View Full Version : Cantilevered routers and other ideas...

05-21-2003, 10:55 PM
Well, I'm slowly but surely gathering up parts, and finalizing my design for my router. I'm working on the Z axis, and various subassemblies right now.


All of the designs I've seen on this forum show a Z axis supporting a router, and they all seem to be cantilevered. The router is on one side, and the supports are off to the other side...analyzing the forces, it's a short cantilevered arm!

Since a router's weight is non-negligible, that means that lateral stress is going to be put on the bearings and guides. This should reduce their useful life, and wear them out faster than they should be. (Of course, cutting drag also introduces lateral forces, but those can't be eliminated very easily)

Has anyone addresed this "problem?" In a real machine, is it even a real problem?

I've had the inspiration to place my Y axis rails next to each other on a horizontal plane, and place the entire weight of the Z axis and router in their center. This will be a relatively simple design to implement, and will eliminate any lateral stresses involved with that axis.

I could conceivably do the same basic thing with the Z axis, mounting it such that the router's center of mass falls between the rails, but it would complicate the design a little bit...just trying to figure out if it's worth the extra trouble.

Of course, by doing this, my drive screw is going to be moved off center -- if the fit on the linear bearings is tight enough, it WILL move straight, but again there will be lateral forces involved. Has anyone come up with a simple design which overcomes this?

I could do it by making my Z axis longer and putting the drive screw in line with the cutting tool, probably above it -- this would keep everything aligned. Another obvious way is to mount the router *inside* the screw drive itself, but that'd require an enormously large diameter screw, and would likely introduce other problems, too. But, something like that could be implemented with relative ease in large diameter PVC... Hmmm...I might have to play with that idea.


Anti-backlash nut design. I think I've come up with a simple design...but it seems too simple. Does anyone see a problem with this basic design?

Premise: The drag on a nut is directly proportional to the area of contact -- in other words, as little of the nut should touch the screw, as is possible, to minimize drag.

Solution: PVC pipe fittings. Internally (female) threaded cap and an externally (male) threaded plug that screw together. The fittings for SCH40 are fairly thick, so they have some structural integrity, but this arrangement results in a hollow box, and only the ends will be in contact with the lead screw.

So, you take this assembly, drill a hole in both ends, and tap them to accept the lead screw -- screw the fittings together, or apart to dial out any backlash. Since the thread pitch of the fittings is different than the pitch of the lead screw, the "nut" will put pressure on the lead screw's threads. Voila...we have adjustability and minimal surface for contact. Should produce a cheap, homebuildable (by anyone) anti-backlash nut. If it slips, it's easy to add a set screw to the outer half, to hold it in place.

I look forward to your comments.

-- Chuck Knight

05-21-2003, 11:09 PM
Hi Chuck,

Your idea for an anti-backlash device is likely a good one. I would caution that too thin of a nut will wear out pretty quickly, because you've got too much unit pressure on the bearing sides of the threads. Most drag is not caused by the nut itself, but by how much pressure is required to move the slide. The nut is merely the fulcrum of the helix. The more thread you have, the lower the unit pressure, and the longer the nut and screw last without premature wear. Excessive length would create drag depending on the viscosity of the lubricant on the threads, that seems to me would be about the only source of drag. Of course, the threads should be smooth, too.

05-21-2003, 11:24 PM
If wear is a concern, instead of tapping the PVC, use a little epoxy and off the shelf steel nuts on each end -- should provide a tougher surface, and as strong as the epoxies are today, that won't be the weak point of the assembly. The PVC is there only to act as the adjustment device...and one that *anyone* can build. My thought is that it could be tapped more accurately, and provide a smaller surface area.

(Or would an alternate metal be better for the threads? Brass and aluminum, for example, are used extensively in the helical focussing mechanisms of cameras, because of their extremely low friction when used together. Steel and brass are used in clockmaking for similar reasons, and the brass gears don't show undue wear, despite the sliding friction present between them and the steel gear teeth.)

Now...how about the Z-axis ideas? I'm working on some final designs, and would really appreciate some feedback on the Z axis design ideas. The cantilevering of the router support just "rubs me the wrong way."

-- Chuck Knight

05-21-2003, 11:35 PM
I don't see any reason why your Z axis plans couldn't be superior for the reasons you stated. Overhanging loads do tend to wear the slides in a particular fashion. The lateral force on the screw is not a major concern, if the slideways are all nice and tight, and long enough to prevent slip-stick locking. If you had the vertical room for it, you could still center your screw drive above the router head, could you not?

Can you make the spindle easily accessible for the tool changes? That would be my primary concern.

05-21-2003, 11:49 PM
This image is a perfect example of what I'm talking about. This *is* a cantilevered arm...you can see where the wear should and will take place. *This* is what I'd like to avoid -- *IF* it is actually a problem in practice. The solutions I've come up with will work, but they add complexity and difficulty to the design.

HuFlungDung says it best in his .sig file...theory and practice aren't always the same thing. Theory says this cantilevered design is a problem... Since I've not yet built my machine, I don't know what practice has to say.

-- Chuck Knight

05-22-2003, 07:57 AM
Seems to me that the design above ( very similar to the machine Iím building) introduces primarily radial loads to the bearings. Iím not an engineer, but I would think that in a round shaft design there would be no lateral loading on the shaft.
As for your Y axis design, Shopbot makes a router with a similar Y axis. From what I have read on the Shopbot forum, there have been some issues with poor torque stiffness in the Shopbot Y axis, but this may be a function of the components used in the Shopbot rather than design.

05-22-2003, 09:26 AM
OK. Real world scenario. At work I run a Biesse Rover 335 CNC PTP router (a cross between a point-to-point borer and a CNC router). The gantry design is cantilevered and depends on having the base of its two supporting rails vertically spaced about 47cm (18 1/2in) apart along the back of the machine. The rails support a gantry which carries 24 vertical / 8 horizontal drilling heads as well as 3 x 7.5HP HSD high speed spindles. In use this cantilevered "arm" is stable and doesn't dip towards the table at the furthest reach of Y-axis (950mm), probably because the two side members are 9 x 3-1/4in channel section steel sitting on a cast-steel Z-axis carraige . The Y-axis carraige baerings are approximately 43 cm (17 in) apart in the Z-axis direction and 64 cm (25 in) apart in the Y axis direction. The Z-axis slides are a two part affair, pneumatically driven (to drop the head to its "zero" position) and servo driven (from the "zero" position to cutting depth). The spindles are bolted onto aluminium face plates some 10mm thick and these in turn have THK bearings running on linear tracks (two of them) with the ball screw in between. Our unit is supporting considerably more weight than a small unit like yours and bearing wear isn't an issue. Lubrication, however, is - we relube twice a week, religiously, with the correct red EP grease and NEVER change manufacturer. I posted this info to give you an idea how full-size machines tackle this problem in the real world.

Having owned a ShopBot in the past I was sceptical when we bought this machine that it would be ridgid enough (ShopBots flex not a little, I believe because the Y-axis gantry is just not ridgid enough). Incidentally, there was a CNC manufacturer here in the UK a few years ago who used to build machines with a "wide base gantry" (rather like the ShopBot) which used a two rail carraige to hold its spindle heads. The motor was bolted onto a chassis which was sort of U-section with a guide rail and bearing sets at each end of the "U".

Hope the above info is of some assistance.

05-22-2003, 10:45 AM

To prevent that problem, full size cnc machines have counter weigths attached tothe back of the "z" axis. Similiars to the ones in an elevator.

05-24-2003, 06:41 AM
Then no one has told either Biesse or Morbidelli about that! The technique used on many of their machines (including ours) is to use an "air cylinder". This uses compressed air to act like a damper on Z-axis movement and assists in raising the Z-axis back to its upper limit stop. I have also seen this system used on some of the older SCMs and Wadkins. I reckon counterweights would suffer from many problems. Firstly you are adding to the weight caried by the Y-axis gantry, secondly you are going to have to secure them in slides to prevent them moving around when the machine is in motion (and ours will happily cut in X- or Y- directions at 14m/minute with jogging at about 60m/min). I think that the speeds at which woodworking CNCs accellerate depending on gravity might be just a little too slow, unless the weight was enormous. ;)