Molniya orbit in KSP.

[DMSP]

Hi, DMSP here with something I just randomly made. With a communications network of some sort coming in 1.2, I was wondering if a Molniya type orbit would be useful. Here's an idea I had so far, 3 Sats in an "Apethellia" Orbit. Do you see the idea? I think it might have potential.

With only three satellites, there's a blackout every 6 hours at a non-servicable spot. But who cares, I only had time to put up three.

Thanks for reading!

[Fallarnon]

It really depends on what you're seeking to cover. That kind of orbit is specifically designed to focus coverage on a specific hemisphere.

[katateochi]

In the past when playing with Remote Tech I've used molniya orbits.  Before that I used to use "perfectly" circular orbits and put 4 sats up on the same orbit, just spaced out, but with timewarping and the fact that you can't get perfectly matching orbits I found that the sats would shift out of relative position and there'd be times when they all were clustered together.  

Using molniya orbits greatly reduces this problem and 4 sats setup like evenly spaced petals on a flower (sorry best desc I could think of) around Kerbin made for a network that didn't need any maintenance and provided constant coverage.  I would also put 2 long range sats in circular orbits that touched the Ap of the others (like a ring around the edge) and even though they'd drift in relative pos to each other it was a very reliable setup. 

[DMSP]

19 hours ago, Fallarnon said:

It really depends on what you're seeking to cover. That kind of orbit is specifically designed to focus coverage on a specific hemisphere.

Yeah. It would work best with KerbinSide or if there was some sort of expedition. It might be useful on Laythe.

16 hours ago, katateochi said:

In the past when playing with Remote Tech I've used molniya orbits.  Before that I used to use "perfectly" circular orbits and put 4 sats up on the same orbit, just spaced out, but with timewarping and the fact that you can't get perfectly matching orbits I found that the sats would shift out of relative position and there'd be times when they all were clustered together.  

Using molniya orbits greatly reduces this problem and 4 sats setup like evenly spaced petals on a flower (sorry best desc I could think of) around Kerbin made for a network that didn't need any maintenance and provided constant coverage.  I would also put 2 long range sats in circular orbits that touched the Ap of the others (like a ring around the edge) and even though they'd drift in relative pos to each other it was a very reliable setup. 

Yes, having four spread out would work well. In reality though, as long as you have 2 communications rings chances are you will never loose coverage.

[Hodari]

I got a contract once to place a satellite in a Molniya orbit once(think it was around Gilly or someplace weird like that, so didn't actually do it, but at least it shows the idea is in the game...)

[Mastikator]

Another option is to have 2 sats in equatorial orbit, one going with the rotation of Kerbin and one going against. And a 3rd and 4rth in perpendicular polar orbit. Any downtime would only last a very short time and be extremely rare.

[RCgothic]

Molniya orbits have a specific use - quasi-stationary above a particular surface location at high latitudes. They're similar to geostationary orbits, where you want to minimise the amount that your surface-based dish has to move. Point at a particular patch of sky and there'll always be a satellite there. Let's discuss these two orbits a bit:

Geosynchronous Orbits

Geostationary orbits (or Geosynchronous Equatorial Orbits: GEO) acheive this by matching the orbital period to the earth's rotation in a circular orbit. Because a geostationary satellite is exactly over the equator it remains perfectly stationary from the earth's reference frame and your surface dish therefore doesn't need any off-axis or tracking capabilities. They also remain perfectly stationary with respect to each other, which simplifies sat to sat communications, making global relay networks simpler.

However there are a couple of disadvantages. Because they have to orbit at a specific height very closely, and because there's a minimum spacing between satellites before you start to get radio interference, there are only so many satellites that can operate in GEO. Yes, countries and corporations actually have arguments about this, and there's an international body that mediates. Secondly, for countries in high latitudes to point your dish at a satellite in GEO means pointing it at or near the horizon, which is very bad for reception because you start picking up ground-based interference.

Molniya Orbits

You can't get a satellite to hover over a high latitude in a circular synchronous orbit. Inclined orbits must necessarily cross the equator, which means your satellite will disappear below the horizon frequently. Worse, it moves rapidly across the sky, meaning you have to track it with your dish. It also shares an orbital height with sats in GEO and thus takes up a slot.

Molniya orbits attempt to solve a couple of these problems for high latitude applications. Because satellites in eliptical orbits move more slowly near apoapsis, if you place them in a highly eliptical orbit they spend most of their time in the higher part of their orbit. Orbits don't need to be circular to have a period of 24 hours, so you can place them in a highly eliptical inclined orbit and they will appear to 'dwell' over a particular surface location at the same time each day, which means your surface dish can again be pretty simple as long as you're prepared to wait.

Or you can discard the 24h requirement and have several sats in still more eliptical orbits. You won't get the same satellite at the same time each day, but as they're further away each satellite will spend even more time in an even smaller area of sky, and by the time it's leaving your area of reception another will be entering it. There are also any number of these orbits, so bandwidth slots are far less restricted.

There are a couple of drawbacks to molniya orbits (as might be expected). Because they go much further away, they require more signal power and bigger dishes to communicate with. Secondly, because the orbits are not circular it's much harder for satelites to stay in contact with each other as the angles between them change wildly over the course of an orbit.

Note: For mission control, having a satellite with a high dwell time isn't a huge problem, because you can just invest in a dish that can track it and hire ground stations over the horizon for when you can't see a satellite directly. Dwell time becomes much more important in commercial applications. 100,000 trackable dishes, paid for by households? GEO and Molniya suddenly look much more attractive.

KSP

Anyway, in KSP even Remote Tech does not model dish orientations, with each dish having perfect 360deg tracking. Tell a dish it's trying to point at KSP, and as far as it's concerned it actually is, no matter that it's physically pointing at Jool at this instant and the angle to KSP is changing at 25deg/s. It is thus highly unlikely that any stock solution will concern itself with which way dishes are pointing. It's a level of difficulty too far for most casual players.

So in KSP we don't have to deal with dish orientation/tracking, and we're not trying to reach 100,000 households with simple dishes. There's just no need outside of a specific contract to put a satellite in KEO or Molniya. They're not more useful than any other orbit.

Me, Personally

When playing with Remote Tech, I use a network of 3 sats at approximately 800km altitude. This gives them enough line of sight over the horizon to allow them some margin for error for phase angle and phase angle drift. In actuality, they go in an orbit with an orbital period of a round number of minutes and a 5/6th orbital period that is also a round number of minutes. They aren't in stationary orbits, but they don't have to be. All dishes in this game have perfect tracking.

I use Excel to work out the target orbital height/period and KER to match those in game.

I launch three sats on the same launcher to the target apoapsis and 5/6th orbital period. I detach one sat and use its on-board monoprop to circularise. I like my sats heavy, but with loads of RCS they circularise pretty easily. Two orbits later the launcher has fallen 1/3 of an orbit behind (because 5/6 period!) and I detach and circularise a second relay at apoapsis. Two orbits later I do the same thing again.

Finally, I disable most of the onboard RCS, and use just one or two thrusters and fine-control mode to exactly match orbital periods using KER. These usually don't end up being perfectly circular! That's far too much of a pain. There may be 100m or so between apoapsis and periapsis. It's only the orbital period that matters and usually I get them down to 0.01s per orbit, which is enough to keep them in phase for ages. High mass/low thrust help do this accurately. Maintenance-free relay network! And even if they do fall out of phase, just deboost the offending sat into a fractional orbit for a few rotations before circularising and fine-tuning again. But I've never actually had to do that yet.

Expanding the network is a variation on this theme.

 

 

Edited April 25, 2016 by RCgothic

[eddiew]

What I think you'll benefit from most is a satellite constellation, rather than trying to make them co-orbital. KSP orbital precision just isn't good enough, it's verging on impossible to plant multiple satellites with exactly the same period, with the result that over several years (say enough time to for a probe to get to Jool) they all bunch up, spread out, bunch up again - and may well be bunched up on the wrong side of Kerbin when you need them most.

By having several (4-6) comsats in staggered elliptical orbits, you move away from precision planning (which KSP is bad at) to the idea that "most of the time one or more of them will have LoS to ground control". It doesn't matter if they're out of sync, since time spent low to Kerbin is very short relative to being high in the sky, and most of them can see at least two of the others most of the time. If they have a 10 day orbital period, then 9 days, 5 hours and 45 minutes of that time is spent high up.

In the picture below, two satellites are out of LoS of ground control (white dot) but all of them can see at least one other satellite that can relay the signal. It takes a matter of minutes (10-15) for each to complete its near-Kerbin pass and get out of the planet's shadow to where it can be seen by the others again.

The bigger the ellipse, the better, as this results in more time with a bigger visibility of the planet, and higher chance of LoS to other satellites. I never got round to it, but it did occur to me that using elliptical polar orbits could allow a satellite to stretch out beyond Mun's orbit without getting picked up by its gravity.

Also, I think it plays better to KSP's inherently kerbal nature that the question "which satellite can I connect to?" should have the answer "one of them"  

*edit* Swapped the image. I'm worried that people will read this thread when comms relays are a thing and not realise the important bit is not trying for precise orbital timing  

Edited April 26, 2016 by eddiew
clarity

[Choctofliatrio2.0]

Interesting idea! I'll try one of these orbits, they look pretty.

[eddiew]

14 minutes ago, Choctofliatrio2.0 said:

Interesting idea! I'll try one of these orbits, they look pretty.

They do make nice flowery patterns, tis true. 6 looks better than 4, but I couldn't be bothered drawing it  

[RCgothic]

If we're still talking about remote tech, I can think of a couple of a few disadvantages to the molniya 'petal' compared to a more traditional relay network:

Constructing it is probably four launches rather than just one. I've previously put up 21 relays in that many launches to LKO.

If your AP is above ~1100km you'll need something bigger than the communotron 16 to ensure connectivity.

Higher orbits leads to increased signal delay compared to a low-orbit relay network.

[eddiew]

With RT, most people would be kitting out their comsats for use at kerbostationary orbit, so they'd all be on a small dish   The eliptical pattern can, but doesn't have to, be smaller than kerbostationary, but the larger each petal, the more chance that the satellite is "up there" vs "under the ground", from KSC's viewpoint. It's a formation based on probability, rather than precision, and my experience was much better than with co-orbital attempts.

I also suspect that Squad's implementation will be more like @toadicus' Antenna Range mod. I personally don't expect speed-of-light, inability to control when lacking signal, and possibly not LoS blocking. (After all, why can kerbals not have 3 dishes on the equator so's to avoid the blind spot?)

I failed to mention in my first post; when creating a flower constellation (as they will henceforth be known!) it is actively beneficial to have the orbital periods a little different, ideally by 15-30 minutes. I recommend picking prime numbers like 17 or 29 to avoid falling into a resonance trap.

Reason; orbits in KSP tend to wander, and you don't have station keeping. Two apparently co-orbital or co-periodic satellites will at some point bunch up. If you have them where their orbits are only 1 second apart, then once they are bunched (e.g. all low at the same time) they are going to stay bunched for several hundred more orbits. If they have significantly different periods, then the bunching will happen more times throughout the year, but only for 10-15 minutes at a time, since they'll all be back to high orbit again and out of step for the next pass.

Just 2p born of my own experience. It may not be optimal, and there may be solutions involving mods and ultra-fine orbital adjustments that can stand the test of time. I just never had any luck with it myself and moved to a flower constellation as a way to combat floating-point errors with pseudo randomness and probability  

Edited April 25, 2016 by eddiew

[Wahgineer]

The only problem is that there is no reason to have a sat network around Kerbin: The Devs have stated that there will be a 'Deep Space Network (DSN)'. This means that any craft with an antennae that is close enough to Kerbin will have a direct link to the Space Center, regardless of which side of the planet it is on. ComSats around Kerbin will only be useful to boost range. My plan is to just put a big satellite in an extremely high polar orbit, so that way I don't have to worry about occlusion.

[Enorats]

They certainly look pretty, but one of the best things about remote tech is how pretty the patterns of moving connection lines are. I can stare at those for quite some time, lol. Heck, I've found myself putting satellites up not for utility but instead purely for aesthetic reasons.

1 hour ago, awsumindyman said:

The only problem is that there is no reason to have a sat network around Kerbin: The Devs have stated that there will be a 'Deep Space Network (DSN)'. This means that any craft with an antennae that is close enough to Kerbin will have a direct link to the Space Center, regardless of which side of the planet it is on. ComSats around Kerbin will only be useful to boost range. My plan is to just put a big satellite in an extremely high polar orbit, so that way I don't have to worry about occlusion.

This seems rather silly. The real life DSN is for just that.. deep space. Get too close to the planet and there are huge gaps in the coverage.

Many actually already model this sort of thing using kerbal konstructs. Multiple launch sites can be connected to just like KSC, allowing for planetary coverage.

[Skystorm]

@awsumindyman I think perhaps you missed a few tidbits in the original blog post.  You may not need to build your own DSN on the easiest difficulty levels, but you probably will on the harder modes.  Plus they have said that it will be configurable in the settings.

Quote

Another area we’ve made more approachable is the default implementation of a built-in deep space network, where Kerbin has an ever-increasing inferred relay network, similar to what we have on Earth. The range of this network will be decreased if the player is on ‘Hard’ mode, effectively requiring the player to set up their own deep space network. On ‘Easy’ mode, relay, and antennas operate as they do today, with your only limitation being the power constraints and packet size of the antennas themselves. Furthermore, the settings will be exposed in the difficulty settings so they can be toggled in Custom Difficulty.

The emphasis above is mine.  Here is a link to the original article for your reading pleasure.

Daily Kerbal Bonus Article: Development Relay

[FancyMouse]

A satellite network with a reasonable drift which doesn't require adjustment for 40 years is definitely possible:

 

[suicidejunkie]

If you really want no drift, a landline network is optimal:

Get an arctic expedition to bring a max transmitter to the pole, and the Kerbol system is yours until plate tectonics are implemented.

All of the transmitters pictured were assembled by hand with KAS at the absolute peaks of mountains, but next time I think I'll just carpet bomb them out of a plane with either parachutes or lithobraking struts.

[I_Killed_Jeb]

1 hour ago, suicidejunkie said:

If you really want no drift, a landline network is optimal:

 

oh god my framerate

[flatbear]

1 hour ago, suicidejunkie said:

If you really want no drift, a landline network is optimal:

Get an arctic expedition to bring a max transmitter to the pole, and the Kerbol system is yours until plate tectonics are implemented.

All of the transmitters pictured were assembled by hand with KAS at the absolute peaks of mountains, but next time I think I'll just carpet bomb them out of a plane with either parachutes or lithobraking struts.

does it lag a lot?

[suicidejunkie]

23 minutes ago, flatbear said:

does it lag a lot?

Not at all.  (large ships and explosions slow me down, but that's unrelated to comms)

I'm pretty sure the comm links don't recalculate every frame anyways; its not like the links change very often.

 

just finished a pretty spiffy air-droppable Comms Dart design for just $1740 (vertical fins stick better for un-guided descent), or the UFO version (fins flipped 90 degrees) which can be flown to a target with just the probe SAS.

With the parachute set to fully deploy at 100m, it gets down in a hurry to avoid physics range issues, and with SAS to help they seem to stick the landings on 45 degree slopes quite well.

Edited April 26, 2016 by suicidejunkie
Pictures

[eddiew]

9 hours ago, suicidejunkie said:

If you really want no drift, a landline network is optimal:

I like your dedication   Back in 0.90 with Remote Tech, I tried something similar, but always had power issues with the ground stations. The big deep-space dishes consumed a lot of power, and getting them through a 3 hour night required large stations with something like 50,000 units of electrical storage. In the end, I opted for a satellite network because they only had to endure 10-15 minutes of being in the shadow.

Not sure whether RT is being more gentle with it's modelling now, or whether there's a low-power state when dishes aren't actively transmitting?

[RCgothic]

I wouldn't attempt to put a dish on a ground station. At least, not every ground station. Keeping one powered through a three hour night is definitely non-trivial in an early career game. Not convinced by the ground station approach, but if you do it's probably better to keep to communotrons for most relays and leave the dishes to dedicated stations. 

 

I tend to put dishes in high polar orbits where they're unlikely to be eclipsed by the mun and they only experience night at certain times of year. If there are three of them then even then only one will ever be eclipsed or in night at once and the others should be powered with decent line of sight. Batteries then become a nice optional extra.

Long-range dishes actually sound like a good idea for a polar molniya orbit - more so than Kerbin relays. It's a good way to keep their availability high, their night requirements will be short as they move quickly past Kerbin's shadow, and the additional signal lag at interplanetary distances is small.

[suicidejunkie]

All the cheap relays have DP-10s, so they use less power than the probe core.

The dishes should be at the poles anyways, there it is never night and a 6x1 solar panel can stick straight up and get full input continuously.

With a DTS-M1 on mun or minmus' poles to reflect back, you can flood local space with excellent coverage.

[dewin]

ReRe: Remotetech and networks that don't drift:

I'd love a proper stationkeeping mod that could handle this with small behind-the-scenes burns (I.e. it would just change orbital parameters and remove a corresponding amount of fuel when the vessel isn't active):

Configuration:

• Allow satellites to be assigned to a group.
• Set a configurable target SMA or orbital period for all sats in the group along with acceptable tolerance.
• Set a configurable max phase variance (see below)
• Set a max dV per correction.

Phase:

• As a satellite completes an orbit, its phase goes from 0.0 to just under 1.0 and then resets.  (Or 0-360 degrees)
• This is based on time into orbital period, not relative position of the planet.  This allows separation to be maintained even when using eccentric orbits.
• Phase=0 is defined as the point where this orbit is closest to all other orbits in the group.  (For perfectly circular orbits, this is arbitrary and doesn't matter as much).  This might be defined as lowest max distance or lowest average distance.
• Satellites in a group aim to be 1/n phase apart.

Orbital corrections:

• Periodically scan the group for satellites out of SMA/orbital period tolerance.  Select the satellite that is furthest out of tolerance, if any.
• If no satellite is selected, select the satellite that is most out of phase tolerance, if any.
• If all satellites are within tolerances, do nothing; otherwise
• Plan maneuvers to bring satellite into tolerance -- either adjusting SMA back into tolerance, or adjusting orbital period until the satellite catches up/calls back into place and and then resetting back to target SMA -- and execute them.  No other adjustments to the group occur while maneuvers are planned.
• If a satellite with maneuvers planned is in physics range, aborts all planned maneuvers (though they'll be replanned later when the sat isn't in range.). If any satellite in the group is in physics range, planning is temporarily suspended.  If a satellite in the group is ever on a suborbital, escape or intercept trajectory, all planning is suspended.  If a satellite is ever in a different SOI from another satellite, all planning is suspended.