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

[softweir]

7 hours ago, MorvarghMatt said:

Hey Softweir!

Thanks so much, totally makes sense now! :-) I had it in my head that I used to put 90° to launch to the east but obviously not. The penalty for coming back to KSP after a while of not playing!

Regards

Matt

You're welcome!

[Jim DiGriz]

On 12/3/2017 at 1:28 PM, softweir said:

It's a confusing convention!

The assumption is that most rockets will launch due east so as to make the most of the extra velocity gained from Kerbin's rotation.

Well not quite exactly.  That box is for inclination of the target orbit, not heading. 

Heading also is not constant during a launch but follows a great circle route (the sine waves around the globe that any orbit other than 0 degree inclination follows).  The only launch that shouldn't change heading at all is a due east burn from the equator on kerbin to wind up in a 0 degree orbit (which arguably is very common in KSP, but it is an unusual exception in reality).

So its not really an assumption.  Making that box be the launch heading wouldn't solve any particularly useful launch problems (you can do it, but it would always be inefficient to hit whatever orbit you wound up in).  It really has to be the inclination of the target orbit to be useful.

[softweir]

39 minutes ago, Jim DiGriz said:

... That box is for inclination of the target orbit, not heading. ...

Thanks for the reminder!

[Nicias]

I was noticed last night that mechjeb wasn't calculating deltaV for a droptank, connected with a small hardpoint. It was assigning all of the fuel to the first stage. Has anyone else had a similar problems? When I get a chance I'll post a minimal craft file with screenshot.

[Thrandir]

3 hours ago, Nicias said:

I was noticed last night that mechjeb wasn't calculating deltaV for a droptank, connected with a small hardpoint. It was assigning all of the fuel to the first stage. Has anyone else had a similar problems? When I get a chance I'll post a minimal craft file with screenshot.

on #763 release, it is corrected i believe.

 

[Nicias]

Still not working.

I seem to be on #768. Screenshot attached:

https://imgur.com/a/Jljjp

 

 

[vossiewulf]

I just used the ascent guidance for the very first time, not sure why but that is one thing I've always done manual. Well I use prograde autopilot after about 400m/s and control the trajectory with throttle, so it's not pure manual.

Anyway, MJ is better than me at everything so I assumed he'd be better on ascent too, flying a perfect gravity turn whereas I'm always pitching around a bit to nail the speeds at certain pitch angles that I've worked out for myself through experimentation.

I was wrong, I'm sitting at 154k with my launch at 9,863Dv whereas MJ's was 9,453. If it was a minor difference I'd be surprised but not asking, but 410Dv difference after circularizing is a big difference. I used the default settings, only set the altitude.

I'm still on 1.30 BTW.

*Warning: don't try this without KJR

Spoiler

 

Edited December 8, 2017 by vossiewulf

[Gilph]

10 hours ago, vossiewulf said:

Anyway, MJ is better than me at everything so I assumed he'd be better on ascent too, flying a perfect gravity turn whereas I'm always pitching around a bit to nail the speeds at certain pitch angles that I've worked out for myself through experimentation.

That was not a good assumption...

As with everything else in MJ, it does things using a particular method, and you tailor that method to what you need. In the classic ascent profile, there is no mention of using a perfect gravity turn. It only refers to turn shapes. There is a gravity turn option, but the parameters need to be set correctly for your particular situation.

I have never been able to use ascent guidance completely unattended, but that's OK. All of the variables needed to make a good ascent on every single rocket type in every planet cannot be accounted for every time. You create settings for each case to give the best auto control, but you still may need to have manual intervention.

[vossiewulf]

1 hour ago, Gilph said:

That was not a good assumption...

As with everything else in MJ, it does things using a particular method, and you tailor that method to what you need. In the classic ascent profile, there is no mention of using a perfect gravity turn. It only refers to turn shapes. There is a gravity turn option, but the parameters need to be set correctly for your particular situation.

I have never been able to use ascent guidance completely unattended, but that's OK. All of the variables needed to make a good ascent on every single rocket type in every planet cannot be accounted for every time. You create settings for each case to give the best auto control, but you still may need to have manual intervention.

Ok, thanks, I guess with all the variables this is trickier than maneuver nodes for Jeb to come up with the perfect solution.

As for gravity turn, you need to tweak the UI then. Because the reason I assumed he was doing a perfect gravity turn is that in his status it says "gravity turn"  He does "vertical climb" first followed by gravity turn. The two things that looked odd to me compared to my launches are that first he climbed straight up until he was doing 180m/s before starting the pitchover, I start mine around 75m/s, and second he was using much higher TWR than I do that resulted in a very vertical trajectory that required a 950m/s circularizing burn. When I fly my trajectory correctly it will be <250m/s to circularize.

[LameLefty]

49 minutes ago, vossiewulf said:

Ok, thanks, I guess with all the variables this is trickier than maneuver nodes for Jeb to come up with the perfect solution.

As for gravity turn, you need to tweak the UI then. Because the reason I assumed he was doing a perfect gravity turn is that in his status it says "gravity turn"  He does "vertical climb" first followed by gravity turn. The two things that looked odd to me compared to my launches are that first he climbed straight up until he was doing 180m/s before starting the pitchover, I start mine around 75m/s, and second he was using much higher TWR than I do that resulted in a very vertical trajectory that required a 950m/s circularizing burn. When I fly my trajectory correctly it will be <250m/s to circularize.

Couple points: first, you can change customize the ascent trajectory to your liking with the GUI by setting velocity/altitude at which to start the turn, and the shape (turn rate) of the gravity turn. Second, if your vehicle has THAT much t/w ratio, you're basically wasting thrust fighting against atmospheric drag and you create a risk of part overheating. You can set maximum dynamic pressure in the GUI as well. MJ will throttle down to keep your vehicle below that maximum.

But on that note, if you find yourself throttling down too much during ascent, it's possible you're using non-optimal engines. Smaller, less powerful (and less massive) engines running at higher throttle settings would generally be more cost-effective if you're playing a Career mode game. 

[Nicias]

17 hours ago, Nicias said:

Still not working.

I seem to be on #768. Screenshot attached:

https://imgur.com/a/Jljjp

 

 

Double checked with a clean 1.3.1 install same problem.

[vossiewulf]

2 hours ago, LameLefty said:

Couple points: first, you can change customize the ascent trajectory to your liking with the GUI by setting velocity/altitude at which to start the turn, and the shape (turn rate) of the gravity turn. Second, if your vehicle has THAT much t/w ratio, you're basically wasting thrust fighting against atmospheric drag and you create a risk of part overheating. You can set maximum dynamic pressure in the GUI as well. MJ will throttle down to keep your vehicle below that maximum.

But on that note, if you find yourself throttling down too much during ascent, it's possible you're using non-optimal engines. Smaller, less powerful (and less massive) engines running at higher throttle settings would generally be more cost-effective if you're playing a Career mode game. 

Thanks, I saw that, the question was less can I make it perform better and more just surprise that I'd been able to do something considerably more efficiently than MJ, and assuming I must have done something wrong. I like flying manual launches, and will probably continue to do so.

I look for 1.3 initial TWR that builds to 2 at around 450m/s and from that point I maintain close to 2 until cutoff to coast to apoapsis. I can't fly with throttle at the beginning with the velocity so low, so I use pitch to fly until I'm happy he's captured the intended trajectory, usually between 300m/s and 450m/s. At that point I kick to prograde autopilot and control with throttle, the logic being that controlling a trajectory with pitch is costly, any time your nose is not pointed pure prograde not all of the energy is going to acceleration. And I'm also a pilot and this is what we do when we land, for other reasons but the concept is the same- controlling the flight path with throttle rather than the flight controls. I've found my launches much more efficient this way, and hell I beat MJ by 410Dv

 

Edited December 8, 2017 by vossiewulf

[Gorby1]

1 hour ago, vossiewulf said:

Thanks, I saw that, the question was less can I make it perform better and more just surprise that I'd been able to do something considerably more efficiently than MJ, and assuming I must have done something wrong. I like flying manual launches, and will probably continue to do so.

I look for 1.3 initial TWR that builds to 2 at around 450m/s and from that point I maintain close to 2 until cutoff to coast to apoapsis. I can't fly with throttle at the beginning with the velocity so low, so I use pitch to fly until I'm happy he's captured the intended trajectory, usually between 300m/s and 450m/s. At that point I kick to prograde autopilot and control with throttle, the logic being that controlling a trajectory with pitch is costly, any time your nose is not pointed pure prograde not all of the energy is going to acceleration. And I'm also a pilot and this is what we do when we land, for other reasons but the concept is the same- controlling the flight path with throttle rather than the flight controls. I've found my launches much more efficient this way, and hell I beat MJ by 410Dv

 

I've found that when using MJ for ascent guidance, you can "gain" a lot more Dv during launch by increasing the "Limit AoA" setting to 8 or 9 degrees, sometimes as high as 15 (basically as high as I can set it without the rocket flipping over during ascent). I often leave everything default except for that, but as others have said those default settings shouldn't necessarily be considered recommended settings.

[vossiewulf]

4 hours ago, Gorby1 said:

I've found that when using MJ for ascent guidance, you can "gain" a lot more Dv during launch by increasing the "Limit AoA" setting to 8 or 9 degrees, sometimes as high as 15 (basically as high as I can set it without the rocket flipping over during ascent). I often leave everything default except for that, but as others have said those default settings shouldn't necessarily be considered recommended settings.

Thanks, but why? Should never need that much alpha on a rocket destined for space. All-aspect dogfighting missile tracking a moving target at short range yes, big old rocket aimed for a specific gravity-turn trajectory shouldn't need anywhere near that much.

[Starwaster]

5 hours ago, Gorby1 said:

I've found that when using MJ for ascent guidance, you can "gain" a lot more Dv during launch by increasing the "Limit AoA" setting to 8 or 9 degrees, sometimes as high as 15 (basically as high as I can set it without the rocket flipping over during ascent). I often leave everything default except for that, but as others have said those default settings shouldn't necessarily be considered recommended settings.

I too like to flirt with high AoA, but changing it mid-ascent can definitely increase the risk of flipping... best to set it prior to launch. That generally ensures that AoA will change gradually and MJ can handle the adjustments with greater ease. Other caveats to be aware of are increased drag and increased surface heating as the craft goes supersonic.

[custume]

Hello to all,

I have the same problem with wobble ship like some in this post.

MJ is just making the ship wobble, is trying to over compensate and start wobbling, I have to manual correct it to stop, after some sec´s its start again.

I try with only one engine but is the same, take longer to start the wobble but as it starts is all over.

I have 1 stage perfect balance ship, it can go with no intervention to space, but if I use MJ it will wobble all the way to orbit, is like its overcorrecting it self all the time.

If I turn off the force roll it will stops rolling to the sides, but keeps correcting up/down and after some sec´s it start the swinging hard from side to side.

I don't even want to try a VTOL

 

Best regards

[custume]

1 hour ago, custume said:

Hello to all,

I have the same problem with wobble ship like some in this post.

MJ is just making the ship wobble, is trying to over compensate and start wobbling, I have to manual correct it to stop, after some sec´s its start again.

I try with only one engine but is the same, take longer to start the wobble but as it starts is all over.

I have 1 stage perfect balance ship, it can go with no intervention to space, but if I use MJ it will wobble all the way to orbit, is like its overcorrecting it self all the time.

If I turn off the force roll it will stops rolling to the sides, but keeps correcting up/down and after some sec´s it start the swinging hard from side to side.

I don't even want to try a VTOL

 

Best regards

Here is more information and a video for you to see how bad it is, is better that words.

I made this new ship/rover just to show you the problem, with larger ships the problem is expediency greater.

The video is 3 launch with and without mechjeb 

the extra info:

 GameData\ModuleManager.2.8.0.dll 
 GameData\000_AT_Utils\Plugins\000_AT_Utils.dll 
 GameData\000_USITools\MiniAVC.dll  
 GameData\CommunityCategoryKit\CCK.dll 
 GameData\Firespitter\Plugins\Firespitter.dll 
 GameData\GroundConstruction\Plugins\GroundConstruction.dll  
 GameData\KAS\Plugins\KAS.dll   
 GameData\KerbalKonstructs\KerbalKonstructs.dll 
 GameData\KIS\Plugins\KIS.dll  
 GameData\MechJeb2\Plugins\MechJeb2.dll 
 GameData\REPOSoftTech\DeepFreeze\MiniAVC.dll 
 GameData\SCANsat\Plugins\SCANsat.dll  
 GameData\TriggerTech\KerbalAlarmClock\KerbalAlarmClock.dll 
 GameData\UmbraSpaceIndustries\Konstruction\Konstruction.dll 

 

The Video : (is uploading, will be ready in 10 min)
 

 

Best regards

Edited December 9, 2017 by custume

[vossiewulf]

1 hour ago, custume said:

Hello to all,

I have the same problem with wobble ship like some in this post.

MJ is just making the ship wobble, is trying to over compensate and start wobbling, I have to manual correct it to stop, after some sec´s its start again.

Is this with RCS on (your video isn't playing for me)? Only time I've seen that is with RCS, MJ can easily get into little oscillations that prevent him from ever auto-warping for example. I solved it by not using RCS unless I will be docking and then only on the payload vehicle to be docked, not the launcher. I'm not sure it's a MJ problem, unless your RCS is very very carefully balanced, unlike reaction wheels it's applying small accelerations and it's easy to have the autopilot go into a mini-PIO. Sorry Pilot-Induced Oscillation if you're not familiar with term.

[custume]

4 minutes ago, vossiewulf said:

Is this with RCS on (your video isn't playing for me)? Only time I've seen that is with RCS, MJ can easily get into little oscillations that prevent him from ever auto-warping for example. I solved it by not using RCS unless I will be docking and then only on the payload vehicle to be docked, not the launcher. I'm not sure it's a MJ problem, unless your RCS is very very carefully balanced, unlike reaction wheels it's applying small accelerations and it's easy to have the autopilot go into a mini-PIO. Sorry Pilot-Induced Oscillation if you're not familiar with term.

Video still uploading, slow internet, give it a couple of min.

Best regards

PS, I will let you know when is done

 

EDIT: oh, and no, its not a RCS problem, I don't even use them, but you will see

Edited December 9, 2017 by custume

[custume]

Video is done and here is a pic of the test rover.

The video is of good quality , you can see in youtube

http://www.image-share.com/ijpg-3651-45.html

Best regards

Edited December 9, 2017 by custume

[sarbian]

I lack the words to express what I am currently feeling. In English AND in French.

[Nightside]

5 hours ago, custume said:

I made this new ship/rover just to show you the problem, with larger ships the problem is expediency greater.

Hi @custume, nice video. I think you may have found a limitation of MJ, not a bug. Your vessel is aerodynamically unstable, and asymmetrical. This will cause it to try to flip to a more stable orientation (heaviest end prograde, with the draggy end behind) as you go fast through the atmosphere. 

MJ works well for more “traditional looking “ crafts that are aerodynamically stable. If you are doing VTOLs and other more experimental designs you might be better served by the TCA autopilot mod, as it will balance the thrust of multiple engines. 

 

[custume]

5 hours ago, sarbian said:

I lack the words to express what I am currently feeling. In English AND in French.

Hello sarbian,

I´m sorry, was something I say

 

34 minutes ago, Nightside said:

Hi @custume, nice video. I think you may have found a limitation of MJ, not a bug. Your vessel is aerodynamically unstable, and asymmetrical. This will cause it to try to flip to a more stable orientation (heaviest end prograde, with the draggy end behind) as you go fast through the atmosphere. 

MJ works well for more “traditional looking “ crafts that are aerodynamically stable. If you are doing VTOLs and other more experimental designs you might be better served by the TCA autopilot mod, as it will balance the thrust of multiple engines. 

 

Hello Nightside,

Yes is true, this ship is a strange designed craft but is stable, I don't need to correct it on takeoff, just the normal corrections, but with MJ is just crazy.

 

Also I made this ship/rover to show the effect, is more pronounce in this ship, I have the same problem with "traditional" ships, even ships with only one rocket.

I can make a video to show how it handles a normal ship, but is basically the same, it start wobbling and some time even turn the ship up side down.

On heavy ships tends to do the same but takes more time, landing with this problem is also terrible, even in no atmosphere.

I don't think is MJ it self, because I don't see this if I use only MJ and remove all the mods, but have to be one of the mods that is making MJ more sensitive.

For more I have another video on my channel , an old one with the version 1.2 ( I think ) and I use a similar rover style but larger and more heavy, also similar mods with MJ and it works great, MJ is capable of handling the craft with no problem.

I will add the video later on of another test with a "normal" craft

 

Thank you and best regards to all

PS, the old video is not HD, but you can see + or - well
 

 

[Nightside]

@custume, that is strange, can find the version of MJ you used to use that worked? Maybe you can revert to that version.

Anyway, like many other mods, this one has specific requirements for support requests listed in the OP.

[maja]

@custume Really? You are trying to fly a brick on Kerbin and blame the MJ when it can't handle it? Make an aerodynamic rocket with rover hidden inside a fairing and COL under COM. MJ will fly it without problems.