Can minor orbit error mess up a relay system? (remote tech)

[ToukieToucan]

So a few days ago I made my first LKO relay system, did some missions and a bit of timewarp and discovered that my 4 evenly spaced communication probes where all on one side of the planet... The Ap of all satellites is 500 km but the Pe (500 as well) can have 7 km changes. Do all the orbits need to be exactly the same for a relay system or is this just an exception?

[Stone Blue]

The orbital period (and possibly semi-major axis?), is really the important parameter to get as close as possible to being equal on all sats...

You'll always have SOME drift on each sat, and will eventually have to fix them... How OFTEN you have to adjust, will depend on how big the differences in periods are... (7km IS quite a lot, especially in lower orbits...)

[Snark]

1 hour ago, ToukieToucan said:

So a few days ago I made my first LKO relay system, did some missions and a bit of timewarp and discovered that my 4 evenly spaced communication probes where all on one side of the planet... The Ap of all satellites is 500 km but the Pe (500 as well) can have 7 km changes. Do all the orbits need to be exactly the same for a relay system or is this just an exception?

Yes, they need to be exact.

The only way a synchronized relay system can work is if the satellites have exactly the same period.  Any different, and they will gradually drift.  7 kilometers is a huge disparity.  You want their Ap and Pe to be within a few meters, not kilometers.

There are a few ways to deal with this:

One approach is to be super-precise about your orbits.  Get them as close to exactly precise as you can.  Some people accomplish this by using HyperEdit (or just a text editor directly tinkering with the save file) to make the orbits exact.  I never do that myself (it feels like "cheating" to me), but I find that as long as I'm using whisper-gentle thrust, like an Ant or Spider with the thrust limiter turned way down, then I can fine-tune an orbit within a few meters.  Even that small amount will eventually cause drift, but it's good enough for a few years and that's typically the span of my KSP career anyway.

Another approach (which I actually prefer) is to be deliberately non-precise about orbits.  Have satellites in a variety of orbits with significantly different periods from each other.  This means that at any point in time, the distribution of satellites around the planet will be essentially random.  As long as you have enough satellites, then you're very unlikely to have a gap in your coverage; and although it's technically possible, even if it happens, the outage will be very brief and not repeated again soon.  (As opposed to an attempted-synchronous formation that's not-quite-right, since when that drifts into a gap situation, it stays that way for a long time.)

It's true that in this random-scattered approach, you need more satellites.  A well-set-up synchronous formation can get by with just 3 or 4 satellites, whereas a random-scattered setup may need twice as many.  But I don't see that as much of a disadvantage, myself.  First, a minimal comsat is really cheap.  If all you need in orbit is a Communotron-16 with a couple of OX-STAT panels, OKTO core, Oscar tank, and an Ant, you can put that up in orbit quickly, easily, and dirt cheap.

So that approach is my personal favorite.  Yes, it takes (very little) more cash, and a few more (very quick and easy) launches... but that's more than made up for by the advantages:  much less hassle, much less finicky tinkering with orbits, and low chance of critical failure.

By all means set up a synchronous network if you like-- it can be a fun challenge in its own right, and it certainly looks really cool once you've got it set up.    Just be aware that it's not the only option.

Edited January 14, 2016 by Snark

[ToukieToucan]

31 minutes ago, Snark said:

Yes, they need to be exact.

The only way a synchronized relay system can work is if the satellites have exactly the same period.  Any different, and they will gradually drift.  7 kilometers is a huge disparity.  You want their Ap and Pe to be within a few meters, not kilometers.

There are a few ways to deal with this:

One approach is to be super-precise about your orbits.  Get them as close to exactly precise as you can.  Some people accomplish this by using HyperEdit (or just a text editor directly tinkering with the save file) to make the orbits exact.  I never do that myself (it feels like "cheating" to me), but I find that as long as I'm using whisper-gentle thrust, like an Ant or Spider with the thrust limiter turned way down, then I can fine-tune an orbit within a few meters.  Even that small amount will eventually cause drift, but it's good enough for a few years and that's typically the span of my KSP career anyway.

Another approach (which I actually prefer) is to be deliberately non-precise about orbits.  Have satellites in a variety of orbits with significantly different periods from each other.  This means that at any point in time, the distribution of satellites around the planet will be essentially random.  As long as you have enough satellites, then you're very unlikely to have a gap in your coverage; and although it's technically possible, even if it happens, the outage will be very brief and not repeated again soon.  (As opposed to an attempted-synchronous formation that's not-quite-right, since when that drifts into a gap situation, it stays that way for a long time.)

It's true that in this random-scattered approach, you need more satellites.  A well-set-up synchronous formation can get by with just 3 or 4 satellites, whereas a random-scattered setup may need twice as many.  But I don't see that as much of a disadvantage, myself.  First, a minimal comsat is really cheap.  If all you need in orbit is a Communotron-16 with a couple of OX-STAT panels, OKTO core, Oscar tank, and an Ant, you can put that up in orbit quickly, easily, and dirt cheap.

So that approach is my personal favorite.  Yes, it takes (very little) more cash, and a few more (very quick and easy) launches... but that's more than made up for by the advantages:  much less hassle, much less finicky tinkering with orbits, and low chance of critical failure.

By all means set up a synchronous network if you like-- it can be a fun challenge in its own right, and it certainly looks really cool once you've got it set up.    Just be aware that it's not the only option.

Ah, the Remote Tech master has returned once again. Anyways, Makes sense that it'd mess up my system, guess I'll need to send a brave kerbal up again...

 

Also, what are some tips to get an unmanned craft to the mun/minmus with constant connection? (As in which antennas and at which height should I place the new relay system).

Edited January 14, 2016 by ToukieToucan

[Snark]

38 minutes ago, ToukieToucan said:

Also, what are some tips to get an unmanned craft to the mun/minmus with constant connection? (As in which antennas and at which height should I place the new relay system).

There are a lot of ways you can slice this problem, here's just the way I like to do it.

First, antennas:  For communicating between Kerbin-Mun or Kerbin-Minmus, you want the smallest/lightest stock dish (I forget its official designation), the one that has a 45-degree cone.

Next, setup:

There are various ways to do this.  I like to use a keosynchronous satellite perched above KSC on a steeply inclined orbit (say, inclination of 70 degrees or so).  It has several of the 45-degree dishes on it.  One is kept pointed at Kerbin; one is pointed at Minmus, one at the Mun, and one is left free for other uses (e.g. linking to another mid-level relay satellite, or whatever suits my needs at the moment).

Then, at Minmus (or the Mun), put a similar satellite with a 45-degree cone dish pointed back at Kerbin, and a Communotron-32.  This goes in a polar orbit that's pretty high up (near the SoI boundary for the moon), but low enough that it's within Communotron-32 range of the surface.  That gives you 50% coverage of the moon's surface.  If you want better than that, add a few more satellites around it with omni antennas.

That's pretty much it.  Any ship that's landing on Minmus (or the Mun) then only needs an omni antenna on it.

You can add more satellites for guaranteed coverage, but I find that it's enormously less hassle just to settle for intermittent coverage that gives me what I need when I need it, and live with occasional dead spots.

The key takeaway: You really don't need 100% coverage.  Interruptions are okay.  Lots of people new to RT get hung up on wanting a 100% guaranteed total connection that will never ever under any circumstances get interrupted.  That's doable, but it's a lot of hassle and in practice really isn't needed.  You'll make your life a lot simpler if you settle for most-of-the-time coverage.  The occasional gaps don't matter, as long as they don't happen in the middle of a landing maneuver, and you can plan ahead to make sure that doesn't happen.  (Note that you can schedule maneuvers with the flight computer to take place at a later time, and it works just fine even if your probe is out of communication at the time of the maneuver-- it will execute it on autopilot.)

[FancyMouse]

I once thought I don't need 100% coverage and bravely just put only two satellites on polar orbit of ~160 degree apart, until for one mission I can't control my ship to land for quite a long period of time which drove me mad. Now I always use three.

Match orbital period isn't hard at all. The trick is to use Ant engine, and use both thrust limit and throttle control to let it output minimum thrust. And it's often easier to match semi-major axis via informational mods rather than directly trying to match period. That's why my network semi-major axis is usually some whole number - 2Mm for Kerbin's, for instance. Final trick is to repeatedly pause - resume until the number is really precise (this is especially needed at low gravity environment), and then go back to KSC and never touches the satellite again.

My Kerbin communication network semi-major axis error is <1m, and it can keep running for 10 years without even a noticable drift, and still reasonably good after 40 years.

[Zhetaan]

When you absolutely cannot have drift, don't forget that you can land your relays:  until they simulate continental drift on Kerbin, a landed sat is a fixed installation.  When you absolutely need Kerbin coverage, you can also land a relay on the near side of the Mun.  I like to put a sat with a KR-7 (the 25-degree antenna) on the near face of the Mun for this purpose.

For Kerbin system coverage, I also like to stack a few dishes to different destinations on a big truss satellite that I put into an extremely eccentric polar orbit.  The periapsis is about 75 km and the apoapsis is somewhere around 70 Mm--I forget the exact figure, but the idea was that the apoapsis would cover Minmus with a KR-7 at maximum possible separation, taking into account Minmus's inclination and orbital distance.  This orbit has a period of something like 35 days, many of which are spent up near the apoapsis.  I lose coverage when the sat is behind Kerbin at periapsis, but the eccentricity means that the satellite screams through that part of the orbit quickly.  For losing coverage for about ten minutes per month, I'm not too upset with this arrangement.  I don't have to worry about needing a lot of battery, either.  The only real trick to this setup is that being in a polar orbit means that KSC doesn't easily connect to it; the solution is either to put another relay in keostationary orbit over KSC or else add some capability to your comsat constellation.

[Kryxal]

I did a few constellations of satellites, I remember setting things up with highly eccentric orbits so as to have a lot of "dwell time" nearer to the apoapsis.  I also inclined things a bit and kept the periapsis in the 150-200k range so as to avoid collision potential, but maybe that's a bit much...

[Snark]

1 hour ago, Kryxal said:

I did a few constellations of satellites, I remember setting things up with highly eccentric orbits so as to have a lot of "dwell time" nearer to the apoapsis.  I also inclined things a bit and kept the periapsis in the 150-200k range so as to avoid collision potential, but maybe that's a bit much...

The bit about having long apoapsis dwell times for highly eccentric orbits sounds good on paper, but in practice I've never found a good use case for it.  About the only thing I've ever come up with is if I were to land a probe somewhere that doesn't move much (i.e. either near the poles, or on a planet that has virtually zero rotation, which in KSP means "Moho") and I want to maximize time over that spot for one single satellite.  And that situation almost never happens for me.  Just about every case where I consider "make a really eccentric orbit" turns out to be better served by just having a very high circular orbit.

And yeah, trying to avoid collisions is a bit over-paranoid.    I've played a lot of KSP over the past couple of years, and have never once had a "random" collision.  Space is big.  And also there's the fact that KSP can only collide objects if they're inside the physics bubble; there's no collision of ships in the background.  This makes KSP collisions much rarer, since it's an O(N) problem rather than an O(N²) problem.  So at least the kerbals don't have to worry about Kessler syndrome...

(I've had plenty of not-so-random collisions, though...)

[FancyMouse]

3 hours ago, Snark said:

The bit about having long apoapsis dwell times for highly eccentric orbits sounds good on paper, but in practice I've never found a good use case for it. 

For me, eccentric orbit is useful for servicing active vessel at any distance. My only relay satellites that are on eccentric orbit have the biggest antenna set to ActiveVessel, and any ship on their way transferring to their destination planet will use that. No need to worry about distance/cone angle/etc. any more, especially when close to Kerbin.

Edited January 16, 2016 by FancyMouse

[ToukieToucan]

Launched a new relay system with the max error being 20 meters, this should hold up for a couple of years right?

[Snark]

2 hours ago, ToukieToucan said:

Launched a new relay system with the max error being 20 meters, this should hold up for a couple of years right?

Yeah, that should last you a good long while.

If you'd like to know how long, you can calculate the orbital period for each of the satellites based on the error amount, If you have the orbital periods for two satellites, then you can find the synodic periodto give you an idea of how long the formation will hold up.

The synodic period is basically how long it takes one satellite to "lap" the other one due to having a slightly shorter orbital period.  If the two satellites have orbital periods t₁ and t₂, then the synodic period t_s is given by:  t_s = 1 / (1/t₁ - 1/t₂).

So if you work out that the synodic period of two satellites is 6 years, for example, that means they'll drift 60 degrees relative to each other each year, so that'll give you an idea of the longevity of your formation.

[Geschosskopf]

On ‎1‎/‎14‎/‎2016 at 9:34 AM, ToukieToucan said:

So a few days ago I made my first LKO relay system, did some missions and a bit of timewarp and discovered that my 4 evenly spaced communication probes where all on one side of the planet... The Ap of all satellites is 500 km but the Pe (500 as well) can have 7 km changes. Do all the orbits need to be exactly the same for a relay system or is this just an exception?

When setting up synchronous satellites (whether geosynchronous or just synched with each other), the ONLY orbital parameter that matters is oribtal period.  These all have to be exactly the same.  if their Pe/Ap and inclination aren't quite the same, it doesn't matter.  The satellites will wobble a bit relative to each other, but they'll stay evenly spaced, at least close enough for government work.  Think of Ike at Duna.  It has the same oribtal period as Duna's day but is in a slightly inclined, slightly eccentric orbit.  Thus, It wobbles a bit around a certain small area of sky, but that area of sky is always over the same place on the ground.,  Thus, if Ike was a relay satellite, it would never stray far enough from where you put it to cause a problem.

So what you do is, put the satellite in as good an orbit as you can, in the proper position relative to the other satellites and/or KSC.  Then point it prograde and use forwards or backwards RCS in small, short bursts to adjust your orbital period until it's the exact value you want.

HOWEVER, due to the inherent limitations of KSP's math, achieving EXACTLY equal orbital periods (or an orbital period exactly equal to Kerbin's day) is pretty much impossible.  When you're flying the ship, you'll see its orbital period fluctuating +/- a second or so the whole time you're focused on it (the more circular your orbit, the less frequent and smaller the fluctuations).  And even if you nail it and leave the satellite so that it's on rails and its period is set in stone, next time you come back to that satellite (say to aim an antenna in a new direction), it leaves the rails, it's period will change, and you'll have to tweak it again.

I find this EXTREMELY frustrating.  To me, the best way around the problem is to use HyperEdit to enter the exact orbital period you want.  Cheating to overcome game limitations/bugs isn't really cheating.  I mean, you fly the satellite to the correct place.  You do everything in your power to put in the proper orbit, but the game won't let you.  So get it as close as the game allows, leave it for another ship or go to KSC, and use HyperEdit to nail it down with the proper orbital period.  Just remember, this only works as long as the satellite is on rails.  If you ever focus on the thing again, you'll have to repeat this.

[Kryxal]

It might be worth going to 5* time warp and checking the numbers before you leave ... that way you're on-rails but still have focus.

Edited January 19, 2016 by Kryxal

[Mighty1]

For Kerbin I usually launch 4 satellites with 6 hour orbits and roughly similar Ap and Pe. Orbital height is not that important as orbital period. My sats have 1 comm 32 and 2 DTS-M1 (1 for Mun, 1 for Minmus) The "32"s have just the range for connecting with other sats as they are spaced 4900 km and I fine tune their period with RCS to maximum of 0.02 sec. difference.