Quality Control in Cabinet Door Fabrication
This conversation starts with a question about adhesive choices for frame and panel door construction, but broadens out into a discussion about the acceptable incidence of defects across a range of door quality measures. December 9, 2013
Question
I have always used yellow glue (TB2 or 3) and Bessey K-Body clamps to assemble standard five piece cabinet doors, but am looking for ways to speed this up. Something like a JLT clamp would be great, maybe in conjunction with a reactive polyurethane hotmelt adhesive such as 3M's PUR system. I'm wondering if that sort of adhesive might give me reduced (or eliminated) squeeze-out, and I like the two minute cure time they claim for their ultra-low viscosity formula (which they show being used for exactly this purpose). Has anyone used this? Or something other than yellow glue and had better results overall? How do the huge door makers such as Walzcraft assemble their doors?
Forum Responses
(Cabinetmaking Forum)
From contributor X:
I use a Ritter clamp, yellow glue and pin the joints. I just stack them up as they get pinned so there is no drying time to worry about.
From the original questioner:
Pinning will help, but I would still let the glue set for maybe five minutes. I'm afraid those 23 gauge pins wouldn't hold the joint tight enough until the glue grabs. Maybe I need to reduce the amount of glue I'm using. I pour some Tightbond into a plastic cup and use a little acid brush to paint some glue into each cope. I get a little squeeze-out on the profile maybe to 2/3 of the time, which has to be completely cleaned. I'm not saying it takes a lot of time, but it is boring and tedious and makes me want a faster method. Should I be less afraid of joint failure and just use less glue? I would love to see how the big door shops do it, from assembly to final trimming to size. The squeeze-out cleaning and final sizing stages are the bottleneck for me.
From contributor B:
The big door companies plan on a certain amount of joint failure. Can you afford to do the same? Sometimes you don't discover the glue starvation problem until the door has been finished and is on site through a humidity change or two. The damage at that point also includes your goodwill in the community. I would guess that the costs of repairing one door in the field would be equal to maybe the time you spend cleaning glue out of 1000 doors?
From contributor X:
I say trust the pins and get the hang of putting in the right amount of glue - that system is as good as any. I unclamp right after pinning and I have never had a joint failure.
From the original questioner:
Contributor X - you've never had a tiny glue line open up from taking pressure off right away? I'm thinking it wouldn't be an issue with paint grade, but with clear maple or cherry (some tight grained wood), even a slight opening of the joint would be an issue.
From contributor G:
Why so concerned with the squeeze-out? Sand it off with the wide-belt.
From the original questioner:
I meant squeeze-out on the profile, not on the flat. I've been keeping the glue back from the profile and a lot of the time I don't get any on the profile, but not all the time. If I let it dry completely it's rock hard and difficult to clean out neatly.
From the original questioner:
While the overall quality level of most outsourced doors does not compare very favorably with doors I might make myself, for the price they are impressive. All the time I hear shop owners saying, "Oh, I can't possibly make the doors for what I can get them for from a door shop." Well, how is the door shop getting the price down? At the high production end I'm sure there are tips for getting assembly and squaring down to a bare minimum of seconds. I thought maybe they're using a reactive hot glue such as PUR to even eliminate the pins (I didn't see any pins in the last Walzcraft batch of cherry doors I ordered).
Though my original question was just about glue, I'm also interested in squaring methods. Walzcraft told me they were CNC squared. Not sure what that means. I doubt it's a CNC table router. Maybe they have a CNC beam saw with some way to very quickly load and cycle a door.
From contributor L:
We have the reactive polyurethane hotmelt adhesive system from 3M. I sure wouldn't use it for doors. Have you priced the adhesive? If your joints are sized properly original Titebond will seize in a couple of minutes. No need for pins. With two clamping stations you can work non-stop. Holding the glue back a bit on the cope works fine. A little squeeze-out on the top seals the end grain and keeps the stain from making it a dark line. Let it get to the rubbery stage and peel off. Don't sand until the next day. No need to wide belt if you machine your parts correctly. Orbiting cross grain scratches out is a pain/time killer, assuming you don't have a high end sanding system. There are several clamping systems available that square the work by pushing it into a square corner. With a bit of ingenuity you could build yourself one - Destaco clamps or air cylinders.
From the original questioner:
I've got my shaper setup working really well, so my fit is excellent between cope and profile. I'm seldom out of flush top/bottom by more than .005" or so (I would guess). Perfect friction fit for yellow glue too. I hadn't thought of eliminating the wide belt entirely, but maybe I should try this, or use it only for a final platen sand with 180g. Currently I'm taking off 1/64" top and bottom at the wide belt.
Space and money rule out two clamp setups for the foreseeable future. I've been thinking of engineering a glue-up table similar to what you suggest, approximately 5' x 11' in size, usually allowing two doors at the same time given the length. Glue-up table on one side, vacuum frame press on the other, and the whole thing flips on pivots on either end. Box beam construction of baltic birch, LVL or similar. I'm sold on sticking with yellow glue.
From contributor M:
Why not just get an RF (Radio frequency) glue drying unit? They are priced well enough and most large manufacturers use them. It has been discussed in the Knowledge Base.
From contributor J:
I also use a Ritter clamping table but I don't even bother with pinning the doors. Two doors can be clamped. By the time I have the third ready, the first glue-up is set enough to take out of the clamps. I’ve never had a failure and the glue line also does not open up.
From contributor R:
You may be missing a point here. TB2 and TB3 are not better versions of TB. They are different glues for different jobs. Use original Titebond or better yet get some super Titebond. You won't go back. Learn what the other glues are for and only use them when you have a reason. They are not better, just different! If you have been using 2 and 3 for edge glue-ups you really made a rookie mistake.
From contributor X:
To contributor R: What is the difference between super Titebond and standard Titebond?
From contributor R:
It sets up faster and stronger earlier.It's not available in less than 5 gallons but that’s not much if you are gluing daily. We used super in my shop and loved it.
From contributor F:
I'm not sold that the big door manufacturers are improving the quality of their doors by speeding things up. I just re-finished about 1/2 a kitchen’s worth of Kraft Maid doors which were a pigmented finish. Every single door and drawer had noticeable and not very attractive lines at the joints. I don't mean barely visible in the right light kind of lines, I mean you could almost see them from across the room kind of lines! I don't know the type of glue they use but I am guessing it's something that allows super-fast production. Unfortunately it also allows the seam to open whether through glue creep or something else. Just putting this out there as an observation. I know I have doors I made ten years ago with plain old Titebond I that still look just as good as when I made them.
From contributor B:
Something that has always amazed me about door manufacturers is their ability to completely ignore basic principles of woodworking. Even a hardwood flooring installer will bring material onsite early in order to let it acclimate for a few days before it is nailed down to a plywood substrate. That door manufacturers somehow think they can go from tree to door in 15 minutes without any time for stabilization is, I think, one of those examples of "we'll deal with that warp if it becomes a problem, it would cost too much money to do it like the flooring guys". Would that be an example of engineering in some risk of failure?
From Contributor U:
There are plenty of small to mid-sized door companies that are doing it right. We put out 250 ply panel doors per week and about half that many raised panel with two-three people. It is very unusual for us to get defective doors returned for remake (and I will personally go pick them up and replace the next day, so there is no incentive not to call).
Typical in-process time is three days. Day 1: rip everything required and glue any solid panels (the rips can move a bit overnight). Day 2: S4S and profile rails and stiles, plane and shape panels, assemble doors. Day 3: widebelt, edge sand/profile, ROS if required by customer.
As far as glue-up goes, I second the TB1 or the higher solid super formula. They set up fast. We have a 5-station rotary door clamp, so they get about 8-10 minutes in the clamp, no pins. We used to have a single clamp and used a WorkRite WoodWelder to zap the joints - five-ten seconds per joint and the door was ready to come out by the time the next one was glued. A tiny bit of squeeze-out definitely helps with the end grain, but you need to wait at least a day before sanding to avoid shrunken glue lines. If you are getting glue in the inside corners of the profile, then you are using too much in the wrong place. We just use a squeeze bottle with small orifice tip. On sanding, I personally find that going to 220 and then backing off to 180 for the ROS is very fast and consistent.
From the original questioner:
I use TB3 because I like the working qualities a lot and am also reassured by the water resistance. But mostly I like the viscosity and how it lays out, just the right amount of open time, etc., (and if that's a rookie mistake I know several fellow rookies with double my 13 year experience who agree.) If memory serves me, last time I used TB1 I found it too thick and sticky for my liking, but maybe I'm wrong. It's been a while. I used to use ProBond for site carpentry because it's really thick and drips less. TB3 I find a little too runny. I've used a few other industrial mail-order yellow and white glues, not preferring any of them over TB3. In any case, I'll look into the glue suggested - thanks.
From contributor B:
The hardwood flooring people use that same warehouse for storing their material. The only difference is that their material is usually milled closer to final dimension than yours. You still have to break wide boards into skinny ones and crosscut long sticks into short ones. This is usually the first time those wood fibers have seen daylight in their entire existence.
The hardwood flooring is going to get nailed to plywood. A cabinet door has five elements and only one of them is secured to anything. If you had hardwood flooring installed at your house would you insist that they let it acclimate first or would you just let all that air drying happen at the warehouse? Most door companies have a caveat in their literature that stipulates "A warp or twist of 3/16" is not considered a defect". Does your literature include anything like this?
From the original questioner:
I would say 3/16" is an enormous twist for a typical sized cabinet door. I would remake it before delivering it and if it happened on site I would replace it without being asked by the customer.
So the gist of the responses seems to be that there's no mysterious ultrafast high-end or high-dollar methods for assembly, and even the really big shops just use big multi-station clamp tables or racks that automatically square and clamp. Maybe they use RF dryers, or more likely just pin them and remove them almost immediately, or they have enough clamp stations that they can afford to wait five minutes per station for the glue to grab. Is this about the size of things as far as anyone knows? If this is the case, I will stick with my plan of a JLT or Ritter table and depending on how rushed I am, use pins or wait about five minutes per door and not shoot the pins. (I haven't noticed any ill effects with the wide belt). I guess this is as fast as one can get without a really big outlay and more floor space than I have as a one plus man shop?
Now let's argue about the best method to square the door to final size after glue-up. I'm really liking CNC'd 1/4" MDF templates with a flush cutting setup using a high fence on the slider. I completed 80 plus doors the other day in 4.75 hours including the time to CNC the dummy panels. I can't find a way to do it faster or more accurately. I know someone who does the same thing but prefers the shaper instead of the slider - less edge sanding. True, but I still prefer the slider.
From contributor L:
Just how deep does a board air dry in a few days? This assumes that you sticker all your lumber. Letting the parts air dry a few days is a bad idea when it comes to gluing. I've been in big plants where they make tons of parts and put them in bins for later assembly, bad idea too. Piles of parts not moving is a sign of sloppy management. We don't often need raised panel doors but when we do we use Conestoga and get them pre-finished. Really nice finish! For dovetail drawers we use Valen pre-finished parts flat packed and have always been very nicely finished and fit together nicely. About the idea that using a water resistant or proof glue is somehow better, lots of other things will fail first if the door gets wet enough for original Titebond to fail.
From the original questioner:
Yes, I agree that there's little if any actual value in the increased water resistance of type 3 for this application. Mostly just makes me feel warm and fuzzy inside.
From contributor Y:
To the original questioner: I don't understand your door squaring system with templates. How does it work?
From the original questioner:
Check out images of the "Aigner Sawboy". You can make your own simplified version that works just as well for this purpose. Basically you're using the slider to cut a projecting edge (the oversized cabinet door) flush with a final-sized template (the 1/4" MDF CNC'd to exact size).
From contributor Y:
When we started using our JLT door clamp we discovered we could build the doors to size and just bump them on the edge sander top and bottom. The stile edges were already run through the widebelt and sanded before assembly. Even works on inset doors.
From contributor X:
I'm doing the same thing as Contributor G. All you have to do is bump the end grains and you’re done! I use a Ritter clamping table and the doors come out perfectly square every time.
From Contributor U:
I’ll third gluing up exactly to width and just a tiny bit long. Just keep glue off the stiles.
From contributor D:
I am a small cabinet shop to but I outsource my doors because of the cost and the quality of the local big box door company is very good. However, I have made some doors in my day. The problem I found was I don't own enough clamps to glue up panels and glue up the rails and stiles for even one job needing 30 doors or so. I used yellow glue because you can take a damp cloth and wipe off the squeeze-out near the profile. Good glue coverage in key because the wood will shrink and leave a crack at that point and potential for a call back.
