Satellite formations - What should I look at?

[Sokar408]

So I'm trying to bring some satellites into as stable an orbit as possible, for minimal maintenance. My usual method is to use cosinus relations so calculate spacing (trigonometry! ), then I bring the satellites into position one by one, and insure that they have the same orbital period. This works to some degrade, but it seems that after a good year or two in the game, they get very visible out of position. I have tried to use to use semi-major and semi-minor axis as that seemed to get more precious numbers. This worked a bit better, but still isn't as permanent as I'd want.

So I'm asking here, has any RemoteTech2/satellite enthusiast/math person figured a better way of getting satellites to stay in position?

[FREEFALL1984]

Base it on orbital periods, I like to release my geosync satellites from the same vessel, make sure your first lifter has an orbital period of 4 hours and an apoapsis of 2868.75km, then release your first sat at apoapsis, then circularize its orbit and make sure its orbital period is 6 hours, then when your main vessel release the next sat and circularize that one making sure it also has a 6 hour period, then do the same for the final satellite, you now have 3 satellites in a geosync orbit all with orbital periods or exactly 6 hours, they might not be perfectly circular orbits but they should never drift out of sync.

if you want 4 satellites make sure your main vessel has an orbital period of 4.5 hours (6-(6/4)), for a 5 sat setup have a period of 4.8 (6-(6/5)) hours and so on

Edited April 25, 2014 by FREEFALL1984

[Sokar408]

Base it on orbital periods, I like to release my geosync satellites from the same vessel, make sure your first lifter has an orbital period of 4 hours and an apoapsis of 2868.75km, then release your first sat at apoapsis, then circularize its orbit and make sure its orbital period is 6 hours, then when your main vessel release the next sat and circularize that one making sure it also has a 6 hour period, then do the same for the final satellite, you now have 3 satellites in a geosync orbit all with orbital periods or exactly 6 hours, they might not be perfectly circular orbits but they should never drift out of sync.

if you want 4 satellites make sure your main vessel has an orbital period of 4.5 hours (6-(6/4)), for a 5 sat setup have a period of 4.8 (6-(6/5)) hours and so on

Wauw I never thought of doing it that way! Like I said I usually do Separate launches, and use cos relation calculation:

a^2 = b^2 + c^2 - 2bc * cos(A)

Lets say I want 4 satellites equally spaced, at an altitude of 100.000 meters, I know that b = 100.000 and c = 100.000. Since there is going to be 4 satellites, and a circle (the orbital plane) has 360 degrades, then I simply take 360/4 (number of satellites) = 90 degrades = A

Now you simply plug and play to find the distances each satellite needs to have between each other to be equally spaced.

Needless to say your method is a lot more practical.

Anyway on the matter of the post. So you simply use orbital time? Like I said I always get inconsistent readouts on VOID/KER/MechJeb. I don't get enough decimals to get the precious needed to stave off drift between the satellites. How do you get around this?

[FREEFALL1984]

Anyway on the matter of the post. So you simply use orbital time? Like I said I always get inconsistent readouts on VOID/KER/MechJeb. I don't get enough decimals to get the precious needed to stave off drift between the satellites. How do you get around this?

There will be some inevitable drift without regular maintenance but you can minimize it if you're prepared to put a single Ion engine on each satellite, then you limit the thrust to 1% and thrust retrograde until the orbital period is 5:59:59 then you turn prograde and apply minimal thrust until the orbital period "just" ticks over to 6:00:00, the instant you see the timer tick, kill your engine, depending on the weight of the satellite (and your reaction times) you should be within one twentieth of a second of a perfect 6 hour period.

You can also do this with RCS but they tend be a little too powerful for smaller satellites

Edited April 25, 2014 by FREEFALL1984

[Sokar408]

There will be some inevitable drift without regular maintenance but you can minimize it if you're prepared to put a single Ion engine on each satellite, then you limit the thrust to 1% and thrust retrograde until the orbital period is 5:59:59 then you turn prograde and apply minimal thrust until the orbital period "just" ticks over to 6:00:00, the instant you see the timer tick, kill your engine, depending on the weight of the satellite (and your reaction times) you should be within one twentieth of a second of a perfect 6 hour period.

You can also do this with RCS but they tend be a little too powerful for smaller satellites

Haven't the changes made to the Ion Engine made this harder though. I tend to use as small probes as possible (Considering they always end up just being huge batteries so they don't run out of power. Speaking of that, does remotetech actually manage this, when the vessel is not active?)

[FREEFALL1984]

yeah the ion engine buff has made it a little less accurate but you'll still only have 0.02kn of thrust so it should be within a few milliseconds of perfect, also if you set up all satellites in the same way they you'll generally have them land within 1 millisecond of each other since you'll likely be more or less the same difference from the 6 hour mark on each satellite

you'll find that if you have only one satellite in contact with ground control, as soon that satellite passes onto the night side of kerbin and its batteries die your full communication setup will die, I remedy this by having 2 sats in contact with ground control at all times, and make sure all sats are talking to the satellites nearest them this means you need at least 4 active dishes on each satellite, one each to communicate with the satellite next to it and one to communicate with the target space craft, and one to communicate with the KSC (although the sat on the opposite side of kerbin only needs 3)

alternatively you could add a whole crapload of batteries on the one above KSC instead

[Sokar408]

yeah the ion engine buff has made it a little less accurate but you'll still only have 0.02kn of thrust so it should be within a few milliseconds of perfect, also if you set up all satellites in the same way they you'll generally have them land within 1 millisecond of each other since you'll likely be more or less the same difference from the 6 hour mark on each satellite

you'll find that if you have only one satellite in contact with ground control, as soon that satellite passes onto the night side of kerbin and its batteries die your full communication setup will die, I remedy this by having 2 sats in contact with ground control at all times, and make sure all sats are talking to the satellites nearest them this means you need at least 4 active dishes on each satellite, one each to communicate with the satellite next to it and one to communicate with the target space craft, and one to communicate with the KSC (although the sat on the opposite side of kerbin only needs 3)

alternatively you could add a whole crapload of batteries on the one above KSC instead

My approach to this is relaying on Omni-Antenna. Sure you can't use them in Geo-Stationary orbits, but considering that 1 antenna can talk to all others in range, it doesn't really matter. I tend to make a low-mid altitude grid of 4-6 satellites, this way you can have your geo sats talk to those sats using antenna, and there by bypass A LOT of dishes

[Pecan]

There are three things to note about orbital drift:

1) It happens in real life too, only more so. Every vehicle needs some sort of propulsion so it can perform orbital corrections.

2) As long as you don't take control of a vehicle in KSP or come within 2.5km it flies 'on rails' so the only drift is from rounding errors.

3) Such drift as there is in KSP is almost impossible to eliminate completely; it's just the way computers are. You can still do better than real life though.

[Sokar408]

Alright! So I have done some testing now, and it turns out to be quite interesting.

I made specific satellites that would barely accelerator with a single Ion Drive, put them up in a grid of 4, and start correcting their orbital timing. I did this by, as previously suggested, I chose an orbital time of 55 mins, 44.1 sec, at an altitude of about 400 km. The way I would do it, would be to speed up and as soon as the orbital time hit .1, I'd stop.

Using this method I warped time at max, from a different vessel to insure they orbited on rails. After 2 years the sat positions was so off they no longer ranged all the way around the planet Kerbin.

I then tried to use the Semi-Major Axis, going for the exact same one on each. Again I used the speeding method as mentioned, but instead did it to get the SMA as close to each other as possible. This time after 5 years, the grid was still almost Identical.

This suggests to me that the SMA, for some reason, is far more important then orbital timing, as it far more accurate. That is not to say orbital period isn't essential, it is, but you can get that far more accurate by using the SMA as measuring tool.

[StainX]

I always used one single uncomplicated method for comsat network:

A kerbin day is 6hours... so i launched every hour to an ~50° inclined HKO (in range of omnidirectional antenna) and same orbital period.

After that same thing with geosync orbit every 1h and sat-dishes.

Not very mathematical i know, cuz i suck at such stuff

[FREEFALL1984]

Alright! So I have done some testing now, and it turns out to be quite interesting.

I made specific satellites that would barely accelerator with a single Ion Drive, put them up in a grid of 4, and start correcting their orbital timing. I did this by, as previously suggested, I chose an orbital time of 55 mins, 44.1 sec, at an altitude of about 400 km. The way I would do it, would be to speed up and as soon as the orbital time hit .1, I'd stop.

Using this method I warped time at max, from a different vessel to insure they orbited on rails. After 2 years the sat positions was so off they no longer ranged all the way around the planet Kerbin.

I then tried to use the Semi-Major Axis, going for the exact same one on each. Again I used the speeding method as mentioned, but instead did it to get the SMA as close to each other as possible. This time after 5 years, the grid was still almost Identical.

This suggests to me that the SMA, for some reason, is far more important then orbital timing, as it far more accurate. That is not to say orbital period isn't essential, it is, but you can get that far more accurate by using the SMA as measuring tool.

you need to remember that the drift is entirely dependant on orbital period, you could have 3 sats with different semimajor axis all have the same orbital period, perhaps inevitable problem of orbital drift is exaggerated by vector rounding when entering timewarp or switching vessels. perhaps if you match the semimajor axis AND the orbital period as described it would help.

[Sokar408]

you need to remember that the drift is entirely dependant on orbital period, you could have 3 sats with different semimajor axis all have the same orbital period, perhaps inevitable problem of orbital drift is exaggerated by vector rounding when entering timewarp or switching vessels. perhaps if you match the semimajor axis AND the orbital period as described it would help.

As far as I can see they are tied together. If the SMA is the same, the orbital time is the same. This could be wrong, considering I already placed the sats in the correct positions, with the correct orbital period, BEFORE I adjusted the SMA.

[Amaroq]

Wiki link confirms that assumption, in fact, you can calculate one from the other: http://en.wikipedia.org/wiki/Semi-major_axis#Astronomy

I am very pleased with the idea, here - adjusting the SMA rather than the time to take advantage of the increased precision. I will definitely be applying that when I next launch a RemoteTech constellation.

As StainX pointed out, though, there's not much reason to avoid redundancy (using more than 3 craft).

Personally, I put my omni-range "repeaters" in stable geosynch orbits, with my interplanetary comsats on a polar orbit, on the theory that there's no way all three of the interplanetary sats are going to be blocked by a moon at the same time.

[Sokar408]

Wiki link confirms that assumption, in fact, you can calculate one from the other: http://en.wikipedia.org/wiki/Semi-major_axis#Astronomy

I am very pleased with the idea, here - adjusting the SMA rather than the time to take advantage of the increased precision. I will definitely be applying that when I next launch a RemoteTech constellation.

As StainX pointed out, though, there's not much reason to avoid redundancy (using more than 3 craft).

Personally, I put my omni-range "repeaters" in stable geosynch orbits, with my interplanetary comsats on a polar orbit, on the theory that there's no way all three of the interplanetary sats are going to be blocked by a moon at the same time.

repeaters? And about the polar orbits. I'd like to have 2 sats in highly eliptical orbites, on the same plane, but with one having its periapsis near the north pole, and its apoapsis over the southern, and vice versa. However I can't come up with a formular to calculate this, and I don't know what to look for. Could you help me out?

[Amaroq]

Ah -- okay, certainly.

I make a distinction between my inter-planetary comsats and my local SOI omni broadcast; I think of my vessels as "Inter-planetary comsat" or "Local comsat" (repeaters).

With my interplanetary comsats, I focus on minimum-occlusion orbits: high polar (85 - 90 degrees) .. but I also make sure to keep them in closer than you might expect: the lower their orbit, the narrower the angle supported by aiming at the planet.

If you set up your KIC's (Kerbin Interplanetary Comsats) at, say, a 75Mm equatorial orbit so that they are outside of the orbits of Mun and Mimus, then you're going to have a problem. When Jool is in a planetary alignment with Kerbin (at a minimum distance from Kerbin), a JIC aimed at "Kerbin" will have too narrow of an angle-of-coverage, and won't pick up the KIC's. You'll experience gaps in coverage.

Instead, I keep them inside the Mun's orbit, but highly polar, so that no matter where the Mun is, at least one of them can see "over" or "under" it, even if the others are occluded (behind Kerbin, blocked by the Mun, etc). That way, my interplanetary sats can aim at the parent body (Kerbin) instead of at individual satellites.

My KLC "repeaters", then, (Kerbin Local Comsats) are the ones which are responsible for making sure that that signal can reach any point on the planet's surface, and I have the classic geostationary satellite network. Signal tends to bounce from KSC to KLC-1 (overhead), to one of the KIC's, and then outbound.

There are other configurations, of course -- in one of my saves, I opted for three main comstations, outside of Kerbin's SOI, but in a coplanar orbit. Rather than trying to make direct connections from Kerbin to anywhere, Kerbin broadcast to those, and they sent the signal further. That covered me for when my target was on the wrong side of the sun, and provided a shorter signal delay in most cases, but made for more manual targeting as I'd have to set up the communication path by targeting the comstation's satellite at Jool, and targeting Jool at the appropriate comstation.

...

To your question, let's start with, what are you trying to gain by the highly elliptical orbits? I'm intrigued.

If I were going to try this, I'd start out by setting up two Kerbal Alarm Clock time markers exactly six hours apart.

At the first KAC alarm, I'd launch polar (e.g., make your gravity turn a little bit to the left of "North"), settling into about an 85-degree North north-west inclination. I'd track that inclination with KER during my launch (don't worry if its way wrong while under 40km, its only once you're a long way into the burn that it starts converging on the true inclination). Burn to a fairly circular orbit at your intended periapsis, leave that craft in a stable orbit.

Take the second one to the launch pad. Set it up with "target" equal to the first craft. Fast-forward to the second KAC alarm. Follow the same launch profile as above. Once you're up & circular, you should be very close to co-planar. I'd adjust one of the two to the other's plane at the Ascending Node or Descending Node, as though you're making a rendezvous, but without actually docking or worrying about your closest approach.

Then plot your burns to create the elliptical orbits, one as it cross the North Pole, one as it crosses the South Pole. As before, match the SMA's. I'm not sure what kind of resonance you'll want (both close at the same time, or opposite each other) so you'll have to pick your own timing for those burns.

Make sense?