Ascent guidance broken with Principia

[drake127]

MechJeb Version

2.15.0

KSP Version

1.12.5

Description

The latest version of MechJeb2 doesn't work properly with Principia and as a result PVG doesn't seem to be working anymore. I have updated both Principia and MechJeb yesterday and I wasn't able to run ascent guidance as it wanted almost infinite deltaV. It was like it wasn't able to fetch orbital info overshooting the orbit by huge amount (not sure if it would stop).

I have downgraded Principia but I cannot tell whether it helped, it behaved funky. I have read that Principia deprecated some external API but when I downgraded MechJeb2 to 2.14.3, it works well even with the latest Principia (2025.11 Laplace).

I can help with some form of debugging if anybody is interested.

Replication Case

KSP RP-1 (RSS, RO, Principia, MechJeb2) in latest versions, attempt to do a PVG ascent guidance to LEO with any reasonable launch vehicle.

[draqsko]

Can confirm, starting new RP-1 career with Principia Laplace and Ascent Guidance doesn't work at all after MJ updated to 2.15, it doesn't even attempt to steer using Classic Ascent module. Previously on the same save it had worked for my current rocket (A-4 photo, which I had simmed before updating) but it broke after the update. Rolling back to MJ version 2.14 with no change to Principia has Ascent Guidance working again at least for the Classic Ascent module, not up to the stage where I can test the PVG module though.

So to be clear:

Principia Laplace + MJ 2.14 = expected Classic Ascent trajectory

Principia Laplace + MJ 2.15 = rocket doesn't even attempt to steer with Classic Ascent, all it does it autostage on Ascent Guidance.

[lamont-granquist]

With the latest version of Mechjeb with PVG you have to set the last stage your rocket uses to get to orbit, and you need to specify the early shutoff stage for an optimized burntime of your insertion stage. Leave that unchecked to do a suborbital flight using all your available fuel to maximize your apoapsis.

Lots of people have also been using that version of PVG in Principia and not seeing any issues with it. There also isn't that much different between Principia and "stock" gravity in terms of APIs, you'll just hit the osculating orbit instead of the Keplerian orbit as a target, but it won't make any difference. There's no known bugs with Principia that have been reported for a while now (there were issues caused by the LAN of targets wobbling around constantly in Principia that happened well over a year ago now, which was the last known bug).

@draqsko classic ascent isn't PVG, and you're describing something completely different, and there's no evidence you've offered that it has anything to do with Principia. You should check that your settings are the same since the configs use different values, so it probably didn't port forward your settings correctly.

[lamont-granquist]

You'd also both probably do better to get onto the RO discord so that other people can help you, I don't have an up to date install of Principia right now and probably don't have the time to deal with it unless I get a really viable bug report:

https://discord.gg/3nqeEnMv

[draqsko]

@lamont-granquist My settings are fine, it literally worked until I updated MechJeb from version 2.14.3.0 to version 2.15.0.0 and rolling back to version 2.14.3.0 allows it to work again. No other mods were changed, just MJ. I used the same settings in both versions and I know I double checked that. There is literally no steering response from MJ with Principia installed, not from Ascent Guidance, not from Smart SAS. It'll autostage and do other stuff but it's like the PID is broken or something. The stock SAS module will work, but MJ won't and I think the issue with PVG is probably related to that as well which is why I mentioned it here.

I'll continue to use the previous version for now but until you can get an updated Principia installed for testing, I don't see how you'd even be able to progress this issue anyways.

[draqsko]

Here's some screenshots showing what I'm describing. But clearly the attitude control module is broken somehow with Principia Laplace installed. And if that is breaking, then anything that depends on it will break as well, like PVG or CAG.

https://imgur.com/a/mj-attitude-control-broken-with-principia-version-laplace-version-2-15-0-0-JHvBW7b

As I said, rolling back to version 2.14.3.0 on CKAN restores attitude control.

[lamont-granquist]

Having attitude control disabled on the pad is deliberate, otherwise it winds up the integral on the PID, that has nothing to do with Principia. See #1954.

[draqsko]

Except it's more than just the pad where it was disabled and it didn't disable the PID in the previous version of MJ as my screenshots demonstrate. However I think I may have found a fix for that issue and honestly, you probably wouldn't have found it without updating an old installation via CKAN.

Delete all the cfg files in /GameData/MechJeb2/Plugins/PluginData/MechJeb2 folder and let MJ rebuild them all from scratch. CKAN didn't delete or alter them after updating so there was probably some garbage or corruption in the cfg files, deleting and letting them get rebuilt fixed the dead PID for me, still not sure on overshoot though.

The irony is that I observed another odd behavior that I was trying to get to the bottom of before reporting it here that led to this fix, I'll report that in its own thread since it is entirely different than this issue.

[draqsko]

Well I was stupid and didn't save a copy of the cfg files and instead blew them away entirely. If it happens again, I'll save a copy but try as I might, I can't break them again right now. It was very repeatable before deleting those cfg files so it had to be some garbage that wound up in there. I'm sure if it is something repeatable, I'll hit it again in the new save I've started and hopefully have something you can pick apart.

[lamont-granquist]

Yes, I said:

"You should check that your settings are the same, since the configs use different values, so it probably didn't port forward your settings correctly."

Or pave them and start over works fine as well.

[draqsko]

"You should check that your settings are the same, since the configs use different values, so it probably didn't port forward your settings correctly."

And I will say it again; I did check that the settings are the same. Both in the game UI AND in the actual CFG files using a plain text editor, specifically Notepad+. Now I know there are characters that don't display in Notepad and that was probably what happened but since I don't have those files anymore, I can't tell but there was nothing that a normal end-user would be able to see that was different between a CFG that was working and one that wasn't. I would have needed a code editor to really check the files, and I don't have one installed on this PC.

And it did more than just break the PID in version 2.15 (but oddly not after rolling back to version 2.14.3). What led me to actually blowing away the files was the fact that after I installed KER for the HUD info displays, and closed out the delta V window for MJ, the lead time value for hot staging and dropping solids early was not being honored, in either version. It would stage as soon as pre and post staging delays allowed if either option was checked in the AG settings UI. But that would only happen if the delta V window was closed and never opened during the flight scene, if it was open at any time after the start of the flight scene everything operated normally, even if it was subsequently closed. So yeah, I got frustrated and blew everything away because none of it made sense. Even looking at the settings between a rocket I know worked and one that didn't, showed no visible difference. They both read like this:

DropSolidsLeadTime { ValConfig = 0.10000000000000001 TextConfig = 0.1 }

How am I supposed to tell when a CFG is actually working correctly and one that is breaking the PID or other functions without a code editor? I mean this was a clean install from a copy of KSP 1.12.5 and they both looked the same visually and somehow updating MJ from 2.14.3 to 2.15 managed to break the config files, it was the only mod that updated that morning and I had been testing the install for over a week without an issue. So why is one working fine without the delta V window ever being opened, and one stages solids immediately if the delta V window is never opened, yet will work fine if it is opened even if it is closed before you ever activate Ascent Guidance? And yes, I tested both versions. When I said I tried to break it, I spent an entire weekend destroying rockets, rolling forwards and backwards with versions, doing every scenario I could remember doing through the past month with every rocket exactly as it was designed originally, and I can't get it break like it did on that update. So, I'd really like to know how I am supposed to tell if it's not something more serious, or hey that CFG looks messed up, delete it and rebuild a new one. Because there doesn't seem to be ANY visible difference.