Jump to content

[1.12.x] Anatid Robotics / MuMech - MechJeb - Autopilot - [2.14.3] [4th March 2023]


sarbian

Recommended Posts

So, I'm frequently encountering an issue with the Primer Vector Guidance mode of the ascent guidance where, at some point during a coast phase, it'll put up the "Optimizer iteration convergence failed" message and suddenly start immediately burning on a different trajectory than it had originally calculated - sometimes only slightly different from its original track (though with the entire coast phase vanished), sometimes at a significant angle (up to 90 degrees) from what it was originally planning, and always on a longer-lasting burn that generally is beyond the rocket's capabilities. Is this a known issue (hopefully with a known workaround)?

Edited by Starseeker
Link to comment
Share on other sites

On another note, here's the product of a few hours worth of testing & data-gathering with the PVG guidance mode: a graph of the minimum "Pitch Rate" setting needed for a 3450m/s-dV launcher on stock Kerbin, as a function of TWR, for various settings of "Pitch Start". Pitch rate settings at or above the ones on this graph should work for a given TWR and pitch-start speed.

vSc8hFN.png


Interestingly, it seems that there's a "sweet spot" around TWR=1.5, where a higher or lower TWR means a higher pitch rate is required. I couldn't get the design I was using for testing to get to orbit at all with a TWR of 1.2, and it wasn't capable of a TWR above 1.7 without altering the design. Here's a screenshot of the craft I was using (don't have a picture of it in the VAB at this moment), using a Boar engine from ReStock+ on the lower stage and a Pavonis hydrolox engine from CryoEngines on the upper:
 

Spoiler

OfNqwX4.png


Hopefully other people find this helpful!


EDIT: Turns out this isn't accurate.

Edited by Starseeker
Link to comment
Share on other sites

KSP:  1.11.2.3077 Windows 10 x64

Problem: Maneuver node execution stuttering when center of thrust locate over centre of mass

Mods installed:

  • MechJeb 2 v2.12.0.0
  • Module Manager v4.1.4.0
  • Trajectories v2.4.0.0
  • Tweak Scale v2.4.5.0
  • KSP Recall v0.2.0.0
  • Kerbal Engineer Redux 1.1.8.3
  • Interstellar Fuel Switch(+Core) v3.29.2.0
  • Community Tech Tree v3.4.3
  • Hide Empty Tech Tree Nodes v1.3.0
  • Community Resource Pack v1.4.2
  • Community Delta-V Maps v2.8

Reproduction steps: Short video

Log: Player.log

Link to comment
Share on other sites

So, I'm somewhat at my wit's end here. I'm trying to use the ascent guidance, but I can't seem to find a consistent pattern in the optimal settings for getting to orbit with 3400-3500m/s dV, with both classic guidance and PVG. I've been looking at launch TWR, upper stage TWR, pitch start & pitch rate for PVG, and turn shape & final angle for classic guidance, and what works for one rocket design doesn't work for another with all the craft parameters I looked at (launch TWR and upper-stage TWR) being the same. I've even had cases where adding boosters, then tweaking the thrust limiters so that launch TWR is the same (and upper-stage TWR stays reasonable), results in a previously-working launcher failing to make orbit, despite the extra dV from the boosters.

Does anyone know what I might be doing wrong? All I can think of is that either there's some variable(s) I'm not looking at properly, or I'm just going to have to trial-and-error every variant of every launch vehicle design I ever make, and I really hope it's not the latter. :(

Link to comment
Share on other sites

8 hours ago, Starseeker said:

So, I'm somewhat at my wit's end here. I'm trying to use the ascent guidance, but I can't seem to find a consistent pattern in the optimal settings for getting to orbit with 3400-3500m/s dV,

I use launch twr about 1.8, Q limit at 40k, pitch start at about 50 ms, pitch 42, target orbit 75k

Works most of the times.

 

Link to comment
Share on other sites

9 hours ago, Starseeker said:

Does anyone know what I might be doing wrong? All I can think of is that either there's some variable(s) I'm not looking at properly, or I'm just going to have to trial-and-error every variant of every launch vehicle design I ever make, and I really hope it's not the latter.

You are trying to get consistancy using an algorithym in a semi chaotic system. You would be better off using your own experience and playing the ascent by feel and intuition. MechJeb is a great guide but never expect it to perfectly do everything all the time. Oh and a slight shift in COM the wrong way can sabotage you 9 times out of 10.

Link to comment
Share on other sites

2 hours ago, Curveball Anders said:

I use launch twr about 1.8, Q limit at 40k, pitch start at about 50 ms, pitch 42, target orbit 75k

Works most of the times.

It's been a long time since I last used MechJeb's Ascent Guidance, but I found with each rocket design you needed to play with the parameters to see how well things performed.  Eventually you get a good set that works even with slightly different rocket designs.

Link to comment
Share on other sites

4 hours ago, ColdJ said:

You are trying to get consistancy using an algorithym in a semi chaotic system. You would be better off using your own experience and playing the ascent by feel and intuition. MechJeb is a great guide but never expect it to perfectly do everything all the time. Oh and a slight shift in COM the wrong way can sabotage you 9 times out of 10.

Unfortunately, I don't really have all that much experience with manual flying to orbit- especially when it comes to the low-TWR upper stages that using CryoEngines tends to result in. The "pitch over a few degrees and hit hold prograde" thing also seems to take a *much* steeper ascent trajectory than MJ guidance does, at least a lot of the time; not sure if I'm just misjudging the initial pitch-over amount though.

Link to comment
Share on other sites

10 minutes ago, Starseeker said:

not sure if I'm just misjudging the initial pitch-over amount though.

Again, if when you try it tends to topple over then it is a COM balance problem. The prograde marker is not a fixed point. It changes as you change direction or pitch. You are best to use the degrees of angle on your nav ball and your speed and altitude to determine when and how much to tilt. If you find your rocket hard to get to do what you want you may need more vernier thrusters or, while in atmosphere, more wing and elevon control surfaces. If you attach your control surfaces to decoulplers you can always jettison them once you leave atmosphere and don't need them anymore. All of this gets easier with practice so, try your rocket designs over and over in Sandbox till you get the hang, before using the design in a career game.

Link to comment
Share on other sites

Just now, ColdJ said:

Again, if when you try it tends to topple over then it is a COM balance problem. The prograde marker is not a fixed point. It changes as you change direction or pitch. You are best to use the degrees of angle on your nav ball and your speed and altitude to determine when and how much to tilt. If you find your rocket hard to get to do what you want you may need more vernier thrusters or, while in atmosphere, more wing and elevon control surfaces. If you attach your control surfaces to decoulplers you can always jettison them once you leave atmosphere and don't need them anymore. All of this gets easier with practice so, try your rocket designs over and over in Sandbox till you get the hang, before using the design in a career game.

Oh, it's got plenty of control authority, and isn't toppling over or anything. Just, for that initial pitch-over before hitting "hold prograde", I'm not sure if should I be going for like, the first mark beyond the vertical point on the navball, or the second, or what. I've heard "5 degrees" said before, but that seems to result in a pretty steep ascent.

Link to comment
Share on other sites

2 minutes ago, Starseeker said:

Oh, it's got plenty of control authority, and isn't toppling over or anything. Just, for that initial pitch-over before hitting "hold prograde", I'm not sure if should I be going for like, the first mark beyond the vertical point on the navball, or the second, or what. I've heard "5 degrees" said before, but that seems to result in a pretty steep ascent.

You don't hold prograde, prograde follows you. You need to ignore prograde and pitch to what feels right. Remember to watch apoapsis and periapsis down in the left corner once you switch it out of docking mode to understand where your angle is taking you.

Link to comment
Share on other sites

Well, manual flying to orbit takes more time - and more importantly, attention - than I'm really wanting to always devote, particularly when it comes to testing launcher designs.

So, perhaps a better-stated version of my original question: what parameters of my rocket should I be looking at when determining ascent guidance settings? And is expecting orbit at 3400-3450m/s always reasonable, or should I be including a buffer of 100-200m/s?

Link to comment
Share on other sites

Start Burn In: fails.

I have a vessel in Tylo SoI, going to the circularization node.
tylo%20circularization%20node.png?dl=1
I execute the node but MJ starts burn 20s late.
Results in an obvious eccentric orbit.

Modify the percentage buttons,
or whatever it is called, modify the start time,
but
it does not matter, form 0% to 100%, the result orbit is always the same,
presuming the vessel direction changes too,
in order to compensate.

mj.jpg?dl=1

So why MJ starts burn so late, forcing me to start the burn manually?
Can I do something so that MJ starts the burn at the exact time?

ok nevermind: I'll manually burn.

Edited by antipro
Link to comment
Share on other sites

On 5/19/2021 at 9:46 PM, Starseeker said:

Well, manual flying to orbit takes more time - and more importantly, attention - than I'm really wanting to always devote, particularly when it comes to testing launcher designs.

So, perhaps a better-stated version of my original question: what parameters of my rocket should I be looking at when determining ascent guidance settings? And is expecting orbit at 3400-3450m/s always reasonable, or should I be including a buffer of 100-200m/s?

I dont know about manually flying takes moar *time*... it may take moar *attention*
MJ Ascent Guidance is not all that great for using on newly designed rockets, unless you already really know what you are doing.
It can not be expected to perform perfect, or even successful launches of a badly designed rocket.

This is how MJ becomes a crutch and/or causes new players to just get frustrated and uninstall the mod, or even quit playing altogether.

Theres no escaping learning basic manual launching and properly designing a rocket. That being said, the rocket doesnt have to be perfect, but if you can get it to orbit manually, if even sloppily, *then* you can try using MJ... It may take a few attempts with MJ, while you tweak the MJ settings to get the perfect, optimized launch you want.

It sounds like you are expecting too much of MJ on a new rocket design, and that you are overthinking it, by worrying about exact parameters. And possibly not even the correct ones, or most important ones, at that. vOv

Getting a rocket into orbit in KSP, is...actually, *NOT* rocket science. :P
if it were, I dont really think this game would be as popular as it is...And the game Orbiter, *would be* :P


 

Edited by Stone Blue
Link to comment
Share on other sites

Uhh.. you delete whatever you did with that folder, then you grab the MechJeb & Engineer For All config that is floating around (stick it anywhere in your GameData folder), or if there is a patch for all craft included in MJ itself, you just have to enable it.

Edited by Stone Blue
Link to comment
Share on other sites

7 hours ago, Stone Blue said:

Uhh.. you delete whatever you did with that folder, then you grab the MechJeb & Engineer For All config that is floating around (stick it anywhere in your GameData folder), or if there is a patch for all craft included in MJ itself, you just have to enable it.

It's meant to be a method to make MechJeb unlock by making an empty folder called MechJebUnlocked. It was added a few patches ago. I was just wondering how to make it work for all crafts. 

Link to comment
Share on other sites

On 5/21/2021 at 8:27 AM, Stone Blue said:

I dont know about manually flying takes moar *time*... it may take moar *attention*
MJ Ascent Guidance is not all that great for using on newly designed rockets, unless you already really know what you are doing.
It can not be expected to perform perfect, or even successful launches of a badly designed rocket.

This is how MJ becomes a crutch and/or causes new players to just get frustrated and uninstall the mod, or even quit playing altogether.

Theres no escaping learning basic manual launching and properly designing a rocket. That being said, the rocket doesnt have to be perfect, but if you can get it to orbit manually, if even sloppily, *then* you can try using MJ... It may take a few attempts with MJ, while you tweak the MJ settings to get the perfect, optimized launch you want.

It sounds like you are expecting too much of MJ on a new rocket design, and that you are overthinking it, by worrying about exact parameters. And possibly not even the correct ones, or most important ones, at that. vOv

Getting a rocket into orbit in KSP, is...actually, *NOT* rocket science. :P
if it were, I dont really think this game would be as popular as it is...And the game Orbiter, *would be* :P


 

Ah.

I suppose you're right; just, not looking forward to having to spend several minutes carefully nursing a craft to orbit when before I could hit a few buttons and then just make sure nothing goes wrong in the next minute or so (thanks to timewarp). :(

Edited by Starseeker
Link to comment
Share on other sites

Umm,,, well, there is Kerbal life before experiencing MechJeb and life after experiencing MechJeb.  Just wished I didn't wait for 1900 hours play time before experiencing MechJeb for first time. Thanks for this mod!  MechJeb breathing new life into my game, keeping me motivated to pass 2000 hours.

I  resisted so long because "Didn't want to automate" the game  to be easy. I still stand by this logic for new people, they should play the game vanilla and learn how to do this manually first. But once you got it down, man oh man start using mech jeb! 

Its not like anyone in human history actually flies a rocket with a joystick like KSP let's you do in game.  There is usually hundreds of people involved in calculating  burn times,  trajectories, planetary transfers, etc.  So in this angle of how humans actually plan and execute spaceflight , using mechjeb is actually more true to human spaceflight then vanilla KSP. 

 

Tip  for people experiencing stuttering and sway issues under autopilot :

Found that really long rockets in zero gravity  and/or poor center of mass rockets will  sway due to overcorrected steering by the autopilot routine.  This autopilot overcorrections causing sway  is very noticeable in zero gravity, less so under gravity. Some of my long giant rockets had so much back and forth sway occurring, thought the rocket was going to fall apart. 

Adding RCS to the nose will usually take care of this autopilot causing sway however, I found a more lightweight(maybe logical?) solution is to simply cut the engine gimbal limit to 50 or less. And definitely only have the center engine gimballed if can arrange.

 

 

Link to comment
Share on other sites

One of the best things about MechJeb, that I rarely hear mentioned, are the customiseable read outs it lets you create. Height and directional speed accurate down to micrometres are some of my favourites.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...