Exterior Door Panel Construction - Can Someone Explain?

smorgasbord

Member
Joined
Jan 7, 2022
Messages
1,214
Start at 5:52 if the link doesn't for you:


Exterior door out of cherry, rail and stile construction.
For the panel, they "glue" up some narrow boards. I put "glue" in quotes because they're claiming their secret is to not use glue, but a flexible silicone. Here's the quote while they're showing the construction:

"This allows the expansion and contraction of the wood pieces, but still maintains a waterproof seal. On a panel this large with a tongue and groove construction, you wouldn't want to physically glue these together because you need them to float so that this panel isn't going to expand so much that it begins to separate your stile and rail joinery."

I don't get this:
1) They are literally "gluing" the long tongues to the long grooves, yes with silicone, but it's not like the tongue doesn't reach all the way into the groove. As you can see in this screenshot after they cut the assembly to length, the cut-off stays together:
Screenshot 2025-07-04 at 7.12.57 PM.png
So, it's really just a glued up wide board.

2) Worse, they use Titebond III to not only glue up the rails and stiles, but also TO GLUE THE PANEL IN. Photo showing the glue along the long edge of the stile:
Screenshot 2025-07-04 at 7.15.27 PM.png

3) Worse yet(!), they cross-grain glue the panel in place:
Screenshot 2025-07-04 at 7.14.56 PM.png

4) So, I'm thinking the idea is that while the edges of the panel is glued to the stiles, the inner pieces of the panel can shrink and still have that silicone hold.

5) BUT - what about if the panel expands? They don't talk about installing this during the wet time of year. And that cross-grain glue is a no-no no matter what.

So, either I'm missing something or this door may self-destruct like the door they're replacing, lol.


Note that I do agree with them on using oil-based finishes rather than polyurethane, even Spar Varnishes. There are no clear finishes that hold up for a long long time to the sun's UV rays, so you might as well go with something that's easier to maintain to avoid having to strip the door completely.

But, this door construction technique seems very wrong to me. I built my own front door 3 decades ago, and all the joinery on it is still tight as the day it was installed. Exterior finish is CPES and Sikkens, but I'm thinking of stripping the Sikkens next time and going with something else.
 
Unless they left a gap between each panel board, gluing that panel in is a terrible idea, and cross grain gluing, insane!

Even with a gap though I'd still never actually glue the panel into the stiles.

And it's not like silicone works extremely well with "waterproofing" wood, water always finds a way in.
 
'Silicone' is a misnomer. It's Dow Corning 732. It's an acetoxy RTV sealant.

As for them choosing to glue the panel to the stile, I think it's for reveal reasons. Because the TG panel itself is narrowly constructed and floating, I don't feel this is an issue. This would be an issue if you TB3 the inner panel together though.'

If one were to complain though, I don't see weep holes in the bottom rail. Usually hard to spot though from that distance so I'll give them the benefit of the doubt.
 
@woodferret I could understand if the T&G boards had some spacing, but as it shows in the video and the pic above they're a very tight fit with a thin bead of the silicone, and they not only use TB3 to glue the panel into stiles, but also the rails? The silicone would make more sense to me to use in the stiles and rails to provide some movement flexibility.

When the boards draw in moisture, how can they expand when they're already such a tight fit, and glued in on all four edges?

And the weep holes as you say are a whole other issue, I couldn't see any at all.
 
Last edited:
Fair enough. While I did see 1mm gaps during the glue up, I missed the part where they were clamping it together. That does seem like a procedural hiccup although I can only hope it was only finger tight - kinda want that squish in the middle ground.

I'm pretty sure though the rails were RTV (ya, I checked 7:43). I think you can tell the RTV vs TB3 by them leaving the RTV as a single bead. For the most part, they squirt and brush spread the TB3. Which leads me to think they've actually used RTV on the stiles as well, despite what the 'narrator' says.
 
OK, so what I'm getting it that they designed it not as a frame and panel, but as a set of T&G boards, each of which essentially floats (due to RTV flexibility) in place. Kind of like one might build a deck - except that they're allowing at most 1mm of expansion space on the side of each board.

Here's a CU of the boards just stacked together:
Screenshot 2025-07-05 at 11.17.30 AM.png

The joint at the right doesn't appear to have a gap. The joint at the left does, but these are just stacked, not clamped yet and any clamping will close that joint down. Notice that the gap at the tongue matches the gap at the "V" notch. So any clamping pressure closes both up equally. At best they clamp and then remove the clamps to let the boards "settle" apart, but one wonders how that would actually happen.

Here we can see squeeze-out as they panel is assembled, not yet clamped:
Screenshot 2025-07-05 at 11.18.09 AM.png

Here they are clamping the panel:
Screenshot 2025-07-05 at 11.18.32 AM.png

And here's the squeeze-out after clamping:
Screenshot 2025-07-05 at 11.19.12 AM.png

And here's after trimming the top curved edge, with the cut-off:
Screenshot 2025-07-05 at 11.24.34 AM.png

I sure don't see any gaps between the boards in the panel nor the off-cut.

Seven boards make up the panel. I'll guess the boards are about 6" wide, so losing 3/4" to the T&G, that's probably about 36" wide. Cherry in Oregon will need about a ½" of total movement accommodated.

Looking at final assembly, it does appear that the final tongue milled on the panel's vertical edges doesn't stick out quite as far as the stub tenon on the bottom rail. Here's a shot, which when zoomed seems to indicate a slight recess:
Screenshot 2025-07-05 at 11.26.54 AM.png

IF they had designed this such that each of the 8 vertical T&G "joints" had 1mm of expansion potential, then it's possible that the total panel inside the frame could expand 8mm, and then contract 4mm. But, since these are joined so tightly together, I don't see the 1mm expansion potential - and certainly there's a limit as to how much the RTV sealant can compress (remember it was already compressed enough to squeeze-out during assembly) as the panels expand.

So yeah, I now do see a theory that might might this work, but in practice here I don't see how it works. Especially with the Osmo UV Protective Oil finish that another thread here on FOG shows isn't very good outdoors.
 
Last edited:
OK, couple more.

Here's the door getting glued up. They're using spacers pressed into the V notches to keep the panel centered:
Screenshot 2025-07-05 at 1.09.20 PM.png


And when they're applying the stain/finish to the door itself, you can see the larger gap around the inside of the door frame while the inside of T&G boards are tight:
Screenshot 2025-07-05 at 11.51.36 AM.png

Maybe that gap is what, 3mm? Is that enough on either side for a 36" wide cherry panel? So, it would seem all that talk about T&Ging the center panel with RTV sealant was for naught after all. The main thing is they have a 3mm (maybe 4mm?) gap on either side of the wide panel, and have sealant in the edge grooves to accommodate wood movement. Is that enough? 36" of cherry moves ½" total in their neck of the woods, so assuming the panel is mid-width moisture content it could be OK. But, maybe not if it's dry while assembled and the rainy season is 6 months away....

Also, looking at the bottom rail in the top photo, they have six dowels in that joint, over at least 10" of rail width, maybe more like 12". And all of those dowels are glued. There's probably about 1/8" - 5/32" of movement that will want to happen in that wide a joint. Is this also a potential problem down the road?

Here's my front door, built over 30 years ago:
Door.jpg

For that extra wide bottom rail, I two-pieced it, with (unglued, unsealanted) T&G.
Each of the two 10.25" wide boards has two big loose tenons (DF700 wasn't invented yet), but only the top tenon of the top board and the bottom tenon of the bottom board are glued. Here's a close-up:
DoorCU.jpg

That gap between them can expand or contract. I did toy with the idea of putting black caulking in there, especially on the outside, but never did. The wood species is Narra, which is about as stable as teak so very little wood movement anyway. I should have made that gap no larger than half its current width, but live and learn. I would not want to glue a 20" wide board across its entire width - this construction gives me a tight bottom edge with the stile as well as a fixed top edge where the glass rests - important to not crack the glass.

Anyway, would you glue a 12" wide rail to the stiles as these guys did?
 
Last edited:
OK, couple more.

Here's the door getting glued up. They're using spacers pressed into the V notches to keep the panel centered:

Anyway, would you glue a 12" wide rail to the stiles as these guys did?
The whole's just baffling to me. As you note they're using spacers between the panel and the stiles, that they then say they're using TB3 to glue the panel in which will inhibit any movement!

And they used a flexible sealant on the panel joints, and then proceeded to completely remove any expansion room.

And then it's glued in the rails, this last one is the worst for me, gluing a wide end grain panel into the rails?

I would have thought a 2mm space between each of the panel members with a bead of that silicone, could then allow 1-1.5mm expansion/contraction safely while retaining the water proofing aspect. No glueing the panel to the rails, and maybe glue in the stiles, although I'd prefer they float with some of that silicone there too myself.
 
My conclusion is, as @woodferret suggests, that the panel isn't TB3'ed in, but is RTV sealanted in. With the spacers, they're getting 3mm-4mm of expansion room on either side of the panel, which might be enough, just barely if the rainy seasons aren't too, well, humid for too long.

The whole thing about the panel glue-up being sealant and T&G, however, is misdirection. They could just as easily TB3'ed that assembly.

As for the RTV sealant on the top/bottom rail, I suppose there's enough flexibility even if the panel edges move 4mm each. And if not, it's likely the internal sealant will just give way and no-one will ever know.

Still wondering about that 12" wide rail glued with 6 dowels though....
 
Back
Top