Having a bit of trouble with interplanetary transfers

[godefroi]

I've successfully made interplanetary transfers (manually to Duna, MechJeb has gotten me to Eve and Duna as well), and I think I'm doing it the hard way.

My normal SOP lately is to put a protractor (mod) on my ship, then wait until protractor says my launch window is up. I park in a 75km orbit, then wait for the angles, and make my burn (which is usually around 1050 m/s). This pretty much never gets me an intercept, though. It doesn't even really get me very close. I play around with maneuver nodes, can't nail down an intercept, and then tell MechJeb to tune my approach. MechJeb then wants to make a 400-500m/s burn to fix my approach. From everything I've read, this combined total of ~1500m/s dV is much more than I should be spending to get into a Duna intercept.

Also, I've tried using Kerbal Alarm Clock to wait for transfer windows (both modes, which don't agree), and it doesn't agree with Protractor on when transfer windows are, and none of them agree with MechJeb on when transfer windows are.

What is it I'm doing wrong, here, and, for my own curiosity, how "wide" are the transfer windows? If I attempt a transfer a few days to either side, how much does that hurt me in terms of required dV?

I'm sure these questions come up all the time, but I wasn't able to find any succinct answers, so I appreciate any help the collective wisdom could provide.

[Lazro]

http://alexmoon.github.io/ksp/

A fairly simple tool for figuring out transfer windows. Sorry for piling another tool to decide by, but I've seen it work really well.

It's often imperative to transfer ASAYOTWO (As Soon As Your Optimal Transfer Window Opens,) and missing it by a few days on either side can be disastrous. Just remember that mods are often slightly inaccurate and what's best is to learn the angles

[Oggula]

If I were you I would set a maneuver node to get an intercept from Kerbin itself, not from outside its sphere of influence when you may be on a divergent path. Then start your burn so that half falls before maneuver node and half after; do not start at the node, start before.

Of course this is quite time consuming. It took me 15 minutes to do a node perfectly the first time I tried, but now it takes around 5. You should be getting to Duna with around 1030-50m/s of dv. It also helps when you know from experience when the transfer window is (around 40 days for Duna) as it can be difficult to tell phase angles from looking.

[godefroi]

I've actually got that tool open in another window here. It says it's based on KSP 0.22 data, I assume that remains accurate for 0.23?

How does one measure the angles accurately other than cutting out triangles and taping them on the monitor?

[Lazro]

Kerbal Protractor should tell you the planetary angles...correct?

[Oggula]

It is the same phase angles in both versions. And I use the phase angle tool on Mechjeb when I need to, but I find I can usually do it visually for all transfers except Moho, as the outer planets have much wider transfer windows

[godefroi]

Protractor tells me angles, yes, I've been using the "adjust X" option where you just wait until it's zero. I didn't realize that MechJeb would tell me angles, I'll have to look into that. Mostly MechJeb seems good at turning my nice rockets into fireballs, I've noticed.

When you say "much wider", just how wide, for example, is the Duna window? Is there some way to graph the dV costs over time before, during, and after the transfer window?

Scott Manley makes it all look so easy, but I guess that's what you get for spending years studying this stuff before you start playing the game...

Also, I tend to use LN-V for transfer stages, which means I don't have a very high TWR, leading to long burns. When I'm making my transfer burn, I start at prograde, but do I keep pointing at prograde through the burn, or do I keep pointing where prograde was when I started the burn? How much difference would that make?

[Lazro]

In my experience you should keep pointing in that direction without changing your attitude.

[godefroi]

Ah, there's me. Protractor says I've got a few days yet to my transfer window, have MJ plot it, and it's 237 days out, and not even close to an intercept anyway. Grr! (15:25 in the video)

[godefroi]

When you say "much wider", just how wide, for example, is the Duna window? Is there some way to graph the dV costs over time before, during, and after the transfer window?

Aha, found exactly what I was looking for: http://alexmoon.github.io/ksp/

The Duna window, assuming you're willing to spend, say, an extra 100m/s of dV, is something like 3 weeks wide, according to that. The other windows are significantly smaller given that kind of dV constraint.

[Lazro]

Aha, found exactly what I was looking for: http://alexmoon.github.io/ksp/

The Duna window, assuming you're willing to spend, say, an extra 100m/s of dV, is something like 3 weeks wide, according to that. The other windows are significantly smaller given that kind of dV constraint.

Just noting I had posted that earlier...

...

[adinfinitum]

Also, I tend to use LN-V for transfer stages, which means I don't have a very high TWR, leading to long burns. When I'm making my transfer burn, I start at prograde, but do I keep pointing at prograde through the burn, or do I keep pointing where prograde was when I started the burn? How much difference would that make?

That might be part of your problem. The maneuver nodes assume that you are applying the change in velocity in an instantaneous moment, which works fine for burns that are a very tiny portion of your orbit period. When you get into longer burns you need to make sure that you time the burn so that it is already halfway done when reach the cross the node. During the burn your ship should always be pointed towards the blue marker on the navball. If you do this your final orbit should get you an intercept fairly easily, without requiring to much correction on the way.

[godefroi]

Just noting I had posted that earlier...

...

Right. When I saw the URI I assumed it was the OTHER transfer calculator tool: http://ksp.olex.biz/

Yeah, yeah, me, umption, all that. Sorry for not paying more attention.

[godefroi]

That might be part of your problem. The maneuver nodes assume that you are applying the change in velocity in an instantaneous moment, which works fine for burns that are a very tiny portion of your orbit period. When you get into longer burns you need to make sure that you time the burn so that it is already halfway done when reach the cross the node. During the burn your ship should always be pointed towards the blue marker on the navball. If you do this your final orbit should get you an intercept fairly easily, without requiring to much correction on the way.

Protractor doesn't give you a node, and the help says only, "burn prograde". I guess what I need to do is actually set up a node when using protractor.

[adinfinitum]

Ah, I had assumed you were setting up maneuver nodes for the transfer. If you don't want to use maneuver nodes, you should still be able to eyeball it fairly well if you know how long your burn will take.

The same stuff from before still applies. Try to time your burn about half and half before and after the ejection angle. For every minute you travel through a low Kerbin orbit, your attitude relative to the prograde horizon will increase by about one degree. For example, with a ten minute burn, you would want to start about five minutes before the ejection angle and facing a little more than 5 degrees below prograde. This should get you fairly close to an encounter.

[godefroi]

Something must be wrong either with my game or with MechJeb, at least. I had it plot a node for a Moho transfer (it agreed with KAC that a window was near), and it burned ~1650m/s as expected, then, once I'd gotten into the sun's SOI, I had it plot a course correction, and it wanted to burn another ~1400m/s. That can't possibly be right.

[Claw]

Something must be wrong either with my game or with MechJeb, at least. I had it plot a node for a Moho transfer (it agreed with KAC that a window was near), and it burned ~1650m/s as expected, then, once I'd gotten into the sun's SOI, I had it plot a course correction, and it wanted to burn another ~1400m/s. That can't possibly be right.

Make sure you aren't time warping too fast through SOI changes. Sometimes the errors when that happens can cause your trajectory to change.

High warp just approaching the moon can change you to an impact or sometime launch you interplanetary.

[Wooks]

I prefer this one: Interactive illustrated interplanetary guide and calculator for KSP BY OLEX, it will tell you how and when to leave on your transfer pretty accurately and your general DeltaV, I´ve gone everywhere and back in the Kerbol system with it.

Find it here: http://ksp.olex.biz/

The calculator was based on this thread: http://forum.kerbalspaceprogram.com/threads/16511-Tutorial-Interplanetary-How-To-Guide

Edited February 21, 2014 by Wooks

[Kasuha]

Something must be wrong either with my game or with MechJeb, at least. I had it plot a node for a Moho transfer (it agreed with KAC that a window was near), and it burned ~1650m/s as expected, then, once I'd gotten into the sun's SOI, I had it plot a course correction, and it wanted to burn another ~1400m/s. That can't possibly be right.

It's hard to tell without screenshot of the trajectory but if the correction was supposed to be an inclination match then you need to do it at the right spot.

[m4rt14n]

This is not really mechjeb's fault. For low, or very low TWR your maneuver nodes would need correction(s) after finishing the thrusts. Try it yourself, create a node to get an encounter or use mechjebs manuever planner, but dont execute it. Then do the burn yourself.

You'll notice that the path of the ship differs from the intended path. The thing to do is after the burn is finished make another mechjeb maneuver planner node to correct the intercept (forgot the name) to a small periapsis. Or better make the node yourself. DO NOT however wait until the end of the path to do this, because the closer you are to the start of the arc, the stronger your corrections will be. I usually do this maybe around a third or half-way. If you have RCS, try to burn using that because it is easy to overshoot with your main engine.

[godefroi]

The thing to do is after the burn is finished make another mechjeb maneuver planner node to correct the intercept (forgot the name) to a small periapsis. Or better make the node yourself. DO NOT however wait until the end of the path to do this, because the closer you are to the start of the arc, the stronger your corrections will be.

That's exactly what I did. MJ wanted to make another burn nearly as big as the original ejection burn.

[alexmun]

Something must be wrong either with my game or with MechJeb, at least. I had it plot a node for a Moho transfer (it agreed with KAC that a window was near), and it burned ~1650m/s as expected, then, once I'd gotten into the sun's SOI, I had it plot a course correction, and it wanted to burn another ~1400m/s. That can't possibly be right.

The problem here is Moho. MechJeb's transfer maneuver generator calculates a Hohmann transfer (just like Protractor, ksp.olex.biz, etc). This works great for Eve, Duna, Jool or other bodies with orbits that are mostly circular with low inclinations. It doesn't work so well for Eeloo, Dres, or especially Moho. MechJeb's midcourse correction maneuver does properly account for Moho's eccentricity and inclination so it has to use a lot of delta-v to fix the bad ejection burn.

It's possible that a midcourse correction somewhere else along your transfer orbit could be substantially cheaper (I've had what would have been a 150m/s burn drop to 7m/s just by changing when the burn was executed), so you might try it at different points along your transfer. However, if you want to create a truly efficient transfer to Moho you're going to need use a different strategy than a Hohmann transfer.

For example, you could use a gravity assist from Eve to help lower your orbit. You could use a pork chop plot generator, like my launch window planner, to calculate a non-Hohmann ejection burn. I've even seen someone recommend just targeting your transfer at Moho's periapsis then waiting however many orbits it takes to get a close encounter.

Finally, be aware that your Moho capture burn can be prohibitively expensive unless you have a very good transfer that ends up with a small amount of relative velocity at intercept. A bad transfer can easily leave you with a 5000m/s or greater capture burn. Also, there's no need to match Moho's inclination. You just need your transfer orbit to intersect Moho's at the point of intercept.

[godefroi]

Here's an example of an attempt I made to set up a transfer to Dres. I'm using VOID here to give me angles. I'm still a ways out from my ejection angle, but my phase angle is just about there, and I've guesstimated the ejection angle and set up a node of roughly the right dV (the only value I found which gives me "closest approach" indicators:

And, here's the closest approach. As you can see, "closest" is a relative term. My closest approach is something like 1/4 orbit away, which isn't really "close", even if you're counting horseshoes and hand grenades.

MechJeb wanted to wait another few weeks for the maneuver, and the node it set up not only didn't have "closest approach" indicators, it didn't really come anywhere very close to Dres' orbit. I played with it for a while, and I got the closest approach there down to something like 600k km, but I was approaching 3k m/s dV by that point, and it wasn't an encounter, just a "closest approach".

Assuming I've gotten this far, what is it I'm doing wrong? I've played with my ejection angle, but nothing I tried got my closest approach any closer. I fear I'm missing something somewhat fundamental here.

[godefroi]

The problem here is Moho. MechJeb's transfer maneuver generator calculates a Hohmann transfer (just like Protractor, ksp.olex.biz, etc). This works great for Eve, Duna, Jool or other bodies with orbits that are mostly circular with low inclinations. It doesn't work so well for Eeloo, Dres, or especially Moho. MechJeb's midcourse correction maneuver does properly account for Moho's eccentricity and inclination so it has to use a lot of delta-v to fix the bad ejection burn.

You, sir, are a god among men. I used your porkchop plot (well, I used the fantastic information to the SIDE of the beautiful plot...) and achieved a 15,000km closest approach (without a course correction, yet). Took a little more dV that your tool said, but hey, it's better than being 1/4 of an orbit away. Once I get out into the Kerbol's SOI, MJ plots me a 6.9m/s burn to get me into Moho's SOI. Fantastic.

Now, to install a mod that gives me ejection angles for my maneuver nodes, and I'll be all set. Thank you for the very helpful information.

I guess where I went wrong was in doing a few transfers to Duna, figuring I had it down, and installing MJ. I let it take me to Duna a few times, and Eve once, and then it completely broke down when I tried to have it get me somewhere more interesting. That'll teach me to take the easy way out.

[godefroi]

And, just a few days later, a near-perfect intercept at Eeloo's AN. Color me very impressed.