Why can't I attach a subassembly on a radial decoupler?

[Zeiss Ikon]

So, the other night, I was redesigning my four-passenger Munar orbit tourist bus (stock career, 1.3).  Because I had a very good-flying and reliable two-passenger space glider orbiter, and the four-passenger version (stretched, and tandem wings) hadn't worked at all well, I decided to just strap two of the reliable ones belly to belly.  I rearranged the strap-on boosters so the booster cores could be close together, then pulled the subassembly, an entire orbiter, out of the storage shed and went to stick it on the radial decoupler on the primary orbiter's belly (planning to build down the booster from there, likely also with a subassembly).

https://imgur.com/a/H5kvP

(Album showing similar orbiter -- added a 100 unit Lf/O tank ahead of the Crew Cabin for Munar service)

Imagine my consternation when no way, no how, would the fully assembled orbiter stick on the radial decoupler.

I satisfied myself that the decoupler's attachment node was working -- I could pull a fuel tank out of the parts rack and stick it right on there, but the orbiter subassembly wouldn't go.  Furthermore, the subassembly would happily mount to an end node on a tank that was rerooted upward from an engine (equivalent to being built from the ground up).

Even odder, I started building up a new orbiter from the fuel tank (22 parts, not that big a challenge), got the storage bay, engine, wings, and tail fin mounted, then accidentally pulled the tank off the decoupler -- and this partial orbiter assembly would only stick onto the decoupler at the joint between a wing and the tank.  It would stick there reliably, on either wing root, but no way would the tank stick back onto the decoupler.

Thinking about it now, I wonder if this doesn't come in some way from having parts mounted in mirror symmetry, disabling radial attachment.  Contrary to this idea, however, the orbiter that was still on the stack gave no trouble in mounting the radial decoupler to its belly.

I did eventually get the orbiter built back up from the primary fuel tank, everything in place, and verified with a reverted launch that the combined vessel handles exactly like the single-stack version (not too suprising, but gratifying none the less).  I was left, however, with this mystery.

Any idea why the orbiter subassembly wouldn't latch onto the radial decoupler, and the partial assembly would latch only at the wing roots?

Edited October 26, 2017 by Zeiss Ikon
Add image showing orbiter

[bewing]

The attachment point of a subassembly always has to be the root part of the subassembly. (AKA the Reroot tool is your friend.)

[Spricigo]

 Bewing already pointed the hard facts of life using subassembles. 

Since there's nothing else to say, I will use the opportunity yo shameful self-promotion: https://kerbalx.com/Spricigo/Selene easy-to-use, 5seat tourist bus to Mun/Minmus orbit. 

 

Edited October 27, 2017 by Spricigo

[Zeiss Ikon]

Aha.  That explains the orbiter subassembly's misbehavior, but doesn't make sense of why the partial assembly (800 tank, storage bay, and Swivel, mirror image symmetry wings and tail fin) would only attach at the wing root, not anywhere else on the (first part attached) tank, after I accidentally picked it off the decoupler.  By this rule, I should have been able to stick it back on in exactly the orientation I'd built it; instead, it wanted to rotate 45 degrees and offset so the decoupler was attached equally on the tank and either wing.

[Geschosskopf]

Subassemblies don't like to radially attach.  They have a stack node at their root part and you're pretty much stuck with that.

HOWEVER, there's a simple workaround.  Build the subassembly, the put a single cubic octagonal strut on the side of it where you want it to radially attach to the main ship's radial decoupler.  Then make that cubic octagonal strut the root part of the subassembly.  Now, when you bring the subassembly to the main ship, it's attachment point will be on the side, at the end of the cubic octagonal strut, and this will attach to the main ship's radial decoupler.  Once you've got that established, you can then use regular struts between the sub assembly and the main ship to reinforce the connection.

[Zeiss Ikon]

@Geschosskopf That sounds like it should work for ordinary stuff, but this isn't an ordinary subassembly.  Here's the complete "conjoined" vessel.

https://i.imgur.com/cxHeEA4

The intention was to stick the second orbiter onto the belly of the first, then add the booster (possibly as another subassembly).

Does pasting images not work in the Q&A subforums?

Edited October 29, 2017 by Zeiss Ikon

[Spricigo]

Have you considered to put the spaceplanes around a central core? Something like this:

Spoiler

As for how I deal with that subassemble behaviour, I just attach it [Somewhere] so it become part of the craft and then take the part I want to radially attach to  [desired place]. It works fine to me (only drawback is usually I need to divide the 'subassemble' in two to move to the desired place) and it surprises me that it didin't worked for you.

[Zeiss Ikon]

11 minutes ago, Spricigo said:

Have you considered to put the spaceplanes around a central core? Something like this:

  Hide contents

 

I don't yet have 2.5 m engines (or other parts -- a couple tanks only, and IIRC a single adapter) in this career, else I'd most likely be sending missions with Mk. 1-2 command pods and Mk. 2 Lander Cans.  As it is, I'm about to attempt a Mun landing with a stretched-tank version of the Taxicab IIIm, which how has landing legs for vertical Mun landing, and landing gear (for horizontal landing on Kerbin -- assuming I can hit land; I hold little hope yet for landing on the runway at KSC).

[Geschosskopf]

1 hour ago, Zeiss Ikon said:

@Geschosskopf That sounds like it should work for ordinary stuff, but this isn't an ordinary subassembly.  Here's the complete "conjoined" vessel.

So where's the subassembly?  Where are the radial decouplers?  All I see is a vast tangle of stuff at a far distance, from which it's impossible to see details.

[Spricigo]

38 minutes ago, Zeiss Ikon said:

I don't yet have 2.5 m engines

My suggestiong don't requires 2,5m parts. Maybe another image exemplify it better:

Spoiler

 

[Zeiss Ikon]

19 minutes ago, Geschosskopf said:

So where's the subassembly?  Where are the radial decouplers?  All I see is a vast tangle of stuff at a far distance, from which it's impossible to see details.

There's a single radial decoupler between the two orbiters, one of which was to be the subassembly I was attempting to attach.  The other eight radial decouplers mount the strap-ons, and behaved precisely as expected. 

Sorry, that's the only image I have of the complete conjoined vessel.  I could upload a .craft file (it's 100% stock) if you want a better look.

17 minutes ago, Spricigo said:

My suggestiong don't requires 2,5m parts. Maybe another image exemplify it better:

  Hide contents

 

Well, that's simpler to build, at least if I know in advance the "root part" requirement for attaching subassemblies.  What I built, conversely, had a previously known acceptable TWR from four boosters on the single orbiter stack, so after conjoining had two cores and eight boosters.  This one, with a longer, heavier core and still two orbiters, has only seven engines ignited at launch instead of ten.  Even if the engines are all Reliants (likely to lead to a bad day, since the original required three Swivels plus steerable fins to keep pointed to prograde during launch), this has less than 70% of the TWR the original Taxicab IIIm had.  Even with a (presumed) two-stage core, I doubt this would leave the orbiters in LKO with 10-15% core fuel remaining (a couple hundred m/s dV without using any orbiter fuel).

[Spricigo]

4 minutes ago, Zeiss Ikon said:

Well, that's simpler to build, at least if I know in advance the "root part" requirement for attaching subassemblies.  What I built, conversely, had a previously known acceptable TWR from four boosters on the single orbiter stack, so after conjoining had two cores and eight boosters.  This one, with a longer, heavier core and still two orbiters, has only seven engines ignited at launch instead of ten.  Even if the engines are all Reliants (likely to lead to a bad day, since the original required three Swivels plus steerable fins to keep pointed to prograde during launch), this has less than 70% of the TWR the original Taxicab IIIm had.  Even with a (presumed) two-stage core, I doubt this would leave the orbiters in LKO with 10-15% core fuel remaining (a couple hundred m/s dV without using any orbiter fuel).

Well, I just strap the parts together to show the idea of a central core, its not even a WIP design. I don't dare to tell it would make the gravity turn nicely as it is but from what KER told me it would have enough TWR/DeltaV to reach the orbit with a nice margin.

In any case if you can make the whole thing aerodynamic stable you don't need gimbals or steerable fins to fly  it to orbit (easily). Right now you have those wings sticking out in the middle of your launch vehicle. Put another two at the bottom and it probably will be much more flyable.

[Zeiss Ikon]

19 minutes ago, Spricigo said:

Well, I just strap the parts together to show the idea of a central core, its not even a WIP design. I don't dare to tell it would make the gravity turn nicely as it is but from what KER told me it would have enough TWR/DeltaV to reach the orbit with a nice margin.

In any case if you can make the whole thing aerodynamic stable you don't need gimbals or steerable fins to fly  it to orbit (easily). Right now you have those wings sticking out in the middle of your launch vehicle. Put another two at the bottom and it probably will be much more flyable.

As long as the pilot aboard has the ability to hold prograde, it'll keep pointed the way it needs to with just the canards on the orbiters and the steerable surfaces on the boosters directly behind the wings.  No other fins needed, the gimbaled engines compensate easily for the vertical tails.  It only gets flippy if you let the nose get outside the prograde marker above around 100 m/s.  Set a 5-8 degree pitch at 35-45 m/s and then go to prograde hold as soon as the prograde marker moves onto the heading, and she flies a perfectly tractable gravity turn.  By the time the boosters with the fins and Swivels stage away (40+ km) there's too little air for the wings on the orbiters to matter.

[Spricigo]

Yes you can  kep the ship in line with steerable fins and gimbaled engines, but  design your ship to not need it and it will open other options.

Take for exemple  the craft I posted early (Selene) It uses. It has 4 exagerated big Wings at the botton, what allow me to launch it with a relliant and two  thumpers and not even worry about the [D] key. A thumper cost 850funds, plus 2wings that is 1800, meanwhile a Swivel+FL-T800 cost 2300funds. If I can do the same job with either (I do) why not save 500funds?