I’m starting to build a torsion box for my LR4, it’ll be 3/4 MDF ribs 4” tall, joined with an edge joint exactly half the height of the ribs (like the pic below).
. The top and bottom will be a 1/4” underlayment type plywood, glued and stapled to the ribs.
(Yes, I know it’s overbuilt, but this is what I’m going with…)
The dado for the joints will be a friction fit - loose enough to join by hand, no hammering needed.
My question is whether I should glue the joints or not, given that the top and bottom will be glued and stapled - my concern is whether I’ll be able to make the box flat during assembly if I glue the joints - lots of potential for something to move a little bit. If I rely instead on the top and bottom sheets to hold everything square and flat, that seems like a better option.
So based on my research during making my torsion box table, apparently the goal is to have an opening for the cross lap joint that is exactly the same width as the thickness of the material.
I glued my joints, but for some reason did not glue my skins. Also my skins were too thick. If I ever redo it, I will use thinner skins, and I will glue both the joints and the skins.
Thanks, @DougJoseph - I’ll tune the dado blade to make a perfect (but not too tight) cut (I’ve got a range of shims to let me get to 0.005” adjustments if needed). An earlier attempt saw me using an exact width dado, which ended up needing a lot of hammering (which, in turn, broke a lot of MDF ribs).
typically, these joints are made with an allowance that makes them a nice friction fit when dry fit, but allow for expansion after glue addition to really tighten them up. Honestly, if its a good friction fit, and you glue the skins, you’ll get a great torsion table. I doubt you would see any additional benefit to gluing the half laps, and may even introduce (probably irrelevant) variance due to unequal glue volume addition and squeeze out.
personally- wouldnt glue the half laps, would glue to skins
The strength of the torsion box comes from the skins being held to the ribs, and the ribs being unable to move. A tiny amount of movement reduces the strength of the box greatly.
As a result, I glue and screw everything.
However, if you glue the skins, it will be fine, because that will hold the ribs and spars in place. Look at Ikeaboard as an example. That is expanded corrugated paper, glued to the skins inside, and it’s plenty rigid.
I would insist on the skins being glued, as that bond is far superior to any reasonable number of screws or nails holding the skin in place, and will result is a far superior torsion box, although screws will still result in one good enough for our purposes.
Remember, we aren’t doing high precision machining. Or if we are, the CNC isn’t doing everything. I’ve seen luthiers doing guitar and violin necks going for 0.001" tolerances, but not from a raw CNC piece. I’m not going to machine a cylinder head on my LowRider either, nor would I suggest it to anyone.
We can easily go way overboard on the table here. It’s nice to have something that’s good, but there is such a thing as “good enough.”
I mostly agree with Dan, it comes down to how the skins are connected and whether you have a top skin only or both top and bottom.
Think about the notch that is cut in that rib, it basically makes it 1/4 as stiff at that one point and any load will either try to pinch the notch closed or bend it open. If you make it a tight fit then it can’t pinch closed but can open easily. If you glue/screw/cleat/whatever it to the crossing rib then it can’t open or close easily.
If you glue/screw the skin on the other side then you’ve braced the open part of the notch and it can’t close or open.
So with no bottom skin or poorly attached skins then that joint is important.
With bottom skin well attached, I doubt those joints even need to touch.
Progress is being made - this is the first half, just loosely assembled to verify everything fits. (you can see my LR4 patiently waiting in the background…)
Second half looks just like it and joins at the obvious location. Overall size is about 64” wide by 116” long.
The plan is to skin both sides with 1/4” underlayment plywood (glue and 18ga staples), then add MDF “runners” on the left and right edges for the LR4 to ride on and a full sheet of MDF as a spoil board. Ought to weight about a metric ton when I’m finished…
So I decided to glue the long ribs together where they overlap, to make final assembly easier. And that’s where I forgot that one side overlaps opposite the other - I glued them all the same, so 3 out of the 5 fit but the other two are no bueno. This is where plans would have helped — I should have ensured consistency; there’s no reason to have made the two sides different.
And at this point, I’m done messing with it. I think I have enough MDF scraps that I can glue and screw to my old workbench where the LR3 lived and make it wide and long enough to handle full-sheet use - I’ll just need to add some “outrigger” legs in the corners and middle for support. It’ll be flat-ish, which is probably as good as I really need (and all I really needed in the first place!)
I second this!
I don’t see anything that disqualifies that picture from being a perfectly usable torsion box center structure.
The ribs inside a torsion box only need to locate and constrain the top and bottom, and that looks like it would be fully the case with what’s shown in the picture.
It was absolutely beautiful during dry assembly! Things went sideways when I had the bright idea to glue the long ribs together and did it wrong, making assembly impossible.
Is it recoverable? Yeah, I could just cut a new set of long ribs and glue them together the right way.
But I’ve already burned through 2 full sheets of MDF that fell victim to other mistakes and are now overflowing my burn bin. My wallet says to quit before we go bankrupt…
So I’m moving on to something simple, making use of as many scraps as I can. Basically, I just screwed the leftovers from my rib-making exercise to my old workbench and turned some of the old ribs into a (doubled-up) support apron around the perimeter. There’s about 15" of overhang on all sides of the old bench, so I’ll be adding legs on the corners for support. Some MDF runners on the left and right for the LR4 to ride on and an MDF spoilboard. And then maybe I’ll be ready to actually cut something…
From the photo a few posts up where I highlighted a couple of overlaps – note that one overlaps on the right and the other on the left. When I glued those long ribs together, I overlapped all of them on the right. So none of the cross ribs would fit without re-making the long ribs correctly…
@RockinRiley - That’s what I thought, too. But rotating 180 degrees horizontally looks just like not rotating at all. And rotating 180 degrees vertically puts the dados for the cross-ribs on the wrong side for half of the long ribs…
@Tokoloshe What I have now isn’t a box - it’s a few pieces of MDF screwed to the top of an old workbench, overhanging the sides by 15" or so. I added a double-thickness “apron” around the perimeter using old torsion box ribs.
Chiming in after a long absence… Torsion boxes get most of their benefit from the rigid connection of the skins to the ribs. Which is why you can have hollow core doors that are 1/16"-1/8" veneers glued to a relatively sparse honeycomb of cardboard, and it gives you a reasonably stiff door. In theory, a torsion box has the same rigidity as a solid of the same volume, but not necessarily the same strength. But it’s critical that the ribs and skins be well connected. If you watch manufacturing videos, hollow core products usually have the skins placed on with a complete coating of glue on one side, and then the whole thing is placed in a press for some amount of time to cure and ensure the skins have bonded to the ribbing material.
And while I don’t know the specifics, I’d think that it’s moderately critical to have both the top and bottom skinned if you want maximum rigidity. If you think about it, the skins prevent warping by limiting the movement of the ribs, and if you only skin one side, then the other side can bow out with relative ease. You want it all over-constrained.
