CNC Router for Hardwoods: Evaluation and Questions

[dmalicky]

Glad it's helpful. Yes, the "face layer then machined" method is one of the best I know of, assuming the face layer is hard enough. Agreed with Gerry that MDF is too soft. For rail mounting, we don't want to yield/dent the material, so traditional "hardness" definitions/tests don't apply (those typically involve a permanent penetration of a ball indenter). The Modulus of Elasticity is a pretty good measure for rail mounting purposes, assuming the material is strong enough to not permanently deform (dent). MDF would probably dent while epoxy may not. Here are some options with MoE (units are Mpsi or psi*10^6):
Self Leveling Epoxy: 0.25 Epoxy.com Self Leveling Seamless Epoxy Mortars Epoxy Grouts Zero(0) VOCs
West System Epoxy: 0.45 WEST SYSTEM | Epoxy Resins and Hardeners - Physical Properties
MDF: 0.5
Tempered Hardboard: 0.75 http://www.fpl.fs.fed.us/documnts/pdf1993/mcnat93a.pdf
Devcon steel-filled epoxy: 0.85 http://www.devcon.com/prodfiles/pdfs/fam_tds_101.pdf
Paper-based phenolic (Garolite X): ~1.3 http://www.professionalplastics.com/...DataSheets.pdf
Glass-based phenolic (G-10): ~2.2 (same source)
Epoxacast metal-filled epoxy: 2.0 - 7.2, evidently they use a lot of filler http://www.smooth-on.com/tb/files/EPOXACAST_655_TB.pdf
Aluminum: 10
Steel: 29

A caveat with MDF and hardboard is that those published MoEs are along the face direction; the perpendicular-to-face direction (what we care about for rail mounting) is probably not as stiff. For solid wood (oak, maple...), it's around 10% the stiffness of the long direction, so ebony, lignum vitae, etc, are of little help. This caveat probably applies to the phenolics, although not as drastically.

If using a jointer to level, the best option appears to be paper-based phenolic. I've never jointed it, but it appears possible: http://www.sawmillcreek.org/showthre...d-how&p=785891

For more surface stiffness, another option is to first bond the face layer (e.g., hardboard) and joint/flatten it, then top that with aluminum (or cold-rolled steel) strips to stiffen the area just under the rail contact lines. E.g., two strips of aluminum, 2"x3/8", bonded with very thin epoxy. Aluminum or CRS strip thickness is normally within a few thou tolerance; if clamped tightly and uniformly to squeeze out all but a few thou epoxy, the resulting surface should be pretty flat. As long as the alum is thick enough, the softer hardboard layer will not have much effect on the overall stiffness, and may provide some damping. For alum vs steel strips, I'd match the strip material to that of the gantry tube, so thermal expansion doesn't cause distortion. If using hardboard as the 'sandwich' layer, I'd segment it somehow (e.g., kerfed every few inches) so that its moisture and thermal expansion don't cause it to debond.

[ger21]

I'm building a 4x8 machine out of mostly wood. From my experience, no wood (or wood product) is hard enough to mount any metal hardware with screws or bolts. The tightening forces are more than enough to crush the wood.
My linear rails are mounted to phenolic plates. I epoxy the phenolic, and CNC machine it flat with a straight ledge.
I wouldn't try to run phenolic on a jointer, unless you have carbide knives. Preferably a spiral head with carbide inserts.

[dmalicky]

Completely agreed that phenolic is a much better choice, like in Gerry's machine. Another issue mounting to a wood surface is the relatively low friction: rails may shift under cutting vibration.

If you have access to a big enough surface plate, or flat mill table (typically 9x49 or 10x54) another option is:
1. Lay down 2 parallel strips of metal bar stock (e.g., 2 x 1/4), spaced the same as the rails will be. The strips must lay flat on the table.
2. Coat with high MoE metal-filled epoxy, pretty thick (as thick as the gantry tube face is not flat). Big gaps (1/8"+) can be mostly filled with metal strip of various thickness.
3. Place the gantry tube on the epoxy and lightly clamp to the surface plate/table. The epoxy squeezes out where needed; the metal strips remain flat. (Before starting, wax any surfaces that you don't want the squeezed epoxy to stick to.)
4. Cure, unclamp, trim squeezed epoxy.
5. If needed, touch-up the metal strip surface flatness (steel blue, mark high spots, scrape or sand).

A limitation of the method (and the jointer or self-leveling epoxy) is it does not give a registration ledge for the master rail location; a long precision straight edge is needed for rail mounting. Gerry's CNC-router machined phenolic method does give that ledge. The ledge is not essential (some commercial machines do not have it), but it does simplify assembly quite a bit.

[wizard]

David
Thank you. You insights are most helpful..
The distortion covers one of my fears.

Nothing to fear here, if you weld on the beam there will be distortion how much depends upon many factors. Ideally if you weld on the beam you would also stress relieve.

I did not want to spend the time welding up the tube and ending up with something resembling an overgrown cheeto.
Epoxy I know is too soft.

Possibly but it depends upon many factors.

Once you add fillers for rigidity, it will not flow for self leveling. I went through gallons of it building my boat.

One common use for self leveling epoxies is to make surface plates. If you make such a plate you can use it to replicate a flat surface from. The approach would be like what David M suggested to use with a Bridgeport table. {as an aside don't assume that old Bridgeport tables are flat themselves}. In other words create your surface plate, treat it with release compound and then sandwich epoxy between it and the beam. Of course the epoxy that you adhere to the bean has to be a harder more suitable resin than the self leveling stuff. You also need a resin that has an open time that is long enough to give you a chance in hell of pulling off this. Epoxy resins have a long history of being used in the machine tool industry, you may want to try the Moglice web site for more information about stuff marketed specifically for such uses.

Milling the steel rail flat is a problem. The length is probably too much for a bridgeport with my skills..

That is what job shops are for! Seriously, if you weld up a beam with pads for the rails and have it milled flat afterwards you probably won't be out that much more money than screwing around for an epoxy solution. Obviously this depends upon the shops the area, how busy they are and how far behind they are on boat payments😂😂😂.

However what about a face layer applied to the tube (e.g. MDF) and then machined?

I'm highly biased against MDF so take this with a grain of salt but I'd never recommend using MDF on a machine like this!!! Not only is it soft, it breaks down over time and is adversely affected by moisture. I've never have had good experiences with the stuff, if you go this route I'd look into other composite sheet goods that would be more reliable.

Epoxy underneath would provide a stable substrate match the tube undulations. The MDF or other outer layer would be easily machined on a large jointer .

You bring up an interesting question in my mind here, would a jointer give you a flatter surface than the beam as is from the factory? I really don't know the answer to that question.

I have been able to do 8' lengths for table top glue ups with wide planks on my ancient 16" joiner (<.005 gap between planks over 8' before clamping). Or is the MDF just too soft as well and will deform like the epoxy?

I would have to say that the MDF is far worst than epoxy. At least with epoxy there are options for much harder resins.

This would provide a reasonable alternative to metal machining and most cities have large cabinet shops that could provide this service for very reasonable $$$.

Possibly if you can get the shops owner to expose is precious jointer to such a beam. I still wouldn't use MDF but that doesn't mean phenolics or other materials wouldn't work. I can just see the look on a cabinet makers face when you tell him you want to run a steel backed phenolic plate over his fine jointer. Beyond all of that there is the issue of getting the face square to whatever mounting pads you expect to use with the beam.

All in all it would probably be easier to replicate off a surface plate. If you take the beam to a good machine shop they should be able to keep the face for the rails square to the mounting pads you will have on the beam. This would make assembly far easier.

[mbronkalla]

So far so good. I went with 6x6x.25 steel tube 66" long. 3 diagonal rods were welded in. Holes cut with plasma cutter after cutting to length
It has mounting rails 1x.25 stee. I ran these through my performax trum sander with several light passes of 60 grit paper to provide a nice finish and tooth for the epoxy. They were then glued on the straightest face with thickened epoxy. I used west system 105 with colloidal silica filler for a peanut butter consistency. The silica will probably make drilling and tapping no fun but is much less of an issue than the high density silica based on past experience
My jointer provided the reference surface. I used a dial indicator to get the tables coplanar (same method I use for setting the knives and outfeed height). Wax paper and packing tape kept everything from sticking together

[dmalicky]

Nice work and nice jointer. Good to hear the method is working. Do you have a way to check the flatness?
You may already be planning on this: capping/diagonalizing the end of the tube is also helpful for stiffness. Else, the joint between the tube and gantry upright/foot is flexy.
Look forward to your progress.

[mbronkalla]

Checking for flatness is hard. Sliding the baem on the joiner or the bridgeport shows very nice flatness but neither are not a perfect reference surface. A precision straightedge / camelback is out of my price range. Fortunately the jointer tops are not very worn. You can still see the original planer marks from the factory (~125 years ago) over much of the tops. I see what you are saying about the end flex. I put the tube on the mill and went to trim the ends. A fair amount flex, even with light cuts and a roughing cutter. I will need to add 2 more diagonals near the ends as you suggest. I want to leave the ends open for bolting it down to the x axis carriages. I only destroyed one drill bit and did not break a tap in over 50 holes for the rails. So the idea of using the colloidal silica (cabosil) seemed to work well.

[dmalicky]

I'd say if both the jointer and bridgeport table show it's flat, then it's probably flat. Wow, that's an old jointer -- no wonder I didn't recognize it. Yes, diagonals are effective, no need for caps per se. Good to hear about the drilling and tapping.

[dmalicky]

This is an update to the prior gantry cross-section FEA work, to include the structure of the Y and Z cars.

First, some context:
- There are at least 10 sub-systems that contribute to deflection-at-the-cutter in a CNC machine (for example, http://www.cnczone.com/forums/diy-cn...ml#post1408740).
- Flex can also occur between any 2 bolted parts, if the joint surfaces are not prepared meticulously and clamped very tight.
- Achieving high stiffness is very difficult because flex always exploits the weakest link: 9 stiff parts plus1 flexy part = 1 flexy machine (though not nearly as bad as 10 flexy parts).
- So far, analysis can be found for most but not all the sub-systems. There will be unknowns. So even if we build our best understanding of a "stiff" design, there's no guarantee the actual machine will be stiff. But if we don't try, for sure it won't be stiff.
- That said, below are some results for 2 more critical parts: the structures of the Y car and Z car. These 2 parts and the gantry tend to be among the worst offenders for flex.
- The table and X system tend to be pretty easy to make stiff, because they don't have to take much bending or torsion (if well designed), and because the cutter load 'splits' into left and right components (at ~50% each) when fed into the uprights/feet/Xrails/table.

As a base case, I reran design #15 above:
- 8x8x1/4" alum gantry with diagonal sheet. 57" long, 8" wide YZ car, for 49" travel.
- ~Rigid Y/Z/Spindle assembly, fixed to the gantry
- ~Rigid endcaps/uprights.
- 100 lb cutter force, directed on the same vector: 40% in X direction, 40% in Y direction, and 20% in Z direction (2i + 2j + 1k).
- Cutter force is applied 10" below bottom surface of gantry. (10" Z clearance, which is a lot and more than I recommend, but is a worst case.)
As before in case #15, deflection is 0.0012", for stiffness-at-cutter of 83k lb/in, for this 1 component:


Next, I added a simple but flexy Y car, Z car, and spindle mount. All materials are alum (for steel, deflections would be 1/3, while stiffness and weight would be 3x).
- 1/2" plates for Y and Z cars
- Center of each Z block vertically separated by 3" (too small)
- Rails are oversize, to roughly account for the stiffness of steel rails.
- Z rails are attached to the Y plate, as is typical (but not ideal). I'll analyze the reverse config later.
- The Z car is ~1.5" from the bottom of its travel. I didn't choose the very bottom since that is best case for the Z blocks. This puts the cutter force 8.5" below the gantry bottom, which applies less torque to the gantry tube.
- A single small spindle mount (somewhat like the typical collar mounts for round spindles).
- For simplicity of analysis, all bearing blocks are fixed/bonded/welded to their rails.
-- For cutter loads in the X direction, this is an accurate assumption (X loads don't cause those blocks to move).
-- It is not accurate for cutter loads in the Y direction, or for Z loads on the Z axis. That is, this model just looks at the structure of the Y and Z cars under simplified constraints; it does not account for how Y and Z cutter forces cause rotation of the cars, sliding of the blocks, and reactions in the ballnut mounts. Those effects require a more complex FEA model. For sure, those effects will lower the stiffness by quite a bit.
Below is the result:
Deflection is a miserable 0.0310", for a stiffness of 3k lb/in. The pic below shows much of the Z plate is hinging due to the single spindle mount and thin Z plate.
Note the deformation of every pic is auto-scaled to show the *type* of deflection; see numbers for actual deflections. At this scale the gantry deflection is not visible -- all purple:


Case 48: include 2 spindle mounts, placed strategically near each end. A huge improvement.
Deflection is 0.0064", stiffness is 16k lb/in. Now there is a local hinge in the Z plate:


Case 49: let's try thickening the Z plate to 1".
Deflection is 0.0034", stiffness is 30k lb/in. Much better, but it's still hinging at the same spot:


As we know from the earlier gantry work, solid plates and bars aren't very stiff in bending, unless they are crazy thick. An I or tube section is ideal for bending, but awkward here. A C-section fits well, is easy, and is pretty stiff -- try that...

Case 50: revert the Z plate back to 1/2" thick, but add 1/2" x 4" legs to make a C-section. A better use of material:
Deflection is 0.0024", stiffness 42k lb/in. The pic below shows those legs have ~zero bending deflection. Now the most noticeable flexy spots are: the spindle moves relative to the Z car, the Y plate bends, and the gantry tube is twisting (note the bigger blue patch due to the auto-scaling):


That relative movement between the spindle and Z car mostly happens because of bending in the Z plate, which occurs because the channel legs are too far from the spindle:


Case 51: Same C legs, but now mounted very close (0.15") to the spindle.
Deflection is 0.0022", stiffness 45k lb/in. Pics show the z plate stays flatter, and the spindle/Zcar relative movement is reduced. Actually the Z legs are bending a little now because we're loading them more directly.



You can see that auto-scaled deformation pictures always show flex somewhere, so it can seem like a tail chasing exercise. Fortunately it's not, since by fixing the one that is currently worst, the next worst one will appear and we'll keep getting stiffer.

The end goal is: For each component, make its contribution to deflection-at-the-cutter to be on target with the overall machine goal.
(Technically stiffness is what we're after, but it's confusing to "add up" stiffness since they add by inverses (1/a + 1/b + 1/c... = 1/total). So I pick a 100 lb load and add deflections.)
- If that overall goal is 20k lb/in (very good for cutting aluminum), then at a 100 lb cutter load, we want a overall cutter deflection of (100 lb) / (20,000 lb/in) = 0.0050", or less.
- That means each component should cause a max cutter deflection of 0.0005" -- the 'fair share' for each.
- So far, we are accounting for 3 components (out of ~10). So ideally, we want those 3 components to cause a max cutter deflection of 3 x 0.0005" = 0.0015". We're currently at 0.0022", not quite there yet.
- If it's easy to make a component stiffer, I do that to give some reserve for other components and the inevitable unknowns. At the same time, it's pointless to make something super stiff if uncontrolled unknowns will ruin the overall result -- and this can easily happen.
- When that's not easy, I think it's most strategic to let each deflect about the same amount -- no obvious weak spot.

To be continued.

[aarggh]

That's some really useful diagnostics you've done there David, very interesting indeed to see real analysis.

I'll be making sure I digest all the data!

cheers, Ian

[mbronkalla]

Dave, this is great. Real analysis and moving beyond opinion. Your previous posts helped me greatly and surprised me in the gantry beam analysis , which was borne out in my machinig. Now I really have to reconsider just how stiff my Y& Z carriages are and add likely some bracing.

[dmalicky]

Glad to hear it's helpful, as well as for the prior work. Some good news is that those C-legs are generally easy to retrofit to an existing machine. You can probably predict the fix for the Y car. More results to come.

[dmalicky]

Case 51 showed the deflection was down to 0.0022", close to the 0.0015" goal for the structure of these 3 components. The pic for Case 51 shows a 'kink' in the Z plate, where the lower Z blocks attach. So let's try reducing the bearing block force by raising the upper Z block (so the 2 blocks have greater leverage, for less force on each).

Case 52: raise upper Z block by 3", for a total of 6" Z block vertical spacing (on centers). Also make the Y and Z cars taller to match.
Deflection = 0.00200". stiffness = 50k lb/in. The pic shows that it reduced the local 'kink' in the Z plate:

While the structure only saw 0.0002" improvement, the deflections of the bearings themselves will go way down with 3" more separation (a different analysis).

Case 53: Add a 3rd spindle mount mid-way between the other mounts. It helped a tiny bit:
Deflection = 0.00196", stiffness = 51k lb/in. Pic is about the same. There is still a bit of bending in the Z legs.

Case 54: Thicken Z legs to 3/4".
Deflection = 0.00188", stiffness = 53k lb/in. Pics show the Spindle is still rotating more than the Z car.

We can see lateral bending in the Z plate, a partial cause:


Case 55: Thicken Z plate to 3/4".
Deflection = 0.00183", Stiffness = 55k lb/in. The Z car has improved noticeably compared to Case 51:

Certainly, we could do more to the Z car (increase legs from 4" to 5", or change the C-section to a box all around the spindle). But the Y car and gantry tube are bigger fish to fry at this point.

Case 56: Add 1/2" x 4" legs to the Y plate, to make it a C section, too.
Deflection = 0.00172", stiffness = 58k lb/in. Pics show the Y legs are very rigid in this configuration (but they'll likely show bending if the Z car is raised up high).


Here are the side and front views, to show X and Y deflections. It's still worst in the X direction; the side view shows the gantry tube is twisting (lower edge moving out):



Case 57: Thicken gantry tube's wall thickness from 0.25" to 0.50".
Deflection = 0.00131", stiffness = 76k lb/in. Yeah. The light blue patch on the Case 56 gantry is back to dark blue (especially impressive since the auto-scale has lowered the threshold between those colors):



- It was a pretty stiff gantry when we started, but after stiffening the Y and Z cars, the gantry then needed more.
- I'd say for cutting aluminum, a 8x8x1/4" steel gantry (with internal diagonal or bulkheads) is also a good option, even though pretty heavy. It would be even stiffer than Case 57.
- Deflection is now under the 0.00150" target for the structure of these 3 components, so this could be a good stopping point.
- Later, I'll try some other variations to see how stiff it can go, and I'll move the Z car up and down to check how that affects results.

[pippin88]

David,

I'm thinking a good Z plate would be steel Z channel, with the spindle mount welded in, spanning between the channels.

Next question is Z rails on Y carriage, or on Z plate? I went with on Y carriage (cars on Z plate) on my current router as I felt it gave me the best travel / fixation arrangement.

[dmalicky]

David,
I'm thinking a good Z plate would be steel Z channel, with the spindle mount welded in, spanning between the channels.

Next question is Z rails on Y carriage, or on Z plate? I went with on Y carriage (cars on Z plate) on my current router as I felt it gave me the best travel / fixation arrangement.

Yes, that should be a stiff Z car. In the US, our steel C-channel is hot rolled and has fairly short legs that taper towards the end (American Standard shape) -- https://www.metalsdepot.com/products...cc=%20&aident=
Hot rolled is not very straight or flat (unless cold finished, which ours isn't), and our short tapered legs are not very stiff in bending.

We can get aluminum C-channel in 2 profiles, and 1 of them is fairly good for bending (Aluminum Association) -- https://www.metalsdepot.com/products...l?page=channel
But the legs are not very long, and the AA profiles have fairly thin web thickness (0.29").

Your drawing shows constant thickness legs, of reasonable length for steel. Sounds like metric steel C-channel has a fairly good cross-section for bending. The web (main plate) thickness of your drawing looks a little thin -- even in steel, it will distort more easily -- that can be minimized by mounting the rails or blocks close to the C legs.

The welded in spindle mounts should be very stiff. The higher the upper mount, the better for stiffness. Or ideally, make it a similar height as the upper Z blocks on the backside -- then the spindle loads go straight thru to the Z blocks.

I'm assuming you would stress-relieve it in an oven, since it's not too big (else it could distort over time, as you probably know), and post-machine all mounting surfaces.


Yes, the rail/block question is an interesting one. I think Z rails on the Z car is overall stiffer, but I've not run that config yet. Z rails on the Y car can work fine if the Y car has long enough legs for its C section.

[pippin88]

David, I'm not sure on the exact specs of our C channel. I quickly googled and one of the results had a section with 12mm flanges (legs), 6mm web. However it did not gave a drawing, and the flanges may well taper.

Steel would require machining. Welded sections are stronger than bolted. I don't have the equipment / skills for welding aluminium.

[pippin88]

Casting an aluminium z car may be option. Then machining.

[dmalicky]

Yeah, 6mm would be pretty thin for the web/backplate.

I agree welding has the best joint quality. After welding and stress relief, the backplate flatness would probably be pretty far off, maybe requiring a lot of post-machining. Also post-machining the spindle mount hole seems awkward. Those are certainly doable, but the overall work may be comparable to the next option...

A combination of bolts and pins should work well for this particular joint. The interface between the channel legs and backplate is mainly in shear, which is relatively easy to control with bolts and pins. (I'm assuming the legs would be mounted to the face of the backplate, not the edge.) Roll-pins are easier than dowels, and have good shear stiffness. I'd probably do a hole every ~inch, alternating bolts and pins. A thin layer of high strength epoxy would make the joint nearly as good as welding, for CNC purposes (where we need stiffness, not strength). With this method, the backplate could start as cast alum tooling plate, which is stable and very flat. I'd do the legs out of 6061-T6, so the tapped holes are stronger.

Casting would be cool, but for me it would be much more work than fabrication.

[dmalicky]

This set is to try more stiffness improvements, and check different Z car positions. Recall Case 57 numbers:
Deflection = 0.00131", Stiffness = 76k lb/in.

Case 58: Thicken Y plate to 3/4"
Deflection = 0.00130", Stiffness = 77k lb/in. At this Z car position, ~no advantage.

Case 59: Thicken diagonal sheet (inside gantry) from 1/8" to 1/4".
Deflection = 0.00127", Stiffness = 79k lb/in. The 1/2" gantry tube wall thickness is already getting thick enough that the cross-section has more inherent stability. A bigger diagonal isn't really needed.

Case 60: Thicken gantry tube's wall thickness from 0.50" to 0.75". This is roughly similar to a 8x8x1/4" steel tube (Esteel = 3 x Ealum) -- but the steel tube will be stiffer since its metal is closer to the 8" outside dimension (the outside surface does the most work).
Deflection = 0.00114", Stiffness = 88k lb/in. A substantial improvement, but not as dramatic as the change from 1/4" to 1/2". There are 2 likely reasons:
1) Adding more material to the inside of a tube is progressively less effective
2) We're now in the territory of diminishing returns for the gantry -- other parts are more flexy than it, at this point.


Case 61: Drop the Z car 1.5" to near the bottom of its travel (cutter load is 10" below gantry)
Deflection = 0.00114", Stiffness = 88k lb/in. It's surprising that it isn't worse! Apparently, while the gantry experiences more torque and must be twisting more, this Z car position puts the Z and Y blocks close to overlapping -- and that is a stiffer config. So the 2 effects happen to exactly cancel.


Case 62: Raise the Z car 4.0" (cutter load is 6" below gantry)
Deflection = 0.00123", Stiffness = 81k lb/in. Now the Z blocks are relatively far from the Y blocks, so there is more bending in the Y car.


Case 63: Raise the Z car 5.0" (cutter load is 1" below gantry)
Deflection = 0.00141", Stiffness = 71k lb/in. The Y car is the main weak link now. The Z blocks are too far from Y blocks, creating a lot of bending in the Y car.

(See the pic for case 65 below, to see where the Z blocks are)

Case 64: Increase Y leg depth from 4" to 6".
Deflection = 0.00122", Stiffness = 82k lb/in. A good use of added material for this config.


Case 65: Eliminate Y legs, just to see how bad it is.
Deflection = 0.00446", Stiffness = 22k lb/in. Clearly, if the Z rails are on the Y car, and the Z car is raised up, the Y car needs a lot of bending stiffness.


Case 66: Thicken Y plate from 3/4" to 1", to show what thicker can do (not enough). A 1" thick aluminum plate would have similar bending stiffness as a 0.69" thick steel plate. (Based on the same E*I stiffness: SteelThickness = 3^(-1/3) * AlumThickness ).
Deflection = 0.00316", Stiffness = 32k lb/in.

Case 67: Increase Y leg depth from 0" back to 6"
Deflection = 0.00116", Stiffness = 86k lb/in. Compared to Case 64, the 1" thick Y plate shows benefit when the Z car is raised up. Also compare to Case 58, where even a 1/2" thick plate was adequate.


Case 68: Drop Z car 9" (cutter load is 10" below gantry)
Deflection = 0.00112", Stiffness = 89k lb/in. Not much different than Case 67. Compare that small difference to the big difference between Cases 61 and 63 (same Z car positions). This indicates that Z rails on the Y car can give consistent stiffness IF the Y car is very very stiff. These pics show the front and side view. X deflection is still dominant.




So, with Z rails on the Y plate:
Assuming the gantry that is quite stiff in torsion, the highest stiffness at the cutter occurs with the Z car in the low ranges of its travel. The least stiffness is generally with the Z car at the top of its travel. If the Y car is very very stiff, the difference is probably not significant. But if the Y car is just a plain plate, there will be a major stiffness loss in the up position.

Next I'll look at Z rails on the Z plate.

[pippin88]

David,

Nice work. At the end of this, I believe you will be able to put forward compelling design elements for the 'best' machine. A lot of the threads in this forum are repetition of basics (usually with quite poor gantry designs), but that is partly due to a lack of good guides, partly due to a lot of opinion without people crunching the numbers.

A lot of your results are not surprising, given the basic principle that the larger a section the stiffer, but that only applies in the direction / vector the section is large in. The great example being the classic Z plate (a flat plate) being a weak area, with the addition of legs / flanges vastly improving it..

With the gantry beam, I think steel wins. I reckon 1/4" 8x8" steel is much more available than 3/4" alu tube (does it exist?)