with a workaround: We can split a rocket stack into three with a tri-coupler. So we took our existing Korona design and turned the lowest stage into a tri-coupled liquid stack. Our new radial decouplers even gave us enough clearance to experiment with tri-coupled boosters, too.

Korona 8

Our first prototype was Korona 8. It had the old Korona command stack, mostly — we added an extra cargo canister that could hold over 300 days worth of life support supplies, because the standard three days just wasn’t going to cut it for a trip to the Mun.

Korona 8 accident

Our numbers showed it had over 5000 delta-v, and an initial thrust-to-weight ratio of 3.61 — this thing was going to launch like a bat out of hell. We figured most of the drag was in the back due to the tri-coupled design, so it shouldn’t need stability fins.

Boy were we wrong. Thankfully, even Jeb wasn’t crazy enough to try to fly a brand new prototype. It began veering off-course right from the start, and as soon as we jettisoned the lowest booster stage, all hell broke loose.

Rocket science is hard.

Korona 9

So, we slap a bunch of winglets on it and call it Korona 9. Should make it a lot more stable.

And, it does. Too stable, in fact. We can’t steer the damn thing! Without steering, the ascent profile is all wrong — it’s going almost straight up the whole way.

There are other little problems, too. Like, the fairings automatically blow at 25km instead of 75km. That’s a problem from around Korona 3 that was fixed a model or two later. Guess what we based our command pod on? The engineers figured Korona 3 was a good match for our current mission profile, but forgot to apply the later Korona model updates.

They called it “the dangers of using an old ship design”. I called them lazy and incompetent. Maybe me dying out here is karma for that, I don’t know.

Korona 10

Korona 10 marked the beginning of our descent into Control Surface Hell.

In a nutshell: Control surfaces on rockets are hard. If you set the deflection too low, you have no control at low speeds. But if it’s too high, the autopilot can’t cope with them at high speed. It overcorrects, and the ship starts oscillating more and more until it tears itself apart.

We break out of symmetry so that we can add two flat horizontal pitch surfaces and one vertical yaw surface on the lowest set of boosters. We also add some smaller roll surfaces on the higher ones.

The result? Massive oscillation. The ship can’t seem to decide what direction it should be going, how it should be rolled — anything. It’s like watching a rodeo, not a rocket launch.

Plus, the fairings still blow early! I take back what I said about karma — those guys in engineering are the laziest bums ever. One of the fairings catches on a booster and starts pulling the ship even further off-course, and it’s abort time.

Korona 11

Most of the oscillation seemed to be in the roll direction, so for Korona 11, we try just removing the roll surfaces.

Ha! If only it were that easy.

Korona 11 rolling

Instead of oscillating, it just goes right into a crazy roll and stays there. We hit the abort switch, but the parachutes refuse to deploy because we’re above the pre-deploy limit.

No problem. It’ll just fall down to 5km and then deploy them as usual, right? Well, no, turns out you need to arm them manually if the abort fails to deploy them.

Good thing we’re still testing these unmanned.


It’s at this point that Bill finally makes a breakthrough.

See, all of our control surfaces have these settings that let you tell the fly-by-wire computers what they should use that surface for. If you want a yaw-only surface, you keep yaw set to 100%, and you set roll and pitch to 0%. Simple enough.

Control surface settings

Well, turns out our engineers had been setting them all the way to the left. That used to work back in older versions, so it’s what they got used to. But somewhere along the line, they changed it so that 0% was in the middle, and left of that was negative.

Setting it all the way left is -100%, i.e. “if you think you should turn this fin one way, turn it the other way instead”. So basically, we’ve been telling our computer that if you want to roll right — or want to stop a left roll — you need to try to roll left even harder. (Removing the dedicated roll surfaces is what made the problem obvious, since we removed its ability to fight itself.)

Major facepalm moment.


We still haven’t eliminated high-speed oscillation, though, and the next several flights are spent testing various configurations and settings. Nothing seems to work.

Eventually, we give up on control surfaces and try just using thrust vectoring. Our ascent path isn’t as flat as we wanted, but it gets into orbit with 2000 m/s delta-v remaining.

Eh, good enough. Plus, Jeb’s getting impatient.

Korona 18

For Korona 18, we redesign the command module for our Munar mission. The probe is replaced with a parachute, and now that we’re more aerodynamic, we decide to try going without fairings.

We add the usual mix of science wedges. The top layer has two Mystery Goos and two Science Juniors, while the bottom has the orbital telescope, magnetometer, barometer, and a bay full of batteries. The hope is that, in the unlikely event of a hard landing, we’ll lose the recording gear but keep the samples.

The engineers tell me they think they have the control surface thing figured out. Something about enabling “FAR ‘DCA’ flight assist” (whatever that is) because it dampens out control inputs at high speed. Even so, we limit the surfaces to 50% size and 1º deflection so they’ll only have a tiny influence.

We haven’t tested this new configuration, but Jebediah insists on hopping into the pilot’s chair. He’s talking about an “end to unmanned flight” and how we need to “take more risks” for progress, but I think our space program needs better psychological screening.

The rocket is crawled out to the launchpad, the countdown starts, we fire off the main engines, and … nothing. Well, lots of smoke, but no liftoff. The boosters start overheating, but run out of fuel before they explode.

Our lead engineer mumbles something about a “bug” and how he’s going to try to “quit and start again” — but whatever he does, it’s not enough: Korona 19 is still stuck to the pad.

Korona 21 accident

We try using launch towers to raise Korona 20 off the pad — you can’t stick to the pad if you’re not touching the pad, right? — but the high-speed abort nearly gets Jeb killed, and even he backs down on the whole “manned test flight” thing. (Never seen him that shaken. Didn’t think it possible.)

Meanwhile, engineering logs something about “downgrading the Kerbal Joint Reinforcement mod”, and it might all be gibberish to me, but it seems to stop the sticking problem.

More unmanned test flights ensue, but finally, we settle on a design that works. Mostly.

Korona 26

Korona 25 is a success, and engineering finally regains Jeb’s trust. He gets on board the identical Korona 26. Five-degree control surfaces (with that “DCA assist” thing) and more struts seem to be enough to keep it stable, and our ascent plan is to ask the computer for a 70º ascent and watch it overshoot to 60º or lower.

The launch goes okay, but it’s not as efficient as we hoped, and we only have about 1500 m/s delta-v. It was also much more inclined than we would’ve liked, making Munar orbit harder since we’re forced to approach it on a high polar orbit.

Mission Control plots some possible manoeuvres and crunches the numbers, but it’s not looking good. It’s 850 m/s to the Mun and at least 250 m/s to orbit it. Achieving low orbit will blow our delta-v budget, and there’s not much point to just high orbit.

So, the mission plan changed: Jeb would do a flyby, collect some science, then come home. Nice and simple. Plenty of fuel to spare.

Everything goes fine up until reentry.

Jeb’s all oriented for reentry, but he gets distracted and forgets to jettison the service module until he’s already in the atmosphere. Drag is low enough that it separates cleanly … for about ten seconds. Then it comes right back and “un-jettisons” itself, mashing itself into the heat shield.

It stays there for a good chunk of the reentry, but it’s being buffeted and is rocking from side to side, and it’s only a matter of time before it falls off. Finally, Jeb says he heard a horrible grinding noise, and it breaks free. He’s intact and on a proper reentry again.

Then, a muffled explosion. Too quiet to be the ship. Must be the booster finally overheating behind him. But that must’ve changed the aerodynamics, because he looks out the window and sees a fireball screaming past the ship! Another muffled explosion ahead of him, and that’s the last he heard of it. Close call.

Mission success. 67 Munar Science. Finally!


At this point, we know that 5000 delta-v isn’t enough — or at least, isn’t enough if you can’t guarantee a perfect ascent profile. We’re going to need to push for 6000 or even higher.

So they make Korona 27 longer. More fuel tanks. Switch the boosters up to the larger BACC model. Lots of struts. It seems to do okay for a bit, but veers off course after the first boosters separate, and they abort.

Korona 28

For Korona 28, they try something new: Move the boosters down, so the tri-stacks are below the main engines, and hold back on the main engines until the second stage. The numbers are showing us that we’ll get better delta-v that way, since we’ll be higher up when we fire them.

We launch it, and … it’s stable! In fact, it’s our most stable rocket to date, while still being quite controllable. We kick off a celebration party that evening, and after quite a few drinks, I make the biggest mistake of my soon-to-be-over life:

“You guys really pulled it off! That rocket is so stable, even I could ride in it!” (I’m known for having motion sickness.)

“Well Bob, funny you should mention that …”

Uh oh.


To their credit, they weren’t kidding about the automation. The launch is quick and efficient. There’s a bit of spin, but they tell me I reached 110km orbit with 3100 delta-v remaining. They never even fired off the middle liquid stage!

Our inclination is good, too, so we intercept the Mun for 830 m/s delta-v and achieve a circular 55km orbit for only 300 m/s more. Almost 2000 m/s delta-v remaining for the trip home. Amazing margins.

(Also, I’m now the proud owner of a “first Kerbal to orbit the Mun” ribbon. Take that, Jeb! Not bad for a first mission, eh? Guess they can pin it on my frozen corpse when I get back.)

Kerbin-rise on the Mun

I orbit until I tell them I’ve seen everything I’m going to see, and then they fire up my engines and we break orbit. They’re going for the most efficient burn, one that ejects me out behind the Mun’s orbit so that I give up the most orbital velocity and have a cheap return.

They’ve got me lined up for a 100km periapsis above Kerbin, at which point we’ll circularise and then descend. It’s not technically necessary, but we’ve got tons of fuel to spare, and going for Low Kerbin Orbit (LKO) before deorbiting means I’ll have the smoothest and safest reentry possible.

My cynical side can’t help but wonder what they’re trying to protect — me, or my treasure trove of just over 411 Science? But, either way, smooth is good.

And that’s when the solar eclipse hits — on the return journey, when nobody’s expecting it.


We all knew there would be periods of no solar power. But the handy thing about orbits is, there’s a limit to how long the darkness lasts. In low orbit, you zoom around the planet back to the bright side. In high orbit, you’re slower, but the planet is smaller and doesn’t block the sun as long.

We did the math. The longest “darkness time” I could expect would be 23 minutes from the Mun, 38 minutes from Kerbin. Four hours of battery? Plenty.

Nobody considered what would happen if you ended up on a Munar return trajectory that just happened to keep the Mun in front of the sun for hours and hours on end.

Back at my lab, with all the trajectory details and my simulations, I could find out how much longer the eclipse would last for. But up here, I’m blind. I have this pencil, this paper, this flashlight — and with the nav computer down, no actual numbers to crunch.

If I ever live to go on another mission, I swear I’ll pay more attention to our orbital parameters next time.

Also, there had better be a backup generator.

Ah well, time to wrap this up before my hands freeze. This is Bob Kerman, signing off for good. Tell my kids their daddy was proud to die doing science.

[… various doodles …]
[… bad poetry …]
[… unreadable gibberish …]

Sunrise

OH MY GOD I HAVE NEVER BEEN SO HAPPY TO SEE A SUNRISE I’M COMING HOOOOOOME

[Log addendum: Korona 29 splashed down at 2.3 m/s due to overcalibrated parachutes. 411 Science recovered. Bob Kerman available for duty.]

Technical notes

If you ever need to know how many batteries to pack in your ship, use the KSP Darkness Time Calculator. I didn’t actually use this before the trip, but my own worst-case results were what I figured: under an hour of darkness. Just didn’t count on an eclipse!

Life support came back on with only 30 minutes remaining until death. I need to be more careful with antennas. While having an active antenna even on a manned ship is useful in case the crew dies and you need to remote control it, I also had two DP-10s that I should’ve automatically disabled when I automatically enabled the Communotron 16 at 75km altitude. Would’ve bought us 45 minutes or so.

I always wondered why anyone might want to use the Universal Storage hydrogen fuel cell generator. Now I know!

The FAR “DCA” flight assist is very handy for damping control inputs at high speeds and pressures, and it works with standard controls, MechJeb “Smart A.S.S.”, the RemoteTech flight computer, etc. In my experience, it’s still better to just avoid control surfaces on rockets altogether — use SAS or vectoring to tip the rocket over at low speed, then rely on fins plus high speed to keep you stable — but I haven’t researched any SAS technologies yet, so I kinda need it.

The “stuck launchpad” bug was almost certainly related to Kerbal Joint Reinforcement. There have been a lot of reports of trouble in the 3.x series, so while I fully expect Ferram will work those out eventually, I’m sticking with the older 2.4.5 for now.

Can we take the Korona 29 design to Minmus, too? Will Bob forgive us for sending him right back out again? What will we do with all this Science? Find out in the next episode!


I've chosen not to host a comments section here, but please feel free to contact me directly if you have something to say. I'm happy to hear what you think — whether it's about an article in particular, or just the site in general.