New SAS functionality and You! [0.23 Update]

[Ratzap]

You're onto the real problem there I think. I ended up removing KSP completely, deleting every last scrap and doing a ccleaner job on the registry (you can never be too sure with windows ) and the fresh vanilla install is flying test craft true (I built the one Vanamonde was having trouble with and it's fine). I'll now start adding the mods I use and retesting after every one.

If you're right (and I think you are, it fits what people are seeing) once I test my 1456 tonne monster built with NovaPunch it'll fail because it's so large the CoM/CoT are not aligned exactly. I rebuilt it using the sub-assembly mod yesterday taking hours to get everything lined up "perfectly" but with that sort of mass and thrust, a strut or two probably makes enough difference to throw it into a fit.

Edit: I just loaded the stock Z-Map and it will not fly east after launch. Even with SAS on it drifts west very slowly, with the SAS off it's not slow. Now I've not tried the stock things before so I don't know if it worked pre 0.21, maybe someone here does.

Edit2: The stock craft look like they don't have any wheels (inlines or whatever we're calling them). The Kerbal X isn't as bad as the probe but that'll be done to the difference in built in. I added an inline wheel to the Z-Map and it's nailed in place - Squad! Update your stock designs please, then people will have a reference to fall back on.

Edited July 27, 2013 by Ratzap

[DMagic]

Edit: I just loaded the stock Z-Map and it will not fly east after launch. Even with SAS on it drifts west very slowly, with the SAS off it's not slow. Now I've not tried the stock things before so I don't know if it worked pre 0.21, maybe someone here does.

Edit2: The stock craft look like they don't have any wheels (inlines or whatever we're calling them). The Kerbal X isn't as bad as the probe but that'll be done to the difference in built in. I added an inline wheel to the Z-Map and it's nailed in place - Squad! Update your stock designs please, then people will have a reference to fall back on.

It's not a lack of torque that's preventing the Z-map from responding, it's those Delta-Deluxe Winglets. Remove them or replace them with the AV-R8 Winglets and it works fine. Even taking off the Deltas and moving the up a bit so that you can align them properly (so that two of them are directly above the boosters) seems to make the rocket work better. It's something about the combination of those winglets and their position that doesn't work right.

Edit: I tried out a version of the rocket Vanamonde and everyone else was using but switched the canards to the Delta winglets. With all of the torque on it works ok, but if I deactivate the torque from the inline reaction wheels and the command pod I can barely move away from the prograde vector. This does make sense in an atmosphere that the rocket would try to stay pointed in the direction of forward velocity, but if I use the AV-R8 winglets instead I can point in just about any direction, with or without the reaction wheel torque on.

I guess the point is that the Delta winglets are terrible for rocket control surfaces and they should probably be removed from the stock Z-map.

Edited July 27, 2013 by DMagic

[rhoark]

Based on my observation of the new SAS behavior, its indeed not PID - it's just P. And the P factor is fairly small. When off course it will continue to apply the same (right direction but too small) control input indefinitely.

[Frederf]

Whatever the old PID numbers are it's clear that the new SAS is just new PID numbers: static, unconsidered, one-size-fits-all numbers. One simply has to spin up a high moment craft with weak reaction wheels and hit T to see the results. You get a weak underdamped oscillation. To test the integration term one can put a slightly offcenter tiny engine and fire it with SAS engaged and see if the I term will eventually counter the torque entirely or will there be a steady state bias.

[DMagic]

Updated the original post. It should, I hope, provide a clearer and more straightforward explanation of the 0.21 SAS system. I also added details about SAS modes, and RCS changes discussed in C7's blog post from a few weeks ago.

[Lojik]

I found that if my CoL is not inline with my CoM I would drift with SAS on and rotate if the CoL is very far out, very minor differences with CoL and CoM did not really have a noticeable effect and would keep my heading in a acceptable range, which sounds right, if the system is designed to be a simple "heading hold" style control then the CoL will have an effect to its control.

My install is windows 7 64bit via steam from version 0.21. all my earlier small rockets have been straight and true as long as I have a stable launch platform, I found out last night that unlined rockets (usually ones carrying payloads) drifted allot and I either needed lots of control surfaces to keep it corrected or heavy manual control.

I'm new to game though so not sure if this is how its mean't to work

[DMagic]

I think the general consensus is that if you have enough control surfaces/reaction wheels you can overcome a mismatch between your CoL and CoM. I've seen shuttle-type design using twenty or so reaction wheels to keep the craft stable despite being very unbalanced. So you should be able to prevent drifting by adding more reaction wheels, but in general it's best to try and keep the CoL and CoM lined up together.

From what I've heard it sounds like the SAS system will be tweaked a little in the 0.22 update, so the new version might be a bit better at accommodating unbalanced crafts, we'll have to wait and see.

[KerbMav]

The other point is that damping mode doesn’t work the way you might think. At around 50% throttle the controls on the pitch and yaw axes are almost maxed out and the system is in locking mode. If you then rotate, the system switches into damping mode. Right away, you will notice that all control on the pitch and yaw axes drop to zero. This will cause the craft to start spinning out of control despite the fact that there is enough control authority to hold it steady. The system will resume controlling the pitch and yaw axes (even if you continue to hold the rotate controls), but that brief interval where the controls drop to zero is enough throw off the system.

So we should never try to rotate our rockets during, just pitch and yaw them on course?

Hampers the ability to safely asparagus stage staging, does it not?

[WafflesToo]

Is this an explanation of how it works or a sales pitch? Because it sounds like a sales pitch.

The ONLY thing that the 0.22 SAS does better than 0.21 is hold the heading. It hunts really badly and has a bad tendancy to oscilate after minor heading changes which is a pain now that electric change is something of a premium.

Argue that you like it better if you must, but don't dictate that it's "better".

[DMagic]

So we should never try to rotate our rockets during, just pitch and yaw them on course?

Hampers the ability to safely asparagus stage staging, does it not?

This isn't unique to rotating, any control input will zero out the controls on the other axes. This is definitely a problem and isn't working quite the way that it should (at least from what I understand about how damping mode is supposed to work).

That said, this was an example of what happens when the control authority is maxed out. The imbalance in this example requires near-maximum input on the pitch or yaw axis, so any interruption in that control causes the craft to spin out of control. For a normal rocket, one not so severely unbalanced, this shouldn't be as much of an issue. It requires more manual input than it probably should, be it's still pretty easy to keep the craft on course, or make a gradual turn.

The ONLY thing that the 0.22 SAS does better than 0.21 is hold the heading. It hunts really badly and has a bad tendancy to oscilate after minor heading changes which is a pain now that electric change is something of a premium.

Do you have any specific examples of how it's not working better. There are certainly cases where a craft designed for the system in 0.21 won't work as well in 0.22, but you can't expect anything made for the old system to work as well or better in the new one (some might, but not all). Of course, new craft designs can have problems too, but much of this can be mitigated by careful flying.

In my opinion the system is better in 0.22 than it was in 0.21. There are some definite improvements, but there are some areas that require more careful design and flying. Overall yes, it does tend to oscillate more, and if flown the same way as in 0.21 a craft can spend a lot of time bouncing around and not settling down on a steady course. I've found that much of this can be alleviated with careful use of manual controls to slow or stop your movement and well timed use of the 'F' key to reset the locked heading.

Electricity use is another matter and one that requires consideration when building your craft, though it shouldn't really be a problem during launch since most of the big engines generate electricity, but for large crafts in space it could be problem.

And yes, it is supposed to be an explanation, and, I thought, a fairly objective one. It is mostly a list of changes made from 0.21 with a few examples of how some of the changes affect mass imbalances. I did write that it handles unbalanced crafts better, holds a heading better, and dampens roll better (though not in the manner described in response to KerbMav) and I think those are all true statements. There are issues, and I brought up some of them in the initial post, but I think overall it's a pretty good system.

[panzer1b]

i really like it, more reliable especially for my more exotic scifi craft that are already as unoptimized as can get given i care more about looks over pure functionality.

its easier to fly off center craft

overall a great improvement, now i just need to learn how to make my ships look good, and actually fly semi well......guess scifi designs arent suppose to work with modern physics (never understood how off center thrusted ships seem to fly perfectly well in many movies), now where is my FTL drive and anti gravity boosters?......

[DMagic]

Finally updated the first post for 0.23.

Not much changed, but they did fix the docking mode translation control bug, so using the WASD keys in docking mode won't trigger damping mode anymore.

Also the new variable max thrust can be used to better balance crafts with unequal mass distribution, as long as you have more than one engine.