[bogan]

As in the title, I'm going for something with a small footprint, so have gone with a bridge style design, and mounted the rails underneath the bed with the carriages being fixed to the main frame. With a slightly complex guarding and swarf catchment this might allow the machine to have 1m Y travel and around a 1m footprint, plus I like the way the rails reinforce the bed and keeps the rail contact points in the same place relative to the spindle.

I've attached a layout design I'm thinking to go with, epoxy filled thick wall steel weldment for the non moving parts, to be stress relieved and then machined. Moving parts to be bolted together machined aluminium.

It's HGR-25 rail, X600, Y1000, Z300 travel. Not sure on the ballscrews, probably 2510 C7 with double ball nuts. Likely servo driven, eventually put an ATC on it, but start with the spindle from my preiovus machine to get an idea on the machine's capabilities before choosing the final spindle.

Attachment 454678

I'm keen to get some critique on the design, things I could change to increase rigidity or precision, etc...

cheers

[routalot]

A couple of things come to mind.First of all if the Y axis is the moving table and has 1 metre of travel-you need 2 metres of space for it.Second,while it looks substantial,I would give some thought to reversing the architecture of the side supports.That way you could attach a brace from the corner of the under table cross beam to the top of the upright for better triangulation.

[catahoula]

I just went through a very similar design process. Have you run any FEA? That's the best way to be able to quickly compare layouts. Yours is generally looking good. Nice job keeping a heavy proportion as the size is increased.

I think you'll find that your gantry needs to be much closer to a box in cross section to get it rigid. So you might not be able to keep the same arrangement with it on the front face of the columns.

Your Z plate is quite thin. The bearing spacing is nice and wide and you've got the angle sections for stiffening, but you'll find through FEA that because the spindle is mounted in the middle, there's a lot of web flex in the center with a plate that thin. I went for a 35mm thick aluminum plate.youll have to play around with it to balance web flex vs your very large rail spacing.

while the moving rail/fixed carriage design is good for keeping the dynamic bearing alignment in a good position, you'll have to decide if it's acceptable to mount steel rails onto an aluminum surface with over twice the thermal expansion. As a hobbyist it's hard to put numbers on precision and accuracy, but I decided to only mount rails to steel surfaces.

Also I think you'll find that if you're looking at servos, moving weight just isn't a huge deal anymore. Minimizing it is good, but the fixed gantry layout is already really good for that already, so I wouldn't stress about beefing up the z axis or table, unless youre trying to cut at 15m/min or something.

Lastly, I would recommend against trying to squeeze every last mm of travels out of the footprint. It will give you lots of headaches when doing the finishing touches. Just add another 25-50mm of rail and ballscrew length in the early stages and give yourself room to breathe. Once you start to fit way covers or ballscrew bellows, and also leave a margin for overtravel (2x screw lead, minimum), any extra distance helps massively.

Cheers

Edit to add travel lengths

[catahoula]

Also, what's the intended use? Based on the size seems like you want to mill aluminum with reasonable accuracy?

[catahoula]

One more; go as heavy as you can on the wall thickness. 9-12 mm for tubes, any plates at 15-20mm. Will make managing weld heat distortion and finish machining so much easier, and also good for general rigidity and vibration management.

[bogan]

A couple of things come to mind.First of all if the Y axis is the moving table and has 1 metre of travel-you need 2 metres of space for it.Second,while it looks substantial,I would give some thought to reversing the architecture of the side supports.That way you could attach a brace from the corner of the under table cross beam to the top of the upright for better triangulation.

Sorry, meant to say it could have 1.5m instead of the normal 2m, this could be done by moving/removing the swarf tray so it's only the machine footprint at Ymax position.

I had looked at a more simple flat base piece with single cross brace, as you suggest, still on the cards but I have access to a CNC which only has around 400-500mm effective Z height, which I'm hoping to use to post machine the mounting/mating surfaces between these parts.

I just went through a very similar design process. Have you run any FEA? That's the best way to be able to quickly compare layouts. Yours is generally looking good. Nice job keeping a heavy proportion as the size is increased.

I think you'll find that your gantry needs to be much closer to a box in cross section to get it rigid. So you might not be able to keep the same arrangement with it on the front face of the columns.

Your Z plate is quite thin. The bearing spacing is nice and wide and you've got the angle sections for stiffening, but you'll find through FEA that because the spindle is mounted in the middle, there's a lot of web flex in the center with a plate that thin. I went for a 35mm thick aluminum plate.youll have to play around with it to balance web flex vs your very large rail spacing.

while the moving rail/fixed carriage design is good for keeping the dynamic bearing alignment in a good position, you'll have to decide if it's acceptable to mount steel rails onto an aluminum surface with over twice the thermal expansion. As a hobbyist it's hard to put numbers on precision and accuracy, but I decided to only mount rails to steel surfaces.

Also I think you'll find that if you're looking at servos, moving weight just isn't a huge deal anymore. Minimizing it is good, but the fixed gantry layout is already really good for that already, so I wouldn't stress about beefing up the z axis or table, unless youre trying to cut at 15m/min or something.

Lastly, I would recommend against trying to squeeze every last mm of travels out of the footprint. It will give you lots of headaches when doing the finishing touches. Just add another 25-50mm of rail and ballscrew length in the early stages and give yourself room to breathe. Once you start to fit way covers or ballscrew bellows, and also leave a margin for overtravel (2x screw lead, minimum), any extra distance helps massively.

Cheers

Edit to add travel lengths

No FEA yet, will do that to add strength with minor design changes later on. Or as you say, compare strength between different layouts.

I was hoping the large separation between horizontal rails, thick wall section, and epoxy fill would keep it rigid across that span, what other gantry layout comes to mind? Would you drop the separation to make it more of a box? The RHS is 125x125x9mm wall currently.

It's the Y carriage that has those support on it, I'm hoping to get away with a thinner one there as the rails for the Z axis site right on the carriages on the X axis. You're right about the spindle carrier plate, I expect to make a couple of different versions for different spindles as that will be the main upgrade once I'm running, so it is largely a placeholder at the moment.

Good point about the thermal expansion, I guess the best way properly avoid that is a steel bed? As those rails are 1.5m long the effects could be considerable.

Some of the mass limit is just about being able to machine and work with the weldments, but I take your point, and a lot of the moving parts aren't anywhere near that limit anyway.

For travel length, those are the rail sizes I selected as they seem pretty standard, I'm fine with the effective area being 20mm or more inside this once the bellows and homing/limits are added.

I was wondering if 9mm was a bit too thick tbh, but will stick with it.

Intended use is aluminium, might be some wood, and lighter cuts on steel would be useful, but the main goal is to have it nicely buzzing through ali, ideally some 10mm 1D sloting at 1000mm/min.

Thanks for all you advice and help!

[routalot]

I may not have described my thought very well.I hope the attached screenshot conveys my meaning better.I was concerned about the lack of triangulation on your original proposal as the uprights are just cantilevered from the base and a diagonal brace would increase rigidity.My quick assembly shows solid ends but could b plates with infill spacers or three pieces of box section with a hole in the middle of the triangle.

[catahoula]

I was hoping the large separation between horizontal rails, thick wall section, and epoxy fill would keep it rigid across that span, what other gantry layout comes to mind? Would you drop the separation to make it more of a box? The RHS is 125x125x9mm wall currently.

Yeah if those are 125x125x9 mm tubes making the gantry I don't think you're too far off, but since gantries have a lot of torsion applied to them, a square cross section is most efficient, where depth is the same as height. I settled on 230x180 for my gantry for a ~580mm span between the columns, using a 200x150x12.7 tube and welding some heavy flat bar for the rail mounting to it. Going for a full square tube of 200x200 with the same treatment would have dropped deflection at the middle of the gantry another 1-2 microns at 300N cutting force. But with stiffness going up with the cube of the diameter, any incremental increase from 100-125-150mm does make a pretty substantial difference. If you don't want to do FEA, I attached a spreadsheet calc that I found somewhere that will give you an idea of the effect of various tube sections. Not sure where this came from but I think it was meant for public distribution, so credit wherever credit is due.

If I did it again I might be tempted to build a torsion box for the gantry with laser cut plates, that way it could have ribs all the way along the length.

[pippin88]

A couple of things come to mind.First of all if the Y axis is the moving table and has 1 metre of travel-you need 2 metres of space for it.Second,while it looks substantial,I would give some thought to reversing the architecture of the side supports.That way you could attach a brace from the corner of the under table cross beam to the top of the upright for better triangulation.

You don't need 2m for 1m of travel. It depends on your bearing spacing.

You won't want the bearings right at the corners of a big table - that would result in the biggest unsupported span.

A rough rule of thumb of 1/4 in from the ends works reasonably.

So you end up with bearings spacing ~1/2 of the table length.

So ~1.5m rail length for 1m of travel.

EDIT: My bad. I read your post as needing 2m bearing length when you were talking about space for the machine. You are correct.

[bogan]

I've started the FEA, looks like the basic design of the upper bridge is pretty sound. 2x 125x9 SHS with 100mm separation (filled with 10mm plate at each side) gives 5 micron deflection for 300N at Z full extension (300mm). For comparison 350x200x9mm square section gives 4 microns of delfection. This is only done with SW2016 Simxpress so not the best, but I think the relatives will still stack up.
Even with simple bridge ends added it only gives 7 microns. And given the way it is deforming I would expect whatever I fill the frame with to have a reasonable reduction on it.

[catahoula]

Even with simple bridge ends added it only gives 7 microns. And given the way it is deforming I would expect whatever I fill the frame with to have a reasonable reduction on it.

Cool, there's some discussion of EG modulus over on the Milli thread; modeling my frame tubes all filled with 20 GPa material brought deflection at the collet down another 1-2 microns. Though my guess is that with a resin heavy mix it'd be pretty easy to lose a lot of that. One commercial product I found cited around 30 GPa

[catahoula]

Also regarding the flipped table layout, are you planning on using the spindle to surface the table relative to the machine? With this layout you won't be able to reach the whole table unless you make the rails significantly longer than the work area. One option would be to leave as is and add a thicker section or piece of aluminum t-slot the size of the actual work area, so you can mill it flat without it being below the height of the rest of the table.

[bogan]

Also regarding the flipped table layout, are you planning on using the spindle to surface the table relative to the machine? With this layout you won't be able to reach the whole table unless you make the rails significantly longer than the work area. One option would be to leave as is and add a thicker section or piece of aluminum t-slot the size of the actual work area, so you can mill it flat without it being below the height of the rest of the table.

I'm not too sure on that one yet, given the thermal expansion factor I'm leaning towards a steel structure to support the bed, machining this flat, then using plate on top with a pretty consistent thickness so I can use the whole bed area. Also have a nice thick bit of ali bolted to the top to do smaller parts with higher precision.

For cost I expect the bed to be ali, though I'll do the numbers and make sure it has the strength to deal with the differing expansion rates, it's about 10micron per degree across the 1.5m length so not too bad I'm thinking.

[cyclestart]

I've built and still use a machine in roughly this layout although my components are not near the quality proposed here.
When in use it's footprint is larger than an equivalent moving gantry but you're aware of that.
The original plan was to use a rotating ball nut to drive the table but that's never happened... so far.
For those who know FEA how are the blocks defined in this diagram ? Assumed to be a fixed connection ?

[bogan]

I've built and still use a machine in roughly this layout although my components are not near the quality proposed here.
When in use it's footprint is larger than an equivalent moving gantry but you're aware of that.
The original plan was to use a rotating ball nut to drive the table but that's never happened... so far.
For those who know FEA how are the blocks defined in this diagram ? Assumed to be a fixed connection ?

I'd do that as a fixed connection, sure it might not be 100% technically accurate, but any simulation benefits from doing them as a sliding connection would be miniscule and likely below material/manufacturing variations anyway.

Are you looking to evaluate bending moments to see if such a large overhang is viable? If bending is a possible issue, vibration from machining could be a more significant one also, especially if it can be generated/seen between the two blocks.

[bogan]

One year in, haven't had the time I was hoping so I've yet to get onto the main bit of the build! Though I have purchased the rail, screws, and servos/drives so I'm locked in...

However I'm soon to finish the supporting frame, with swarf tray, electrical cabinet, a lots of heavy duty drawers for tooling/stock.

Bed design has changed to be all aluminium, a 20mm thick plate, with 32mm square bar bolted to the bottom in a simple reinforcing pattern.

Column/bridge design I'm still working through, mainly from a manufacturability point of view, how to machine/grind the faces to the best precision with tools I have available.

[pippin88]

For column mounting to base you don't have to have precise mounting faces. You can use screws to level then inject special epoxy.
See Stefan Gotteswinter 'Tramming a milling machine with epoxy' on YouTube
https://youtu.be/U7Qs-J2swIc

[bogan]

Interesting, I hadn't considered that, thanks!

[bogan]

Haven't updated this is awhile, been doing some work on it lately though, finished the main base/bridge weldment design and got the parts tube laser cut, fairly pricey and I've just noticed there was creep on the two bridge 'span' bits so the holes don't line up with the linear rail (it's all about 5mm longer than it should be on a 1m long part).

Machine base has drawers in it, offset to the one side that won't be jammed in the corner, in that side will be the electrical cabinet and coolant pump/reservoir. Swarf tray on top.

Then to the important bits, the main machine frame is in two parts, the base, and bridge. Those were the tube lasers parts, all from 9mm 125 box section, with some 8mm flat welded to it in various places.

I pre-heated all the weldement as it gives my dinky 180amp welder a bit more penetration, tried to post weld treat it but nowhere near hot enough for parts that size.

Got most of the precision surfaces cut on a CNC mill at work, but couldn't quite reach some areas, so cobbled together a guided grinding rig out of my old cnc machine to finish those off, which came out pretty good actually.

For the spindle I ordered anFM30 package version from CNCDepot, which is 3hp ATC, so that'll be fun! Not cheap (budget wise, it seems pretty cost effective for an ATC spindle) but also wasn't an area I wanted to skimp on, figure there's already about 3k of linear gear, 3k of tube lasered parts, 7k on the spindle's not even half the machine price!

[luke peter]

Looks like its going to be a strong machine.

How exactly does a laser cut part creep by 5mm/m? Even on heavy wall my service providers will be within 0.3mm...