[ger21]

At some point there must be a trade-off with high lead screws ?

With stepper motors, you trade away resolution as the lead gets higher.
Most high end routers ($100K +) use 25mm pitch screws, but use servos, so resolution isn't an issue.

[Jon.N.CNC]

Moment of inertia? We're not walking a tightrope, but if we did, we'd want a GFS extrusion of similar size, since it's heavier and thus has a higher moment. Deflection is a different story.

Not following you? Moment of inertia is required to calculate deflection. Mass alone means nothing. The Gfs is slightly lighter in comparison to two 9090's

2 x 9090, 1500mm beam, 69gpa. 750-750 100kn load, moi 598cm4 - maximum deflection -0.017 m

100200 calculation of 100, 1500 beam, 69gpa. 750-750 100kn load, moi 870000mm4- maximum deflection -0.012m

As you can see there is a direct correlation between moi and rigidity. The larger the moi the more rigid.

So the GFS extrusion is Infact more rigid, alone Its okay. It can be better though. I use the 9090 spaced apart in a ladder like frame for the gantry and strapped together with 6mm 90x90 channel in 5 places on the back of the gantry on my 8x4ft, yes not exactly the most pretty arrangement but this improved the strength considerably and its performance machining aluminium improved considerably. Spacing of the two profiles increases gantry total height to over 200, the larger the spacing the larger the overall moment of inertia.

Using 2 100200 profiles in a 200200 arrangement would be ideal. But I expect would be very expensive. But two 100100 spaced apart is better than a single 100200. Short lever vs long lever scenario.

[dmalicky]

I don't know the 5083 alloy, but it appears this is a cast product similar to our Mic-6. The main issue is the temper designation. If it is condition "O" (annealed), it will be stress-free and suitable for machining flat (or maybe it can be bought ground flat, like Mic-6). Some other tempers may have stress or not. Your supplier would know. Gummy should not be a big problem for the little machining it needs; also use WD40 or similar.

Agreed with others about faster ballscrews for X and Y. 2020 is also available and zippy, but would give up acceleration unless you have big motors. The choice depends on what you cut: for detail work, I'd get 2010. For big sweeps, 2020. 1605 is fine for Z.

The new design looks good. If you have Z blocks on the Y car, the Y car probably does not require the C-section (assuming the Z blocks are within a few inches of the Y blocks, and the Y plate is thick. I haven't done the FEA yet, but based on prior similar models, as long as the Z blocks are near the Y blocks, a thick Y-plate (3/4") is probably adequate -- but let me do the FEA to confirm. Note that to clear the cutter tip over a tall workpiece, the Z blocks need to be about 2" above the bottom of the Y car. So direct overlap of the Y and Z blocks is usually not possible.

[dmalicky]

The Ix and Iy (Moment of Inertia) of the 9090 is 300 cm^4, or 300*10^4 mm^4.
Two 9090s bonded together will have an Ix of 600*10^4 mm^4.
Two 9090s bonded together will have an Iy of 2200*10^4. (This is calculated using the parallel axis theorem... I' = I + Ad^2, where I=Ix, A=AreaOfA9090, d=45mm). This assumes perfect epoxy bonding between them; bolted will be lower.

The Ix of the 100200 is 874*10^4 mm^4.
The Iy of the 100200 is 3180*10^4 mm^4.

So in both bending directions, the 100200 is better by about 50%. Its outside dimensions are only 10-20mm bigger than a 90180, but Ix and Iy are proportional to Size^4. (10/9)^4 = 1.52. So its 50% improvement is explained by that 10-20mm growth in size. In other words, the only thing that beats a big tube is a bigger tube.

Two 9090s spaced apart by C-channel get a high Iy, but not a high J (polar moment of inertia). A high J is only produced by a large closed-section. J is very important because the X-forces at the cutting tip cause a torque in the gantry, causing it to twist. That will be the flexy mode, and deflection-at-the-cutter always exploits the weakest mode. For high Ix, Iy, and J, a single big tube is best.

To calculate deflection-at-the-tool accurately, we usually need an FEA model. As seen in the Hardwoods thread, the gantry cross-sections tend to distort due to the Y bearing block loads. The classic deflection equations assume the cross-section stays the exact same shape, with loading distributed across the entire section, instead of the 'point' loads from bearing blocks. T-slot extrusion distorts less than thin-wall tubing, and the 9090 has good internal diagonalization, so it should be pretty stable.

I looked at combining two 3060 extrusions (3"x6", each oriented flat for better X stiffness) separated by a C-channel, in the Hardwoods thread (case 4):
http://www.cnczone.com/forums/diy-cn...ml#post1413406
I also tried a number of other configurations not reported, but did not find a stand-out solution (considering effort, flatness, cost, and weight). The twin 9090 + C-channel is a good solution, but with the 100200 available now, I'd pick it for a new build.

Yes, we could keep going by combining 100200s into more complex cross-sections. But as mentioned, it's getting expensive. The gantry design depends on what we want to cut. If mainly wood and light-moderate cuts in aluminum, a single 100200 should be good, and is super easy. If higher stiffness is needed:
- Twin 100200s separated by a single C channel would probably give in the ballpark of 68k lb/in stiffness. Case 4 (Hardwoods thread) is like that one, but using a 3060. So rough scaling suggests: (Case4/Case1 is 2.7x) * (single 100200 is 25k) = 68k lb/in. But it's expensive and heavy for that stiffness.
- Twin 100200s separated by twin C channels (forming a tube inside them, like Case 5) would probably give in the ballpark of 120k lb/in. Again, heavy and expensive, but more efficient use of material.
- An 8" x 8" tube, with diagonalization or bulkheads gives 80k or 240k lb/in (alum or steel). This is a far better use of material and $, but of course more work.
As usual, there's no 'right' answer, but there are better and worse choices depending on aims.

[louieatienza]

Not following you? Moment of inertia is required to calculate deflection. Mass alone means nothing. The Gfs is slightly lighter in comparison to two 9090's

2 x 9090, 1500mm beam, 69gpa. 750-750 100kn load, moi 598cm4 - maximum deflection -0.017 m

100200 calculation of 100, 1500 beam, 69gpa. 750-750 100kn load, moi 870000mm4- maximum deflection -0.012m

As you can see there is a direct correlation between moi and rigidity. The larger the moi the more rigid.

So the GFS extrusion is Infact more rigid, alone Its okay. It can be better though. I use the 9090 spaced apart in a ladder like frame for the gantry and strapped together with 6mm 90x90 channel in 5 places on the back of the gantry on my 8x4ft, yes not exactly the most pretty arrangement but this improved the strength considerably and its performance machining aluminium improved considerably. Spacing of the two profiles increases gantry total height to over 200, the larger the spacing the larger the overall moment of inertia.

Using 2 100200 profiles in a 200200 arrangement would be ideal. But I expect would be very expensive. But two 100100 spaced apart is better than a single 100200. Short lever vs long lever scenario.

Spacing two extrusions apart in and of itself does not necessarily offer more rigidity. Ladder braced maybe helps a little but does not prevent the separate extrusions from translating in relation to one another. The best way would be to connect the two via I-beam or ribbed C channel, with way more than 5 connecting points, and ideally welded or fused together somehow. But when you space them apart, then the obvious thing is the MOI is related to the cross section height, but somehow I missed that detail about the spaced extrusions in your early post.

Also there is the cross sectional area of the extrusion. The thicker the extrusion walls and webbing, the greater the MOI as opposed to one with the same outer dimensions but thinner walls and webbing.

[Jon.N.CNC]

The Ix and Iy (Moment of Inertia) of the 9090 is 300 cm^4, or 300*10^4 mm^4.
Two 9090s bonded together will have an Ix of 600*10^4 mm^4.
Two 9090s bonded together will have an Iy of 2200*10^4. (This is calculated using the parallel axis theorem... I' = I + Ad^2, where I=Ix, A=AreaOfA9090, d=45mm). This assumes perfect epoxy bonding between them; bolted will be lower.

The Ix of the 100200 is 874*10^4 mm^4.
The Iy of the 100200 is 3180*10^4 mm^4.

So in both bending directions, the 100200 is better by about 50%. Its outside dimensions are only 10-20mm bigger than a 90180, but Ix and Iy are proportional to Size^4. (10/9)^4 = 1.52. So its 50% improvement is explained by that 10-20mm growth in size. In other words, the only thing that beats a big tube is a bigger tube.

Two 9090s spaced apart by C-channel get a high Iy, but not a high J (polar moment of inertia). A high J is only produced by a large closed-section. J is very important because the X-forces at the cutting tip cause a torque in the gantry, causing it to twist. That will be the flexy mode, and deflection-at-the-cutter always exploits the weakest mode. For high Ix, Iy, and J, a single big tube is best.

To calculate deflection-at-the-tool accurately, we usually need an FEA model. As seen in the Hardwoods thread, the gantry cross-sections tend to distort due to the Y bearing block loads. The classic deflection equations assume the cross-section stays the exact same shape, with loading distributed across the entire section, instead of the 'point' loads from bearing blocks. T-slot extrusion distorts less than thin-wall tubing, and the 9090 has good internal diagonalization, so it should be pretty stable.

I looked at combining two 3060 extrusions (3"x6", each oriented flat for better X stiffness) separated by a C-channel, in the Hardwoods thread (case 4):
http://www.cnczone.com/forums/diy-cn...ml#post1413406
I also tried a number of other configurations not reported, but did not find a stand-out solution (considering effort, flatness, cost, and weight). The twin 9090 + C-channel is a good solution, but with the 100200 available now, I'd pick it for a new build.

Yes, we could keep going by combining 100200s into more complex cross-sections. But as mentioned, it's getting expensive. The gantry design depends on what we want to cut. If mainly wood and light-moderate cuts in aluminum, a single 100200 should be good, and is super easy. If higher stiffness is needed:
- Twin 100200s separated by a single C channel would probably give in the ballpark of 68k lb/in stiffness. Case 4 (Hardwoods thread) is like that one, but using a 3060. So rough scaling suggests: (Case4/Case1 is 2.7x) * (single 100200 is 25k) = 68k lb/in. But it's expensive and heavy for that stiffness.
- Twin 100200s separated by twin C channels (forming a tube inside them, like Case 5) would probably give in the ballpark of 120k lb/in. Again, heavy and expensive, but more efficient use of material.
- An 8" x 8" tube, with diagonalization or bulkheads gives 80k or 240k lb/in (alum or steel). This is a far better use of material and $, but of course more work.
As usual, there's no 'right' answer, but there are better and worse choices depending on aims.

Really good reply. Very Indepth. My only question would be is do you believe a machine will apply force equally in all directions? X and y certainly. J in this instance I would presume in your ladder frame description would be in the z direction.

One thing that I'm not sure is considered in these calculations is mass, if j is Infact rigidity in the z. Then deflection say in + z is required to overcome the weight of the gantry. Also when machining the contact area of the bit is in most cases going to be larger on the side of the bit than the bottom so most forces will Infact be in the x and y. So is it necessary to have equally balanced moi's?

Sorry just re read, yes the x moi would be the same as double the single profile without the c section. Bolting the c section to the back instead of placing between the profiles would increase the moi considerably however.

I already have this arrangement with the 9090 + c section modelled, will have to run them through nastran in cad with like you say the forces applied at the cutting tip, to better understand.