Precision Burns - Before the Node, After the Node?

[Bosun]

I'm currently working on docking maneuvers, and cannot seem to do a precision burn...precisely.

Following the tutorial, I execute the burn on the manuever until there is EXACTLY 0.0 dv left. I look at the orbit, and my new trajectory is well off from the projected trajectory the maneuver was supposed to afford me.

Currently, on a 10 sec burn, for example, I've tried:

1) Burning AT the node, until I get to +10 secs from Node

2) Burning DURING the Node, from -5 to +5

3) Burning BEFORE the Node, from -10 to 0.

Which of these is the preferred, and accounted method? I cannot get an intersect because every time I place a perfect intersect node, no matter how precisely I burn, it never ACTUALLY lines up with itself.

Thanks for any tips!

Cheers,

Bosun

[FancyMouse]

2 is suggested (even by a stock tutorial?) as it's the most precise among these 3.

[Zephyriel]

You can never do the exact maneuver you calculated, because this would imply that you would get the entire delta v at the precise moment the timer hits zero, and not progressively over a certain period of time.

[Vanamonde]

To make the math manageable, all the formulas you'll see assume an instantaneous velocity change, which of course is not possible. The best option is your #2, to split the burn, half before and half after the precise time/location.

[ajburges]

Actually, given that TWR goes up during the burn, you want to burn more before than after to average the dV on the node.

Unless you have a mod that with do the calculus for you, just divide by 2 and give rounding remainder to the burn side. That will also allow you time to finish the burn on the dot with a lower throttle setting.

[katateochi]

Also adding my vote for No2.

But if you want to bulls-eye a target a bit of tweaking after the manoeuvre is often needed. I treat manoeuvre nodes as a rough guide and one thing I often do (especially for rendezvous) is in the last few m/s of the burn I will cancel the node so the intersection markers switch to being live rather than the projected result of the node. As I cancel the node I put the mouse over the intersection markers and watch the separation distance and kill the burn based on that. Then it's time to fine tune; I orientate prograde (and roll to 0 degrees) and then do some experimental thrusts with the RCS while watching how that changes the separation distance. If your manoeuvre was fairly good, then by doing some sideways RCS thrusts (which if you're oriented prograde & rolled to 0, equate to normal/anti-normal & radial/anti-radial thrusts) and some prograde/retrograde thrusts you should be able to get a much better separation distance.

[GluttonyReaper]

It's also worth remembering to hit Z to go straight to full throttle. Throttling up slowly can have quite an effect on accuracy with these shorter burns.

[Drew Kerman]

It's also worth remembering to hit Z to go straight to full throttle. Throttling up slowly can have quite an effect on accuracy with these shorter burns.

On this note, I would slightly counter that full throttle is not always the way to go. If your maneuver is only a few m/s (like correcting for that perfect Minmus encounter enroute after leaving LKO) consider a longer burn at only a small percentage of your maximum thrust by tweaking your thrust setting.

[ajburges]

Tangently related: what's better for pro/retrograde nodes. Burning on target marker or burning on pro/retrograde?

[-ELF-]

I commonly burn at pro/retrograde, cause in my view it will flatten out movement along orbit...

Also, you do not actually need to burn until 0.0, it is better to cut off as soon as your maneuer marker moves away from point where you burn to. And also - do not try to follow that marker when it moves away. just cut your burn and see what you got.

Hint:

Smaller maneuers are better done with RCS instead.

[Sharpy]

Half before, half after. The higher your thrust (shorter the burn) the more precise the effects.

If you burn for a purpose: circularize orbit, change orbital plane, lose all velocity (suicide burn), burn at node (prograde, normal etc) - as it will be followed during the burn keeping the thrust true to its given purpose.

If you burn for a target: encounter, specific orbit, specific landing location, burn at target node, as the node calculations describe the effects of this exact burn direction, regardless of how markers shift during the burn.

Also, when doing Oberth Maneuver close to planet surface, doing axial+prograde, split it into two phases: prograde before periaxis (which should be the maneuver node), axial after periaxis. (or you'll find yourself in for a face full of dirt as the axial component shortens your periaxis.)

[Laie]

You can never do the exact maneuver you calculated, because this would imply that you would get the entire delta v at the precise moment the timer hits zero, and not progressively over a certain period of time.

Yes, but: maneuver nodes have some built-in logic to compensate for this. During the burn, the maneuver marker on the navball is constantly updated to account for errors, be they because of imprecise steering or otherwise. Note that this is mitigation, not a solution -- but it goes a long way.

In low Kerbin orbit, interplanetary ejection burns of up to two minutes (burning one minute at either side of the node) still have near-perfect results. Interestingly enough, that two minute rule holds up even if you've been periapsis-kicking before and cover a lot more distance in the same time.

Edited June 15, 2015 by Laie

[Warzouz]

Yes, as Laie, I'm quite puzzled. I don't have that problem by burning 50% before and 50% after. Sure if you plan a Eeloo encounter, you'll priobably won't succeed, there is always a mid-course correction (<10m/s).

Sometimes I'm not too concentrated and I miss the 50%/50% and THEN my trajectory is a mess.

If you have a hard time with burning, use the "Execute Node" MechJeb's feature. It's quite precise.

Also beware that the stock burn time shown in KSP can be awfully wrong, especially if you're staging during your burn.

[John FX]

Also adding my vote for No2.

But if you want to bulls-eye a target a bit of tweaking after the manoeuvre is often needed. I treat manoeuvre nodes as a rough guide and one thing I often do (especially for rendezvous) is in the last few m/s of the burn I will cancel the node so the intersection markers switch to being live rather than the projected result of the node. As I cancel the node I put the mouse over the intersection markers and watch the separation distance and kill the burn based on that. Then it's time to fine tune; I orientate prograde (and roll to 0 degrees) and then do some experimental thrusts with the RCS while watching how that changes the separation distance. If your manoeuvre was fairly good, then by doing some sideways RCS thrusts (which if you're oriented prograde & rolled to 0, equate to normal/anti-normal & radial/anti-radial thrusts) and some prograde/retrograde thrusts you should be able to get a much better separation distance.

this gets my vote. Burn most of your burn then go map and do little test burns (or RCS) to see which direction you need to do your adjustment burns to refine your intercept. Just doing one burn I am lucky to be within 10km of my target. Doing this method I usually get within 100m of my target (in LKO for docking)

This scales well. I use the same method for interplanetary burns. Do most of the burn, then go to the map, focus on the planet you are aiming for and adjust until you get as close as you can from your current position. If you cannot get closer than a certain distance then redo the fine adjustment half way there (usually).

Sometimes you just can`t get close to your target in one burn due to the way orbital mechanics works.

[ajburges]

There is three exceptions to the split rule for nodes: circularization, rendezvous, landing.

Unless you want to circularize to a particular orbit, you want to do the circularization fully before Ap and you want to do it prograde instead of on node. The reason is to optimise your burn for kinetic and potential energy.

You can use a node to calculate the burn time needed to zero velocity during target intercept (useful if you regularly acquire 100 m or less intercepts ). It is fairly obvious why you would want to finish such a burn before you reach closest distance.

If you use a node to plan a suicide burn, you also want to complete the burn before the node. The reasons for this are also obvious.

[Leibniz]

Even intercept burns and suicide burns should be finished after the node time.

During the burn you will be slowing down; in fact your average speed will be halved across the burn time (average of initial speed and zero end speed). In theory you should split the burn half before and half after so you finish the burn at the node position. In practise you should start a bit earlier for extra control and safety.

I use this myself so I can stop and land at the node position from a ground-hugging ~5km orbit over the Mun.

For orbital manoeuvres you could try http://forum.kerbalspaceprogram.com/threads/88982-Java-Long-Burn-Calculator

I haven't tried it myself but it claims to convert long burns into efficient prograde/retrograde burns with accurate start and end times.

[Yasmy]

in fact your average speed will be halved across the burn time (average of initial speed and zero end speed).

There are two errors here, which can become significant depending on how much of your total mass you use in the maneuver:

1) If you are using a non-trivial amount of fuel, and not adjusting your thrust to account for decreased mass, your acceleration is not constant.

2) If your acceleration is not constant, your average speed is not the average of initial and final speed. (It is the time integral of your speed, divided the by the final minus initial time.)

Thus generally speaking, if you split the burn half way in time, you are spending more delta-v after the node than before the node.

If your total mass change is small compared to your total mass, the effect can be ignored.

Edited June 16, 2015 by Yasmy

[Yasmy]

Let's consider splitting the delta-v 50/50 around the maneuver node instead of splitting the burn duration 50/50 around the maneuver node.

Question: Given a burn of dV which takes time T, how long (t) does it take to burn dV/2? Let the burn consume total mass mf at a constant rate dm/dt = mf/T.

Note that dV is not your total rocket dV and mf is not your total fuel mass, but just the quantities used in this burn.

Full burn: dV = Isp * g0 * ln(m / (m - mf)) = Isp * g0 * ln(m / (m - dm/dt T))

Half burn: 1/2 dV = Isp * g0 * ln(m / (m - dm/dt t))

Solve for t, using the identity a*ln( = ln(b^a):

t = m/(dm/dt) (1 - sqrt(1 - dm/dt T)) = T (m/mf) (1 - sqrt(1 - mf/m))

Now approximate this for mf << m:

t = T/2 + mf/m T/8 + higher order corrections, all strictly positive

By inverting the rocket equation, mf/m = 1 - exp(-dV/Isp/g0):

t ≈ T/2 + (1 - exp(-dV/Isp/g0)) T/8

Starting your burn at time t before the maneuver node splits the delta-v 50/50 around the maneuver node.

I won't say which is better (yet). I'm just doing the math in case anyone wants to try another option.

[Vanamonde]

During the burn, the maneuver marker on the navball is constantly updated to account for errors, be they because of imprecise steering or otherwise.

Does this work properly, though? (I'm asking, not arguing.) I've tried both, and I seem to get better results by keeping my original heading throughout the burn than if I chase the maneuver marker as it moves. The burn seems to come out more closely to the original plan if I hold the original heading.

[Laie]

Does this work properly, though? (I'm asking, not arguing.)

Good question. I frankly do not know. From observation, it seems to work well to a degree, but it can only compensate for so much. Example:

That's 1750m/s split over five burns of about 2 minutes. After that was done, I needed a 3m/s correction burn. You may notice that I wrote a kOS script to place the nodes; I've been using it on several occasions an learned that timing is more important than direction. A cumbersome vessel may not be able to turn quickly enough between nodes, yet this doesn't matter -- just open the throttle at the right time and let MJ find the maneuver marker as quickly as it's able to. Not only will the current maneuver marker move to compensate, but the consecutive maneuvers will be updated, too, and in the end things work out quite well.

I have no idea how it really works. My attempts to willfully outsmart a maneuver had the oddest results. But it seems that maneuvers can compensate for some amount of error, and that going through several small maneuvers keeps the error within a threshold they can handle.

[xtoro]

Maneuver nodes do update, but not accurately. I think the position of the node adjusts according to the burn you originally wanted when you set it, and before you passed it. However, it does not update according to the changing mass/TWR of the ship, which I noticed by decoupling a large heavy stage (the node did not move). Therefore it does use the original mass/twr of the ship when you first set the node. For very long burns (using nukes for example) I will burn until I'm about a minute past the node (depending on how long I have left to burn), delete the node, and create a new one ahead if me and continue. It's annoying to recreate nodes but it's more accurate in the end. Also, if I ditch a stage midburn, I will delete and recreate the node.

However, if I'm feeling especially lazy, I'll just finish the burn and do a correction burn halfway through. But I do prefer the set-it-and-forget-it approach so I don't have to pay attention to that ship until the destination SoI...