Decoupled/Undocked satellite disappears or is moved to a pole and falls

[hazelnut2]

I have been fruitlessly trying to design a craft that would, in the same run, place a satellite in a polar orbit on Mun and also have a main, manned cockpit for other chores around the Mun. My issue arises when I decouple or undock the little satellite on top. When I used a stack separator the satellite did not become controllable. I selected the "debris" from Mission tracker, however it became offset to what seemed like a Mun's pole (way far from the original part) and it just fell straight to the ground.

I, then, used a docking port (actually two, at first, but noticed they would become stuck, so decided to use just one!) and performed the decoupling using the port in Kerbin's suborbital flight. Now the satellite would first be controllable, however as soon as I moved back to the main craft it would just disappear.

I am trying to understand what is happening: is this a bug with the game, am I incorrectly building such a craft or is there something else I'm not seeing?

Game is vanilla, no mods, Steam under Linux.  Version 1.11.0.3045. Otherwise no issues. In the accompanying image, the satellite is right on top of the Mk1 capsule (which is manned).

 

Edited January 5, 2021 by hazelnut2
Added more platform details

[paul_c]

A satellite needs the following:

A controller module of some kind
An aerial (this might be the built-in one in the controller module if its close enough to a relay or Kerbin)
A battery (the built-in might be enough)
A source of power, eg solar panel(s)

Also if you're not even including an engine/fuel tank, then bear in mind that there will be no control over its orbit. In theory you can launch it in precisely the right orbit but in practice, there is the ejection force of the decoupler adding a good few dV and also, because satellites are relatively light and only maneouvre in orbit, you don't need much fuel and the engine can be tiny.

I can't see the pic very clearly but I am wondering if its lack of comms, lack of battery power, or both.

Top tip for power: I've tried skimping with 1 solar panel......if its shaded by something, you're stuck because a chicken/egg situation develops where the satellite dies in the shaded position, then you can't move it to put the panel into sunlight. So for even the smallest/lightest satellites I now put 2 panels on.

[hazelnut2]

6 hours ago, paul_c said:

A satellite needs the following:

A controller module of some kind
An aerial (this might be the built-in one in the controller module if its close enough to a relay or Kerbin)
A battery (the built-in might be enough)
A source of power, eg solar panel(s)

My design has all of those (provided that by "aerial" you mean an antenna). The controller is the Probodobodyne Stayputnik, there is an antenna right on top of the controller, there is plenty of battery available and there are two solar panels. Yet, the above problem persists. Can anyone reproduce this?

Edited January 5, 2021 by hazelnut2

[VoidSquid]

I don't know where the thread is, but I'm pretty sure that I have read about the exact issue before...

[jimmymcgoochie]

I’ve had that “detached parts teleport to the South Pole and fall vertically” bug before, sometimes when loading a space station and it decides to break itself in half but on other occasions when I was deploying a relay network using the orbit cheats to space them evenly, some satellites would decouple then immediately vanish whereas other identical satellites on the same deployed would work exactly as expected. It’s a bug, no idea why it happens but you’re not the only one that’s had this issue.

Edited January 5, 2021 by jimmymcgoochie

[hazelnut2]

To me it seems this is a very serious issue that should be addressed, given the fact that I am not running any mods, nor using any cheats. It's as stock as it gets. I have just read in a different thread that there is a developer console that could perhaps be useful in case it does catch something. I shall give that a try tonight.

[VoidSquid]

Never had that happened myself, I assume it's a new issue since 1.11.0 then (I reverted to 1.10.1 after a short test of 1.11.0)?

[Spricigo]

12 hours ago, hazelnut2 said:

is there something else I'm not seeing?

Yes.  The command pod is a lot heavier than heat shield and material bay,  that will make the craft flip during reentry. (you may have better luck with the engine attached, but then the heat shield is useless anyway)

[hazelnut2]

3 minutes ago, Spricigo said:

The command pod is a lot heavier than heat shield and material bay,  that will make the craft flip during reentry.

I've balanced out the CoM pretty well (already landed with similar design). Either way, my question is actually about the bug, rather than good ship design. Thanks, Spricigo!

[paul_c]

Can you re-run the thing you did, but keep the focus on the detached satellite, then time-warp forwards a few days? I am wondering if its something design-related with the satellite; or a bug which makes it blow up/disappear. I've had issues where lander pods, when time-warped for a while, will just blow themselves up. I think they call it "The Kraken".

[hazelnut2]

52 minutes ago, VoidSquid said:

Never had that happened myself, I assume it's a new issue since 1.11.0 then (I reverted to 1.10.1 after a short test of 1.11.0)?

I've only ever played 1.11 (Steam, under Linux) since I'm very new to this. Also, I don't know of a way to roll back a game to a prior version in Steam (I'd certainly play a 2 years old version of DOTA )

Quote

Can you re-run the thing you did, but keep the focus on the detached satellite, then time-warp forwards a few days?

I'll certainly keep experimenting more tonight.

Edited January 5, 2021 by hazelnut2
appended another reply

[Spricigo]

33 minutes ago, hazelnut2 said:

I've balanced out the CoM pretty well

Seems you have it covered then. It also seem there is nothing with the satellite design that should be causing the issue.

Maybe a corrupted file,  a validation may solve the issue if that is the case. Failing that, i'm out of ideas.

[hazelnut2]

It looks like I'm onto something. I've reduced the design to a simple engine and Okto core connected by a decoupler to a Stayputnik and a service bay. Took off, waited for space (70km), decoupled, and as soon as I switched view ("]") to the decoupled probe the debug console filled in with errors:

[FLIGHT GLOBALS]: Switching To Vessel Untitled Space Craft Probe ---------------------- 
(Filename: ./Runtime/Export/Debug/Debug.bindings.h Line: 35)
[PlanetariumCamera]: Focus: Untitled Space Craft Probe
(Filename: ./Runtime/Export/Debug/Debug.bindings.h Line: 35)
CheckEncounter: failed to find any intercepts at all
(Filename: ./Runtime/Export/Debug/Debug.bindings.h Line: 35)
CheckEncounter: failed to find any intercepts at all
(Filename: ./Runtime/Export/Debug/Debug.bindings.h Line: 35)
dT is NaN! tA: NaN, E: NaN, M: NaN, T: NaN
  at System.Environment.get_StackTrace () [0x00000] in <ad04dee02e7e4a85a1299c7ee81c79f6>:0 
  at Orbit.GetDTforTrueAnomaly (System.Double tA, System.Double wrapAfterSeconds) [0x00000] in <8ed6226a17824027b61a78aa9c35f6f0>:0 
  at PatchedConics._CalculatePatch (Orbit p, Orbit nextPatch, System.Double startEpoch, PatchedConics+SolverParameters pars, CelestialBody targetBody) [0x00000] in <8ed6226a17824027b61a78aa9c35f6f0>:0 
  at PatchedConicSolver.Update () [0x00000] in <8ed6226a17824027b61a78aa9c35f6f0>:0 

I uploaded the full log here: https://drive.google.com/file/d/1YrPwZnrTKniF5zO0J9QH714k06V5DVIn/view?usp=sharing

The decoupled probe showed a vertical speed of 0, however seemed to move upwards with respect to the vessel left behind (which naturally had positive vertical speed still).

 

Know a good place to submit bugs?

 

Edited January 5, 2021 by hazelnut2
added text

[VoidSquid]

11 minutes ago, hazelnut2 said:

Know a good place to submit bugs?

The one and only: https://bugs.kerbalspaceprogram.com/projects/ksp/issues  

1 hour ago, hazelnut2 said:

Also, I don't know of a way to roll back a game to a prior version in Steam

See here: 

 

[hazelnut2]

Well, looks like 1.11 is broken hard:

• https://bugs.kerbalspaceprogram.com/issues/26977
• https://bugs.kerbalspaceprogram.com/issues/26971
• https://bugs.kerbalspaceprogram.com/issues/26926

These seem to be all related and pretty severe.

[VoidSquid]

I was just about to mention 26971, you were one minute faster.

But yes, I reverted after 2h KSP 1.11.0 to 1.10.1, too many game braking bugs for my taste. And that was on December 19th, you can see yourself how the bugtracker has filled up since.

If I were you, just starting with KSP as a new player I assume, I'd get myself 1.10.1, that version runs pretty fine, lots of excellent mods available, and the few issues it has, there are known workarounds.

What do you have to lose?  

Edited January 5, 2021 by VoidSquid

[hazelnut2]

8 minutes ago, VoidSquid said:

What do you have to lose?

Hm, nothing, just some 150 hours of career

[VoidSquid]

150??? Oh... and here I thought, just from the visuals of that craft, you'd maybe playing 20 or 30 hours. Beg your pardon.  

At any rate, I'd much prefer a stable game that works than have to expect more such like bugs you already had.

Edited January 5, 2021 by VoidSquid

[hazelnut2]

3 minutes ago, VoidSquid said:

150??? Oh... and here I thought, just from the visuals of that craft, you'd maybe playing 20 or 30 hours. Beg your pardon.  

What can I say - I don't use time warp a lot. Was intuitively afraid of breaking the physics (the irony of that!)

[VoidSquid]

Normal time warp is physic-less, works fine, just be a bit careful of a change of situation, namely change of SOI or entering the atmosphere, with high time warp the vessel's trajectory has a tendency to change a lot or shoot right through the atmosphere w/o switching to physics and real time.

Physics time warp, just watch the display in the upper left corner, if it's red, the game has issues to fully compute everything. Ascending within atmosphere, as well as critical moments when landing (opening parachutes and touch down in particular), I'd go for max. 2x physics warp.

Edited January 5, 2021 by VoidSquid

[James M]

It's a bug for sure. I built my own Duna satellite launcher the other day that housed 8 satellites neatly packed on engine mounts and placed into a payload fairing. Once i got there to deploy, same issue for all of them. I'd hit space, they'd just sit there, I'd switch to them and "nan" errors would display, and if I timewarped they'd disappear. Needless to say, I was a little frustrated after that. I think it had something to do with either the engine mounts I used to pack them, the microgirders used to give them space from the mounts, or the micro decouplers. 

Edited January 28, 2021 by James M

[vv3k70r]

On 1/5/2021 at 1:49 AM, hazelnut2 said:

When I used a stack separator the satellite did not become controllable.

Guiadance? Electric power? Antena?

On 1/5/2021 at 1:49 AM, hazelnut2 said:

however it became offset to what seemed like a Mun's pole (way far from the original part) and it just fell straight to the ground.

If straight - bug. If slope - decopuling aply force. On low gravity body it is importnat what direction You decouple and what force aply. On Minmus You can find satelite on far away orbit then planed just by those few meters per second.

On 1/5/2021 at 10:18 AM, hazelnut2 said:

The controller is the Probodobodyne Stayputnik, there is an antenna right on top of the controller

If You can bounce signal back to Kerbin. Do You have already other comunication in range? Because when You get line of sight covered by da Mun You are about to loose conection.

I'm not sure if built in antena is enough for every level of tracking station.

On 1/5/2021 at 1:50 PM, jimmymcgoochie said:

It’s a bug, no idea why it happens but you’re not the only one that’s had this issue.

World stabiliser?

On 1/5/2021 at 4:07 PM, hazelnut2 said:

The decoupled probe showed a vertical speed of 0, however seemed to move upwards with respect to the vessel left behind (which naturally had positive vertical speed still).

It is posible if You get exact over the pole in area that fit in empty/full (some combination of 0 an 1, but not all) vector from origin angle with precision od word (It could be 32 bit). When You load game (as for example switch vessel) these anglea do not return corect values for direction. It is a problem of transforming some special values from vectors to directions with given data precision. It was a common problem in older 3d engines when seting camera on top (it rotates). There is some issues with camera in game so I have suspicions of this kind of problem.

On 1/5/2021 at 5:26 PM, hazelnut2 said:

Hm, nothing, just some 150 hours of career

Easy. There be lot more trouble and You will lose more vessels (or find them no more uneusefull).

 

Try to deploy this sat on same orbit in front of Kerbin a bit later (best if daytime for solar panels). Switch visibility of comsat in map view, see if there is any green line to this stayputnik.

 

Edited January 28, 2021 by vv3k70r