Multiple reaction wheels

[cephalo]

The first time I made a rocket with two reaction wheel modules, they seemed to either fight each other or fight with my rocket fins. I was getting a weird translation from side to side and not able to turn well. However, all the older stuff on KSP reaction wheels I'm searching on seem to suggest that it doesn't matter at all where you put reaction wheels relative to CoM or how many you use. Is that still true?

I'm at the point where I need a bit more torque, but the next size reaction wheel part is a ways away on the tech tree. Can I stack these in 1.0?

Edited May 12, 2015 by cephalo

[Reddeyfish]

you can stack them, but they will wobble a bit if you combine reaction wheels with engine gimbal or control fins.

[cephalo]

you can stack them, but they will wobble a bit if you combine reaction wheels with engine gimbal or control fins.

Well, I always combine the single unit with gimbals and fins, and I haven't noticed this behavior. I should also say that I had put them rather far apart from each other, as in one just below the command module and one in front of the last stage. Trying to tip east made me translate west about ten feet before snapping back.

[THX1138]

How would you even know if you had translated?

[cephalo]

How would you even know if you had translated?

I think there is a delay between the position change of your craft and the camera so you can see these kinds of movement. The craft appeared to move sideways relative to the middle of the screen. Also there was some flexing in the middle. I suppose the whole thing could have been an optical illusion, but I wanted to see if anyone else is having issues with reaction wheel placement.

[Red Iron Crown]

Multiple reactions wheels work fine, though you're better off spreading them throughout the vessel if you can rather than having them close together as it reduces bending.

Also important is where the ship is being controlled from. If there's a pod atop a wobbly rocket with reaction wheels far below, it can get into a destructive oscillation as the pointing of the pod differs from that of the main body.

[cephalo]

Also important is where the ship is being controlled from. If there's a pod atop a wobbly rocket with reaction wheels far below, it can get into a destructive oscillation as the pointing of the pod differs from that of the main body.

That's interesting. Would it be better to have the command pods near CoM?

So let me make sure I understand, in space, you can spread reaction wheels throughout the craft and they will work together?

What if, on a basic rocket, I accidentally inverted the reaction wheels during assembly, with one upside down? Maybe I'll test it later.

[Red Iron Crown]

That's interesting. Would it be better to have the command pods near CoM?

The best solution is to have a reasonably rigid ship, then it doesn't matter where the pod or reaction wheels are. If you have to have a wobbly ship then yes the control point is better closer to the CoM. Note that it doesn't necessarily have to be the pod that you're controlling from, a docking port facing the right direction will do, too.

So let me make sure I understand, in space, you can spread reaction wheels throughout the craft and they will work together?

What if, on a basic rocket, I accidentally inverted the reaction wheels during assembly, with one upside down? Maybe I'll test it later.

The amount of turning force imparted by the wheels will be the same no matter where they are. Inverting them has no effect. What can make a difference in a wobbly ship is to spread them out, so that the forces are distributed throughout the ship instead of concentrated where they'll make the ship bend more.

[Geschosskopf]

The first time I made a rocket with two reaction wheel modules, they seemed to either fight each other or fight with my rocket fins. I was getting a weird translation from side to side and not able to turn well. However, all the older stuff on KSP reaction wheels I'm searching on seem to suggest that it doesn't matter at all where you put reaction wheels relative to CoM or how many you use. Is that still true?

I'm at the point where I need a bit more torque, but the next size reaction wheel part is a ways away on the tech tree. Can I stack these in 1.0?

It is a huge myth in KSP (at least in recent times, since the last major SAS overhaul) to think that multiple reaction wheels "fight" each other. This is simply not the case. Any wobbles you experience are the result of either 1) a weak, wobbly rocket, or 2) a bad harmonic in the ratio of available control force vs. external force resulting in a dynamically unstable, self-reinforcing oscillation. Assuming you build a rigid rocket, then the problem is #2.

No matter what means of attitude control you have on your rocket (torque, fins, RCS), it only operates in 2 situations. First is to change the rocket's orientation by your control inputs. Second, when SAS is on, is to counter whatever outside forces are trying to point the rocket in a different direction from what you want. It's this latter that can sometimes results in bad oscillations when you don't have enough control authority for SAS to react fast enough to shifting external forces, so it usually gets out of phase and thus the rocket wobbles more and more. The solution is to add more control authority in whatever form you think would work best in that particular flight regime. Torque, RCS, or fins.

Note: I typically put reactions wheels on each stage of my rockets, and also put RCS tanks on the transfer stage and thrusters on all lower stages. I have zero wobbles.

[Pax Kerbana]

Reaction wheels work best close to the center of mass or at equal distance from. Don't forget you can turn off pod torque if needed. Pod torque (when combined with reaction wheels) is often responsible for making your ship rotate in weird ways.

If your reaction wheels aren't equidistant or worse, multiple far from the COM, your ship will try to pivot around the COM but will not behave the way you want it to.

IMO if reaction wheels cause your ship to wobble so much that it's a concern either you have too much torque or the ship is not rigid enough. 2 of the 1.25 m wheels will adequately move a fuel station with 8 orange jumbos without shaking it apart (even if tanks are only connected by clamp-o-trons). I have little use for the large reaction wheels except for massive sandbox monstrosities... They just aren't needed.

[Red Iron Crown]

Reaction wheels work best close to the center of mass or at equal distance from. Don't forget you can turn off pod torque if needed. Pod torque (when combined with reaction wheels) is often responsible for making your ship rotate in weird ways.

If your reaction wheels aren't equidistant or worse, multiple far from the COM, your ship will try to pivot around the COM but will not behave the way you want it to.

Reaction wheels apply exactly the same turning force no matter where they're located. No need to worry about them being equidistant from CoM or close to CoM. Ideally the ship is rigid enough for it not to matter, with wobbly ships it's better to distribute them around and have the point of control near the CoM.

[Geschosskopf]

Reaction wheels apply exactly the same turning force no matter where they're located.

This is what I'd always thought. But there's an easy way to test it. Build a test rig and HyperEdit it into orbit. The test rig could be a stack several orange tanks so that there's a reaction wheel in the middle and on each end. The CoM thus remains constant. You disable the torque on 1 end and leave if off (it's just there to balance) and alternate rotating the stack with either one of the other 2 wheels (central or the one on the other end). This will show you if there's any difference in effect with the wheels at different distances from the CoM.

[cephalo]

I have since used multiple reaction wheels without issue. The craft where I was having problems was actually a very rigid craft, but the problem was occurring in the Kerbin's atmosphere on the way up. It must have been a problem with my aerodynamics, but it was very strange. It wasn't a wobble but a shift. I did solve that problem by removing one of my reaction wheels and just living with a slow turning craft. I don't really know what was going on I can't replicate it.

[Red Iron Crown]

This is what I'd always thought. But there's an easy way to test it. Build a test rig and HyperEdit it into orbit. The test rig could be a stack several orange tanks so that there's a reaction wheel in the middle and on each end. The CoM thus remains constant. You disable the torque on 1 end and leave if off (it's just there to balance) and alternate rotating the stack with either one of the other 2 wheels (central or the one on the other end). This will show you if there's any difference in effect with the wheels at different distances from the CoM.

The effect was demonstrated to me in one of Kasuha's experiments here: http://forum.kerbalspaceprogram.com/threads/75916-If-something-has-an-SAS-torque-of-20-what-does-that-actually-mean?p=1080585&viewfull=1#post1080585

And another experiment of his, showing how reaction wheel placement affects wobbly ships: http://forum.kerbalspaceprogram.com/threads/82091-SAS-modules-can-fight-each-other?p=1197907&viewfull=1#post1197907

[MacroNova]

So there's no benefit to putting reaction wheels far from the center of mass to get increased torque?

[Red Iron Crown]

So there's no benefit to putting reaction wheels far from the center of mass to get increased torque?

No. Though RCS thrusters and engine gimbals benefit from being further from the CoM.

[MacroNova]

Haha, wow. Reaction wheels are even more magical than I realized.

[Pax Kerbana]

Reaction wheels apply exactly the same turning force no matter where they're located. No need to worry about them being equidistant from CoM or close to CoM. Ideally the ship is rigid enough for it not to matter, with wobbly ships it's better to distribute them around and have the point of control near the CoM.

I understand what you said and are saying.

The wheels apply the same force, that is correct, but the moment arm about the COM will be different if they are not spaced the same. In general this isn't usually an issue but if you have a large or long ship with wheels at the bottom or otherwise not symmetrical it will lead to problems and the ship will not move the way you want it to. Sometimes this isn't perceptible, sometimes it is.

Unless you are saying that regardless of placement of the wheels, the game rotates the ship around the COM anyway. If that is the case I would be very surprised because I do not think it works that way.

[Red Iron Crown]

Unless you are saying that regardless of placement of the wheels, the game rotates the ship around the COM anyway. If that is the case I would be very surprised because I do not think it works that way.

That is what I am saying. I was imprecise before, reaction wheels don't apply force, they apply torque directly, so moment arms don't matter. (This is in contrast to RCS thrusters and engine gimbals, which do apply force and moment arm is very important.) Have a look at the experiments I linked in post #14.