To Mech-Jeb or not to Mech-Jeb, that is the question...

[Hojo_Norem]

I started playing KSP when the map screen was just an idea and palm trees could regularly found on the way up into orbit. My rockets then were stuffed with SAS to keep them straight. Then when ASAS arrived that was the first part to be placed under the capsule and with a combination of RCS and rudders I still kept my rockets straight. I found the less I had to manually keep then that way the better they flew. As for making a stable orbit, I managed that twice and only twice before it got 'meh, nothing else to do'.

As for the Mun, I managed a successful landing or two. Even managed to get back on one occasion. Then persistence arrived. My next failed mun landing left survivors. I had to rescue them. A few more attempts later and those survivors were enough to play a few games of 5-a-side footie.

Whats this got to do with MJ? As stated earlier on it allowed me to shift over to the mission control aspect more and when I saw the accuracy of MJ's landing autopilot pretty much every craft I throw into the sky has one.

My last few mun landings have been on semi manual (mainly because the autopilot don't like landers with barely enough dV land). Things like the translatron's 'keep serf' function for descent control and even super-simple things like kill-rotation, pro. and retro-grade alignment are a god-send. Even the orbit normal- and + functions make shifting my orbit around a process that doesn't involve sending by craft into the cold black yonder or accidentally lithobraking into the munar surface.

[trekkie_]

I totally agree with this statement. If you reduce your throttle 50% you cut your acceleration 50% therefore it takes twice as long to get to the same speed. At no point do you somehow magically increase the specific impulse efficiency of an engine allowing you to somehow travel farther on less fuel.

Any craft made in version .16, unless you downloaded the hacky fuel consumption fix posted somewhere around here, is experiencing a fuel throttle bug. They all are. Any throttle less than 100% is abusing the bug, and the lower the throttle the more you gain from it, because of the code that was accidentally left in place, as stated by the Devs.

You say my premise is silly, but you obviously have a misunderstanding of how specific impulse, thrust, and delta-v all relate to each other. Because thrust has nothing to do with delta-v requirements. At all. Specific impulse is the engines efficiency rating and that is what is used in delta-v calculations.(with fuel amount and total mass) Thrust only tells you if your craft has enough thrust to push the mass it is carrying to escape the affects of atmospheric drag and gravity. That is all. As long as you have a T:W above 1.5, you can then forget about thrust unless you need to know how long to burn your engine for a particular maneuver.

You would not be 'saving' fuel, you would be halving the fuel rate. There is a difference. The second craft at 50% throttle would merely be consuming fuel at the same rate as the first craft, yet they would travel the exact same distance on the same amount of fuel (all other things being equal). What you are saying is that somehow throttling down increases your fuel efficiency, and it doesn't. Throttling down would just increase the time it takes to perform the maneuver, yet it would still use the exact same amount of fuel.

You keep talking about 'saving' fuel. You aren't saving fuel. You are reducing thrust, which increases the length of your burn time, which is all proportional. It doesn't matter it still takes the same amount of fuel to get a specific mass amount into orbit. You can waste fuel by trying to accelerate too fast against the much thicker atmosphere, but once you pass Max-Q it doesn't matter. You can accelerate fast, or slow, as long as you maintain momentum to reach orbital velocity it will use the same amount of fuel.

I don't even know why you are trying to explain this. I completely understand how delta-v, mass, amount of fuel, and specific impulse all relate to each other. They aren't separate, they're all tied to each other. Changing one affects the other. Decreasing mass, Increasing amount of fuel or Isp, increases the delta-v a craft can achieve. Increasing mass, decreasing fuel, decreasing Isp, all decrease the delta-v a craft can achieve. You can muck about with the amount of thrust all you want but it won't ever change the delta-v a craft can achieve, because that is based on total mass, fuel, and Isp.

and here I will provide examples to prove you incorrect.

both test flights were done following the same exact course. straight up until 10K meters, then a turn that ends at 80K meters for a 100K meter orbit altitude. the only difference is that one was at full thrust the entire time, while another was at 75% thrust after hitting 10K meters.

Total delta-v expended: 4438

velocity:2491

gravity losses: 24.5%

drag losses: 21.2%

steering losses: 2.4%

speed gained: 52.3%

altitude: 51K meters

===

Total delta-v expended: 5433

velocity:3300

gravity losses: 21.2%

drag losses: 19.1%

steering losses: 2.8%

speed gained: 57.2%

altitude: 76K meters

same amount of fuel, different throttles, different heights, different delta-v's. NO fuel bug.

what this obviously means is I can reach the same delta-v and velocity using less fuel, which means more fuel to play around with for orbital operations and transfers. as long as you find the right balance of lowering thrust enough to keep velocity rising at a decent rate, you can use less fuel to obtain the same delta-v because you're taking advantage of your kinetic energy provided by previous thrusts and weight loss from fuel consumption.

http://en.wikipedia.org/wiki/Delta-v_budget

"Delta-v is a particularly useful measure since it is independent of the mass of the space vehicle. For example, while more thrust, fuel, etc. will be needed to transfer a larger communication satellite from low Earth orbit to geosynchronous orbit, the delta-v required is the same. Also delta-v is additive, as contrasted to rocket burn time"

don't quit your day job, NASA will not be looking for you any time soon

Edited August 19, 2012 by trekkie_

[ferram4]

Kerbtrek, I think you've just provided absolute proof that you are taking advantage of the fuel bug. That rocket should have the same delta-v no matter what throttle you use. Form wikipedia:

where V_exh = Isp/9.81 m/s^2. Your argument boils down to "Isp increases to infinity at low throttle" which is only possible using the fuel bug. Try using the Hacky Fuel Consumption Fix:http://kerbalspaceprogram.com/forum/showthread.php/16439-0-16-0-Hacky-Fuel-Consumption-Bug-Fix plugin and redo that test. I think you'll get a different answer.

------------------------------------------

On topic, I like Mechjeb for testing rocket designs or launching massive, lag-tastic space stations. Otherwise, I prefer manual control, since it can be more efficient, since Mechjeb doesn't do proper ascent or descent optimization, and because it's more fun to lithobrake anyway.

Edited August 19, 2012 by ferram4
Messed up the URL

[Tigga]

NO fuel bug.

YES fuel bug. The fuel bug exists if, at any point, you run with less than 100% throttle. By running at 2/3 throttle you decrease your fuel usage to 4/9 of max usage (2/3 squared). You're therefore effectively increasing the specific impules of your engines by 50% by using the fuel bug.

[softweir]

And to take it further, at 1/10^th throttle you consume fuel at 1/100^th the rate, getting 10 times the ÃŽâ€v out of the fuel.

This is a known, reported bug that has been fixed for release 0.17.

[Ziff]

and here I will provide examples to prove you incorrect.

both test flights were done following the same exact course. straight up until 10K meters, then a turn that ends at 80K meters for a 100K meter orbit altitude. the only difference is that one was at full thrust the entire time, while another was at 75% thrust after hitting 10K meters.

No, you did not, what you did was just prove that version .16 has a major fuel bug. You should probably now read this thread here-> http://kerbalspaceprogram.com/forum/showthread.php/15196-0-16-MAJOR-Fuel-consumption-is-wrong-at-low-throttle/

You might want to read that first post, and then the third from the last post where HarvesteR confirms it's a bug.

If you are a more visual person, you can watch a great explanation of the fuel bug by Scott Manley here->

Just skip to the 3 minute mark where he fully explains what is going on.

You can also try downloading the demo, version .13, and try the exact same thing you just did and you will see how it really works.

http://en.wikipedia.org/wiki/Delta-v_budget

"Delta-v is a particularly useful measure since it is independent of the mass of the space vehicle. For example, while more thrust, fuel, etc. will be needed to transfer a larger communication satellite from low Earth orbit to geosynchronous orbit, the delta-v required is the same. Also delta-v is additive, as contrasted to rocket burn time"

don't quit your day job, NASA will not be looking for you any time soon

The line above just explains how it doesn't matter how large a vehicle is, or how heavy it is, it still needs to change it's velocity (delta-v) by a certain specific amount in order to perform a specific orbital maneuver. However, you must also understand that the delta-v a ship is capable of achieving (which is the exact wording that I used) is a completely different story my friend. The delta-v a craft can achieve is wholly 100% based on the total mass, fuel amount, and engine Isp. Changing fuel, mass, or engine Isp changes the delta-v it can obtain. This is how NASA creates delta-v budgets to figure out if a spaceship can actually get where it needs to go. They actually use, you know, math, to figure this out. You too can use Tsiolkovsky's rocket equation to figure this out, which you can also read here-> http://en.wikipedia.org/wiki/Tsiolkovsky_rocket_equation

On Topic After reading another thread about the Apollo computer, I would absolutely love to see an in-game version of this. A very basic computer that can be programmed in a simple manner to perform basic orbital maneuvers would be neat. However, I still love MechJeb for all the information it provides. My skill as a pilot greatly increased when I could focus all my attention at the MechJebs informational windows. Before I used MechJeb I had more than once accident where I crashed into the surface of Minmus because the mountain was over 2km tall!

For me, the best part about MechJeb was it allowed me to concentrate on important things, like canceling horizontal velocity during landing, rather than trying to manipulate the camera and see how close the surface was.

[Binky]

I generally use MechJeb for the launch phase sometimes and for various bits of info, and for the Kill Rot function which is IMO the best thing in the universe ever. But it isnt flawless.

When I launched my Mun base I used the Ascent autopilot, which worked well, and tried to use the landing function on the Mun (during my orbit I found my first mun arch and wanted to land near it) but MechJeb had a total meltdown and couldnt seem to do anything right.

So I landed manually 2km from the arch and it was the easiest Mun landing I've ever done. Mun Base Alpha flies really stably by hand, but its buggy in odd ways* which was causing the autopilot issues I think.

*The 47 ton station used 0.1 tons of fuel between LKO and Mun landing, and another 0.25 to return and land at KSC.

And Jeb got stuck in the crewtank when trying to go on his 3rd EVA (hatch obstructed. by the space kraken?)

[Maxed-Rockets]

I like the orbital info and twr's at my fingertips. I used to be mechjeb heavy, and it is still included on my ships, but ever since 0.16 I seem to be the only force in the universe that can keep anything flying in a straight line. The landing autopilot is also buggy, many a munar station have been lost.

Handy tool if I feel lazy, especially since ship design is my favorite pat of the game, but you need to...

1. Land on the Mun

2. Land on Mimmus

3. Make basic orbital maneuvers including circularizing orbits, changing pe and ap, and rendezvousing with an object...

Before you get a mechjeb license. You never know when a stray bit of debris might accidentally knock you mechjeb capsule off.

EDIT: I have been known to use mechjeb when lag is excessive to the point that my control inputs are showing up too late to be useful.

Edited August 20, 2012 by Maxed-Rockets

[Polkavenger]

MechJeb is pretty much a 'must' for me if I'm trying to put something big into space. I'm on an old laptop and during the bigger launches it's not at all unusual to get 1 frame per second or less.

But I do understand the sentiment of doing things yourself instead of letting a computer do it. I do things (orbits, landings) manually for the first time (in a new version or after a break from KSP). But when I start bigger projects or start approacing "real" science instead of kerbal science, I use computer assistance.

Plus, I find MechJeb to be a wonder itself. The first time I used the automatic landing on the Mun I was almost as nervous as the first time I landed manually. I couldn't believe it worked, just like that. I imagine if I looked at the source code of MechJeb, I'd just see swirling torrents of magical energy.

[Nachtwind]

WIthout reading the previous posts:

For me Mechjeb is a perfect tool for understanding some maneuvers. When i was new to KSP i tried to just get my craft into orbit by speeding as far as it got and when therespeed some more to get a perfectly elliptic orbit that goes beyond mimmus but still circles arround Kerbin.. Some friends pointed me to Mechjeb and i didnt want to use it since i thought it was sort of cheating. An "Autopöilot" to do whatever i want.. well, after some talking i installed mechjeb and hit engage to get into orbit..

i did crash into the ground.. a few times..

I still have to build better rockets for Mechjeb to work, but once a rocket makes it into space i was able to see how IT makes it go and speed etc.. Now i can get a stable orbit easily myself but keep using Mechjeb for ascending due to the simple fact that i can easily test new designs without having to worry too much about steering it.. if the design is ok it will work.. if not.. well.. i know soon enough then and cant say if it was my skill of flying or not..

So imho Mechjeb is a great tool

[drinniol]

My mechjeb landing autopilot is set at 'CFIT' setting haha

[BurningSky93]

My personal reasoning for using MechJeb is thus:

Real rockets do not fly without computer assistance, why should I?

As far as I'm concerned, MechJeb isn't really cheating. You still need to design and build a rocket that will do what you want it to do, MechJeb won't tell you whether or not your rocket can complete the mission. MechJeb also isn't perfect, it does mess up maneuvres and can be very heavy handed. I've seen it go overboard with gravity turns on ascent autopilot, sending rockets into a tumble; even if they recover, then precious delta-v has been lost. I've used the injection autopilot only to be put into collision courses when I wanted a periapsis 500km out.

I don't see it as taking away from the fun of controlling the craft yourself. You still need to input values into MechJeb for it to work, not to mention I find fully manual control tiresome and somewhat frustrating (as I don't have a joystick, I'm stuck with purely digital controls which are imprecise). The Smart A.S.S. function allows you to put the craft on the heading you want without having to see-saw and muscle about with manual controls.

Finally, as I already understand most of the theory that comes with the maneuvres, I don't feel like MechJeb has really "taught" me what some might feel you need to learn on your own.

Edited August 20, 2012 by BurningSky93

[illectro]

Look MechJeb should become part of the main install, and the zero weight radial components might be worth dropping. And hey in campaign mode it can be a whole tech tree on its own.

Seriously, I can fly any mission by hand if I want to, I've demonstrated that I can fly missions that would actually be impossible in MechJeb because of its flawed Ascent and Landing autopilots which have a habit of burning too much fuel. MechJeb sometimes causes problems when you use it, my 700+ prototype interplanetary rockets have a habit of shaking themselves apart if I trust the launch to MechJeb, and it burns up all your RCS fuel if you turn it on at the same time as RCS is enabled. The original parts also add mass so you pay a price for the extra convenience.

When I'm making videos I like to have MechJeb up even if it's just for the orbital/surface information, it makes it easier to explain whats going on and saves me having to fly from the map screen. Sometimes I have to hold a microphone in one hand (because headset mics sound terrible) so flying with one hand is a lot easier if I can just push buttons on the Smart A.S.S. - and hey, last night I burned all the fingertips on my left hand so I can't really play well with 2 hands right now.

[Tigga]

I wouldn't want to see MechJeb as it stands as part of the main install, nor do I ever expect it to happen. I do expect some improvements in the automation, such as a more stable ASAS, or one that can control thrusts levels on engines independently to compensate for imbalances. I also expect some of the information that MJ gives to be transferred to the default HUD in some way.

All the more advanced auto-pilot functions - launching, landing, transfers, exact orbiting etc - I'd prefer to see them left out of the core game. As a gamer one could argue "but it's optional, so why not include it?", however one of the skills of being a good game developer is working out what is needed and what is not. You have to put in place restrictions, otherwise the sense of achievement for a gamer playing the "stock game" is diminished. If the "add everything because it's optional" argument is valid surely we should have infinite fuel tanks, or indestructible components available as default? Optional, but very likely to ruin the fun for a lot of new players.

[AncientAstronaut]

and here I will provide examples to prove you incorrect.

NO fuel bug.

what this obviously means is I can reach the same delta-v and velocity using less fuel, which means more fuel to play around with for orbital operations and transfers. as long as you find the right balance of lowering thrust enough to keep velocity rising at a decent rate, you can use less fuel to obtain the same delta-v because you're taking advantage of your kinetic energy provided by previous thrusts and weight loss from fuel consumption.

http://en.wikipedia.org/wiki/Delta-v_budget

"Delta-v is a particularly useful measure since it is independent of the mass of the space vehicle. For example, while more thrust, fuel, etc. will be needed to transfer a larger communication satellite from low Earth orbit to geosynchronous orbit, the delta-v required is the same. Also delta-v is additive, as contrasted to rocket burn time"

don't quit your day job, NASA will not be looking for you any time soon

Oh my gosh, fail... Fail, fail...

That's not how Delta-V works. Specific Impulse will increase from vacuum to atmospheric pressures, but throttle has nothing to do with it and the increase isn't too much. For example, Isp of a non-vectoring engine is 320 in an atmo. and 370 in a vacuum.

The formula for Delta-V says nothing about throttle. So if you want to continue with what you're saying, you're saying Tsiolkovsky was wrong and via a GLITCH in the game, you've proved it...

Other people have already given you the "Rocket Equation" so I doubt I have to here... But seriously... Everyone here is right. And you are wrong. And what the heck is up with this crack?:

don't quit your day job, NASA will not be looking for you any time soon

Look who's talking.

[TheCardinal]

[quote name=Tigga;242451

...

All the more advanced auto-pilot functions - launching' date=' landing, transfers, exact orbiting etc - I'd prefer to see them left out of the core game. As a gamer one could argue "but it's optional, so why not include it?", however one of the skills of being a good game developer is working out what is needed and what is not. You have to put in place restrictions, otherwise the sense of achievement for a gamer playing the "stock game" is diminished. If the "add everything because it's optional" argument is valid surely we should have infinite fuel tanks, or indestructible components available as default? Optional, but very likely to ruin the fun for a lot of new players.

I consider that an invalid argument. If a player wants to have infinite fuel or indestructible components the player merely has to alter the part.cfg of the required components. There is nothing which prevents the players from doing so. The developers left that choice completely to the players.

Restrictions can enlarge the sense of achievements, true. However if the difficulty exceeds the capabilties of the players (or makes the learning curve to steep), players tend to loose interest and move on (to a different game). The game provides sufficient challenges, even with MechJeb.

I would applaud when Mechjeb would become a part of the main install. It wouldn't deminish the challenges, but it certainly would make it more likely that MechJeb functions correctly when a new KSP version is launched.

[Tigga]

I consider that an invalid argument. If a player wants to have infinite fuel or indestructible components the player merely has to alter the part.cfg of the required components. There is nothing which prevents the players from doing so. The developers left that choice completely to the players.

To use MechJeb the player merely has to download it off the internet and follow some simple instructions. I'd say probably a lower barrier than editing a file, but they're fairly analogous. At the moment if you want to play with these things to make it easier you have to go out of your way to do it. Having it so that the default is easier is very different.

It all comes down to the effort required to enable a change. If it's in by stock no effort is required, everybody will be playing with MechJeb, a large amount of the restrictions will be removed from default play and players will be missing out in the challenge. A lot of people will design a rocket, land on a far away planet using MechJeb, and put the game away saying it's too easy. Not all people, as people find different things fun, but a significant proportion of people. As soon as there is a barrier - explicitly downloading something to change the game, or editing the parts configs - you are deliberately altering the game. This is very different from having things on as default. In one the game is too easy, while in the other you are making the game easier for yourself. My main point above was that arguing that "options are good and if you don't want it don't use it" while sounding rational, doesn't actually take into account how humans work.

[rkman]

All the more advanced auto-pilot functions - launching, landing, transfers, exact orbiting etc - I'd prefer to see them left out of the core game.

Then you might be disappointed by what the final game will apparently have to offer in the form of rocketry learning aids, including mission planning with capabilities somewhat similar to real space mission planning, which is not "flying by the seat of your pants". Then again, it looks like using those tool will be optional, not the default.

If the "add everything because it's optional" argument is valid surely we should have infinite fuel tanks, or indestructible components available as default?

I'd say it's not exactly optional if it's the default.

Rather the option would be to make parts not indestructible and to not have infinite fuel.

[Tigga]

Then you might be disappointed by what the final game will apparently have to offer in the form of rocketry learning aids, including mission planning with capabilities somewhat similar to real space mission planning, which is not "flying by the seat of your pants". Then again, it looks like using those tool will be optional, not the default.

You can do mission planning without anything external controlling the direction your ship faces, the staging or the thrust levels you're at. Icons on the navball, or a target trajectory in the map mode (ie. tell the mission planner you want to get to a place, and rather than doing it for you it shows the trajectory on the map mode, for you to try to achieve). Still up to you to make sure the nose is pointing the right way, to make sure the burn is for the right amount of time, to stage, to make mid-course corrections if you've slightly mistimed or misdirected something, ... basically to control the craft efficiently. Planning is completely different from an autopilot.

I'd say it's not exactly optional if it's the default.

Rather the option would be to make parts not indestructible and to not have infinite fuel.

Well you could have the current ones, and the silly ones and be able to pick between them. Of course it's a silly idea, but it illustrates my point about MechJeb being available by default.

[SNapple]

I can only imagine it would be good for testing out things.

[rkman]

You can do mission planning without anything external controlling the direction your ship faces, ... Still up to you to make sure the nose is pointing the right way...

Well you could have the current ones, and the silly ones and be able to pick between them. Of course it's a silly idea, but it illustrates my point about MechJeb being available by default.

As i said: full automation (and if Squad wants to be silly, cheat parts) will (very likely) be optional, not default.

As long as the player can choose not to use it, i don't see the problem with including autopilots etc.

[Bandock]

MechJeb I have found after some usage can be quite handy for even manual maneuvering as you get additional information that normally would require calculating by hand. Have not used any form of auto-piloting from MechJeb as of right now, but did use the information features as stated in my first sentence to be handy.

It is very likely we'll need some form of auto-pilot for at least tasks like orbital rendezvous (considering you have to know where the orbiting objects are at, even without a map!), reaching the moon, returning to the space capsule from a lunar module upon rendezvous, etc.

[CobraA1]

I guess it depends on the plans of the developers and how they perceive the role of the player - do they see the player as a mission planner, or do they see the player as a pilot? Or perhaps both?

One thing I've noticed practicing orbits around the sun is that I really don't do a good job of planning orbital transfers. I can see already that planning interplanetary transfers will be much harder than planning a moon transfer. I don't think I could eyeball it. As this game scales up, it's going to very much require some automation if they want to keep the game accessible to most people.

[r_rolo1]

I guess it depends on the plans of the developers and how they perceive the role of the player - do they see the player as a mission planner, or do they see the player as a pilot? Or perhaps both?

That is pretty much the crux of the argument. I can understand how people that play KSP mostly to pilot rockets and have the necessary motor skills and maybe fine tune hardware for that will probably have more fun doing a full human controlled mission, but the people that have inate or hardware related issues controlling the rockets would be heavily frustrated with the game fast if there isn't any way of removing that stump block of their path and the persons that play the game to build rockets would probably dislike that emphasis in hard-stick piloting. And let's be honest, RL automation is here for a long time ... in fact the overwhelming majority of the global lot of space missions made so far was made in a automated or at least in a semi-automated fashion, even if you don't include the routine sat launches :/

One thing I've noticed practicing orbits around the sun is that I really don't do a good job of planning orbital transfers. I can see already that planning interplanetary transfers will be much harder than planning a moon transfer. I don't think I could eyeball it. As this game scales up, it's going to very much require some automation if they want to keep the game accessible to most people.

Agreed. Orbital transfers are a precision issue by itself and only the fact that the relatively big angular dimension of the Mun and Minmus SoI seen from Kerbin makes it possible to use eye balling with consistent success for most people ....

[Bluejayek]

As per mechjeb being available in the stock game in its full form, I would figure it should be the similar to how cheat codes exist in many single player games. They exist for those who find it more fun to use them, but there should be some slight barrier to using them, perhaps having to edit one line of the settings file where it says "Mechjeb="OFF"" to on, and then it will appear in the VAB. This would prevent ruining new players experience with a part that takes away the fun of flying, while allowing those who know a bit of what they are doing to use mechjeb when they wish.