[TeeTomTerrific]

Hi TTTTTT - sign again - it has to overlap the gantry and the base. If you transverse bend this design where will it bend? In the gap you have left between the web and the base element.. Peter ask yourself if your column & web were a piece of cardboard and you pushed it sideways where will it bend? Then you will get why the web has to overlap....It can be on flat if you like but have it overlap....Peter

I think this will work.

[ihavenofish]

I'm not 100% sure that rib is worth much in stiffness. If I couldn't make the uprights thicker plate, (money etc) I would do a vertical rib top to bottom, 1" wide and 2" deep, screwed from the inside. what you are trying to fight here is the gantry skewing and having a little rib on an angle doesn't seem the best use of the material.

[TeeTomTerrific]

both layouts have plusses and minuses.

The ones that matter in the end are:

- the moving saddle has much less weight to move, which means you need less servo power
- the moving saddle is also more challenging to get high Z clearances and long z travels
- the moving saddle can be much simpler to assemble and align

If you want big Z travel and clearance, you probably want the moving slide. In any other case, I would pick a moving saddle.

Honestly the sadddle version I had was more complex, had mor pieces and would have been heavier. I think this actually improve upon what I originally designed.

[TeeTomTerrific]

I'm not 100% sure that rib is worth much in stiffness. If I couldn't make the uprights thicker plate, (money etc) I would do a vertical rib top to bottom, 1" wide and 2" deep, screwed from the inside. what you are trying to fight here is the gantry skewing and having a little rib on an angle doesn't seem the best use of the material.

The upright pieces are 20mm thick. The rib is 30mm thick. I'm having a hard time imagining that 20mm thick 6061 aluminum is going to bend much less with the rib. Its getting late and I've been at it all day making modifications. I'll think about widening it but making it deeper overnight.

[ihavenofish]

The upright pieces are 20mm thick. The rib is 30mm thick. I'm having a hard time imagining that 20mm thick 6061 aluminum is going to bend much less with the rib. Its getting late and I've been at it all day making modifications. I'll think about widening it but making it deeper overnight.

My little machine has 20mm (3/4") uprights, which are very short (100mm clearance over table). The larger machine is 25mm, with additional 25mm plates for bending resistance. That machine has 225mm clearance. bot are adequate, but not what I consider "stiff". Most factory routers have very large 3d box or pyramid section uprights. 1/4" steel plate and 12" up to 20-30" box sections. (huge). When I redo the little machine for production (to sell) it will have 32mm uprights. For me its easier to buy thicker plate than create a complex bolted together design that adds more points of error.

There are many much weaker point in your machine though, so, kinda got to draw a line somewhere.

[TeeTomTerrific]

My little machine has 20mm (3/4") uprights, which are very short (100mm clearance over table). The larger machine is 25mm, with additional 25mm plates for bending resistance. That machine has 225mm clearance. bot are adequate, but not what I consider "stiff". Most factory routers have very large 3d box or pyramid section uprights. 1/4" steel plate and 12" up to 20-30" box sections. (huge). When I redo the little machine for production (to sell) it will have 38mm uprights. For me its easier to buy thicker plate than create a complex bolted together design that adds more points of error.

There are many much weaker point in your machine though, so, kinda got to draw a line somewhere.

I think that for this machine I could easily go to 30mm then with 160mm of clearance and 110 inches of Z travel. I would prefer not to use the rib. and be done with it. Great thing about a DIY machine is you can always change it later.

[ihavenofish]

One day you'll even get to use it :P



So, this is the result of my murdering 150KH of aluminium.

The weakest point at this point is now the uprights, but not in the lateral direction so much. I am using old Xzero uprights and the linear bearing are places very close together - 110mm outside race to race. A wider spacing would eliminate the chatter in the Y cuts.

[routalot]

That web is not worth the effort.It contributes very little to maintaining the shape of the gantry with the loads applied by a transverse cutting move.I felt a need to be blunt because the message wasn't getting through.It isn't just a matter of adding stiffness,you need to triangulate the structure.A web extending out perhaps 5 inches at the top and a couple of inches at the bottom would do it.You need to use a spanner as an analogy-the tips of the spanner are basically triangulated so that they can't spread and because your steppers/servos are at the top beam,pushing sideways with considerable force it is imperative that you design to resist this force.Attachment 478346

I hope the attached screen shot clarifies things.

[peteeng]

Hi TTT - The point is that you have to look at every every component and try to maximise its stiffness (so if your having the columns machined consider steel 3x !! the stiffness for the same size). Adding flanges, webs, extra thickness. Every opportunity to add stiffness to a part must be taken. That's because bolted joints are 50% efficient, Construction extrusion is inefficient (too many little flanges and webs for its overall geometry), flat cantilevers are springy (your columns) laminated parts are less efficient then solid parts. All of these factors work agin the machine laddy. I realise its tough redesigning and redesigning but its easy in CAD and rough in the real world, but your close to the Grail, your at the crux. Have a review and a mental break then move fwd. your 1st machine is always the toughest and its never perfect... neither is no2 or 3 Hmmm maybe 4 as well, or variants omicron? Peter

[luke peter]

Hi TTT.

Man, I am right here with you in design analysis paralysis... My old mentor used to tell me, if you have done some maths, and are more than 50% sure, JFDI (Just f'ing do it), and see what happens

[ihavenofish]

[QUOTE=routalot;2503210you need to triangulate the structure.[/QUOTE]

Triangulation saves materials, but lets not confuse things. Wider = stiffer. If 20mm wide = 1 for stiffness, 40mm wide = 4. a 30mm rib might equal 2 (would need to do the fea). a 120mm rib might = 5, and therefore be stiffer and less material than just 40mm plate. But you introduce more surfaces that need to be machined very precisely, more holes, threads, more work. 5" in triangulation might be a 4.8, saving lots of material over the rectangle plate with basically the same stiffness but with all the same complexities.

The trick becomes, what is the min stiffness you need (deflection and resonance come in to play) and what is easier, paying for thicker plate or making a complex precision assembly.

When you look at mine, the 'stiffness" for the cutting load is actually fine - deflects a few tenths with 20kg force sideways, but resonance creeps in in the front to back direction. You do not necessarily have to deal with the resonance with thicker material, but often this is the easiest way. In my case though the right way is to make the upright bases wider, no the plate thicker. In his case being a much bigger taller machine than mine, thicker plate will help.

But as mentioned before, there may be other weaknesses more important on his machine to address first.

[dixdance]

I'm a complete newbie to router design, been following this design and have learned a lot, thank you all. Have a small Techno router, thinking to go larger DIY

I'm wondering if having the spindle cantilevered out so far beyond the trucks on the y axis slides doesn't contribute to resonance and vibration? I'd be tempted to increase the distance between the 2 trucks to reduce the moment arm, even if it required lengthening the bed. Or am I underestimating the stiffness of the trucks and rails?

[ihavenofish]

Your leverage is from the nose of the spindle which will be the same on both.

Hypothetically both would be identical stiffness - all you've really done is reverse the positions. Likewise You can (and people often do) screw up either. In this case, his is pretty decent.

Overhang is a big deal though. My machine is optimised for as little overhang as possible. Compared to the original Xzero axis, the small machine is 50mm less overhang and the big machine is 72mm less overhang. I accomplished this mostly by making the YZ components one piece instead of 2. The downside was that big Z slide cost a ton of money both in material and machining time. I also made my spindle clamps to pull the spindle close as well, to the point there is only 4mm between the top of the spindle and the servo mount.

The "tilt" back of his uprights reduce effective overhang as well. Leverage from the spindle nose to the bearings at the bottoms of the uprights is far more important than from the nose to the Z blocks, or Y blocks which is easily dealt with.

[dixdance]

My concern is with what I think you are calling "overhang". It seems to me that on many of these machines the spindle center is pretty far in front of the footprint of the bearing blocks on the uprights (Y axis, yes?).

On the OP's design, if the side plates were wider at the top so the gantry could be further back to eliminate this overhang, would that be an improvement or just a trade off for another source of flexibility?

[ihavenofish]

This is what I'm talking about (I call the bottom axis X sometimes so it could get confusing).

What i'm saying is the 2 styles of z should make no difference.

Setting back the uprights is a trade, you reduce leverage but you weaken the upright. This is why you use FEA, to find the balance.

[dixdance]

FEA is way above my pay grade!

If I were to build a machine, I think I'd make the uprights thicker and wider, space the bottom axis trucks further apart, and live with a slightly longer footprint.

[ihavenofish]

FEA is easy and free these days. If you are building a machine without FEA then building machines is above your pay grade :P

[TeeTomTerrific]

My concern is with what I think you are calling "overhang". It seems to me that on many of these machines the spindle center is pretty far in front of the footprint of the bearing blocks on the uprights (Y axis, yes?).

On the OP's design, if the side plates were wider at the top so the gantry could be further back to eliminate this overhang, would that be an improvement or just a trade off for another source of flexibility?

If I replaced the X axis extrusion with a custom machined beam the same design as my Z axis using the exact same columns the spindle would be moved back 50mm to close to dead center of the Y axis bearing trucks and likely a lot stiffer. A possible future improvement for me to explore, but for now I'm locking the design down with that last change to 30mm wide columns and building it. No design is perfect and it is a hobby machine and part of the hobby will be the learning by doing.

One of the constraints I was working with for this machine is how much can be built using stock parts such a extrusions and the plethora of Chinese CNC parts available. The learning was there are limitations with that approach, but it is a less expensive way to go for a first medium sized DIY CNC project. When all the hardware and shipping is added up it will be under 5K. It would have been 1k to 2k more if I bought extrusions and machining domestically in the US, but that part is also coming from China. (That does not include the controller and software. That expense is yet to come.)

[ihavenofish]

How are you making these parts, uprights etc?
I have some suggestions, but it will vary depending on how you have to make them.

[dazp1976]

Hi TTT.
My old mentor used to tell me, if you have done some maths, and are more than 50% sure, JFDI (Just f'ing do it), and see what happens

JFDI at only 50% !?.
That's incentive.
Lets go snap some bits