Underside diagonal reinforcements excluded as in LDraw part 51262.
The underside horizontal cylinders do not look perfect, but perhaps good enough without adding a lot of small triangles. The geometric challenge is the contact area between two perpendicular cylinders. I could not find a nice primitive to cover such a case. Feel free to edit if you find a better solution.

as a joiner of the cylos use this sample:
1 9 86 4 -20 0 0 2 0 19 0 -1.5 0 0 box4-1.dat
1 9 86 4 -60 0 0 2 0 19 0 -1.5 0 0 box4-1.dat
1 9 154 4 -20 0 0 -2 0 19 0 1.5 0 0 box4-1.dat
1 9 154 4 -60 0 0 -2 0 19 0 1.5 0 0 box4-1.dat
1 9 154 4 -100 0 0 -2 0 19 0 1.5 0 0 box4-1.dat
1 9 154 4 -140 0 0 -2 0 19 0 1.5 0 0 box4-1.dat

1 9 104 7 -40 0 -20 0 0 0 10 10 0 0 2-4cyli.dat
1 9 104 7 -40 0 4 0 0 0 10 -10 0 0 1-4cyls2.dat
1 9 104 7 -40 0 4 0 0 0 10 10 0 0 1-4cyls2.dat
1 9 84 7 -40 0 -1 0 0 0 10 10 0 0 2-4ndis.dat
1 9 84 7 -40 0 -1 0 0 0 10 10 0 0 2-4edge.dat

3 9 84 17 -30 84 17 -50 84 24 -84
4 9 84 4 -84 84 24 -84 84 17 -50 84 7 -50
2 14 107.8085 4 -50 84 4 -50
3 9 107.8085 7 -50 107.8085 4 -50 104 7 -50
3 9 84 7 -50 84 4 -50 84 4 -84
4 9 84 7 -50 104 7 -50 107.8085 4 -50 84 4 -50

2 14 104.761 16.239 -36.173 104 17 -40
2 14 105.2176 14.7044 -33.8768 104.761 16.239 -36.173
2 14 107.2999 10.827 -30.761 107.8085 7 -30
2 14 105.851 14.071 -32.929 105.2176 14.7044 -33.8768
2 14 105.851 14.071 -32.929 107.2999 10.827 -30.761
2 14 104 17 -40 104.761 16.239 -43.827
2 14 104.761 16.239 -43.827 105.2176 14.7044 -46.1232
2 14 107.8085 7 -50 107.2999 10.827 -49.239
2 14 105.2176 14.7044 -46.1232 105.851 14.071 -47.071
2 14 107.2999 10.827 -49.239 105.851 14.071 -47.071
2 14 84 4 -50 84 4 -84
2 14 107.8085 7 -30 107.8085 4 -30
2 14 107.8085 4 -50 107.8085 7 -50
2 14 84 4 -50 84 7 -50
2 14 84 4 -30 84 7 -30
2 14 84 4 -30 107.8085 4 -30
3 9 107.8085 4 -30 107.8085 7 -30 104 7 -30
4 9 107.8085 4 -30 104 7 -30 84 7 -30 84 4 -30

Try to avoid TJunctions on flat surfaces

there are gaps between the top cylinder ends and the surfaces

the joiner looks quite detailed and complex then, should that get subfiled into an eigth?

@rene, that would make sense of course, I leave that up to the author.

This T-shape at the corners which touch the duplo studs shoule be there:
https://i.ebayimg.com/images/g/bN4AAOSwO8JgWfIp/s-l1600.webp

add duplo castloe keywords since i appread most in duplo castloe sets

Thanks Gerald! Most helpful comments.
Also included Jeff's request for keywords.
I finally solved the annoying dot at the end of the horizontal cylinders. It was a superposition of the conditional edge of the bottom stud and the centre of a 2-4disc embedded in the 2-4cylc. I solved the first by removing a hidden section of the outer wall of the bottom stud and the second by swapping the 2-4disc for a 2-4chrd.

Instead of cutting it in the middle of the side, it is better to keep the long side complete and add half a corner. And then rotate that subfile 4 times.
Dividing a quad into 2 symetrical halves is not good. Only adds unneeded vertices.

How do one create a half cylinder/stud/understud on a diagonal? Aha! I could include the whole cylinder/stud/understud on one side, but not on the other, i.e. *not* cutting the whole piece at the same line. Thanks Magnus!

Subfiling is fun.
There might be one problem. Including the studs in a mirrored/rotated subfile might give a logo in the stud a bad orientation.
But we don't have any logos in the duplu studs. Yet. And maybe we never will, since there are 2 known different logos.

Corrected file name, removed redundant keyword 'duplo', and moved studs on top from subpart to here.

If this should work as intended, with logos in the stud primitives, they need to all have the same orientation.
Looks like you have rotated the 2x2 groups.

Corrected rotation of two 2x2 stud groups.

* Should come with a HELP giving indication how the trap door gets positioned:

1 16 0 8 40 1 0 0 0 1 0 0 0 1 51706.dat

Can that be done with a header edit?
Will one of the possible 4 trap door placement orientations be enough?

yes, can be done via headeredit (ask an admin)
because the part is symmetrical, one is enough (rotating the whole plate is easier...)
do we need an assembly here (both color 16)? which way should the trap door be?

Added help for placement of trap door.
Reordered subpart numbering. Now s02 is used in s01 and main part uses s01.
Removed T-junctions visible from top (originally in 51705s02, now in s01).
Header edit in smaller subpart (originally s01, now s02)
Apologies for resubmitting a part with certify votes

Updated HELP info

* No need to re-submit a part just for a HELP. Just ask one of the admins to add info via header edit.

Release 2025-04