Multi-Ply Construction Details for Raised Panels on Exterior Doors

Woodworkers develop ideas for back-to-back panel construction on exterior doors. November 14, 2011

I need opinions on raised panels for exterior doors. This is for a solid lumber glue-up of sapele. In most cases we have fabricated exterior doors made out of hickory and alder. We typically create two raised panels sandwiched to make one panel for an exterior door. This allows for movement with the variant of temperatures inside and outside of the house. We also flip the grain to reduce warpage issues. So for a 1-5/8" panel, each panel would be 13/16" thick. We have found that the panels still have a tendency to warp. If we make the raised panel as a 1-5/8" 1-piece glue-up, it has a tendency to check more. Does anyone have any suggestions?

Forum Responses
(Architectural Woodworking Forum)
From contributor M:
On moderate size doors with say two panels we do a three ply layup with 1/4" acx in the middle. We seal the end grain of all panels and have had decent results. On the monster single panels we have had some checking issues using plywood in the center when there is any exposure. Lately, and with good success, we have used a series of horizontal slots while using a router bit and short screws to lock the front and back panel together, still allow for expansion. We do this on the CNC.

From contributor D:
Interesting thing you're doing there Contributor M. We also build back to back like you guys, and instead of a butt joint we use a glue joint. Lots of the majors have switched to this method. I cut many panels out of our contractor's doors to put in different lite configurations, so I get to dissect many doors from many companies. I see a trend going thinner with the panels, down to 1 1/8" for two back to back.

Definitely with the thinner panels a warp doesn't have as much power. Flipping the grain we believe wholeheartedly is absolutely essential. Every single crack we've ever had can be traced back to endgrain patterns cupping the same way adjacent to each other. I know some people think all the flipping is a waste of time. In our shop it is one of the artistic type skills we value greatly, rearranging your stock into a beautiful, natural looking transition across a glueup that is also correct on the endgrain. Another thing that helps is narrower pieces, less than 3" wide. Wood movement is volumetric - the bigger the piece, the more the movement. We do go wider if the grain is vertical. Large panels are an Achilles’ heel for doors and our sales staff tries to discourage them. I too am interested greatly in any tricks others use in panel construction, as we also feel your pain.

From contributor O:
This is interesting. I have made doors for many years, and have tried all sorts of different methods from 'cross you fingers and hope' to multi-ply thickness, back to back, no glue ups with natural widths, butt joints, and even rimmed stable panels, and more. We currently do panels in solids, unless the door is in an unwarranted exposure and we are called in to repair. Then we do a three ply glue up with the core 90 degrees to the faces. I used to do a three ply with two faces of solid onto a 1/4" luan core. Either three ply works well, with only slight glueline failure or checks in the face in heavily exposed, dark finished doors. No cross grain movement is seen, which is a benefit.

As an evolutionary step, we have also done a three ply with cross banded and face veneered panels - a seven ply assembly. This worked well, but is expensive. With solid panels the only problem we ever see is that different pieces of wood in the glued for width panel will expand a bit differently from their neighbor. Not a crack, but a raised seam where one surface is a few thousandths higher one side of the joint. Not really a problem, but one I would prefer to not see or hear about. These occur even in our pattern grade Honduras panels. TB3 is the preferred glue, but we also use plastic resin for panels.
My experience with back to back was short. The outer panels would like to split at the thin edges where they met the rail/stile. Being thin, I think they also cupped more easily.

As for 'rip and flip', I have not done that for many years. When wood moves the cupping we see is mostly due to one side taking on more moisture than the other, regardless of ring orientation.

Regarding Contributor D's remark about narrower pieces (3") moving less than larger pieces, I would argue that a 12" natural width panel of flatsawn will move the same as four 3" flatsawn pieces glued for width. The number of parts is irrelevant to movement. Even if it were accurate, one could gain more advantage by using quartered stock to minimize movement and get stability. The best method I have found is a seven ply panel with splined rim raise mold applied and splined miters at the corners. Tedious and expensive, but they are about bullet-proof.

From contributor D:
No question ply’s with a mitered surround is better, and we've done it as well, but it sure is much more tedious. I think you aren't envisioning the same scenario as I am. A 12" wide flat sawn board will cup, and maybe a quarter of an inch or more at that. A 3" wide board may only cup a sixteenth. Quartersawn will have movement in width only. Two severe cups adjacent to one another will crack. Ideally no moisture variation and no cup is better, but I hear stories of house painters who show up with car wash brushes and TSP, then rinse everything down with a high pressure nozzle to prep the surface. I stand firm and believe opposing ring orientation is cheap insurance and due diligence, and in the worse case at least can't hurt.

From contributor C:
I have less experience building doors than probably anyone on this thread but the door I am currently working on has flat panels which I built with two 3/8" edge glued panels and a 1/4" plywood core, much as others described. My million dollar idea was to kerf the back side of the panels. I can't say for sure it is going to make a difference as this is the first time I have tried it.

From contributor D:
To contributor C: I like it - I like it a lot. That fresh perspective is not new - it's used commercially on the backs of jambs and works wonders but I've never thought to relief cut the backs of panels! This is seriously a revelation. I have an order for 12" jambs and a couple dozen oval cutout big panel white oak doors that will get this treatment right now. I'm even thinking one better. Instead of kerfing the backsides of panels with a sawblade, how about a v-groove? More stock removal and movement stresses.

From contributor C:
I'm pretty new to building doors (two years) but I used to work in a cabinet shop that pre-hung a lot of doors so I did get the idea from door jambs. I almost didn't post because I thought maybe it was just a stupid idea. It's nice to get some validation.

From the original questioner:
We are going to try kerfing the backs of the panels. That should reduce or eliminate any warping issues with the panels with minimal labor. Splicing the profile to the panel is a very effective way to eliminate these issues, but it is more time consuming than what we would like to spend on our projects.

From contributor R:
To contributor C and D: Just so I get this right, are you kerfing the plys’ and then gluing them up or just kerfing the backs of the panel?

From contributor C:
I'm making separate front and back panels and kerfing the backsides, then assembling with a piece of 1/4" ply in between so it's a 3-ply layup. I think I understand your question. Are the kerfs visible – no.

From contributor D:
I've been thinking about this and looking at flooring, mouldings, jamb, and etc. where this technique is used. I think I'll forgo kerfs in favor of dimples, using a core box bit on the CNC. Same idea - just no sharp corners. Furthermore, I won't be running them all the way out to the edges, keeping any stock removal 2" from the edges, where the hip of a raised panel gets thinner. Kind of like stop fluted casing using a 1" core box but going only 1/4" deep. More of a gradual washboard effect rather than sharp square kerfs

From contributor O:
Playing the devil's advocate here, I am wondering that if all production is changed to this 'stop dimpled' technique how you will know if it is successful? In production of 120 exterior paneled doors a year, we have problems in two or three - two percent. Of those, it is almost always unclear as to what the manufacturing defect is. We are slow to change everything on what we can best describe a possible cause. Not even probable.

How will you know the new technique worked? If you have no failures or fewer failures, how can you be sure it is not due to some other cause? The tricky part about science is that it has to work both ways, and be verifiable. If only one way, then bad science arises and the next thing you know, the guys making the panels have to wear their lucky socks when gluing panels.

I will be the first to say we often do a three ply panel, and I realize it breaks the rules for solids applied over a stable core (the solid faces should be no more than 1/8" or 1/4" depending on whom you are reading). However, we get away with it for the most part, with no problems. So, what causes the one or two failures?

From contributor C:
Contributor D - I understand not going all the way to an edge when doing raised panels but what made you not want square corners? I've already got my door assembled so it's too late for this one but I might change on the next one if you can share your reasoning.

Contributor O - you're absolutely right but what I do know for sure is that my panels for this door were starting to cup before I put the kerfs in and they have kept themselves completely flat since.

To contributor D:
Never mind I think I figured it out. A round bottom would distribute stress more evenly if the panels get shocked.

From contributor D:
More important than anything, I believe anyway, is that in any panel glueups’ there is a "Drop test". The chemists at major glue producers insist on this before they test their glue joints. Basically, it's just a drop of water placed on any glue surface. You watch it, and it should dissipate in about 15 seconds. If it doesn't, sharpen your cutters. This is a better way to judge when your cutters are dull- if they feel like they're not cutting as well as they should, they're probably already too dull, and burnishing/laying over the grain slightly.

Anyway, me wanting to use dimples now is probably more a judgment call than anything. There's plenty of fluted casing I've seen that's been 50 years old and looked pretty good- can't say the same for the doors they've been around though. It's our mission to improve. The jamb and moulding producers can't be doing this just for kicks. Maybe it helps save one panel like the one out in Hawaii where only a few contractors are and they've got a little agreement amongst each other and won't replace a panel for less than $500. Flying one of your guys out there from North Carolina in the winter would be good for morale, but not profitable.

Maybe it's all for naught. One of the things I took away from years of commercial fishing is if you think you have an advantage over the next boat by using a pink glow stick rigged every 100 fathoms on a long line then do it. Knowing you're doing the best you can given your circumstances is good for peace of mind.