Question
Does anyone here do raised panel work on their CNC? I know that Thermwood sells a package for the purpose, but it seems too much of a hassle fixturing just one component at a time. I don’t actually own a CNC yet, just planning ahead for the future.
Forum Responses
From contributor N:
We do our own solid doors, both hard and soft timbers. I was reluctant to purchase the Lietz type full-on package, as our volume for in-house work is 50 doors a month, at most. We just use a modified hand router bit with the ball bearing guide removed bit its place a straight cutter for doing styles and rails. With arched heads I run with a straight cutter to get the shape first, which saves bandsawing.
For the center panel, we had a custom cutter made to the maximum diameter that would clear the other tools in the tool change at about $200. This is a standard carbide bit. We run the center panel cutter around the panel about 5 times, bringing it closer each pass. This puts less strain on cutters and lessens the chance of movement on our suction pads. More vacuum, the better for the smaller items.
Time-wise, about an hour per door for timber prep, assembly, light sand. So cost-wise, it may not be worth it, plus it does hog the machine for a good 1 1/2 days, which drives people mad. I do it because I have control over door color and quality. Also, I have the doors when I need them and not on my door suppliers schedule.
Also, if fixtured properly, your setup time becomes near zero. You can make doors in any order and number you want without having to justify the setup time for each machine. You simply cannot justify the time to set up the shaper just to do one raised panel, but a CNC can do a raised panel from rough blank to finished without any setup time. (And writing the program takes only a few minutes, and only the first time.)
Lastly, you save a whole lot of time on the milling of the blanks. As an example, the glued up panel for the center only needs to be sized for thickness when it comes to the router. You can then raise the panel and trim it to size and shape all in one operation.
The key to all of this is the fixture to hold the frame parts. I have helped many companies design simple shop built fixtures that can mill all four at once.
One argument for processing doors on a CNC is that the operator could actually glue the door frames together while the machine is cutting parts. This aspect could save a tremendous amount of labor. A door might take a few minutes to machine, if you have a fixture to run all five pieces at once, giving the operator time to set up the next door and possibly sand or assemble too.
To contributor B: Based on the original question, I was working on the assumption that the questioner's operation was not very large. If he is only making a limited number of doors for his own use, then investing in three dedicated shapers may not be practical, and many of my customers simply cannot justify that much floor space for dedicated machines. As always, different needs will present different solutions.
Someone mentions that programming is quick, but door sizes vary from job to job. Can you really reprogram that quickly? My next reservation is with the cost of good quality router bits. Do you really want to machine all sides of a panel? You get into a decent oak or maple job and that can be a lot of panels. How do you test your setup on the CNC? A couple of mis-cut and raised panels can cost you a lot of money and more importantly, time.
These setup, fixturing and programming times always seem to get underestimated. Anyone who has said I'll just use my computer to "save time" knows what I mean. Click, click, click of the mouse and pretty soon hours have gone by. There is a local shop in our area who bought a dedicated door machine (Unique, I believe) and it has been all but set aside in favor of the old shapers because even it is too slow!
Also, consider that once the fixture has been built and the program written and proved, the setup is done forever. Any time I need that set of parts again, all I need do is set the fixture onto the machine and call up the program. The fixtures are typically used for most sizes of doors, so I can literally run a door, call up a different program and run a different door, then go right back to my original program with no setup time in between.
Again, please understand, I am not saying that a CNC is the only or even the ultimate way to make 5 panel doors. What I am saying is that small and medium sized cabinet shops that want to make their own doors would do well to seriously consider doing it this way. To not explore ALL of their options is not in their own best interest.
I also love it when a mistake is made with incorrect door size, as we can fix it straight away and use part of the materials from the wrong door size. Whereas, if an incorrect door is ordered, I'm worried about the time factor, extra door cost plus if the timber will match. Here in Australia, one of our popular timbers is Jarrah, which looks a bit like rosewood. This timber ranges from light pink to burgundy red, depending on the soil structure. Thus, I get the colour request of not too light, not too red. Buying the doors in can be like Russian Roulette. However, that is my personal way of doing things for my small, 8-man shop.
Here's a tip for holding the rails and stiles in place. We use a 220 grit paper on a spoil board, then hold parts in position with fully concealed hinges. Use the door adjustment to hold them down. We remove the door cup part, so we have the levers as the hold down point.
As far as dedicated shapers are concerned, this assumes that you’re just using the same profile all the time. We don’t, so since the spindle and fence would have to be adjusted for each different job, it would virtually be like starting from scratch each time. Even in the case of using the same profile, the milling of arched rails necessitates the moving of the fence.
Here’s my dream scenario. Please tell me if I’m nuts.
Enter the door sizes in to a computer program. Using parametrics, the program would generate the code needed for all the different sized doors. The program would also automatically vary the height of drawer front rails, depending on how high the drawer front is (i.e. I’ll usually make my rails 1 1/2" if the drawer front is 6" high).
Straight line the material and S4S using the moulder, planning to exact thickness, but making each strip about 1/8” extra wide.
Upload the code to a Tiger Stop. The Tiger Stop would cross cut the blanks, let's say 1/4" long and print a bar code label for each. Each door would be designated by a large capital letter, which would be printed on the bar code label for later sorting of the door parts.
Use a spoil board with dedicated hold-downs for parts. Have another bar code label stuck to the spoil board alongside each hold-down.
Lay the parts down on the hold-downs in any order, (just taking care to machine the stiles and rails separately). Then, using a scanner gun, scan the parts bar code, followed by its hold-downs bar code, so the CNC knows which part it’s dealing with and where it is on the bed.
Start the cycle by first cross-cutting each piece to length with an aggregate saw. Follow this with taking 1/16" off one side of each part (to ensure straightness) and then doing the cope and pattern work.
A quick question about vacuum systems. One manufacturer told me that if I used his “rotary vane pump”, it would achieve a virtually perfect vacuum (29.9mm mercury), resulting in holding power of about 10 lbs. per square inch. Any comments? If need be, I wouldn’t object to having some sharpened screw points coming up though the spoil board a bit to take care of any lateral movement.
My Weeke software would certainly generate the parametric programs easily. The hardest stumbling block to overcome would be holding the stiles and rails firmly. Also, toss the saw and use the router for the entire format cut. This can be done chip-free with the right tooling and geometry.
The fixture board would require some setup as the parts change size. You would need some special gasketing or homemade pods to set up an efficient fixture board. You can't hold a 2" wide rail on a flow-through and expect it to stay put with a heavy profile rout.
You can change shaper cutters faster than sticking labels on all those parts. As far as resetting for other profiles, how could it be slower than reloading the toolchanger or holders on the router? Can you afford tool holders for all the different patterns of sticking and panel raisers? I assume this router would be processing your cabinet components also.
Our basic shaper setup is three machines. We rip 1/16 oversize and cut to length on the cut off saw (tiger stop would be great here). The coping is run right off the fence with no trimming to length on the shaper, and the center cutter removed. With all the parts coped that need it, we stick on a shaper using an outside fence and a chip breaker. The beauty of this method is no snipe on the front or back of your part and we now remove the extra 1/16 to remove saw marks. Setting the width of the cut is quick. The only fixturing we reuse is the cope fences that we have plunged the cutters through, and this becomes the basis of setting up the cutters if we go to another profile. Arched work is definitely a little more involved, but all the rub bearings stay on the machine as they do not effect the way we shape the straight stock. I have the luxury of about 9 shapers, so we can spread the setups out, but the three basics machines are still at the core.
Contributor N questioned the cost of shapers, but I think if you compared to what tooling, hold downs, and other custom fixturing for the router cost, you can get a lot of shaper(s).
I think you get caught in the situation of the router is fast enough for the custom shop but too expensive, and too slow for production shops.
We had a sliding table shaper setup to do end grain cuts on drawer fronts and doors on the slider side and the long cut with the grain on the other side of a homemade double-sided fence. Two cutters were mounted on the same spindle and it was raised or lowered to change profiles. Fence inserts were used because the two cutters were different diameters. The shapers were all in a cluster to keep handling to a minimum.
The only trick I see is preserving a setup with reference devices (go/no go gages). Easily done! For light work, just use sealed wood; aluminum for production work. Done right, there is no need to widebelt the doors, they come out flush. Unless you buy a very expensive widebelt, the extra time it takes to get the cross grain scratches out is better spent on good setup. No CNC needed and you can make as many doors as you like any time, little or no set up, might even start making for other shops.
And yes, you can produce doors a lot cheaper than purchasing them. Plus, how do you tell your customers "Oh yeah, I just buy your doors, I don't really make them"?
Our company has been researching CNCs for about three months now. We have talked to most and even traveled to their plants to see as much as we can before we spend all that cash. Everyone must do the research first.
To use the shapers as you have described, the stock MUST be S4S prior to reaching the shaper. Again, if your shop already has this process in operation, then the shaper is a good way to go. But if your shop is small and does a lot of custom stuff, the CNC has advantages.
Using the CNC, I can work with S2S stock, only needing it sized for thickness. Otherwise, the stock only needs to be rough cut to approximate size.
For a smaller shop with limited space and budget, this is a big savings. Otherwise, in order to cost-effectively mill the doors in-house, a moulder is also needed. A basic moulder to do nothing better than S4S will add an additional $20K to the project.
I am currently setting up a demo on doing 5 part doors on a 3 axis CNC. I will be milling the inside edges of the stiles and rails and shaping the panels in one operation. The door is then assembled and reset onto the machine where all outside edges are finished. This procedure means that a lot less time and effort are required in the assembly phase, as any joints that don't perfectly line up get milled off flat.
