Orbit changes over time

[CanadianDutchman]

KSP 0.90

Windows 32bit not steam

every orbit I have is changing over time.

I have no thrust on, no RCS and my periapse keeps increasing and my opoapse keeps decreasing

this is a fresh KSP install I downloaded right before I launched this and made these screenshots

the craft is just a command pod, fuel tank and lv-909

it stops changing when I accelerate time

Edited February 4, 2015 by CanadianDutchman

[vexx32]

Is this change continual, or does the orbit simply fluctuate by a few tens of metres? Can you provide a log file for us to look at?

[CanadianDutchman]

http://imgur.com/a/Si14K

- - - Updated - - -

http://imgur.com/a/Si14K

it does seem to fluctuate over time

I know big space station or other floppy crafts may do that but this is a really tiny craft.

I also noticed on another save that it can change the inclination. maybe that also fluctuates back but this is quite annoying if you want to make precise changes for contracts. I have not yet visited other planets but it might also ruin those encounters with this

[vexx32]

What I meant is... does it continually increase/decrease, or does it simply increase for a bit and then decrease for a bit?

Either way, though, it is looking a bit odd. Can you upload a copy of your persistent.sfs file to http://justpaste.it/ so we can take a look?

[CanadianDutchman]

I hope I did this right http://justpaste.it/j8ql

[CanadianDutchman]

I have now decided to just go with it.. it seems to be less now for some reason. might be because I am ignoring it now. if there is a fix for this some time I would like to know

[DuoDex]

I have now decided to just go with it.. it seems to be less now for some reason. might be because I am ignoring it now. if there is a fix for this some time I would like to know

Doesn't seem like a major thing to me - the largest fluctuation I saw was ~30 meters.

[georgir]

I'm seeing this too and it is a major thing in this case. A small ion/solar powered probe on its way to minmus is having its APO reduced quite quickly and visibly over time, maybe by as much as 500-1000 m each second if I switch to the craft. Currently it is 48 120 000 m so as a percentage the drift is not too huge, but it still matters because if I set a maneuver node for my circularization it becomes visibly offset from the orbit in a minute, and after that if i try to edit it by a tiny amount it jumps up and shows wildly different resulting orbit...

Now, I can just wait for the probe to get closer to Minmus without it being active, i.e. in the tracking station. So it is not a problem that'll break my game in this particular case. Still it is something that'd better get fixed somehow... What is it due to? Some accumulating rounding errors? But when I dont have thrust, why isn't the game doing the calculations in the same manner that it does for inactive objects?

[CanadianDutchman]

I know this thread is a bit old now but it's still constantly doing it.. I just did a transfer to Minmus. the first screenshot is right after the maneuver node is done and the second one is 3 minutes later. yes I do now have mods installed but it's been doing this in stock too so the mods are not the issue here http://imgur.com/a/dPs7Q

[sal_vager]

There is nothing we can do about this issue CanadianDutchman, sorry, it's floating point errors when dealing with such large numbers, though the developers have tried to improve things and Unity5 is supposed to offer better floating point accuracy.

You will not have any issue with this when dealing with a contracts orbit, none of them are so precise or at such large eccentricity that these fluctuations will prevent them from begin completed.

[CanadianDutchman]

well.. it's gonna mess up my contract if I slam into Minmus.. but at least I finally have my answer why this happens.. thank you.. *waits for Unity 5*

[sal_vager]

Oh you'll have plenty of time to do a corrective burn if you need one, but as long as your burns are as accurate as possible you should find you're not too far off from where you want to be.

But at those distances a half second either way will have a large difference to your final orbit, and can be more of a difference than the floating point error.

[OddFunction]

Couldn't there be a check somewhere?

[Table of all crafts and their Ap and Pe]

IF

engines = off and rcs = off

AND

Orbit is not being modified by player (craft not selected)

AND

There is a change in 'Ap' or 'Pe'

THEN

Revert orbit back to value found just before the last time a player has selected the craft (an additional check goes here to make sure its correcting by an order of meters, so it doesn't do something wild like correct your orbit by a few KM to somewhere totally different).

Edit: I'm asking not because I realistically expect Squad to do it, but more because I'm studying computer science and this is an interesting thought-experiment for me

Edited June 5, 2015 by OddFunction

[sal_vager]

Well I dunno, your AP/PE is calculated based on your altitude and speed, and the more eccentric your orbit the greater the error can be.

[Greenfire32]

I have noticed that if a craft has a pretty substantial "wobble" to it, its orbit will behave in much the same way you've described simply because the "control from here" point is wobbling so much that the game can't pin it down accurately.

The best example I can think of is a probe core or capsule on the end of a station arm.

[r3_141592654]

This is a massive pain for trying to get geosynchronous satellite networks in place. I had one which was all set up at 2863333 give or take a dozen or so and Period 5h 59m 9s. I needed to do a 2 minute phase shift so I set up a node on Ap to burn retro for a 5h 57m 9s period. when I came back after a number of hours to do the burn, I noticed that my satellite was midway between Ap and Pe, but still at the correct Altitude of 2863km. I did notice that my Ap had shifted up to 2873km and Pe was like 2816km. Its almost as if Kerbin was moving while my orbit was stationary in reference to some absolute point in space. The Ap was continually getting larger very slowly and the Pe was continually getting smaller very slowly. I also notice that my period was increasing by about1s per couple of minutes. I was indeed recording video, but not sure where to post it and link to it. Maybe I can put it on Imgur? Odd thing I noticed was I had a lot of lag, like the MET timer was almost constantly yellow for whatever reason. I know it has nothing to do with time warping because I have never used time warp in the game ever. I even tried shutting down the LV-909 Terrier thinking maybe it was leaking or something. I was just brainstorming. My satellite is quite small, just the Terrier, a parabolic fuel tank and the 100G Relay Antenna, and 4 of the simple retractable 1x6 solar panels.  I put #2 up in the exact same way and didn't really notice any drifting and manage to get 120 Deg, 2863km, 5h 59m 9s period spot on. For some reason #3 is having these drifting issues. I first noticed this slight drifting in LKO, which I thought might have been some residual atmospheric drag since I was only at about 70.5km to 71km. I thought maybe they added a little tiny drag close to the cutoff of 70km. It didn't really matter since I am not in LKO for very long. These geosynchronous satellites need to hold position for a very long time since it is a bit of a pain to adjust and if you are trying to hit Ap or Pe while they are drifting it is near impossible.

 

Edit: Ok, Imgur has a 60 second length on video so I cannot use that.

Edited April 1, 2022 by r3_141592654
Add detail