4' x 4' x 9" Welded Steel Frame Fixed Gantry CNC Router

[Dillsnap759]

*Disclaimer - this is my first post here and if I've put it in the wrong thread I apologise. This may seem more like a question but I plan on updating this page with budget information and pictures and documentation as the build progresses*



Hello CNC zone members and viewers. I've fallen in the later category right up until recently today. I've been lurking in the darkness collecting the information I needed from this forum and other corners of the internet to create my own CNC router. Today I would like to present my machine expectations, model, and some of my parts. I'm going to proceed with the build with a plan of completing the build before April 20, 2021.

So lets get into it:

I've attached some screen captures of what my machine should look like, that should help add context to what I'm planning to build.



This will be my first machine, I've never done anything with a scope quite like this in the past. Here the boxes I would like to tick. Some I know are a little beyond my components, but I'm aiming for upgradability.

>250"/min or 9.8m/min Rapid (and ideally when cutting light material like soft wood)

High accelerations as to be proficient with more complex and precise CAM tool paths (I'm under the impression that this will decease cycle times of complex parts far more than faster cutting speeds)

Accuracy to be limited by ball screws. Hoping for a resolution of less <0.02" or 0.5mm but I haven't calculated my resolution yet. (I'm working on a budget, but should I fall further in love I want to know I can upgrade my components for greater precision)

Cutting hard wood, aluminum, and brass is the current goal. Limping the cutter through steel may also happen on occasion.



THE FRAME

I plan on using the 2.5" x 2.5" x 0.125" steel for the frame of my machine. I settled on a larger cross-section and thinner wall, after doing some FEA in solidworks and reducing the wall thickness until the deformation on the frame rails was within allowable tolerances. I started with .250" steel and that increased the cost and weight of the machine significantly. I plan on filling the frame tubes with sand to help dampen vibration and add mass.

I've attached a screen capture of my simulation deformation results. This test was done by fixing all 4 legs of the machine and putting a 500N force on the face of the gantry. I felt as though this would give rough, but generous estimation of the strength of the machine as the table feeds into the spindle. I've not yet taken any classes on Simulation so the results may not be worth much, but I tried.


THE GANTRY

This machine has taken inspiration from ThisOldTony's router build, and Wade'O Design's Mill. I Really like the idea of have the ball screw within the structure of the machine, but I didn't like the added complexity of using a C-Channel for the gantry. I've decided on a 6" x 4" x 0.25" A500 Steel Section for Gantry Beam. I'd like to bolt it in place to make it removable should I ever need a larger distance between my Y axis rails should I upgrade it in the future.

The Gantry's total length from end to end will be ~55". My model is still set to change as I actually procure parts for to start building.


THE TABLE

The table of my machine will be mounted to the 2 Square tubes spanning under the gantry. I'm planning to use 0.25" steel tubes for those rail supports, and they will be mounted much as ThisOldTony's are, bolted in place, trued up, then tacked in place. I've been considering adding additional supports perpendicularly between those supports as I need cross members to allow the leveling epoxy to flow, and it would allow be to support the rails from the bottom with an additional "middle Leg".



LINEAR MOTION

These decisions are entirely budget driven. I feel like I've managed to source the best deal in my area of linear motion supplies.

All linear rails will be mounted to an area flattened with Precision Epoxy Product, except maybe the Z axis which I may machine myself on my grandpa's mill.

I plan on purchasing the 4' x 4' CNC kit from Bearings Canada. All parts included in the kit are in the link below.
Link: https://www.bearingscanada.com/4-X4-...-guide-kit.htm

I am worried that 1605 ball screws will have potential for whip. But I've calculated 1605 ball screws over a 53" span with a safety factor of 1.25 at 1070RPM.

Linear Speed in m/min = RPM x Screw lead in meters/rotation
= 1070 x 0.005m
= 5.35m/min

I'm aware that this number is below my goal of 9.8m/min, I'm not sure whether or not I'm using the correct value for the lead of the screw, or if I can try to get a 10mm screw from the supplier with the kit insead of the 5mm as shown online.

This has been my major dilemma because I need to know my lead before I can appropriately size motors.


MOTORS

I've bought into the marketing behind closed loop stepper motors. The main draws for me are the quieter operation, lower power consumption, and the impossibility of losing steps ruining an entire part. My gripe is that purchasing good closed loop steppers is beyond my budget, and I'm not aware if cheap chinese closed loop steppers are any better than decent quality open loop steppers that are sized appropriately.

I haven't even gotten into sizing motors for the moving table, but I've used Rexroth Indrasize and have attached screen captures of how I setup that calculation. I used 9.81 m/s^2 for my acceleration. I'm completely unaware is that number is even remotely reasonable. I feel like if the axis moved as fast sideways as it would if i dropped it that would be quite fast, but I'm not sure. This calculation implies that I need to have a stepper with at least 2.2Nm of holding assuming the running torque is 70% of the holding torque.

I've been on ebay and have found 3 Axis nema 34 closed loop stepper kits with drivers and power supplies for less than $800 dollars. This is attractive to me, but seems to good to be true.
Link: https://www.ebay.ca/itm/3Axis-US-Wan...EAAOSwCnpcf142

Any assistance on sizing motors appropriately, and insights about chinese closed loop steppers would be appreciated.


SPINDLE

I plan on using a chinese 2.2kw spindle from ebay as well. I explicitly want water cooling for reduced noise, and I feel like this is the most power/money I'll be able to come by.
Link: https://www.ebay.ca/itm/CA-2-2KW-ER2...fpPei#shpCntId

I believe I read somewhere that lower rigidity can sometimes be remedied with more power, and I feel like the chinese spindle is the best I can afford.

CONCLUSION

This ended up becoming quite a lengthy post, but I'm really hoping to get some insights from others, and I would hate for someone to see this thread and decided they don't have enough context to help me.

I will try to be as active as possible and have set notifications to instant email. As I've said above, any help is appreciated. I'm even willing to send someone money for there time if they bring enough insight and experience.

Ultimately this is a learning experience and the goal is to learn as much as I can. I'm planning on documenting this project and using it as a resume skill for my upcoming co-op term December 2021. I also just want my own machine but refuse to spend $3k for an 12" by 12" aluminum extrusion machine from china.

Similar Threads:

• New Machine Build- Building Steel Frame Gantry Router for wood, MDF, and maybe more
• Welded steel frame router build
• New Machine Build- Welded frame fixed gantry CNC router
• New Machine Build- Lessons in overkill! Welded steel, fixed gantry router
• Welded Fixed Gantry Router/Mill

[pippin88]

Maybe your post got cut off?

What travels do you want? Why do you need those travels?
What materials will you be cutting?

You need diagonal bracing.

It is easy to make a fixed gantry more rigid - go bigger and add diagonal bracing to the uprights.

What is the size of your gantry beam?

[Dillsnap759]

Curious of how much of my post you can see, I have full caps sections in my post. I see you posted from a mobile device, could that be why it got cut off?

EDIT: nevermind they all say mobile device

I'm wanting to cut half sheets of mdf, ply wood, and I also want the vertical freedom to cut thicker material.

My gantry beam is a 53" 6" x 4" x 0.25" A500 Steel beam.
Its in my considerations to add diagonal bracing to the vertical gantry, ideally only the back side. Likely using the same 2.5" x 2.5" x 0.125" tubing as the frame.



The rest of the diagonals on the frame I could probably brace will flat bar welded across the corners.

[Dillsnap759]

So I quickly added some more reinforcement in the form of diagonal bracing for the gantry. I also added the under table supports I mentioned in my post. It's starting to look like overkill. Does that mean I'm approaching an appropriately strong frame?

[ger21]

>250"/min or 9.8m/min Rapid (and ideally when cutting light material like soft wood)

You really want double that, or even more. 250ipm is not even a fast cutting speed.

You do not want to use 5mm pitch screws on a stepper powered router.They are just too slow.10mm-20mm pitch is a much better choice.

I used 9.81 m/s^2 for my acceleration. I'm completely unaware is that number is even remotely reasonable. I feel like if the axis moved as fast sideways as it would if i dropped it that would be quite fast, but I'm not sure. This calculation implies that I need to have a stepper with at least 2.2Nm of holding assuming the running torque is 70% of the holding torque.

You'll be lucky to achieve 1/4 of 1G acceleration on a stepper powered machine, even with much larger than 2.2Nm steppers.
When a stepper starts spinning, it's torque starts dropping. A low inductance stepper will usually lose at least 2/3 of it's rated holding torque at 1000 rpm.

I've bought into the marketing behind closed loop stepper motors. The main draws for me are the quieter operation, lower power consumption, and the impossibility of losing steps ruining an entire part

If you push a closed loop stepper beyond it's ability, it will lose steps, and the drive will fault. If you have the control configured to stop all motion when a fault occurs, it may save your part. Depending on what happens when it stops, it may not.

[peteeng]

Hi Dill- Since you mention cutting metals in your brief and since the gantry is stationary I'd use a 6x6" SHS with very thick wall 1/2". This will also provide a machining allowance to true the front face where you will mount your rails. Do you intend to stress relieve the frame? You will need to construct it in pieces that can be machined after welding and preferably stress relieved before machining. These things change shape after welding and after machining. SR fixes this issue. You do mention epoxy levelling but some people have success and some don't. I feel epoxy levelling is a compromise that you should not make. SR costs similar to the epoxy and a machined surface is much better then a plastic surface ... Peter

[Dillsnap759]

Thank-you ger,

Ball Screw Pitch

I was worried the 5mm pitch would be too low, but I assumed that a company selling a kit wouldn't have put a 5mm pitch on a kit if it would be painfully slow. I'll take your word for it. The numbers don't lie. I will have to contact the supplier and see if they can substitute a different pitch into the kit for me, and for future buyers.



Acceleration

I'll try to use the 0.2m/s^s while do the indrasize calculations. Thanks for that insight.


Motor Torque

2/3 of the holding torque isn't too far off what I had found to say 70%, however 70% of holding torque means nothing without RPM over the torque curve. So thank-you for that as well.

[Dillsnap759]

Hey Peter,

Thanks for your thoughts on the gantry. I'm trying to complete this project for less than $3000 canadian, thats whats driven my decision to go with epoxy.

My steel supplier (metals pro in ottawa) that I'm looking at doesn't seem to carry sizes that substantial. The largest profile I can find that isn't prohibitivly expensive is 8" x 3" x 0.375". I can consider other suppliers however metals pro seems to be the only place around that wants to deal with me.

I had no intentions to stress relieve the frame, and I had made the assumption that getting parts machined would be prohibitively expensive for my project budget.

Would I be wrong to say that a more substantial gantry and proper machined surfaces would be bottlenecked by the small size linear rails?

Maybe machining metals is beyond my current budget

Dylan

[peteeng]

Hi Dill - There are several threads in which cars jam when screwed to std sections even before welding. the long face of an RHS is usually cupped inward. Welding will warp it somehow. For instance to epoxy level the table its in the right attitude but to level the gantry you have to turn it on its end then you have to ensure (absolutely, scouts honour, two virgin marys etc etc) the gantry is orthogonal to the table. You will have to subdivide the structure so you can tweak everything to true. Linear rails aren't stiff and will conform to its foundation in both directions. Maybe a machined subplate for the rails that screws to the gantry so you can make the motion subplate true on a bench then screw and epoxy set to the gantry of table? You do need to investigate SR costs and machining costs vs assume things. Sometimes its surprising what you can get done nearby. Peter

Re gantry I'd go 1/2" min on the gantry then. at the biggest you can get. Plus I'd go to the supplier with a straightedge and pick a piece vs someone pulling the first bit off the shelf and chucking it on a saw thinking its something for a trailer....

If you have a budget do not buy anything until you have sourced & costed all parts and drawn everything in CAD. Its only then that you have a real picture of where the costs are and how to redistribute value/performance if you have to.

[Dillsnap759]

Hey Peter,

I understand what your saying and absolutely agree that is certainly going to ensure a more accurate machine, and I may have to go with the scouts honor method of squaring the gantry. I'm going to try to make that as easy as possible for myself and measure 50 times. I'll approach some more local machine shops for quotes but so far I've had a hard time finding anyone interested. I appreciate all the advice and insight, and will surely consider all of it in my build. I share my results when I have completed that phase of the build.

Dylan

[peteeng]

Hi Dylan - I thought Canada was metric? Your gantry is 1325mm (53") this is long and wobbly under high loads. That's why its depth needs to be more (its height is good) plus the Z axis torques the gantry and square sections are torsionally stiffer then RHS sections. The gantry is sort of the heart of the machine structure. You can't make it too stiff... Maybe two 6x4's so its 8" deep...Save money by not using closed loop and put it into the structure...and I think Nema 34 is overkill. It drives all costs up, bigger drives, bigger power sources it goes on Nema 23 will spin faster and do the job. But don't think about motors too much until you have settled the structural and moving parts then you can calculate the motor size required accurately....Peter

In simulation you are aiming at a machine static stiffness of 5-10N/um (1um = 0.001mm) in all axis with the Z at max height. If you aim at 10 you will cut metal. But your FE is about 50% efficient considering bolted joints and various things so you have to aim at 20N/um in the modelling. I use 1000N for a design load then convert result to N/um. If you can apply a forced deflection you can force the tool to move 0.001mm and the FE will calculate the load to do this. Saves some maths. Keep at it and have fun. Peter

[routalot]

I might be old fashioned but I regard routers as wood/plastic cutting machines and if you have any intention of cutting metals,you ought to be giving consideration to getting hold of a manual mill and converting it.A mill is a much more substantial machine and will have a range of spindle speeds to suit metal.Operating a router motor at the sort of speeds that are required for metal will risk stalling it-unless you are only tickling the surface.

One other thing I don't see is what will be holding the workpiece in place.Double sided tape might be OK for model aircraft ribs but larger projects need something more rugged and dependable.I also think you ought to give due consideration to adding support to the middle of the table as this is where the majority of the sagging will occur.

[ger21]

Motor Torque

2/3 of the holding torque isn't too far off what I had found to say 70%, however 70% of holding torque means nothing without RPM over the torque curve. So thank-you for that as well.

I said it would lose 2/3, so you'd be left with 1/3, or ±30% of the holding torque.

My preference would be a small (3Nm) open loop Nema 34, with very low inductance, rated around 6 amps. Using a 60V power supply, these will tend to have much more high speed torque than similar rated Nema 23's. Don't skimp on drives and motors, when th3 cost difference may only be about $200-$300, as it will have the biggest performance impact.

[Dillsnap759]

Hey Peter, sorry about the late response.

I'm starting to realize I'm definitely not ready to start ordering parts yet. I've done some FEA with the 6 x 6 x 0.25 vs 6 x 6 x 0.5 and there seems to be a difference in where the stresses are within my machine. With the thicker profile most the deformation is within the machine frame. I believe that should be what I'm after. I'm going to call some more local suppliers to see about getting a larger gantry beam in the realm of 0.5" thickness.

As for Canada being metric, it is, but it's also the upstairs neighbor to one of the largest imperial economies in the world. As a result most of our raw material is sold in imperial sizes. I've actually yet to find metric steel sizes in my searching. I've been under the impression that most precision work, even in the USA, is done in metric with metric stock and most of the less precise work is done in imperial because the measurements are more tangible to most people (can visualize 30 feet, but not 10 meters).

In relevance to the machine static stiffness, I mentioned that I'm fairly new to this FEA thing. I'm using solidworks. I'm not sure how I would go about simulating for static stiffness in N/um, or doing forced deflections. Do you know of any resources I could use for that information?

I am grateful for all the information I'm getting from everyone. These were things I just couldn't find online but seem to be absolutely essential to a quality build.

Dylan

[Dillsnap759]

Hey routalot,

The plan is to do almost exclusively wood and plastics, but I would like the ability to do some light machining on aluminum, brass, and maybe the ability to cutout sheet steel. These won't be the everyday uses, but I want to know I can without being to far outside the ability of the machine.


As for workholding I've not quite figured out how to fixture my parts. I know that I will have an MDF spoilboard, but I don't know what to do underneath for rigidity. I'd like to keep it reasonably light as to keep the table accelerations high (not sure if weight is essential for accelerations but it seems to make sense). I've seen people use aluminum extrusions for there DIY projects, but aluminum extrusions are pretty expensive. I was thinking I might just make a skeleton frame from steel bars, and then fix plywood to that, and my spoilboard on top of that for wood routing. I figure I'd bolt a vise directly to that steel skeleton frame for any metal work.

Dylan

[Dillsnap759]

Hey Ger,

Yeah that's certainly a large difference in motor torque at higher RPM, sorry I read that wrong.

I was worried my budget wouldn't support closed loop stepper and was ready to accept that. Is there any particular brands or specs I should look at when sourcing for drives and motors? Is there any companies that discount or support the forum?

Dylan

[peteeng]

Hi Dylan - re FEA. Use 1000N as the design load. This is 100kgf. Place the load at the collet nose or make a dummy tool. If you use a tool say 12mm diameter with 50mm stick out, place a 1000N load in 3 loadcases at X Y and Z directions at its end. (this is to be consistent) . Make the tool very stiff eg use 5000GPa for its stiffness you want to measure the machine stiffness not the tool deflection. I also make the spindle very stiff to remove it from the deflection. My FEA has a "rigid" material selection ie it makes the part infinitely stiff. SW may have something similiar or create your own material. This is useful for isolating what is moving and what is not. Set up the machine in its "worst" condition, centre of gantry and max Z. Say the tool deflects 0.0301mm 1000/30.1= 33.22N/um which is a very stiff machine. So calculate the X Y and Z stiffness and then you have a "normalised" comparision of different designs. Also the FE model is about 50% efficient as it does not include friction connections, bolts, bearing compliance. So if you aim at a 5N/um machine the model needs to be 10N/um to get to the 5 in reality....

You mention cutting steel and this is a whole new ballgame, does not matter if you want to do it 5mins a year or every day, steel is E=200GPa and requires more power and more stiffness. To cut timber, plastic foam requires a machine stiffness in the order of 1N/um while steel needs about 10N/um and that 10x stiffness is a struggle unless you use massive parts. VMCs are about 100N/um....

Get along the road a bit more and you will find out things. Like a router spindle is way too fast to cut steel with. Look up chip loads for tools and feed speeds etc. Hybrids will lose steps just the same as open loop if overloaded. Hybrids seem to be able to run smoother and a bit faster but for design ask the supplier for the torque curve of the motor/driver they will have it. Peter

[Dillsnap759]

Hey Peter,

Thanks for once again getting back to me. I've done my simulation in the Y direction (parallel to the gantry) using 500N and a rigid dummy tool. I appears that most my deflection is in the Spindle plate. I have a max deflection of 0.110mm

So:

500N / 110mm = 4.5 N/um theoretical in Y Axis

4.5 N/mm X 0.5 = 2.27 N/um real life expectations in Y axis


Is that in the realm of what your getting at?

I'll do my next simulations at 1000N. If most my deflection is in my spindle mounting plate should I be redesigning it? or should I make it rigid as well to test the deflection in the gantry?



I'm aware of the difference in power and cutting speeds required for steel. The only reason I even mention it as a possibility is because I've seen some ThisOldTony content on youtube where he seems to do some very careful and slow machining of steel parts, and his router appears much less substantial than my proposal, however his machine is also much smaller. What I tried to imply when I said occasionally cut steel was that it doesn't have to do it well, just enough to sort of tickle through a part once in awhile, or layout holes for me to drill with larger manual equipment without destroying it self. Now that I have some baselines for N/um I should be able to do some cost/benefit by myself.

[peteeng]

Good morning Dill (at least its morning here) - Your calculation is good. Just missed the 110mm is 110um. Engineers use a term called structural loops. Any machine is broken down to these. In a gantry cnc the loops are :
tool to Z axis bearings (Z axis assembly)
Z axis bearings to gantry bearings (saddle)
gantry bearings to column bearings (gantry assembly)
column bearings back to the job (machine base assembly)

Generally in the early stages we support the frame at its feet but the tool load does not go to the feet it returns thru the table to the tool. or restrain at the column feet similar thing.

So you can restrain the Z axis bearings do the run and you will find its say 200N/um
Restrain at the saddle its 100N/um
restrain at the column bearings and its 50N/um
(made up numbers) you can then say which loop is contributing to the compliance the most and then work on that loop. In the prior numbers the gantyr contributes 50N/um to the stiffness/compliance so I'd work on that and rerun. If your aiming at 10N/um then roughly the z axis will be 3, the gantry will be 3 and the rest will be 3N/um. Its basically a stiffness pyramid.

Regards Peter

[Dillsnap759]

Hey Peter, good morning to you, it's almost dinner here

I do want to mention that my Z axis rails are mounted to my spindle mounting plate. This should increase my rigidity substantially when my z-axis is in its top 2 or 3 inches of travel. I plan on building a riser for my work surface to facilitate this. Assuming my gantry is rigid this would be much more likely to allow machining of metals would it not? I understand that this would also have to be paired with an appropriate spindle, but it leaves room for upgrading.

I'll post some simulation screen captures with some more details at a later date once I get some simulations done. I'd like to give my conclusions on the results, and hopefully someone on the forum can verify my conclusions are on the right track. After that I'll probably do 100 redesigns and present some more completed designs.

Thanks Peter. Your insights have been invaluable