Rocket starts rolling and pitching right after I start turning east

[hellix08]

Hello,

there is this issue that I've been trying to solve for like 4 days and I still can't figure out what's wrong with my rocket.
It's just a very simple and small rocket but, for some reason, when I launch and start turning east as soon as I hit 100ms/s, it starts rolling and pitching and goes completely off trajectory which is very annoying.

I've recorded a short video to better explain my point:
https://www.youtube.com/watch?v=MojJDt_Es-k
 

Also here's a link to the .craft:
https://mega.nz/file/2JJGxCaT#FyZ2BwMTUfI1BDdeRdthWf3vRTxBAkoL7TyNWoYuHHo

I've already asked this question on Reddit before:
https://www.reddit.com/r/KerbalAcademy/comments/hcu6q2/why_is_my_rocket_so_unstable_it_starts_rolling/?utm_source=share&utm_medium=web2x
They gave me some nice suggestions but none really did explain or provide a solution to my issue.
Here's what I've already tried:

• Disabling SAS
• Disabling roll control on the wings
• Adding 6 basic fins along the center core
• Attaching the wings on the center core instead of the boosters
• Using 4-way symmetry for the wings
• Using 3-way symmetry for the wings and the boosters
• Making sure the engine has gimbal capabilities (is a Swivel and not a Reliant)
• Using smaller boosters (Hammer)
• Removing the science parts
• Removing nose caps
• Putting the instrumentation and the probe inside a service bay to reduce drag aereodynamics
• Going full thrust with the engine to make full use of the gimbal potential

And a mixture of those. None really worked or improved the stability.

 

I am clueless and desperate. If you have any idea please please write that down :).

 

[SPACE-MAN]

You need wings that extend perpendicularly to the ones in the picture. Try using smaller winglets, or making rocket taller. Having wings far below center of weight is good design for stable rocket. 

[bewing]

The way you've designed the top of the rocket has very high drag. High drag at the front of the rocket makes it unstable. To try to make it more stable, you need to add enough drag to the back end so that it has even more drag than the front end. Or, you need to redesign the front end to have less drag.

Additionally, I suspect you are not using autostruts. Those radial decouplers do not form firm connections. Those SRBs on the side will twist when they are under thrust. This makes their thrust go off-center, which will often make your rocket go off course. Unlocking struts, and turning on autostruts will help tremendously with this problem.

So, what else can you do? You must use SAS. You need to add more aerodynamic control surfaces (rudders are probably the best). And you can use rocket engines that have gimbal, and you can turn the gimbal "authority" up as high as you can. These things will allow SAS to steer your rocket as straight as possible. Additionally, you can launch straight up. Yes, doing a gravity turn is somewhat more efficient for fuel, but a failed mission is 100% inefficient. When you launch straight up, your rocket will automatically be much more stable in that position.

 

[OHara]

Welcome to the forum, @hellix08

I find that I need 4 of those large steerable fins (of which you have 2) very far back on the rocket, overhanging the main engine, to make your craft pleasant to fly.  When I drop your boosters, I am still in the atmosphere, so I put those 4 fins fins on the main stack.  I put them at 45° so they fit between boosters and launch clamps.

It is the flat front surface of the Science Junior that is giving you a lot of drag in the front.  Swapping the battery and probe core helps a tiny bit, because the larger probe core covers a bit more of the Science Junior.  Putting things in a bay or fairing would help a lot, but big four fins in the back gets the job done.  With those steerable fins you can fly a smooth turn using alt-D to bias the controls to the right (assuming of course that you leave SAS off) and then alt-X to re-center before you get into space.

For the roll, I found that a couple EAS-4 struts from the nosecones to the main stack stopped the flexing and the roll.   You might prefer those struts at the bottom of the boosters, and to move the separators up, so the separators push the noses of the boosters away upon release.

[Streetwind]

As the others mentioned, your rocket is violating the golden rule of rocketry ("heavy stuff at the front, draggy stuff at the back"). To fix this, you have attached large fins. This works... tenuously so. But it also causes a cascade of interdependant problems that you would not have if you designed your rocket differently from the get-go.

For example, there are two main forces that fight each other over the heading of the rocket: your center stack is trying to reverse itself because its front is extremely draggy, but the fins on the side boosters are trying to keep the rocket straight because they are also extremely draggy. Two strong forces that push and pull against each other. And guess what's in the very middle of them? The radial decouplers that attach the boosters to the center stage. The center stage literally wants to turn 180 degrees, and the boosters don't want the center stage to turn at all. So the radial decoupler... flexes. Just a tiny bit. The center core is now just a miniscule, impercetible amount off-center. Or actually, since the center core is the one that creates the reference frame for the direction of the rocket, the boosters are off-center. Also, the force that presses the decouplers into flexing is not static. Your rocket's acceleration changes constantly, the aerodynamic forces change constantly, you sometimes have player steering inputs and sometimes not, and so on and so forth. That means the magnitude of the flexing also changes constantly. It makes the boosters wobble. And when the nozzles of the boosters no longer point in exactly the same direction - for example, if one points a little up and the other points a little down - they start inducing roll into your craft.

These effects now meet other shortcomings. For example, your rocket is wide, but not long. That means there is less distance between the center of mass and the center of aerodynamic pressure, which lessens passive stability. You are launching at an incredibly high acceleration (TWR >2.0, maybe even 2.5, it seems from the video), which causes you to hit the sound barrier very low in the atmosphere, and the wave drag just hits your rocket like a sledgehammer, magnifies all its control issues, and fights your own control inputs. Then, your pair of fins gives you stability in the pitch direction, but not in the yaw direction. Which is counterproductive, as you want the rocket to pitch (follow its gravity turn), but not yaw (deviate from 90°). Finally, you are fighting high drag by adding more drag. This is fuel inefficient and forces you to build a larger rocket with more engines and more fuel than is strictly necessary - and attaching these extra engines and extra fuel opens up construction traps like the flexing decouplers.

There are a few common best practices for rocket building that should help you avoid these (and other such) issues:

• Try for passive stability. Attaching fins is a band-aid, not a goal to strive for. Each fin you add adds dead weight and aerodynamic drag, reducing the performance of your launch vehicle. Passive stability is a natural result of the golden rule of rocketry, but you can help it along by having a nice, long, realistic looking rocket rather than flat, wide designs. Also by adjusting the fuel flow priority of your launch stage so that the tanks drain bottom-first. Having an engine with thrust vectoring on the core stage is not strictly speaking passive stability, but it typically removes the need for fins as well. If you must add fins, use the smallest size you can get away with. Also, use four of them. Not three, not two, not six. Four. Attach them to the core stage only. And align them in the four cardinal directions. 45° offset works in a pinch (if you must attach boosters) but is less efficient.
• Try for single-stack designs. Attaching radial boosters is a band-aid, not a goal to strive for. You can use them to push a heavy first stage off the pad, but do not make them your main propulsion stage. If you do use multiple parallel stacks: do not give the boosters any control authority if you can in any way avoid it. No reaction wheels, no thrust vectoring engines, no fins (especially not active fins). And if they're big and heavy, then strut them tightly to the center core!
• Try for single-size designs. Something like a tapering rocket where the first stage is wider than the second stage is fine. But do not randomly change to a different-size part in the middle of a stack and then back. This causes massive drag around that location, and additionally, smaller parts have smaller, less stable joints, which will cause your rocket to flex around a smaller part. If you wish to downsize or cluster engines, or attach small parts like probe cores in the middle of a larger stack, there are parts specifically made for this purpose: service bays, structural tubes, engine plates, and even fairings. (Some of these may be DLC only.) When building fairings, try to minimize their bulging out past the core stage's diameter - and if forced to do so, do it as smoothly as possible.
• Try for smooth stacks. Anything you attach to the outside of the launch vehicle is going to create drag. Anything you want to carry to space with you will have to be attached to the upper stage or spacecraft, which means right at the front, which violates the golden rule of rocketry. Just like you can hide undersized parts inside service bays and similar larger parts, you can hide your science experiments, antennas, batteries, and even solar panels in them, instead of sticking them to the outside. If you wish to have them exposed, encapsulate the payload in a fairing.
• Try for sane acceleration values. A surface TWR of 1.4 to 1.5 is recommended. Not only do such rockets need less control authority to steer, but they also experience much lower maximum dynamic pressures while transsonic. Additionally, having a high TWR is an indication that you brought too much engine for the job at hand. And since engines are dead mass, this reduces the performance of your launch vehicle.

You will frequently encounter situations where you will have to compromise on at least one of these, but as long as you keep them in mind and try to avoid doing the opposite, you should end up with fairly stable, benign, and capable rockets.

Edited June 24, 2020 by Streetwind

[AHHans]

In addition to what was already said here (O.K. @Streetwind just mentioned that): your first stage has waaay too much thrust. With the swivel at 100% thrust it has a TWR of 2.7 on the launchpad. Good values for the start at the launchpad are between 1.3 and 1.5. A TWR of larger than 2 anywhere during the ascent in the atmosphere should be avoided. That means that the rocket accelerates too fast while still deep in the atmosphere which causes lots of drag. And because most of the drag of this rocket is at the top that makes it also more flip-happy.

I actually got you craft into orbit: I throttled up only to 30% on the launchpad (so that I have at least some gimbal control), pitched a few seconds after launch, throttled up to max at/just before the SRB separation, and never(!) went too far from the prograde marker in the atmosphere (well, when I did I fought the rocket back to prograde as soon as I noticed). That resulted in a very steep (read: ineffective) ascent with > 1000 m/s needed from the terrier in the circularization burn. My main issues were getting it to turn at all while the SRBs were still running, and that it really wanted to flip once the SRB were detached.

I modified the craft somewhat: I changed the wings to 4-way symmetry attached to the central core, included an aerodynamic adapter between the science junior and the probe core (which meant that I had to re-arrange all that small stuff up there), and set the thrust limiter for the SRBs to 50%. No when launching at 30% throttle it is a much more well behaved craft (throttle up at SRM separation), and got me to orbit with much more dV left than the previous version (even though there is more mass in the payload). You can find my version here: Moon Satellite AH.

Another comment: The science junior is very heat sensitive, and it is directly connected to the heat-shield which gets rather hot during reentry. It isn't unheard of that the science junior explodes when reentering in this configuration. But with closed doors on the science junior and careful flying (keep retrograde to "surface" during reentry) you should be fine (the craft it lightweight so it should decelerate fast). But I personally am lazy, so I stopped trying to reenter unprotected science juniors.

[jimmymcgoochie]

That video doesn’t really show what the problem is, it looked like a fairly standard launch although you pitched (yawed?) over too far, too fast and you’re trying to head-butt your way through the air too quickly.

A few things that might help:

1. Drag. You have too much of it! Add a fairing just under the decoupler below the heat shield and use it to make that blunt front end pointy and smooth which will reduce drag at the front of the rocket.
2. Fins. Those two fins on the boosters aren’t that helpful, take them off and put four of them on the centre stack, right at the base, offset by 45 degrees to the boosters which will provide more than ample stability and control at low altitudes.
3. Control. You have almost none since the probe core’s reaction wheels are far too weak to control a whole rocket during launch, SRBs have no gimbal and the swivel engine is throttled down low so its gimbal is ineffective. You could add a reaction wheel to the probe and a larger one to the top of the first stage, or just dial back the SRBs and use the Swivel at full thrust.
4. Thrust. It’s overkill to use two Thumpers at full power for a rocket that small, reduce their thrust to ~30% so you don’t speed up too much when the atmosphere is thickest. If your main engine is burning too much fuel when you do that, add small fuel tanks to the top of the boosters and set ‘enable crossfeed’ on the radial decouplers so that those tanks drain first and can be dumped with the boosters- it’s more efficient than adding them to the stack
5. Struts. Turn on advanced tweakables then add autostruts between the boosters (and all the parts attached to them) to their grandparent parts. That will keep them steady and stop them wobbling which can add unwanted rotation.
6. Rotate the probe core 90 degrees so you’re pitching rather than yawing, it might make it more intuitive. If that doesn’t help, try using the mod GravityTurn to fly your launches for you; you might learn from it and it’s usually more efficient than trying to do it by hand.
7. That science junior could cook on re-entry followed by the rest of the probe, stick an experiment return unit onto the probe core and use that to collect the science experiments before re-entry, then dump the dud science parts and you have a much smaller and lighter re-entry section. I would stick the ERU on top of the probe core with the parachute on top of that so that even if the probe gets cooked on re-entry the ERU will survive with the science inside it.

Edited June 24, 2020 by jimmymcgoochie

[hellix08]

Thank you all so much for all these replies! They look very promising and helpful. I will read everything carefully and try applying what you've suggested as soon as I get home from work!

[hellix08]

Thank you again so much for taking the time to reply @SPACE-MAN @bewing @OHara @Streetwind @AHHans @jimmymcgoochie I really do appreciate it.
I'll probably have to come back and read this thread a lot more times, the information you've provided is gold to me at this stage.

I've redesigned my rocket to carry the same payload and here's the result (testing in sandbox but will launch in career so I only used the parts that I have available there):

Maybe it's not perfect but I think it's already somewhat before than before!
It took some time to learn how to fly properly with a good ascent curve but I think it is pretty stable.

I used SAS only until I got to 100m/s. After that I used ALT+W to turn steadily, I planned on reaching 45° at 10.000m.
I really wanted to use SAS but I don't know how else to do a smooth ascent trajectory other that using ALT+W and leaving SAS off.
It took the first stage + 15s burn time of the upper stage to reach a 70.000m orbit. I think that might be enough to put into a Mun orbit (my objective) and maybe even come back and land if I wanted?

Your suggestions were great and I tried to remember and apply as many of them as possible so here's what I did to design this new rocket:

• I tried making the rocket taller and slimmer
• I removed the boosters because they were overkill for the mission and the TWR was too high although I enabled Advanced Tweakables and played around with auto-struts a little bit so next time I need boosters I know one more trick to try out
• I put the probe and instrumentations inside a service bay to reduce drag at the top
• I rotated the probe so that pressing W makes the rocket go east
• I tried bringing the CoL as low as possible with fins and wings (using as few as possible!). I found that no fins at all would make controls too weak. So I went with the smallest and lightest controllable wings I have access to in career mode.
• I used fuel priority (available with Advanced Tweakables) to make sure the tanks are drained bottom-up to keep the CoM as high and far from the CoL as possible.

Also I tried keeping the rocket pointing inside the prograde marker to avoid too much drag.

All of this resulted in probably the smoothest and most efficient ascent to orbit I ever did so far (not that I did that many ).

I look forward to building a rocket that uses radial boosters to apply more of the theory and the tips you've provided to make the rocket more stable.

Thank you all again for the tremendous help!

Edited June 24, 2020 by hellix08

[king of nowhere]

to get into mun intercept from low kerbal orbit you need some 850 m/s of deltaV. then, if you also want to orbit around mun instead of just doing a flyby, you need some more 150 m/s to enter orbit. from there you can come back to kerbin with only a small acceleration.

that's assuming you do all the manuevers perfectly, at the most convenient times, with the most optimized trajectories. i would want at least 1200 m/s to try that, to stay safe.

i assume your orbital stage uses a terrier, so the deltaV in vacuum should be about quadruple of that shown there. if i'm eyeballing this correctly, you should have around 1500 m/s left, which is enough for your purposes

[jimmymcgoochie]

There’s a nifty little mod called Where Can I Go that shows you where you can go with your current design in the VAB and also in LKO. It can also calculate the distances assuming the whole rocket gets to orbit, which is useful for getting your payloads designed right. I find it’s a useful tool to get a ballpark figure when designing my rockets, although it doesn’t work as well with spaceplanes because their ascent profile is totally different.

[Streetwind]

8 hours ago, hellix08 said:

I really wanted to use SAS but I don't know how else to do a smooth ascent trajectory other that using ALT+W and leaving SAS off.

Great to hear that you're making progress In regards to this particular issue:

What you are unwittingly describing is called a "gravity turn". This is a maneuver where your rocket tips over sideways on its own, pulled by gravity, but the rate at which this happens is perfectly in sync with your acceleration so that by the time you have levelled off horizontally, you are already out of the atmosphere and close to completing orbit insertion. This is how IRL rockets fly. In contrast to IRL rockets, however, the player in KSP doesn't have a NASA supercomputer aiding in the construction of their rocket. So it's actually really hard to get the exact amount of thrust you need, combined with the exact moment after leaving the pad during which you need to initiate the pitchover, the exact magnitude of that pitchover, and the precise throttle control all the way to orbit. Still, people sometimes go through that effort. Rockets have been built that sit on launch clamps on the pad at a very slight angle, so all you need to do is throttle up, stage, and watch the thing fly itself. Even without SAS.

That is really high-skill construction with a lot of time and trial&error involved, though. Most of the time, people don't bother, and just use SAS instead.

Pilot Kerbals gain advanced SAS capabilities as they level up. Just going to orbit once and back will get them to level 1, which teaches them the modes "SAS Hold Prograde" and "SAS Hold Retrograde". These are invaluable tools, more than any of the other SAS modes. It will make your rocket automatically stick to the prograde marker on ascent, and it lets you re-enter unwieldly and unstable spacecraft because the pilot is so much better at sticking exactly to retrograde than you ever could be.

Probe cores, however, do not learn new capabilities. They can only do what is built-in from the start. And lower-tech probe cores have only limited SAS functionality. The Stayputnik has none at all, the OKTO has only SAS on/off. The HECS is the first probe core with SAS Hold Prograde/Retrograde. If you want to launch an uncrewed rocket that is easy to fly, this is the first core that lets you do so.

[Spricigo]

On 6/23/2020 at 6:02 PM, hellix08 said:

I am clueless and desperate.

Actually the way you posted the question made it easier to help you. For he keen aye of the experienced helpers there a glance to that picture was enough to identify several problem but there was also a craft file and a explanation of what you tried to fix the problem yourself. Would be nice if you also had included some explanation about how far of the tech tree you unlocked and if any low level facility still limit you but we can figure it out that well enough.

well, let's add my thoughts to the good advice is already there:

Getting my hands in the craft Immediately did 3 adjustments for a first test flight:

1. replaced the Delta wings with 4 AV-T1 Wiglets  in the central stack
2. removed all gimbals
3. tilted the entire rocket 10 degrees towards east

 I put it in the launch pad, set throttle to max and hit space. Waiting to see it flip or at least burn in the atmosphere, to my disappointment it made it to orbit with 1600m/s left of deltaV. What brings me back to something people told you to do:

On 6/23/2020 at 6:02 PM, hellix08 said:

Making sure the engine has gimbals capabilities (is a Swivel and not a Reliant)

The Reliant is superior to the Swivel as a launch engine in every aspect. It have higher Thrust, weight less, is more fuel efficient and  it never steer the craft out of course because it have no gimbals. 

A well designed Launch Vehicle needs no gimbals or movable fins to be "controllable", it only need to be pointed in the right direction and will fly itself into orbit. Either you do it with a single nudge shortly after the take off or, ideally, set it up in the editor. Adding a probecore capable of holding prograde and a bit of control(fins, gimbal, reaction wheels) to keep in line a rocket that is about to lose it is OK but if you need to steer it yourself you designed it wrong.

Rant over, back to your craft.

8 hours ago, hellix08 said:

I think that might be enough to put into a Mun orbit (my objective) and maybe even come back and land if I wanted?

My suggestion is to not try to land it back (go ahead if you want but don't seem to be worth the trouble). If there is more than 600m/s left try to land it on the mun or let it in Mun orbit and wait for a "position Mun Satellite in a adjusted orbit" contract. A few tricks I use for satellite contracts: 

• rename the satellite with the remaining deltaV at the end ("Mun Satellite" becomes "Mun Satellite 735") to know If there is enough juice left when the contact is offered. 
• If you are launching the satellite to fulfill a  single contract include the agency on the name (e.g Zaltonic Mun Satellite.) to identify it better.
• Add an HG-5 antenna to make it a "Gap Relay"
• Use a decoupler or docking port to drop any' Contractual Obligation Part' you don't need after completion.
• Unlock Propulsion System ASAP, and design a "Contractual Obligation Satellite" with the essentials and powered by a single ant engine (C.O.S. will pack over 2km/s deltaV with easy since it weight almost nothing)

 

Also an advice for general craft design: design it by stages focusing on what the particular stage will do but start with the stage you will use last working your way to the launch vehicle. After the initial design do another pass refining it.  Keep a safety margin you feel comfortable but strive for a craft where each stage do only what is intended to do without leftovers.

 

[hellix08]

39 minutes ago, Spricigo said:

Getting my hands in the craft Immediately did 3 adjustments for a first test flight:

1. replaced the Delta wings with 4 AV-T1 Wiglets  in the central stack
2. removed all gimbals
3. tilted the entire rocket 10 degrees towards east

 I put it in the launch pad, set throttle to max and hit space. Waiting to see it flip or at least burn in the atmosphere, to my disappointment it made it to orbit with 1600m/s left of deltaV

It did that with no SAS ? No roll at all, perfectly stable?

This way of launching rockets with no gimbal or control surfaces seems quite interesting but I'll also have to look into using SAS following prograde to do a gravity turn. They both seem very interesting.

[Spricigo]

31 minutes ago, hellix08 said:

It did that with no SAS ? No roll at all, perfectly stable?

I think you need to see for yourself https://kerbalx.com/Spricigo/hellix08s-Moon-Satellite

 

[hellix08]

@Spricigo That was very very impressive indeed

[king of nowhere]

On 6/25/2020 at 10:38 AM, Spricigo said:

My suggestion is to not try to land it back (go ahead if you want but don't seem to be worth the trouble). If there is more than 600m/s left try to land it on the mun or let it in Mun orbit and wait for a "position Mun Satellite in a adjusted orbit" contract. A few tricks I use for satellite contracts: 

• rename the satellite with the remaining deltaV at the end ("Mun Satellite" becomes "Mun Satellite 735") to know If there is enough juice left when the contact is offered. 
• If you are launching the satellite to fulfill a  single contract include the agency on the name (e.g Zaltonic Mun Satellite.) to identify it better.
• Add an HG-5 antenna to make it a "Gap Relay"
• Use a decoupler or docking port to drop any' Contractual Obligation Part' you don't need after completion.
• Unlock Propulsion System ASAP, and design a "Contractual Obligation Satellite" with the essentials and powered by a single ant engine (C.O.S. will pack over 2km/s deltaV with easy since it weight almost nothing)

 

i add an alternative path there:

- outfit your contractual obligation satellite as a deep space probe

- also add the best relay antenna you have

- pack a lot of deltaV

- send your satellite off to your target orbit

- as soon as the contract is fulfilled, manuever again and go on another planet to perform all the science.

- once you've carried out all the science you can, your probe is just another useful relay.

 

i advise only doing that late in the tech tree, though: the first probes i sent this way i had big plans, only to realize that they became obsolete very quickly as i unlocked new instruments

[Spricigo]

5 hours ago, king of nowhere said:

i add an alternative path there:

- outfit your contractual obligation satellite as a deep space probe

The thing to consider: if you let the satellite there, the eventual Adjust Orbit contract will pay for the deep space probe, which will be more efficient since it goes directly to where is intended to go. 

For a equatorial Orbit the loss of efficiency may be small enough to be offset by the convenience of launching all in one go. Personally I don't mind an extra launch (it's mostly automated anyway)