Aluminum router build - Could use some input

[Tumblebeer]

I'm about to start a build of an all aluminum router. It's not my first build, but the previous ones were MDF abominations that should have been birthed squatting over a bucket of water. Let's not mention them again.

Right now I'm in a design phase, and I'm very open to input. My build is constrained by rather small space, complete lack of decent tools, severely lacking metalworking skills and a tight budget.
I've uploaded an album with some pictures of the build, and a few FEA simulations giving a hint of where the weak areas of the construction are.

http://imgur.com/a/F77V6#0
Edit: Revised build after getting some tips - http://imgur.com/a/pM3GU
Edit: Third iteration: http://imgur.com/a/kszzz
Edit: Fourth iteration: http://imgur.com/a/zyBVw

12715

It's designed using 60x60 aluminum profiles and 20mm 5083 alu plates. I know 5083 is less than ideal, but it's what motedis.co.uk is carrying, which is where I plan to buy the bulk of the materials.
The linear parts are designed from a kit: http://www.aliexpress.com/item/3-SBR20-300-600-1000mm-set-3-ballscrews-RM1605-ballscrew-support-3-BK-BF12-3-couplers/522559269.html (http://www.aliexpress.com/item/3-SBR20-300-600-1000mm-set-3-ballscrews-RM1605-ballscrew-support-3-BK-BF12-3-couplers/522559269.html)

My questions are:
Will I be severely crippled by using 5083 instead of 6061 or 6083? If so, do you have a recommendation of an alternative to motedis? Apart from that, I am very impressed by them, without actually having bought anything yet.
I'm considering going dual ballscrews on the x-axis. But I only have 3 steppers+drivers, meaning it would be a pretty hefty investment to get everything extra for that. Is it worth it? Is it essential for success?
I'm having some trouble finding a good way to mount the ballscrew supports, any ideas?
I'm going to use a 2.2kw water cooled spindle. Any ideas on a cheap, closed loop cooling system? Also, anyone tried the "Sanven" spindles on amazon.co.uk?

Apart from that, I'm looking for any and all input on how the design could be improved. I'd like this one to be near perfect before I start buying stuff.

[routercnc]

You haven't mentioned what you want to cut, but with a machine like that you will be limited. Wood and plastic would be reasonable if accuracy was not critical - e.g. engraving, or general shape cutting.

Beyond that there are many features which limit this machine, the high sides and single ballscrew being the immediate ones.

Looking at your lateral stiffness analysis this shows 1000N gives 0.39mm. This equates to about 2500N/mm. Although this would be a reasonable start point if you could achieve it, it's not clear if the force is being applied at the spindle centre line, or closer to the gantry on the Z plate. If it is on the Z plate then the real figure would be much less. This analysis is also idealised assuming ballscrew bearings etc. are all solid. Putting this analysis to one side I would say based on your comment about 'near perfect before buying . . . ' this needs more work on the general design before you worry about material spec (e.g. 5083 vs 6061).

Your other analysis showing maximum stress of 18MPa for 1000N is not that useful since all it shows is that the machine is massively strong enough and will not break during cutting. For example yield for 6061-T6 is about 240MPa and I think your cutting forces will be much less than 1000N. But don't confuse this with stiff enough because this will be a very basic machine as mentioned above.

[Clive S]

I'm going to use a 2.2kw water cooled spindle. Any ideas on a cheap, closed loop cooling system? Also, anyone tried the "Sanven" spindles on amazon.co.uk?
I think this is the least of your worries as the spindle will only need about 2-3 litres of water in a tank to keep it cool.
Don't buy anything until you have put some sketches on here and had feed back from the very knowledgeable people on here and take the criticism in the spirit it which it is intended.
Good luck with your build. ...clive

[Jonathan]

If you want to go down the FEA route, then make sure you get the boundary conditions right - for example it looks like you've got the X-axis bearings on round rails, but constrained to the rail not free to rotate. Also, you've applied a force to the spindle mount and fixed the machine base, but it's the deflection between the spindle (tool) and the workpiece that matters - so draw a block of material on the machine bed and constrain that.

Co-incidentally I've just been working out stiffness (torsional and axial) for ballscrews including the bearing mounts - the cheap 16mm ballscrews get around 50N/um. That's an order of magnitude greater than the stiffness of the gantry, so it could reasonably be neglected.

[Washout]

If you are using ali extrusion for your frame then you may be able to get some ideas from my build here: http://www.mycncuk.com/threads/5139-It-s-begun
.
Some caveats though - my machine is compromised by using the round supported rails. Its not useless by any means, but not as good as using profiled guides/rails and bearings (especially on the Z and Y axis). I think your rails should also be mounted differently (especially X axis).
.
Most guys on here also go for a welded steel frame, as it is cheaper, generally stronger, but of course you may not have someone who can weld it for you (I also did not at the time).
.
Oh also consider using twin X axis ballscrews (one on each side).
.
I'm sure you get tons of advice and best of luck with the build.

[Tumblebeer]

Thanks for the feedback, much appreciated!

I'd like to be able to mill aluminum, otherwise I'd feel like I failed. The building materials I'm using can handle alu at the precision I'm happy with, as Washout clearly shows in his build videos. It's up to me to make sure it does.

As for the FEA, I'm not really looking at the numbers at this point. I know I've made too many simplifications and false assumtions for that. It was mostly there to get a visual feel of where the weaknesses were.

I've redone large parts of the construction, trying to listen to the advice routercnc gave me. I've also taken some inspiration from Washouts machine.
What kind of precision are you getting with yours when milling alu?
Also, what kind of problems are the round rails giving you? Chatter, instability, binding?

It's out of my league working with steel I think, even alu is stretching it.

I've uploaded another album with some pics of my revised build.
http://imgur.com/a/pM3GU

It was done in a bit of a hurry, I have to admit.
What do you think? Did I improve it? Have I missed anything, apart from the obvious mounting of the ballscrew supports on the y-axis?

[Washout]

Here's the last issue I sorted out on my machine: http://www.mycncuk.com/threads/7420-Help-on-a-couple-of-niggles-with-my-machine
.
Some profile rails rather than round rails sorted that on the Y axis. I will likely upgrade the z axis as well at some point and Jazz pointed on my thread, profile rails are a lot better than round due to better resistance to play.
.
I didn't have many issues with alignment/binding once I had taken the time to do it properly. I did have a job drilling the ballscrew holes off centre on the left hand corner posts, through the extrusions, but luckily the holes were not critical in diameter and I oversized them so I had some adjustment room when aligning the ballscrews.
.
Repeatability is good on my machine and on a part I am milling, which has 56 identical pockets on it, each pocket was within 0.02-0.04mm of each other dimensionally.
.
At the risk of "I told you so" from Robin and a debate about capabilities ;) , I am happy with the machine for wood, plastics and ali work. However, there are limitations in using >8mm cutters due to general rigidity and although I have cut 1mm slots accidentally through some holding steel washers, I am planning to buy a commercial milling machine (Warco GH) to convert to CNC and do steel and other harder materials. So I would think carefully about how much ali and metals you want to cut generally as a milling machine conversion maybe better (although you will need custom brackets etc made up, which needs CNC to start with).
.
Best of luck with your design and I'm sure others more in the know will give finer advice on your design, which looks much better to my layman's eyes than your first.
.
Cheers
.
Chris

[routercnc]

To help you get more feedback I've dropped a screengrab from your site into this post - hope this is OK.

Here is your version 1:
12595

Here is your version 2:
12594

If you are keen to stay with supported rails and aluminium extrusion then this is a pretty good layout, and much better than the first one. My view is that you will be able to cut wood easily and accurately, and the occasional bit of aluminium.

But if you end up doing a lot of aluminium then you will find that this machine is not enough. An upgrade to profile linear rails on at least Y and Z is recommended, the gantry sections are too small and not joined to each other (e.g at least a plate over the back), and the X axis stepper motors are mounted off long brackets.

With the layout you have now I would look at moving the X ballscrews outwards and down, level with the supported rails. Then extend the small horizontal plate which the gantry sits outwards and screw the ballnut to the underside of it. There are other variations but you get the idea. This will enable you to mount the motors nearer to the frame, effectively stiffening the X direction when cutting. If this is too big a change then at least make the small plates on the outside much bigger and triangulate up to the motor position.

[Tumblebeer]

I'm pretty much convinced that I should switch to profile rails instead. I guess 15mm should be enough?
Something along the lines of: http://www.aliexpress.com/item/China-made-good-quality-6-sets-linear-guide-HGH15CA-3-sets-SFU1605-ball-screw-kits-for/1489001007.html or http://www.aliexpress.com/item/6-x-HIWIN-HGH15-Square-Linear-guide-sets-3-x-SFU-RM1605-Ballscrew-sets-EK-EF/934543282.html

What should I consider when switching to square rail? Can I mount the rail straight to the extrusion t-slot? Any special considerations?

[JAZZCNC]

What should I consider when switching to square rail? Can I mount the rail straight to the extrusion t-slot? Any special considerations?

If mounting directly on profile then you'll need 20mm because 15mm won't span the slot properly. To be honest 20mm are not much more money and they also give other advantages due to there bearings being slightly higher and giving more support. This esp true when you come to build the Z axis.!

[routercnc]

For info a number of us have bought profile linear rails from 'fa-system' on ebay. They are used but the ones I bought still had plenty of life in them and have been running for a few years without problems. You'll have to cost it all up and make a choice.

Here's an example . . .

http://www.ebay.co.uk/itm/THK-SSR20-Linear-bearings-rails-L938mm-cnc-nsk-router-block-/271301904801#ht_1899wt_1006

[Tumblebeer]

http://imgur.com/a/kszzz

There we go, third iteration. I've switched to square rails, 20mm. The frame is now a mix of 60x60mm and 80x80mm alu profiles. The alu plates are 20mm, except for the backpiece spanning the gantry, that one is 10mm.

The steppers should be more securely mounted, the gantry should be stiffer and hopefully I think it might be slightly easier to assemble.

What do you think? For some reason it just looks "wrong" to me, can't really put my finger on it.

[Tumblebeer]

Made some smaller adjustments, mostly cosmetic, but should have made it slightly more stable as well. Especially the table.

http://imgur.com/a/zyBVw

Any input is most welcome, I'm about to start ordering parts unless someone has any objections and/or improvements...

[routercnc]

Looking pretty good now in my opinion. Again, if you are intent on lots of aluminium with a really good finish and short job times then the raised X-axis designs are really the way to go (i.e. not this layout).

However, with this design as it stands you should be able to cut wood, plastic, and the occasional aluminium without any problems. I'd say it fulfills your brief and will make a nice machine.

I'll just make this comment, which you can take or leave! The extrusion pieces between the gantry sides, rails, and lower plate - I can see why you've done it like that and it will help alot with the corner stiffness. But they look a little unfinished, if that sort of thing bothers you, then one of your first cnc jobs could be to cut out some 6-10mm aluminium plates to go over the ends and tidy it up a bit. The rear end cover plate could easily be larger and join all the gantry, side and lower plates together. In your link the rear view is shown in the 5th picture.

Good luck with the build and post back as you go.

[Tumblebeer]

I got the impression that I'd need either massive amounts of aluminium or preferably a welded steel frame for a raised x-axis to work. Neither would work particularly well for me (since I can't weld, nor work steel), so I think I'll go with this gantry design. I wont be doing massive amounts of milling, just the occasional part.
I'm one of those people that build mostly for the sake of building, not really with any clear goal in mind.

What kind of doc and feed speeds do you think I could expect from this? Something like 0.5mm and 1000mm/min, doing alu and still getting good finish and precision?

As for the aesthetics, I completely agree with you. As a matter of fact, I intentionally left that extrusion end open to have a fun first project to route. I would have liked making the rear end cover cover more (but not so much as to bump in to any high parts I'm cutting), but I'm working mostly with standard lengths and parts, meaning this was way cheaper and in my mind a small cosmetic tradeoff...

Thanks for the tips, I'll probably start ordering next week, need to make sure I'm completely decided first. The building will drag out a bit until the summer is over though.

[JAZZCNC]

This machine will be ok for medium duty aluminium cutting and will cut at those feeds and speeds but you will need some form of cooling with aluminium to get good results with finish. either Mist, flood or blown air.

One thing I would do is get rid of those openended profiles has the offer very little strength and they will actually make building harder.? How will you access the Bearing bolts.?
You will get the same or better strength from some simple 90deg triangle shaped plates bolted thru the gantry sides and into the bearing plates. Something like in the pics but made better looking this was just quick throw together.
This will also allow easy access to the bearing bolts and make simpler to build and adjust etc. You could drop that plate at rear or just use some Thin 2mm plate as a blank if you want to close gantry but I wouldn't and would leave open so chips fly thru.!
1271912720

Also I wouldn't connect direct to the screws with steppers but use timing belts and pulleys. This will help with resonance and give other advantages by allowing ratio to be applied if needed. It's been discussed many times on forum so do a search if you need to know more.

[Tumblebeer]

I've actually started building now. I won't really have time to maintain a buildlog here, but I am posting some status updates with pictures on my site/blog.
Currently got the main structure done, need to get the motors on there and a lot of fine tuning.

www.tumblebeer.com

The rails aren't binding, but it's not super smooth. Got any tips on how to alleviate that?

[routercnc]

The rails aren't binding, but it's not super smooth. Got any tips on how to alleviate that?

If the bearings are used they may need a strip-down and clean, followed by a re-grease - have a search in this site on how to do it and avoid loosing the balls!

But if they are new (which I think they are) they may need some grease.

More likely however is misalignment - how did you align everything? That wooden table does not look flat!
Which axis is not running smoothly X, Y, or Z? Most builders go to great lengths to get this alignment as good as possible including epoxy leveling prior to rail fitting (X axis, sometimes Y axis as well). Some builders use a DTI from a known flat reference surface, or a DTI from one rail to try and align the opposite rail.

Basically you have to remember that these types of rails are designed to be used by a professional machine tool design company and set up using machined datums etc. Have a search for the Hiwin fitment guide (either here, or on their website) to understand the lengths they expect the machine tool company to go to. You need to spend time on the alignment as best you can to get these running smoothly. As a start you can loosen one of the pairs of rails, run it back and forth, slowly tightening the rail bolts again. If this doesn't help then you may have a twist problem on the mounting surface. You may be able to help this with shims but I think you are better off getting the surface machined or epoxy levelled.

But do persevere with it because once set up accurately these units help give you a really high performance machine.

Well done on the actual build. I wouldn't have made all the design choices you went for (unconventional Z axis!), but I will say that you've made a good job of turning it into something real and that in the end you have ended up with a pretty capable machine.