MechJeb 2 - Patch test bed release (October 10)

[Climberfx]

Auto rendezvous has been in every version since at least 2.08. If it's missing for you you have a corrupt cfg file in gamedata/mechjeb2/plugins/plugindata

Not sure which one but I think it's the mechjeb_settings_global_.cfg. Delete the file and when you run KSP it will be recreated and all of your mechjeb settings should be restored in game.

Thanks Sojourner. I gonna try and let you know!

[sojourner]

Is there a way to get back the ability to have your vessel keep a pro grade orientation rather than swapping ends for a retro grade one while on the coast up to AP during ascent? Or have the ability to toggle what you would like to happen after engine shutdown?

Mechjeb2 orients the spacecraft for the next maneuver as soon as the last maneuver is completed. The reason for this is to make use of time available before activating time warp. It's a smart piece of programming considering some craft are large and turn slowly. If it waited for the last minute to orient there could be no time to correct for errors.

[Jasmir]

Bug found

Version: MechJeb 2.0.9 and your dll from today.

I have a huge problem with the ascent guidance in this version, belonging to the AscentPath:

Tested on the launchpad.

I can't set the "Turn start altitude" to 0. If i try it, the real turn start still at ~7km even if the value in the "Turn start altitude" shows me the 0 i've entered. The plotted graph under the value-fields also shows me the turn start later than 0km. If i enter a higher value then 7 in this field, the graph starts recognize my wishes.

[MilkDust]

You probably have a high turn shape %

[Cyda]

Thank you, we all appreciate this!

[drtedastro]

Hello,

I am having trouble with the "Landing Guidance" module. I created a base, NO-MODS version of 0.21.1, then followed the installation instruction on this thread to load your updates. (full install of 2.0.9, then replaced the .dll and replace the part.cfg). I then created a stock craft to put a lander into orbit and then using the Landing Guidance mod, targeted KSP and told it to land at target. All goes fine until just before landing when it decides that it just want to hover. I have repeated this three times now, with same effect each time. If i abort the landing just prior to 500m elevation and select land anywhere, this does not happen...

I am sorry if this has been discussed before, but i missed it while going through the posts.

any explanation would be great and i can still land, i just have to remember to abort, and change to land anywhere....

Thanks.

[drtedastro]

I have been working with the docking autopilot and it is so much better than it used to be... I dont know what to do with all of the rcs fuel i have left........ a couple of requests, if possible.

1. would it be possible to have a data variable to let us set the distance that we want to be from the target when we have final alignment, prior to making final approach? This would let the user specify how much space / distance to allow for maneuvers to be completed prior to final vector push for docking. if there is nothing in the way of the docking ports, then this could be a very small number, like 10 meters, however, if the approach to the docking collar is very congested, then we might want to make sure that we are all aligned when we are 300 meters out.

2. The variable final docking speed, seems to only be valid for the final push to docking... again, it would be good if there was a way that we could tell the autopilot to 'go quicker' if we knew that there was no objects to get in the way or if there was an emergency such as running out of elec, and we need to get there NOW....

\

again, thanks for all of the work you have done and are doing. you are making this mod a lot better.....

[Thorbane]

I'm having a horrible time with translatron and the landing autopilot. The translatron when set to kill H/s just spins in circles over-correcting and eventually piles me into the ground sideways. When I was attempting to land on Pol it was actually firing downwards at a -20 pitch due to overcorrections.

[Tex_NL]

Since the patch I have some nasty issues.

Whenever I use Smart A.S.S. to hold my craft and then try to execute a maneuver with the maneuver planner MJ starts fighting itself. Before the patch the Smart A.S.S. would shut down to let the planner execute the maneuver. Now the Smart A.S.S. stays active. The orientation from the Smart A.S.S. seems to take precedence over the Maneuver planner but the engines are switched on at the correct moment, resulting in a burn in completely the wrong direction.

[Thorbane]

using the Landing Guidance mod, targeted KSP and told it to land at target. All goes fine until just before landing when it decides that it just want to hover.

That's probably the new precision landing thing, It put me within a meter of the center of the launchpad at the cost of ~400 delta-v. Would be nice to be able to tell it to just land within x distance from target, if your target is say a kethane mining rig landing directly on it is, well, not good.

[sarbian]

New version !

- Version Number will change automatically for each build( 2.0.<year><dayofyear>.<hour><minute> ).

- I reverted Meumeu landing precision patch. Ship with "high" TWR where hoovering over the landing spot, and I need some time to understand his change. Meumeu if you read this ...

- Fix for reaction wheel. MJ thought the ship still has RW when there was no electricity left. It lead to bad attitude control ( Thanks Raf04 ).

- Fix SmartASAS not switching to AUTO while an other module use the attitude control.

- Removed SAS automatic switching.

[Starwaster]

I patched the .dll, but after that my ships went crazy. I must changed to the "original"!

Whatever problem you had we could have helped you with. just saying that you had a problem and switched back is useless to you and us.

[Ratzap]

Could you put a link in for the new dll please? The one under install is the file from the 27th of July.

[Starwaster]

Mechjeb2 orients the spacecraft for the next maneuver as soon as the last maneuver is completed. The reason for this is to make use of time available before activating time warp. It's a smart piece of programming considering some craft are large and turn slowly. If it waited for the last minute to orient there could be no time to correct for errors.

You're not understanding what he said. During ascent, there is a point where engines cut off and you coast to edge of atmo. During that time, NO maneuver is plotted and the ship is uncontrolled; no control inputs are provided by MJ and none are accepted by you. What happens during this phase depends on your craft design and ascent profile. What SHOULD happen is either SAS engages (can be done manually) or MJ actively Maintains orientation. Preferably prograde as this is much closer to any maneuver node plot that circularizes your orbit.

Consequences for current behavior again depend on your ship. A large ungainly craft that gets set slowly turning end over end might result in you not going to space today. I have a number of such craft that I have to launch several of for my upcoming Duna trip and if I dont watch MJ like a hawk as MECO approaches so I can disengage autopilot if needed and make sure my ship doesnt start tumbling, then... well, it tumbles and because it (or rather the payload) is a fully fueled nuclear drive section for Duna, it is hard to get it back under control before I pass AP and begin a descent back to ground.

[HeadHunter67]

Version Number will change automatically for each build( 2.0.<year><dayofyear>.<hour><minute> ).

"237" does not mean "23rd of July", it means the 237th day of the year... which would be... August 25th. That's today. Get it?

Sarbian, I noticed an issue on a fresh install of your .dll and thought I'd bring it up. The ascent profile started at -5km and ended at -70km. Sounds like a recipe for a crash landing to me.

Edited August 25, 2013 by HeadHunter67

[Hesh]

From the main MechJeb thread:

Suggestions for people suffering a couple of problems....

MJ can't handle a launch without spinning out of control sideways within meters in spite of the control part being correctly aligned: Try deleting all of the craft-specific config files in KSP\GameData\MechJeb2\Plugins\PluginData\MechJeb2 - they're called mechjeb_settings_type_*.cfg and there is one for each craft you've flown. Sometimes they seem to have crap data in them, and the defaults work a lot better.

This is happening to me frequently since installing the testbed patch/new AR202 cfg, but never before. Even with the root part oriented correctly and the navball pointing the right way, every rocket tries to tip over almost immediately after launch, and always in the direction which would place a radially mounted AR202 "right side up." I think it's only happening with new/modified designs but I haven't done enough launches of pre-patch designs to say for sure. As it says in the quote, if I close KSP, delete the .cfg file, and start it up again it seems to "fix it" but anyone with more than a few mods and more than a couple flights in the air at a time knows this is a tedious process.

[sarbian]

Had to use day of year since 121231 was over 16bit limit of version id

It must be something with BloodyRain2k auto ascent profile. I need to take another look at it.

[Jasmir]

Currently, i see 2 Issues, both in version 2.0.13237.1443:

Issue 1, Ascent Guidance-> Ascent Path Editor

Turn start altitude can't be set to 0km. It still starts at 7km on Kerbin, even if i put a 0 in this field.

Picture:

Here you can see, the shape didn't start at the bottom as it should at a "Turn start altitude 0km"

Leaving this will means that mechjeb is unable to start bigger vessels with the ascent guidance, or you have to set up a very inefficient ascent. This mechjeb manage to stall every bigger vessel on ascent because it try to make a abrupt direction change in 7km height with a decent speed. The normal 2.0.9 version don't do this, because you can use a 60% turn shape which starts at 0km. This results in a much smoother shape without abrupt direction changes. So this issue is 2 issues: It don't honor your "Turn start altitude" if you put it below 1/10 of the Atmosphere, and it makes abrupt direction changes. And throught the first issue, you can't workaround the 2nd one

Issue 2, Ascent Guidance

On the Ascent, the buring direction of the vessel seems to be strictly oriented on the current height, not at the current Apoapsis. This means(i used the same setting you can see the picture above), if i had a current height of 45km and a current apoapsis at 70km, the vessel is still aiming very upwards and not at 0°. This result in very inefficient ascents, because i get very less lateral speed.

[sojourner]

You're not understanding what he said. During ascent, there is a point where engines cut off and you coast to edge of atmo. During that time, NO maneuver is plotted and the ship is uncontrolled; no control inputs are provided by MJ and none are accepted by you. What happens during this phase depends on your craft design and ascent profile. What SHOULD happen is either SAS engages (can be done manually) or MJ actively Maintains orientation. Preferably prograde as this is much closer to any maneuver node plot that circularizes your orbit.

I just did a test launch and think I know what your talking about. I use the default ascent profile and that coast period between the MECO and mechjeb reorienting the craft before entering timewaro only lasts about 3 seconds for me. So it hasn't been an issue. How much time is passing for you before mechjeb plots the next burn, orients the craft and then gos to timewarp for the coast to apoapsis?

note: this was with the previous .dll from Sarbian. haven't tried today's .dll yet.

[Starwaster]

I just did a test launch and think I know what your talking about. I use the default ascent profile and that coast period between the MECO and mechjeb reorienting the craft before entering timewaro only lasts about 3 seconds for me. So it hasn't been an issue. How much time is passing for you before mechjeb plots the next burn, orients the craft and then gos to timewarp for the coast to apoapsis?

note: this was with the previous .dll from Sarbian. haven't tried today's .dll yet.

Approx 1-2 minutes. (EDIT: sorry misunderstood... not sure actually. 1-2 minute figure was until apoapsis. Not sure exactly when it plots the course but of course it can't warp because the rocket is out of alignment and already rotating away) That seems like it should be enough time to get back into position, but once the tumble starts it takes awhile. The payload (IIRC) is over 100 tons. So I have to actively monitor to make sure it doesn't start tumbling.

The cause of the tumble btw is probably the final series of pulses when apoapsis is finalized. (and apparently spellcheck does not recognize apoapsis as a word... nor apparently does it recognize spellcheck. 'kay....)

I suspect I could also prevent the tumble by switching off engine gimbaling... or at least mitigate it somewhat.

I'm grabbing today's DLL now.

Edited August 25, 2013 by Starwaster

[johnsonwax]

I just did a test launch and think I know what your talking about. I use the default ascent profile and that coast period between the MECO and mechjeb reorienting the craft before entering timewaro only lasts about 3 seconds for me.

That depends a lot on your TWR and what limiters you put in place. With a ship that is always limited to terminal velocity and 40m/s2 acceleration my coast can begin as early as 35km. Turn off the latter, and I would probably be coasting starting at 30km or less. That's a lot of coasting, and every so often one of my launches will get a little lurch right as the engines are coming off and I'll hit the top of the atmosphere pointing the wrong way. And depending on your ship design, you may be making other tradeoffs to tumble stability. My large lifter (2000t on the pad) is nearly a SSTO - it sheds the boosters halfway through circularization, so it's quite a beast getting up there, and its balance is always questionable depending on what's being lifted. When I'm lifting asymmetrical station components, even if it's seemingly balanced in the VAB, by the time 90% of my fuel is gone, the stability that all of that fuel mass gave me is mostly gone and any small imbalances on the pad gets magnified quite a lot.

I agree that the ship should turn on SmartASS turn prograde.

[Ratzap]

For those wondering, the new dll appears to be here:

http://www.sarbian.com/sarbian/MechJeb2.dll

[sojourner]

That depends a lot on your TWR and what limiters you put in place. With a ship that is always limited to terminal velocity and 40m/s2 acceleration my coast can begin as early as 35km. Turn off the latter, and I would probably be coasting starting at 30km or less. That's a lot of coasting, and every so often one of my launches will get a little lurch right as the engines are coming off and I'll hit the top of the atmosphere pointing the wrong way. And depending on your ship design, you may be making other tradeoffs to tumble stability. My large lifter (2000t on the pad) is nearly a SSTO - it sheds the boosters halfway through circularization, so it's quite a beast getting up there, and its balance is always questionable depending on what's being lifted. When I'm lifting asymmetrical station components, even if it's seemingly balanced in the VAB, by the time 90% of my fuel is gone, the stability that all of that fuel mass gave me is mostly gone and any small imbalances on the pad gets magnified quite a lot.

I agree that the ship should turn on SmartASS turn prograde.

Yeah, the biggest launcher I've ever created could put about 120tons into orbit and I use the default ascent profile in Mechjeb, so this hasn't been an issue for me. But in your case I do see the need for it.

[Mihara]

In the current version, the "Use SAS when available" option is gone, but node autopilot will still lock up attempting to turn when SAS is enabled -- i.e. it won't disable SAS by itself, and it still won't kick SmartASS into auto mode either.

[Payload]

The whole ascent autopilot needs a revamp anyway. It's only just descriptive enough that the operation can be carried out in a repeatable way. It is far from optimal. Here is to hoping that one day we can actually use that nice graph to plot our ascent. What would be really nice would be to be able to click and add points on the graph and to be able to plot our course if you will. Currently we have to give up gravity losses to gain aerodynamic efficiency. That of course will always be true but there is much more DV to be had on both ends. The current system only allows us a best fit to the middle of the efficiency range.