How can I get MechJeb to guide an Atlas-type 1 1/2 stage rocket?

[Zeiss Ikon]

I'm playing 1.6.1 with RSS, Realism Overhaul, and RP-1 (the latter probably relevant only in that it's responsible for my rocket design).

I've built a rocket that's a pretty good likeness of an Atlas -- not the early SM-65 ICBM (that was the first version), but the next generation that actually launched John Glenn et. al. into orbit in a succession of Mercury spacecraft.  I have verified that it can make orbit, at least without a payload, by manually launching in simulation; I can fly it to a marginal orbit (apogee too high, perigee well down in the atmosphere) using only Smart A.S.S. and a simple gravity turn (establish 5 degrees pitch at appr. 50 m/s, then follow SVEL+).  I know MechJeb can do it better than that.

At the point of staging away the booster engines, I'm above 2500 m/s horizontal velocity, and immediately after staging, MechJeb's dV display shows above 6000 m/s remaining in the tank (orbit is appr. 7800 m/s).  TWR is just above 1 at staging, 2;15 after launch, with (current iteration) 3:20 remaining with the center sustainer engine.

If I attempt to engage MechJeb Ascent Guidance (RSS/RO version, what used to be call PEG) on the launch pad, however, the autopilot will not advance the throttles and displays "target unreachable" -- even when I have the target set to 142 km apogee and 0 perigee (signifying circularize), "current" inclination (28.6 or so degrees, launching from Canaveral), so about the least dV requirement you can have (barring moving the launch site to the equator or similar).  Auto staging is off (doesn't work well with low tech rockets, which often require things like hot staging and always need ullage management).  I've been assuming this is because MechJeb is failing to calculate the dV correctly due to boosters and sustainer running off the same tank, hence no cue for MJ that I'll be ditching a bunch of mass when the TWR starts to get high.  Clearly, the rocket is short dV without staging the boosters -- it shows only about 8660 m/s in the VAB, and my recent experience is I need around 9200 there to reach a stable orbit (gravity losses etc.).

Is there a known way to get MJ to correctly handle stage-and-a-half launchers (image below is the ICBM version, but the new one differs only by having a longer tank and higher thrust)?

Edited November 10, 2019 by Zeiss Ikon
duplicated phrase

[Laie]

8 hours ago, Zeiss Ikon said:

Is there a known way to get MJ to correctly handle stage-and-a-half launchers (image below is the ICBM version, but the new one differs only by having a longer tank and higher thrust)?

Probably not.

You can try to take off yourself, then activate the MJ steering after staging -- but knowing MJ a bit, I suspect it's first order of business upon taking command will be killing the throttle. There goes your one and only ignition.

However, for manual launches,

• a) don't follow the gravity turn all the way to orbit. IIRC (it's been a while) an Atlas needs to start on a relatively shallow turn, but then has to pitch up for a while after booster separation.
• b) Quite generally, lacking a throttle, you can still control time-to-apoapsis by pitch -- try to make it so that it's close to zero towards the end of your fuel. It's pretty easy to launch into a near-circular orbit that way, the hard part is hitting a certain altitude. I was happy if I got altitude to within +-10km.
• c) remember the verniers (LV-101?) -- even the original plans for Atlas involved the sustainer being cut a few m/s before reaching the desired velocity, and finalizing it on the verniers. Which will provide perhaps 0.1g rather than the 3+g you get from the sustainer, so it's quite leisurely.

[Zeiss Ikon]

Thanks, Laie.  I've never gotten a decent orbit manually in RO -- in the stock(ish) game I've always done it by pushing apo up to the desired height, then setting up a node to circularize while I coast up.  Obviously, can't do that with a single ignition.  I find it hard to watch time to apo, fuel, and velocity all at the same time, but it sounds like that's what's needed.

I presume you'd use an action group to shut down the LR-105 and let the LR-101s burn.  When I've gotten MJ to fly this booster to orbit, it was with a second stage (which was heavy enough to be required for orbit), and MJ handles that fine, though I'm not sure how much better it'd do if it actually planned for the big drop in TWR at staging.  But yes, MJ pitches up a bunch (25-30 degrees?) after staging, then pitches the next stage down pretty sharply once it's burning.  I've been presuming the big pitch-up was to compensate for the sudden drop in thrust/acceleration, since the curve up to staging was based on about five times the thrust and exepecting conventional staging after less than three minutes.

I understand the Prime Vector Guidance code is based on actual NASA guidance code, but last I read (a year or so ago) the devs were working on upgrading from Gemini-era code (for Titan II etc. conventional staging) to Shuttle-era code (accounts for thrust drops before MECO, but via booster burn-out rather than timed shutdown; that is, MJ can "know" before launch when thrust will drop and how much).  I might have to put in a feature request to support timed staging, but it's probably not worth the devs' effort to support a sort of niche booster (the low reliability engine stage is fairly short in RO -- from first launch to early manned orbit, but really only matters from around 1957-1963 equivalent -- and the Atlas equivalent is the only booster type that needs this, since the Russian line went with conventional side boosters).

Honestly, if I'd known this was a problem, I'd probably have just bypassed the whole stage-and-a-half setup, by using tanked booster like the Soviets did -- but it does perform significantly better than conventional staging for unreliable engines, and saves funds on engines and tanks compared to the Soviet method.

[Laie]

1 hour ago, Zeiss Ikon said:

I find it hard to watch time to apo, fuel, and velocity all at the same time, but it sounds like that's what's needed.

I think I spent a whole sunday on failed attempts until it started to get better; after that, results improved quickly. In hindsight, it was fun; not sure if I'd have said the same then and there.