How to activate the separator

[JackBush]

Here is the rover atop my ship. It is attached by a TR-18D stack separator which I first put on the command pod.

http://imgur.com/kWNtQL7

I got the ship with 3 Kerbals safely to Mun but don't know how to activate or trigger the separator so that it will pop or fling the rover off and onto the ground.

Here is the description of the separator:

http://imgur.com/gtmqUvn

I have no idea what "Crossfeed toggles in Editor, Flight default off" means. What is the editor? How would I turn default to "on"? 

I've been struggling with this whole attachment business for about 10 days now. I'm now the swiftest learner around but I can put that ship together in about 60 seconds now, I've done it so many times trying to get things right. This program is the least intuitive I've ever used, going back to when I began using computers (1980s). People like me are fortunate to have a caring community such as the knowledgeable ones on this forum. I like the game a lot, but it's frustrating as hell. And I'm grateful for the help I get here. 

I look forward to whatever answers you can provide to the questions above. I have 3 Kerbals sitting on Mun, waiting patiently to find out if they're going to get to ride around on their rover. Thanks in advance.

[Pecan]

First, I see no separator between the docking port and rover in that first picture.  If you put one on it goes in the staging sequence just like a decoupler - the only difference between a separator and a decoupler being that the former separates from both parts of the vehicle, not just the one pointed to by the arrows on the latter.  If not in the staging sequence you can right-click a separator (or decoupler) and click the button in the GUI to activate it.  If you didn't stick one on at all then you can right-click and decouple the docking-port - but there won't be much of a push on the rover.

Crossfeed = fuel can cross the separator/decoupler/docking-port when enabled.
Editor = thing you use for building ships = VAB or SPH.
You do not change the default, it's the default.  What you do, if you wish, is change the actual behaviour, which you do by ... right-clicking the part in the editor and changing its setting.  As it happens, you can do that in flight too but it only affects that flight.

ETA:  And for your next questions:

• Once separated, in whatever way, you switch between vehicles using '[' and ']' to control the one you want
• Use RCS or SAS and a rocking action to get your rover to fall off the rocket when it doesn't on its own (docking port)
• Be prepared to catch it when it flies off into space if your force is too high (separator/decoupler)
• Use RCS or SAS and a rocking action to turn it back upright once it hits the ground
• If it survives.  Engineers can repair broken wheels.
• Expect 'fun' driving around

 

Edited December 23, 2015 by Pecan

[Snark]

Also, when it says "Crossfeed toggles in Editor, Flight", what that means is "you can turn the crossfeed on/off either at build time in the VAB/SPH, or while you're flying the ship by right-clicking on the part."

 

[Plusck]

Agreed with what Pecan said. There is no separator visible here. Assuming you don't have something clipping into the parts there, you would have to right-click the docking port and chose "decouple node" (or words to that effect).

However there will be no significant decoupling force. Your rover will wobble on top of your command pod until gravity wins, at which point it will fall and maybe break.

Rovers are not an easy option, because getting them on the gound the right way up and in one piece is hard. It's tempting to think of a rover as something that trundles off your lander, but that is probably the hardest thing to do. I did that early in career game but it was massively non-optimal. I'll try and dig up a pic of it so you can point and laugh...

What is easier is to consider the rover to be the only thing that matters, and all the rest is sacrifical to it. So your final stage lander is basically just big enough to get the rover to 1m above the ground (at whatever angle) and then you cut the engine, tip the rover onto its wheels and abandon the rocket stage.

With your current setup this isn't going to be very possible - you can't just tip the lander over and even if you did the rover would probably end up upside down.

However, add a couple of puffs + monoprop or a droppable pair of oscar tanks and twitches to the sides of the rover (maybe with small reaction wheels too, so that CoM issues are less of a worry) and you'd be able to lift off the lander and drop the rover next to it.

Edited December 23, 2015 by Plusck

[JackBush]

1 hour ago, Plusck said:

Agreed with what Pecan said. There is no separator visible here.

I feel like a damned fool. What it is is a Clamp-a-Tron docking port. Is there a way I can separate the rover from that while on the Mun?. Sorry to be a muddlehead.

And what would be a better way to attach that rover, if this arrangement is bad?

[Snark]

39 minutes ago, JackBush said:

I feel like a damned fool. What it is is a Clamp-a-Tron docking port. Is there a way I can separate the rover from that while on the Mun?. Sorry to be a muddlehead.

And what would be a better way to attach that rover, if this arrangement is bad?

I don't know for sure how to fix that, have never actually done that.    Does the docking port have a "decouple node" option?  The thing that makes me wonder whether it will work is that you're not docked to the docking port, you're attached to it, and I don't know whether it knows how to let go of something that's not docked.

As for what's a better way:  don't use a docking port, use a separator.  (i.e. the short answer to your question "how to activate the separator" is "make sure it's a separator") 

You may want to consider slinging your rover underneath your lander rather than perched on top.  This works if your lander is compact enough that your lander legs reach past it.  That makes getting off the lander easy:  just decouple the separator, then drive out from under the lander.

Another option is to give the rover an engine and small fuel tank, just enough to land itself from low Munar orbit.  Then you can separate from the lander in orbit, land the lander, then on the next orbit land the rover.  That makes the design of the rover more simple, but on the other hand it requires somewhat better piloting because you need to be able to land the rover approximately where the lander is, if you don't want a tediously long drive.  The good news is that it doesn't have to be exact, as long as you get within a dozen kilometers or so, the driving's not all that bad.

[Pecan]

6 hours ago, JackBush said:

I feel like a damned fool. What it is is a Clamp-a-Tron docking port. Is there a way I can separate the rover from that while on the Mun?. Sorry to be a muddlehead.

And what would be a better way to attach that rover, if this arrangement is bad?

A. Already answered in my original post.

B. Two on the opposite sides, so they fall off naturally or, better, one underneath with radial engines on the lander itself to keep them  out of the way.

[Plusck]

6 hours ago, Snark said:

I don't know for sure how to fix that, have never actually done that.    Does the docking port have a "decouple node" option?  The thing that makes me wonder whether it will work is that you're not docked to the docking port, you're attached to it, and I don't know whether it knows how to let go of something that's not docked.

Actually, docking ports make perfectly decent connectors, which you can release by right-clicking and then pushing the "decouple" button. I use and abuse them all the time, especially when I want to keep a docking port for later on top of my main ship and have some sort of single-use probe (which doesn't need one) on top.

The arrangement below worked perfectly well for my Eve lander. The only problem is (a) it looks silly because the docking port isn't visibly connected to the heatshield, even though it is, and (b) the shroud on the heatshield stays magically connected to the main ship when the lander is released, but only for as long as the ship remains loaded (so after landing, I went back to the main ship and the shroud had disappeared).

Spoiler

 

[JackBush]

10 hours ago, Snark said:

 

. the short answer to your question "how to activate the separator" is "make sure it's a separator") 

You may want to consider slinging your rover underneath your lander rather than perched on top. 

Ah, Snark, you made me laugh out loud at the "short answer". Beautiful.

Now, a variety of you say put the rover underneath the lander. Where underneath? My fuel tank and engine are there between the landing struts.

Thanks everyone for your help and suggestions, and for not laughing too hard.

[Foxster]

You could try putting a small reaction wheel and a probe core on your rover... 

You can detach the rover from your spacecraft by right clicking the docking port and choose "Decouple node". Then you will be able to flip the rover off the top of your lander by using the WASD keys. You can also use this to right the rover if it lands upside down.

You will probably find the rover undriveable with the torque from the reaction wheel, so right click it and select "Toggle Torque" off unless you need it to right the rover again. 

A driveability tip for these simple rovers is to right click on the rear wheels and click "Lock steering".

Edited December 23, 2015 by Foxster

[Foxster]

...but then, a rover in a Service Bay will be low drag at launch and can be driven out onto the surface on arrival...

(Note the necessary "roll-cage" ). 

Edited December 23, 2015 by Foxster

[Plusck]

Neat!

This is why I like this forum. I would never have dreamed of putting the service bay between the rocket and the fuel, for example. Just wouldn't have crossed my mind to do it that way.

And I'm glad the bigger rover is fine after falling off the roof like that. Again, I would probably never have put myself in a position to risk it.

[Snark]

3 hours ago, JackBush said:

Now, a variety of you say put the rover underneath the lander. Where underneath? My fuel tank and engine are there between the landing struts.

Where underneath?  Underneath underneath. 

This generally involves putting fuel tanks and engines off to the side.  For example, here's a fairly common lander type I use in early career when I want to strip-mine the Mun or Minmus for science:

The details change around based on mission profile, but you get the basic idea.  Mounting the tanks & engines on the side of the lander gives it a super squat profile, which means it can land pretty much anywhere without tipping over; I've set down a lander like this on a 45 degree slope with no problem.  The design is nicely flexible:  add a probe core up top if I want to fly it with a scientist instead of a pilot; add parachutes if I want it to be a Kerbin-return vehicle as well; sling an extra fuel tank underneath if I want to boost the dV (or even replace the 1t tanks with 2t).

The aerodynamics aren't great for launch, so I'm more likely to use a design like this for missions of the "orbiting fuel-tanker mothership, lander makes multiple trips back and forth to low orbit" variety, where the benefits of multiple landings offset the one-time cost of atmospheric launch.  (Plus, for such a mission, I'm likely launching the lander atop a massive mothership anyway, so the crappy aero of the teeny lander is less of an issue and I can put it in a big ol' fairing.)

The point is, it wouldn't be hard to adapt this for a rover slung underneath; just use longer landing legs.  The aero of the lander isn't great anyway, so adding a rover underneath is just more of same. 

Another option is to just let the lander be the rover, like this:

...which can be kinda fun, though if I want a rover that can land itself, I'm more likely to just use an axially-built one, they're more practical.

Edited December 23, 2015 by Snark

[Snark]

1 hour ago, Foxster said:

You could try putting a small reaction wheel and a probe core on your rover... 

You will probably find the rover undriveable with the torque from the reaction wheel, so right click it and select "Toggle Torque" off unless you need it to right the rover again. 

Reaction wheels on rovers are a great idea, especially on low-gravity worlds.  However, I'd like to comment on the latter remark, about "undriveable with the torque from the reaction wheel."  I find exactly the reverse:  rovers are undriveable without the torque.  The bump, flip, and tumble all over the landscape.

In my experience, folks who find reaction torque to cause rover-driving problems are Doing It Wrong™.  (Speaking as someone who Did It Wrong™ for a long time until I figured out the problem.) 

There are two classic mistakes that cause reaction wheels to give grief on rovers.

The first problem is control-from orientation.  It's fairly common for folks to build their rovers sideways rather than axially (that is, the direction the rover drives forward is at right angles to the center axis of the rover, rather than parallel), with a probe core mounted in the middle so that it's "looking" straight up at the sky rather than forward in the direction the rover drives.  This is bad even if you don't have any reaction wheels at all-- it makes the navball very misleading and hard to read when you're driving around, since it's pointing up at the sky.  The natural orientation of the navball should be in the direction the rover is pointing.  (Rovers that are built axially rather than sideways don't have this problem.  For sideways-built rovers, such as the ones pictured in this thread, the problem is easily solvable by mounting a probe core or docking port on the front of the rover, and then choosing "control from here" when driving the rover.  Or if you have a command chair on the rover, do control-from there and it will be facing the correct way.)  If reaction wheels are present, the control-from-is-pointing-the-wrong-way problem gets worse, because then the rotation controls make no sense:  you want to turn the rover to the left, but instead of the reaction wheels turning it left, they cause it to roll over sideways, or even tip over forward, depending on how the control-from point is aligned.  It's almost impossible to control, it plays games with your brain.  Getting the control-from point so that it's pointing in the same direction as the rover drives solves the problem by making the reaction wheel orientation intuitive; the rover drives like a ground-bound airplane.

The second problem is one that's not really players' fault, but Squad's.  The thing is, there are torque rotation controls (pitch, yaw, roll), and there are rover steering controls (turn left, turn right, forward, backward), and by default, Squad has mapped those controls to the SAME KEYBOARD BUTTONS.  Yeesh.  It's completely unintuitive and causes no end of grief for players-- "I want to steer my rover but the act of steering causes the reaction wheels to flip it all over the place."  The solution to this is to remap the controls so that they don't collide.  For example, I like to remap the rover controls to use the numpad, so instead of using WASD, it uses 8456.  That way, it's completely clear:  WASD applies reaction torque, 8456 drives the rover.  If I want to do both at the same time, I use both hands (for example, when I am steering left by holding down the 4 key, I can also use the reaction wheel to help it by yawing left with the A key).

As long as you've taken care of these two problems, then a reaction wheel can be your rover's best friend on a low-gravity world.  You can leave SAS turned on and it really helps prevent tumbling.  You can control the orientation to land properly when you jump.  And so forth.

[Foxster]

3 minutes ago, Snark said:

Reaction wheels on rovers are a great idea, especially on low-gravity worlds.  However, I'd like to comment on the latter remark, about "undriveable with the torque from the reaction wheel."  I find exactly the reverse:  rovers are undriveable without the torque.  The bump, flip, and tumble all over the landscape.

In my experience, folks who find reaction torque to cause rover-driving problems are Doing It Wrong™.  (Speaking as someone who Did It Wrong™ for a long time until I figured out the problem.) 

There are two classic mistakes that cause reaction wheels to give grief on rovers.

The first problem is control-from orientation.  It's fairly common for folks to build their rovers sideways rather than axially (that is, the direction the rover drives forward is at right angles to the center axis of the rover, rather than parallel), with a probe core mounted in the middle so that it's "looking" straight up at the sky rather than forward in the direction the rover drives.  This is bad even if you don't have any reaction wheels at all-- it makes the navball very misleading and hard to read when you're driving around, since it's pointing up at the sky.  The natural orientation of the navball should be in the direction the rover is pointing.  (Rovers that are built axially rather than sideways don't have this problem.  For sideways-built rovers, such as the ones pictured in this thread, the problem is easily solvable by mounting a probe core or docking port on the front of the rover, and then choosing "control from here" when driving the rover.  Or if you have a command chair on the rover, do control-from there and it will be facing the correct way.)  If reaction wheels are present, the control-from-is-pointing-the-wrong-way problem gets worse, because then the rotation controls make no sense:  you want to turn the rover to the left, but instead of the reaction wheels turning it left, they cause it to roll over sideways, or even tip over forward, depending on how the control-from point is aligned.  It's almost impossible to control, it plays games with your brain.  Getting the control-from point so that it's pointing in the same direction as the rover drives solves the problem by making the reaction wheel orientation intuitive; the rover drives like a ground-bound airplane.

The second problem is one that's not really players' fault, but Squad's.  The thing is, there are torque rotation controls (pitch, yaw, roll), and there are rover steering controls (turn left, turn right, forward, backward), and by default, Squad has mapped those controls to the SAME KEYBOARD BUTTONS.  Yeesh.  It's completely unintuitive and causes no end of grief for players-- "I want to steer my rover but the act of steering causes the reaction wheels to flip it all over the place."  The solution to this is to remap the controls so that they don't collide.  For example, I like to remap the rover controls to use the numpad, so instead of using WASD, it uses 8456.  That way, it's completely clear:  WASD applies reaction torque, 8456 drives the rover.  If I want to do both at the same time, I use both hands (for example, when I am steering left by holding down the 4 key, I can also use the reaction wheel to help it by yawing left with the A key).

As long as you've taken care of these two problems, then a reaction wheel can be your rover's best friend on a low-gravity world.  You can leave SAS turned on and it really helps prevent tumbling.  You can control the orientation to land properly when you jump.  And so forth.

Excellent points. 

I was keeping it simple though for the OP who sounded relatively new to KSP. The rovers above are driveable without SAS and the act of putting a Kerbal into a seat facing forward sorts out the probe orientation issue. 

[JackBush]

I love all this stuff. I should have hung on and waited for these latter posts to arrive. Instead I went ahead and built another ship with the rover on top and a SEPARATOR and it worked like a charm. Flipped that rover off and it even landed right side up. And then I EVA's one of the Kerbals, the scientist, and once I got her (?) out and over to the rover, the rover took over. I couldn't figure out how to get back and get my engineer out so he could travel with her (and fix the broken wheels that I believe Snark warned about some time ago). There must be a way to be able to access whichever unit (rover or lander) that you want, isn't there?

I'm surprised to see you can put a service bay between the fuel tank and the engine. Do you then run a fuel line from tank to engine that I'm not seeing in your pictures?

[Foxster]

1 hour ago, JackBush said:

I'm surprised to see you can put a service bay between the fuel tank and the engine. Do you then run a fuel line from tank to engine that I'm not seeing in your pictures?

Unless the part's description says "No fuel crossfeed" (which is very few parts) then fuel gets piped automatically through all other parts in kinda hidden ducting. 

1 hour ago, JackBush said:

There must be a way to be able to access whichever unit (rover or lander) that you want, isn't there?

If I understand you correctly then you want the [ and ] keys. They cycle focus through nearby vessels/Kerbals. 

Edited December 23, 2015 by Foxster

[Snark]

23 minutes ago, Foxster said:

If I understand you correctly then you want the [ and ] keys. They cycle focus through nearby vessels/Kerbals. 

+1 to this.  The existence of the [ and ] keys is one of those simple, silly little things that's totally easy to miss as a newbie.

They work for everything within a range of 2.3 km.  Also comes in very handy for docking (i.e. so you can briefly switch to the target vehicle, to rotate it to a friendlier orientation, before switching back to the ship that's coming in to dock).

<rant>
I spent months playing KSP before I discovered those... grumbling all the while to myself about "grrrr, they should have an easy way of switching."  If I wanted to switch, I would go back to the space center and then pick the other ship in the tracking station, time after time after time.  Incredibly tedious and annoying.  (And from reading this forum a lot, I know a lot of other KSP players also got bitten by this).  And then one day I find out about [ and ] and think, what?  I could do this all along?  Like Dorothy and the ruby slippers?  Oh, for the love of Mike.
</rant>

[Pecan]

5 hours ago, JackBush said:

I love all this stuff. I should have hung on and waited for these latter posts to arrive. Instead I went ahead and built another ship with the rover on top and a SEPARATOR and it worked like a charm. Flipped that rover off and it even landed right side up. And then I EVA's one of the Kerbals, the scientist, and once I got her (?) out and over to the rover, the rover took over. I couldn't figure out how to get back and get my engineer out so he could travel with her (and fix the broken wheels that I believe Snark warned about some time ago). There must be a way to be able to access whichever unit (rover or lander) that you want, isn't there?

I'm surprised to see you can put a service bay between the fuel tank and the engine. Do you then run a fuel line from tank to engine that I'm not seeing in your pictures?

HELLOoooo?

Can you see my posts at all?
Did you READ any of them?
Since this is the third of your threads I've addressed in two days while being ignored I'm starting to think not.

[JackBush]

1 hour ago, Pecan said:

HELLOoooo?

Can you see my posts at all?
Did you READ any of them?
Since this is the third of your threads I've addressed in two days while being ignored I'm starting to think not.

Every one! But which item are you referring to?

9 hours ago, Snark said:

Where underneath?  Underneath underneath. 

This generally involves putting fuel tanks and engines off to the side.  For example, here's a fairly common lander type I use in early career when I want to strip-mine the Mun or Minmus for science:

The details change around based on mission profile, but you get the basic idea.  Mounting the tanks & engines on the side of the lander gives it a super squat profile, which means it can land pretty much anywhere without tipping over; I've set down a lander like this on a 45 degree slope with no problem.  The design is nicely flexible:  add a probe core up top if I want to fly it with a scientist instead of a pilot; add parachutes if I want it to be a Kerbin-return vehicle as well; sling an extra fuel tank underneath if I want to boost the dV (or even replace the 1t tanks with 2t).

The aerodynamics aren't great for launch, so I'm more likely to use a design like this for missions of the "orbiting fuel-tanker mothership, lander makes multiple trips back and forth to low orbit" variety, where the benefits of multiple landings offset the one-time cost of atmospheric launch.  (Plus, for such a mission, I'm likely launching the lander atop a massive mothership anyway, so the crappy aero of the teeny lander is less of an issue and I can put it in a big ol' fairing.)

The point is, it wouldn't be hard to adapt this for a rover slung underneath; just use longer landing legs.  The aero of the lander isn't great anyway, so adding a rover underneath is just more of same. 

Another option is to just let the lander be the rover, like this:

...which can be kinda fun, though if I want a rover that can land itself, I'm more likely to just use an axially-built one, they're more practical.

I just landed this one on Mun. Identical. I haven't begun to drive it; that's tomorrow's fun! Thanks for all your help.

[Pecan]

13 hours ago, JackBush said:

Every one! But which item are you referring to? ...

Oh that's alright then, I thought I might be on ignore list or something.

Just the various posts here, where you've responded to various others but not mine, plus the 'locked-VAB' thread and something else that I can't even find myself now.

[JackBush]

On 12/23/2015 at 7:18 PM, Pecan said:

First, I see no separator between the docking port and rover in that first picture.  If you put one on it goes in the staging sequence just like a decoupler - the only difference between a separator and a decoupler being that the former separates from both parts of the vehicle, not just the one pointed to by the arrows on the latter.  If not in the staging sequence you can right-click a separator (or decoupler) and click the button in the GUI to activate it.  If you didn't stick one on at all then you can right-click and decouple the docking-port - but there won't be much of a push on the rover.

Crossfeed = fuel can cross the separator/decoupler/docking-port when enabled.
Editor = thing you use for building ships = VAB or SPH.
You do not change the default, it's the default.  What you do, if you wish, is change the actual behaviour, which you do by ... right-clicking the part in the editor and changing its setting.  As it happens, you can do that in flight too but it only affects that flight.

ETA:  And for your next questions:

• Once separated, in whatever way, you switch between vehicles using '[' and ']' to control the one you want
• Use RCS or SAS and a rocking action to get your rover to fall off the rocket when it doesn't on its own (docking port)
• Be prepared to catch it when it flies off into space if your force is too high (separator/decoupler)
• Use RCS or SAS and a rocking action to turn it back upright once it hits the ground
• If it survives.  Engineers can repair broken wheels.
• Expect 'fun' driving around

 

Hi, Pecan,

I've just gone through this post and noted several points of value. I owe you an apology for not recognising them (crossfeed explanation and "[" & "]" specifically). Finally, I've experienced the "fun" aspect, which underlines the importance of the F5 key. 

Merry Christmas!

[Waxing_Kibbous]

Obviously, nobody here knows how to deliver a rover properly

(quite possibly the ugliest craft I ever built- it did work though!)

[JackBush]

On 12/23/2015 at 10:55 AM, Snark said:

Where underneath?  Underneath underneath. 

This generally involves putting fuel tanks and engines off to the side.  For example, here's a fairly common lander type I use in early career when I want to strip-mine the Mun or Minmus for science:

The details change around based on mission profile, but you get the basic idea.  Mounting the tanks & engines on the side of the lander gives it a super squat profile, which means it can land pretty much anywhere without tipping over; I've set down a lander like this on a 45 degree slope with no problem.  The design is nicely flexible:  add a probe core up top if I want to fly it with a scientist instead of a pilot; add parachutes if I want it to be a Kerbin-return vehicle as well; sling an extra fuel tank underneath if I want to boost the dV (or even replace the 1t tanks with 2t).

The aerodynamics aren't great for launch, so I'm more likely to use a design like this for missions of the "orbiting fuel-tanker mothership, lander makes multiple trips back and forth to low orbit" variety, where the benefits of multiple landings offset the one-time cost of atmospheric launch.  (Plus, for such a mission, I'm likely launching the lander atop a massive mothership anyway, so the crappy aero of the teeny lander is less of an issue and I can put it in a big ol' fairing.)

The point is, it wouldn't be hard to adapt this for a rover slung underneath; just use longer landing legs.  The aero of the lander isn't great anyway, so adding a rover underneath is just more of same. 

Another option is to just let the lander be the rover, like this:

...which can be kinda fun, though if I want a rover that can land itself, I'm more likely to just use an axially-built one, they're more practical.

Could you elaborate on what an "axially-built" rover would consist of? Or post a picture? I got this very model to the Mun but didn't have enough fuel to get it back. 

[Snark]

1 hour ago, JackBush said:

Could you elaborate on what an "axially-built" rover would consist of? Or post a picture? I got this very model to the Mun but didn't have enough fuel to get it back. 

So there are two ways to build a rover.  I refer to them as "sideways" and "axial".  That's just my own terminology, not anything "official"-- I made 'em up 'coz I have to call them something.

"Sideways" is how most people end up building them.  I don't know why, it makes no sense to me.  It makes the rover (IMHO) harder to launch and harder to control.  Something about the sideways design just seems to pop into people's heads first, and once it's there, it's as if it fully-occupies the rover-design slot in their brain and keeps the axial approach from having any chance. 

"Axial" is how I like to build them.

I'll get to the screenshot in a moment, which will immediately make everything clear, but first a couple of terms I'll use:

• Flight axis:  This is the up/down axis in the VAB.  It's the axis that things get radially arranged around if you turn on symmetry in the VAB.  It's the direction that probe cores and command pods point by default.
• Drive axis: This is the direction that the rover drives when it's on the ground.  It's the axis that the wheels' "roll direction" is parallel to.

In the above screenshot of a rover, for example, the flight axis is up/down, whereas the drive axis is left-right.

With those definitions in hand, then, I can succinctly describe what I mean by "sideways" versus "axial" rovers:

• A "sideways" rover is one in which the drive axis is perpendicular to the flight axis.
• An "axial" rover is one in which the drive axis is the flight axis.

Here's an example of what it looks like in the VAB:

...and "in flight":

I'm not holding this up as an example of an ideally designed rover ... it's just something I slapped together just now.  But it gets the general idea across.  The example shown is unmanned, but you could just as easily stick a command chair on top of it, or put a command pod in front of it (the Mk1 airplane cockpit works well).  I've shown an example with an engine on it, so it can land itself on the Mun.  This example is using a structural fuselage as the main axis of the rover; you could just as easily use a girder, or a long fuel tank, or a stack of octagonal struts, or whatever.

Perhaps the easiest way to think of an "axial rover" is this:  Imagine that you're designing an airplane.  Now add rover wheels instead of landing gear.  And lose the wings.

I prefer axial rovers to sideways ones for several reasons:

• It's a more natural shape for a rover, makes it easier to drive.  A rover should have a long wheelbase (i.e. is long front-to-back), otherwise it tends to flip over when braking or accelerating hard.  With an axial design, you can make it as long as you want and it's still easy to launch, since the long axis is parallel to the flight axis.  Sideways-built rovers are perpendicular to the airstream when you're launching them, which means they have a choice between "undriveable" and "unflyable".
• It's a much more natural fit for the navball.  When your rover is sitting on the surface, you want the navball to be "looking" in the same direction as the drive direction-- i.e. the center crosshairs should be on the navball horizon.  This makes it natural and intuitive to drive.  An axial rover just naturally gives you that orientation by default, so you can both fly it and drive it without having to play any games.  A sideways rover, on the other hand, typically has a probe core aligned along the flight axis, which means it's staring up at the sky when the rover is sitting on the ground.  This makes it hard to drive.  Yes, you can fix that by adding a probe core or docking port on the "front" of the sideways rover and then choosing "control from here" ... but it adds complexity to the design, doesn't look great, and means you always have to switch modes if you change between driving and flying.
• It's much easier to build a rover that can fly.  An axial rover just sticks an engine on the back.  If you have a sideways rover, you need to sling the engine underneath, which can be difficult to do (hard to get enough ground clearance).  And the axial design means you can use the engine not just to fly, but as a boost for driving up steep hills.  That's something that a sideways design has trouble with.
• More aerodynamic = easier to launch.  It's built like a normal rocket, has a skinny profile.  The only really horribly un-aerodynamic bits are the wheels themselves.  It's even possible to put the whole thing in a fairing, though usually I don't bother with that and just eat the drag loss.

About the only aspect in which a sideways design works better than axial is if you want to sling a rover underneath a lander, i.e. a skycrane model.  However, IMHO that's just not worth it.  Axial rovers can land themselves just fine, and I've never seen a benefit in skycrane/lander designs.  I'd rather just separate my lander & rover in orbit, and then land each one separately.  (This does mean having the skillz to land a rover reasonably close to where you've already landed the lander, so you don't have to drive forever... but I don't find that a problem, and neither will you after you've had a little practice.) 

The one slightly finicky bit about axial rover design is making sure that the CoM lines up on the central axis of the rover, since otherwise it becomes a bit awkward to fly.  Doesn't have to be perfect, just try to get it reasonably close.  The off-center CoM happens because the wheels need to be mounted underneath in order to give decent ground clearance, and the wheels have mass.  The way to offset this is to mount massy parts (such as solar panels) above the rover to balance out the mass.  That, and have hefty reaction torque to compensate (not a problem, since it's generally a good idea for rovers to have plenty of torque anyway).

I generally design my rovers so that they have just enough fuel to land themselves, not to go back to orbit.  If I want to have a surface-roving, land-and-return mission, then I have a rover plus a lander.  The lander comes home, the rover stays behind.

Edited December 28, 2015 by Snark