Torsion box router with a 4th axis.

[TheWire]

After taking in suggestions from forum members, here is my lastest gantry design. It would consist layers of 12" x 44" MDO plywood laminated together to end up with a beam that is 12" high about 3.5" thick. The bottom 3 inches would have the center sawed out of it, but turn down at the ends to meet the linear carriages. I'm thinking the sides (green) would be 3/8" aluminum plate. This design should be very resistant to any gantry side to side motion.

In the interest of time, I didn't put the details of the ball screw supports in yet.

Also, instead of sawing a notch in the beam for clearance for the X screw behind the spindle, I'm looking at putting the X ball screw on top of the gantry beam. My concern is pulling the Z axis along from above both X rails instead of between them. But numerous ball screw and R&P commercial and DIY routers have the axis driven from the top of the gantry beam. I'm looking for advice as to if this could prove to be problematic.

Thank you,
Jerry.

[peteeng]

Hi Jerry - If the bearing pattern is geometrically stable technically the saddle can be driven from anywhere with square bearings (its different with bushes). Its conventional if possible to drive a saddle or component through the friction center of the bearing system but ball cars have very little friction and have no stick slip so the requirement to drive thru the friction center is not that important. For instance you are driving the gantry miles away from the friction center and it will be fine because of the rolling balls in the car. Try doing it with plastic or metal bushes and it will be a disaster.

Also by putting the gantry drive on top you have sidestepped the issue of the mismatch in stack hts of the rail+car and the ball screw system. Well done; Now I just have to convince you to get rid of the construction extrusion on the Y axis. You are bolting a not so stiff extrusion to a not so stiff thin wood. Again if you remove the extrusion it will be much stiffer, less parts and connections. You got the torsion box flat so I'm sure you can get the wall flat.

Having the rear wall removable is a must. I did a machine with a rear wall that was permanent and I have had two occasions where I want to put something thru it and I can't

What accuracy are you aiming at with this machine? Regards Peter

[ger21]

2 LVL on their sides 0.008" Surprising as this is the way they are supposed to be strongest.

Strength and stiffness are two different things. Stiffness is the more important in this application.

I've been too busy to read much in the last few weeks, or I would have posted more in this thread.
If you haven't seen it, you may want to read through my very long thread on building a large wood construction router.

My 68" gantry beam will be an MDF torsion box, roughly 8"x8", and 68" long. To make it stiffer, the MDF panels are laminations of 1/4" MDF.
Calculations using Beamboy show a deflection of about .001" with a 200 lb load on it.

Anywhere on my machine where I will be attaching metal to wood, I epoxy phenolic plates to the wood, and machine it flat. This minimizes the wood compressing when bolting metal parts down. It also gives me perfectly flat and straight surfaces for mounting the linear rails to.
Of course, it's much easier to do this when you have access to a large industrial CNC router.
Here's a link to my parts for my beam, which I cut many years ago. I should be starting assembly this winter, once I get the garage in my new house organized.
https://www.cnczone.com/forums/cnc-w...ml#post1168468

[TheWire]

I made progress on the base this weekend and got the side walls on. The walls are 4 layers of 1/2" MDO plywood glued & screwed. I indicated the height of the walls from the top of the torsion box as I was gluing & screwing them on and had things within 0.005". Its the first wood working project I ever used a dial indicator on . After the glue dried and I painted it with 3 coats of paint, I have about 0.015" of variance in some spots. As I'm hoping to mill wood & aluminum, I'd like to get the variance with 0.002" or higher if possible. Input of how realistic this is welcome.

The torsion box came out very flat. Still, I want to ensure I have an accurate reference in which to dial in the machine as it goes up. My plan is to pour self leveling epoxy on the surface of the torsion box to create very level work platform. Tentatively, I would shim the aluminum extrusions on top of the walls to match them to the epoxy.

There were concerns raised about the 1.5" x 1.5" 80/20 extrusions at the top of the walls leading to a lost of ridgidity. I realize this maybe the case and I'm using as thick of extrusion as I could find. However, I feel I need some method of adusting out that last several thou and to provide an interface between the wood and the linear rail. This is my 1st build and I'm willing to step back and make changes if needed. If the 80/20 becomes an issue, I can fill it with epoxy once its dialed in or remove it and fill in the notch with another more suitable material on which to mount the rails.

This design continues to improve with much-appreciated input from members here.

Thank you,
Jerry

[TheWire]

Hi Jerry - If the bearing pattern is geometrically stable technically the saddle can be driven from anywhere with square bearings (its different with bushes). Its conventional if possible to drive a saddle or component through the friction center of the bearing system but ball cars have very little friction and have no stick slip so the requirement to drive thru the friction center is not that important. For instance you are driving the gantry miles away from the friction center and it will be fine because of the rolling balls in the car. Try doing it with plastic or metal bushes and it will be a disaster.

Also by putting the gantry drive on top you have sidestepped the issue of the mismatch in stack hts of the rail+car and the ball screw system. Well done; Now I just have to convince you to get rid of the construction extrusion on the Y axis. You are bolting a not so stiff extrusion to a not so stiff thin wood. Again if you remove the extrusion it will be much stiffer, less parts and connections. You got the torsion box flat so I'm sure you can get the wall flat.

Having the rear wall removable is a must. I did a machine with a rear wall that was permanent and I have had two occasions where I want to put something thru it and I can't

What accuracy are you aiming at with this machine? Regards Peter

Pete,

Thank you for the heads up on the removable wall. With it removed the router could handle 3/4 sheets of plywood. Have you ever done work where you need to move the workpiece as the router can't reach the extents of the workpiece? It's good to see you think the ballscrew on top of the gantry beam should work will linear rails and bearings. If it doesn't I can still cut that notch in the gantry beam and place the screw between the linear X rails.

I mentioned the aluminum extrusion in the build progress post I just made. I realize it may become a loss of rigidity and I'm fine with changing it out if it becomes a problem. But right now, I don't want to give up my security blanket of adjustability . I'd like to get 0.002" accuracy. How realistic do you think that is?

Thank you for your help,

Jerry

[TheWire]

Strength and stiffness are two different things. Stiffness is the more important in this application.

I've been too busy to read much in the last few weeks, or I would have posted more in this thread.
If you haven't seen it, you may want to read through my very long thread on building a large wood construction router.

My 68" gantry beam will be an MDF torsion box, roughly 8"x8", and 68" long. To make it stiffer, the MDF panels are laminations of 1/4" MDF.
Calculations using Beamboy show a deflection of about .001" with a 200 lb load on it.

Anywhere on my machine where I will be attaching metal to wood, I epoxy phenolic plates to the wood, and machine it flat. This minimizes the wood compressing when bolting metal parts down. It also gives me perfectly flat and straight surfaces for mounting the linear rails to.
Of course, it's much easier to do this when you have access to a large industrial CNC router.
Here's a link to my parts for my beam, which I cut many years ago. I should be starting assembly this winter, once I get the garage in my new house organized.
https://www.cnczone.com/forums/cnc-w...ml#post1168468

Gerry,

Thank you for the input. I downloaded the Beamboy program. I Input some LVL info in there, but my current plan is to build by gantry beam by laminating up layers of MDO plywood to 3.5" thick. I'll have to look up what to feed into Beamboy if there are parallel members/laminations. BTW, have you looked at using MDO for your build. It costs twice as much as MDF but its just as easy to work with and it's moisture resistant.

The wood to rail interface has been something I've pondered quite a bit. There are some wood builds that setup dams & mini aqueducts across both rails and pour in self-leveling epoxy. But it seems many of them end up having to sand down the meniscus that forms with thin pour widths. I like your idea of gluing on phenolic plates and machining them in place, but as you said that requires an industrial-grade machine to do that. If my extrusion idea doesn't work, I could see if I can locate big mill or accurate large scale router to do that on my build. I think it would be a fairly quick operation.

Thank you, Jerry

[ger21]

BTW, have you looked at using MDO for your build.

No.
Most of the material I used was free, as it was leftover scraps I got at work. The side rails are laminations of 10ft long fir plywood.
The main table is a baltic birch torsion box.

[peteeng]

Hi Jerry - By 0.002" (0.05mm) do you mean that's the absolute accuracy or do you mean +/-0.002". Adjustability is a two edged sword better to make parts to the required accuracy (or as accurate as you can) then try to wiggle it out at the end (Imagine assembling a car engine with adjustability?) . What grade MDO are you using? Structural building products have their mechanical properties specified quite well. The issue with engineering plywood is that being a laminate the 90deg plies to the strain are ineffective. This is designed via proper laminate theory (complicated unless you have correct software) or by averaging or understanding the layered inertia of the laminate.

Also what are you going to coat the timber in to stop moisture ingress egress? Yes I have indexed a sheet through my machine. I set up a fence that is square to the axis then set up a zero that is in both segments. Then cut the first segment, slide the piece through to the next index run the second segment...Peter

attachment is in metric

Most beam calculators are for isotropic materials. In iso materials there is a relationship between bending modulus (youngs mod) and shear modulus G. In timber this is not the case. The shear modulus is quite a bit lower then the iso relationship. This means if you use such formulas it will under predict the torsional stiffness of a timber beam. Take F22 LVL its bending modulus is 16000Mpa nd its shear mod is 800Mpa the iso prediction is 11200Mpa for its shear mod. So an iso formula will under predict the torsional deflection by 800/11000= 7% by a long way. I liked your test beam approach its real, you don't have to test full scale either. You can scale with the math...cheers Peter

[peteeng]

Hi All - Correction. The equation I pulled off the net for shear mod was wrong. It was fishy so I looked it up correctly. The correct eqn is E=2G(1+v). This means a material with E=16000Mpa (eg F22 LVL) should have a G= 6100MPa. 800/6100=13% the shear stiffness not 7%. Now a beam such as a gantry has bending deflection and shear deflection. In metals the shear stiffness is very high and the shear deflection is commonly neglected as its a second order property. But in timber it should be included due to its low G. So a physical test is very useful if your not up for the math.

if you look for a calculator that allows you to include the G then its good for gantry design. Calculating the torsional deflection of a beam is not easy especially if it has a high aspect ratio. Even FE systems get this wrong with some beam elements....Have to use plate or solids to get it right. Regards Peter

[TheWire]

Hi Jerry - By 0.002" (0.05mm) do you mean that's the absolute accuracy or do you mean +/-0.002". Adjustability is a two edged sword better to make parts to the required accuracy (or as accurate as you can) then try to wiggle it out at the end (Imagine assembling a car engine with adjustability?) .

Hi Pete,
I'm thinking +/- 0.002 inches but this is my 1st build so I'm not sure what to expect. I'm guessing that would be fine for wood carving but I'm not sure about what I'll need in the aluminum parts I make. Such as plates for more routers or to convert my mill and lathe into CNC.

What grade MDO are you using? Structural building products have their mechanical properties specified quite well. The issue with engineering plywood is that being a laminate the 90deg plies to the strain are ineffective. This is designed via proper laminate theory (complicated unless you have correct software) or by averaging or understanding the layered inertia of the laminate.

Here is a link to the MDO I'm using. https://www.roseburg.com/Product/duragard-premium-mdo/. Not a lot technical data other than "APA - Manufactured to meet or exceed APA — The Engineered Wood Association performance standards PS 1-09"

I think what you are saying is that, by laminating the plywood in layers from the front to the back of the router, the force from the spindle is pushing against the plywood in a direction that it is out of line with the plywood's strongest direction? I haven't purchased the wood yet so I could change it. I was thinking of laminating 2 LVLs or a Glulam beam as alternates. I could also make a box beam, but it seemed just as easy to laminate plywood instead of building a box beam. Do you have other suggestions for wood gantry material? Would metal skins on the beam help?

Also what are you going to coat the timber in to stop moisture ingress egress? Yes I have indexed a sheet through my machine. I set up a fence that is square to the axis then set up a zero that is in both segments. Then cut the first segment, slide the piece through to the next index run the second segment...Peter

I will have 3 coats of Zinsser® FastPrime™ 2 Interior Water-Base Primer & Sealer on the torsion box, so probably the same for the beam. I'm also planning on pouring self leveling epoxy on the bed and covering all but the bottom and the bed, including the gantry beam, with high pressure plastic laminate (counter top material) so I'm hoping it will be protected from moisture variations. The MDO plywood also has a phenolic resin glue and is said to have good dimensional stability in regards to moisture variations.

attachment is in metric

Most beam calculators are for isotropic materials. In iso materials there is a relationship between bending modulus (youngs mod) and shear modulus G. In timber this is not the case. The shear modulus is quite a bit lower then the iso relationship. This means if you use such formulas it will under predict the torsional stiffness of a timber beam. Take F22 LVL its bending modulus is 16000Mpa nd its shear mod is 800Mpa the iso prediction is 11200Mpa for its shear mod. So an iso formula will under predict the torsional deflection by 800/11000= 7% by a long way. I liked your test beam approach its real, you don't have to test full scale either. You can scale with the math...cheers Peter

I might just make up whatever beam we feel is best and test it. It seems to me 7 glued layers of 1/2" plywood would be pretty stiff.

Thank you again!,
Jerry

[peteeng]

Hi Jerry - To actually achieve +/-0.05mm +/-0.002" means you have to design to achieve 1/10 of that. Means buying and making very precise parts. I expect you will achieve better then +/-0.1mm easily. +/-0.004" so you will see when built. Its got to be very stiff to get to +/-0.01mm. Stiffness and strength are different you need to get your head into that. Where your going is fine, beam calculators underestimate timber stiffness especially in torsion. Gantries are shear stiffness dominant at the ends and beam calculators can't figure this. I'll keep the engineering simple. Do a physical test include some torque, do it small scale. When you use latex paints I think their called in USA (we call them acrylic paints) dilute them with water so they penetrate better. I use 2parts paint to one part water. Can use more water. Especially on end grain. cheers Peter

[ger21]

I could also make a box beam, but it seemed just as easy to laminate plywood instead of building a box beam.

The box will be more rigid.

It seems to me 7 glued layers of 1/2" plywood would be pretty stiff.

Not as stiff as a torsion box beam, at least in one direction.

[pippin88]

Don't epoxy the whole bed. Waste of $$$

You can just cut it flat once you have your linear rails flat / adjusted.

[peteeng]

Hi Jerry - If you are going to laminate plywood together consider using some aluminium sheet on it. Say 3mm (1/8") on the outsides. This will create a sandwich which will be very stiff. Calculation shows it will double the beam stiffness in the short direction (on your 3.5" deep beam). ie the skins are the same stiffness as the timber middle bit.

Some 6061-T6 sheet will be very hard so takes bearing loads of bolted bits to it stops water ingress and you don't have to paint! Maybe some metal polish to brighten it up. To double the beam stiffness in timber it would need to be about 112mm wide vs the 93mm. That's why wide is good. Peter

Re-box beam - If you made the same beam as a box using 12mm 1/2" ply it would be 64% the stiffness of the solid. I'd go solid.

[ger21]

Hi Jerry - If you are going to laminate plywood together consider using some aluminium sheet on it. Say 3mm (1/8") on the outsides. This will create a sandwich which will be very stiff.

It basically becomes a torsion box with the plywood core taking the place of the ribs.

Re-box beam - If you made the same beam as a box using 12mm 1/2" ply it would be 64% the stiffness of the solid. I'd go solid.

If you build the box with ribs every 4"-6", then you have a torsion box which should be stiffer than the solid lamination. I'd go at least 6" thick, though. With a torsion box, the further apart the skins are, the stiffer it becomes.

[peteeng]

Hi Gerry - The "box" can't be any stiffer then the solid geometry if its the same material. The box also can't be any stiffer then the "box" geometry which in this case is 64% of the solid, the webs only help in achieving that 64% stiffness by stopping the box from lozenging (sometimes called section efficiency). For the effort I'd go solid vs the webs (unless weight is critical) . Jerry's current beam is 12" high and 3.5" wide. I agree to make it wider. The main cutting loads are in the XY plane which is this direction. Dead weight and plunging loads are in the Z axis which is the 12" dirn which is heaps of stiffness. Peter

[ger21]

The "box" can't be any stiffer then the solid geometry if its the same material

When you are dealing with wood products, it most certainly can. Probably due to the elasticity of wood.

Consider a cheap hollow core door, with the skins being 1/8" (3mm) masonite, and the core usually a paper honeycomb. It's considerably stiffer than an equivalent thickness masonite board.

Or take a lamination of 3 layers of 1/4" MDF, which is considerably stiffer than 3/4" MDF. (I've tested this in my build thread)

[peteeng]

Hi Gerry - if the "hollow" door was say 1.5" thick with 1/8" masonite skins I think you would find that the 1.5" thick solid masonite was quite a bit stiffer (but I wouldn't want to lift it). I do agree with the MDF example. Thin MDF is stiffer then thick MDF. Thin MDF is about 3800MPa stiffness whereas thick is about 3200Mpa. This is due to the compressed skin on the MDF. This info is in the MDF design manuals (see attached). When MDF is rolled into shape the outer surface gets compressed so is denser and more stiff then its centre. This means the thin MDF is relatively stiffer in bending then thick. Laminating thin panels together gives a stiffer product.

By the way a filled epoxy is about 4000MPa stiffness so the glue can add to the stiffness as well. cheers Peter

[TheWire]

I didn't check the forum for today's reply's before drawing this torsion box gantry beam. It would be all the same plywood, I just varied the wood color so you could see the direction of the joints. I need to span about 44" total. The router is fitting in a tight space in my shop, so I was hoping to keep the gantry beam 3.5" thick, but I can deal with 4.5" thick or more if I want to give up more floor and bench space. I attached a pic of my thoughts of a torsion box.

I was also thinking about aluminum skins if it wasn't stiff enough. It looks like they make plywood with thin aluminum verneers that is crazy strong. So my current choices seem to be:

Build the torsion box 4.5" thick , 9" tall with 1/2" skins.

Laminate multiple layers of plywood into a 3.5" thick beam. I would laminate 0.125" aluminum plate to each side if it wasn't stiff enough.

Laminate 2 1.75" thick LVLs together. I would laminate 0.125" aluminum plate to each side if it wasn't stiff enough. The LVLs would be a little easier to laminate than the plywood, but not as compliant to get a very flat side for the linear rails.

Both the laminating options would lend themselves to stacking on more laminations if needed for stiffness.
Thinking outside the box, I wonder if a layer of fiberglass mesh or even carbon fiber between each lamination would help?
I should find someone with a laminating press, But will likely end up gluing it up with clamps and screws using my torsion box base as a flat reference.

Thank you for all your guidance!

[TheWire]

Hi Jerry - If you are going to laminate plywood together consider using some aluminium sheet on it. Say 3mm (1/8") on the outsides. This will create a sandwich which will be very stiff. Calculation shows it will double the beam stiffness in the short direction (on your 3.5" deep beam). ie the skins are the same stiffness as the timber middle bit.

Some 6061-T6 sheet will be very hard so takes bearing loads of bolted bits to it stops water ingress and you don't have to paint! Maybe some metal polish to brighten it up. To double the beam stiffness in timber it would need to be about 112mm wide vs the 93mm. That's why wide is good. Peter

Re-box beam - If you made the same beam as a box using 12mm 1/2" ply it would be 64% the stiffness of the solid. I'd go solid.

Pete,

Thank you for those in-depth calcs. Its good to know the comparison between the laminations and a box beam, especially in the thinner width I'm looking at. Can any of the results be expressed in deflection terms such as N/um?