[ScottMcCormick]

Hey CNC junkies,

Let me first start off by saying that there is no build that is impossible. That being said, I've heard a lot of people state that this build is impossible, but CAD models and FEA suggest that it may be feasible within our desired tolerances (+/- 0.003"). I plan on using this thread as a method of documenting the design for others to collaborate, and ask questions about the design as we run into problems.

Our shop is looking to design and fabricate a 4'x4'x30", 5-axis CNC router that will do 70% foam, wood, composite work and 30% light aluminum work. The proposed budget is currently at ~$10K-$15K. We are in the aerospace industry, so we work with aircraft tolerances, which means this machine has to be very rigid. We are using 4"x4"x0.25" ASTM A500 Grade B Steel for the frame, and 2"x2"x0.120" steel tubing for the structural members. The bed will be made out of a dense wood or composite material, and will include a lattice structure style torsion box. We will be implementing a dual-driven gantry, also made out of steel, and we plan on using rack and pinion drives (the kit from cncrouterparts.com) for the X, Y axes that drive the gantry and Z-axis assembly. The Z-axis itself will be driven using a 48" ballscrew. The gantry and X-axis will traverse along Hiwin Profile Rail guides, and the Z-axis will traverse along continuously supported shafting. The 4th and 5th axis capabilities will come from the Doughty Drive B/C rotary head. I am confident in our abilities to make this system rigid, and for the mechanical and structural systems to be sound.

What I am less confident in, and is the problem we are currently facing, is the electronics and wiring system. We plan on using NEMA34 motors to dual drive the gantry along the Y-axis, to drive the Z-axis assembly (cutting arm) along the X-axis, and the Z-axis ballscrew itself. We will use NEMA23 motors to drive the B/C axes on the head itself in which the spindle sits. That gives us 6 axes to control, and I'm unsure of what control board we will need. I have done some research, and found a 6 axis control board, but I'm worried that this won't allow for the 3 axes + the slaved axes for the gantry, as well as the 2 rotary axes. Does anyone have any experience with this, or know if this will work? We are going to use Mach 3.

In summary, I have a few questions for those of you who've done this before:
- Will the 6 axis control board I've posted a picture below handle what we are trying to achieve?
- Suggestions for spindle head (air cooled, water cooled, 2.2kW?, etc.). Completely newbie to this part of the design.
- Software for generating the toolpath and gcode for 5-axis capabilites (I've heard cnc-toolkit, AutoDesk Fusion)?
- Other tips for the electronics side of things; again, complete newbie with this aspect of the design

Any constructive criticism or suggestions are welcome and appreciated. This is turning out to be a huge endeavor, but I'm enjoying it and I believe that the final product will pay off. I've posted a few visuals from our current design iteration to hopefully help clear things up.




Note: the models are very preliminary, most of the mates and geometry are there, but some components like the rack and pinion drive are not in their final places.

Please help if you can!

- Scott

[Andrew22]

If you loosen up your tolerance requirements you might get by. Nothing wrong with rack and pinion for some machines, but you can't expect to hold the same tolerances as a ball screw. I also wouldn't use mach3. I used UCCNC for my router and it's excellent for most work, but I've probably come up to its limits with high speed 3D surfacing. It can control up to 6 axis, but I imagine any open source controller will have some issues running 5 axis tool paths depending on how fast you want to run.


Sent from my iPhone using Tapatalk

[NIC 77]

You know what? I just wasted an hour of my life writing you a long message telling you all about what is wrong with your design and making suggestions…….

Then I looked at it from a fresh perspective and realized that you are not being honest.

Are you telling us the truth about having a shop that does work in the aerospace industry? The more I’m looking at this, the more I think, OMG! I just wasted an hour trying to help someone out who’s having a laugh at my expense! No one who had a viable business in the aerospace industry would come up with a design as bad as this one and expect to cut aluminum in 5 axis with it!

So no free advice for you!

[Jim Dawson]

Given your requirement for working to somewhat tight tolerances and you have a reasonable budget for materials, you may want to concider a more industrial motion controller. Take a look at Galil Motion Control products, most of their products are compatible with Mach3 and other CNC software. Not cheap, but they will do the job very well. Also open loop stepper systems are not the best for accuracy. Take a look at hybrid steppers, AC servos or better yet, put 1 or 5 micron magnetic encoders (about $130/axis) on the load to close the loop through the motion controller, works with steppers or AD/DC servos.

[awerby]

I can see this cutting foam pretty well, wood if you do a good job building it, but aluminum not so much. If aluminum is really a must, then you probably want to build something a lot more massive. As Andrew22 points out, your tolerances are a bit unrealistic; the rack and pinion alone will contribute about that much slop. Once you get out on the end of that long lever, when the Z axis is fully extended, deflection will be a bigger issue. I wouldn't recommend using round rail anywhere in the design, supported or not. Get the best electronics you can find, not just the cheapest, but wait on all that until you've actually built your frame (things can change, and you can spend the time learning about all the alternatives). The frame could use more bracing too - at least corner braces, if you don't want to block access with cross-bracing. I'd also suggest attaching the Z axis assembly directly to the gantry rather than offsetting it on those stubs. The gantry itself would be more rigid if those 2 tubes were connected along their lengths.

Are those Doughty drives backlash-free? If not, there will be additional error introduced there. If you really plan to use one of those Chinese spindles, I'd suggest paying more and buying it from a US-based supplier who will make sure it really works, and doesn't have the internal ground missing, as often happens with some of these things. As to the software, I don't think that Fusion supports 5-axis toolpaths - you'd need HSM Ultimate, which goes for $3700 per year (or whatever they decide to charge in the future). The Gmax- CNC toolkit approach is difficult if not impossible to run on modern computers; I haven't heard of anyone having much luck with it lately. My best suggestion would be VisualMill Premium, which isn't cheap, but I can get you a good deal on it.

[Jim Dawson]

Another thought comes to mind since you are wanting a very rigid machine. Maybe concider a moving table rather than a moving gantry, a much more robust design. Would add a couple feet to the overall length, but I suspect that in your case floor space is not a critical concern.

[ger21]

First, I would not recommend any chinese controller for use with Mach3. There are several high end, high quality Mach3 motion controllers.
At the lower cost end, there's the UC300ETH. Stepping up in cost, there are very good controllers from CS Labs and Vital Systems.

However, I would not recommend using Mach3 to anyone building any type of new machine. There are many better options, often for less money.

LinuxCNC or Pathpilot, with Mesa control boards.
UCCNC, with the UC300ETH.
KMotion with a KFLOP.
Planet CNC, which has several different boards.
EdingCNC, which also has a few different boards.

All of these are arguably better then Mach3, have better support, and are all still currently being developed. Mach3 development stopped 5 years ago.


As others have mentioned, I seriously doubt that you can get any kind of decent finish on aluminum, or come close to your tolerances, with this design.

[louieatienza]

You know what? I just wasted an hour of my life writing you a long message telling you all about what is wrong with your design and making suggestions…….

Then I looked at it from a fresh perspective and realized that you are not being honest.

Are you telling us the truth about having a shop that does work in the aerospace industry? The more I’m looking at this, the more I think, OMG! I just wasted an hour trying to help someone out who’s having a laugh at my expense! No one who had a viable business in the aerospace industry would come up with a design as bad as this one and expect to cut aluminum in 5 axis with it!

So no free advice for you!

Pretty derogatory remarks from someone who has also asked for advice here on the forum.

__________________________________________________ __________________________________________________ _________________


Scott, some thoughts on your design elements:

1) The design as it stands does not look very rigid at all. Take a look at the majority of 5 axis portal type CNC machines, and look at how they are designed, particularly the Z carriage, and it should become readily apparent how they're designed and why; there's also no reason why NOT to use profile rail on the Z. I also don't think a wood or composite wood table would lend itself to accurate parts, especially on a day-to-day basis. I would use phenolic sheet as a minimum and preferably Blanchard ground aluminum or steel plate. You then could use it as a reference surface to machine or form (via epoxy) the mounting surfaces for the rails. The Doughty Drive site claims zero backlash, though I'm a bit skeptical of those claims. They also are either slow or non-existent to respond based on several attempts at contacting them, unless you have better luck.

2) You should be able to slave any axis with whatever controller you choose. However if accuracy is paramount, I would not use steppers. I definitely would consider AC servos, and they can be had at reasonable prices nowadays. You would have to modify the CNCRP assembly if you decide to use them. But it would be better to source some servo grade planetary gearheads and mount the pinion directly to that. I would also heed Gerry's (and others') advice and forget about Mach3 and that controller.

And some direct answers to your questions:

1) Like we say in NY - "Fuhgeddaboudit!" Between the spotty documentation, questionable components, and lack of direct support, you'll pull your hair out every night till you go bald. Then you'll be anxious till you have a nervous breakdown, praying that the day that drive fails in the middle of a 24-hour run - on a deadline - never happens. Buy something reliable and reputable.

2) Since you're primarily cutting foam, I'd suggest a water-cooled spindle. The fan from the air cooled spindle will blow foam dust all over.

3) If you buy 5-axis CAM, it may cost as much as your entire machine budget. That said, SprutCAM, BobCAM, and VisualMill may be the most affordable solutions, and even they are not cheap. I would suggest checking out their sub-forums here and see what others have to say about them first, and download a trial to see if it will work for you.

4) I'd suggest incorporating some type of riser table for doing the majority of your work whenever possible, this way your Y/Z will be at a more rigid configuration. I would use ballscrews on all axes, with double-nut configuration. And since you don't have to move these axes at lightning speed, 5mm pitch would be fine, and you wouldn't need a gearhead to drive the screws - which would lend to a more rigid drive system. Again, do a search on 5-axis designs, and look at how they're designed.

[NIC 77]

Pretty derogatory remarks from someone who has also asked for advice here on the forum.

I’ve always appreciated your input whenever I’ve had a question. I admit I was a bit harsh in my response. I certainly don’t want to make enemies on this forum. If I’ve offended you Louie, which I think I have, then I sincerely apologize to you.

When I took a second look at it, all the talk of CAD models and FEA analysis, cutting aluminum, confidence in abilities to make the system rigid, and a working shop that manufactures for aerospace, my BS sensor engaged.

And then I looked at the design again and there are two bearings for the Z axis attached by 2 inch square tube at a right angle, and I thought, OMG, this can't be true, I've been had.

I started wondering if this was perhaps posted by someone in China who was looking to advertise their product without overtly doing so.

I certainly wouldn’t have made the comment that I did if this was posted as a beginner making a 5 axis hobby machine.

Look, we all make mistakes and sometimes say things that are too harsh. I’m human. I do this sometimes. This is probably one of those times for me.

Scott, if it turns out that I am wrong, then I will apologize to you as well.

Good luck with whatever you do. Louie has more experience than me, and has given you good advice.

The only other thing I will say is that if you’re for real then perhaps it's better to buy something and adapt it to your needs. I was looking at a thread earlier about a “Saturn” CNC machine from finelineautomation. Looked very nice. Perhaps you could ask them to make you one where the table sits much lower.

Of course, I don't know if they ship to China.

Apologies to everyone for the harsh nature of my earlier response.

[ger21]

3) If you buy 5-axis CAM, it may cost as much as your entire machine budget

Not if you go with Fusion 360, at $1500/year.
A single 5 axis job should be able to pay for that.

[awerby]

Not if you go with Fusion 360, at $1500/year.
A single 5 axis job should be able to pay for that.

Have you been using Fusion 360 for 5-axis work? From what I can glean from their website, it only offers 3-axis capability at this point: https://www.autodesk.com/products/fusion-360/features To generate 5-axis toolpaths with an Autodesk product, it seems that one has to go to HSM Ultimate, which is currently priced at $3700/year : https://www.autodesk.com/store/produ...pport=advanced Or am I missing something?

[ger21]

No, I haven't had time to follow the development.
I just looked, and 3+1 and 3+2 has been available for several months now, and continuous 5 axis is coming. Might be a year, maybe longer, but should be available within a year or so I would think. Autodesk said they were working on it 6 months ago from the roadmap I saw.

https://www.youtube.com/watch?v=cHVfTIhosC0

[daniellyall]

there's 2 5 axis tool path at the moment, plus a beta that comes and goes depending if they break it after updating it, the rest can be used for indexing, there's 3 toolpath and 1 bata for wrapping there are some updates to wrapping coming ones they have it working on apple in windows.

[louieatienza]

No, I haven't had time to follow the development.
I just looked, and 3+1 and 3+2 has been available for several months now, and continuous 5 axis is coming. Might be a year, maybe longer, but should be available within a year or so I would think. Autodesk said they were working on it 6 months ago from the roadmap I saw.

https://www.youtube.com/watch?v=cHVfTIhosC0

I thought a 1 year license of 5 axis Fusion came with the PocketNC machine...

Another option would be DeskProto, but it's only 5th axis positioning not continuous...

[daniellyall]

It's meant to come with fusion ultimate Lou, startup under 100k is free, hobby is free per year, student is 3 years free this is all ulitmate. you can only get standard by paying for it, its 3 axis only.

You can ball**** the wrapping useing sketches to do continuous on certain geometry types.

[wizard]

Your reply might have been harsh but i think the BS detector was right on. The reference to doing FEA and the indication that the machine would be rigid enough had me shaking my head and im nit even a FEA expert.

Now that being said i think we can get them on track to a successful design. We do need to cut through the BS though.

I’ve always appreciated your input whenever I’ve had a question. I admit I was a bit harsh in my response. I certainly don’t want to make enemies on this forum. If I’ve offended you Louie, which I think I have, then I sincerely apologize to you.

When I took a second look at it, all the talk of CAD models and FEA analysis, cutting aluminum, confidence in abilities to make the system rigid, and a working shop that manufactures for aerospace, my BS sensor engaged.

And then I looked at the design again and there are two bearings for the Z axis attached by 2 inch square tube at a right angle, and I thought, OMG, this can't be true, I've been had.

I started wondering if this was perhaps posted by someone in China who was looking to advertise their product without overtly doing so.

I certainly wouldn’t have made the comment that I did if this was posted as a beginner making a 5 axis hobby machine.

Look, we all make mistakes and sometimes say things that are too harsh. I’m human. I do this sometimes. This is probably one of those times for me.

Scott, if it turns out that I am wrong, then I will apologize to you as well.

Good luck with whatever you do. Louie has more experience than me, and has given you good advice.

The only other thing I will say is that if you’re for real then perhaps it's better to buy something and adapt it to your needs. I was looking at a thread earlier about a “Saturn” CNC machine from finelineautomation. Looked very nice. Perhaps you could ask them to make you one where the table sits much lower.

Of course, I don't know if they ship to China.

Apologies to everyone for the harsh nature of my earlier response.

[louieatienza]

Your reply might have been harsh but i think the BS detector was right on. The reference to doing FEA and the indication that the machine would be rigid enough had me shaking my head and im nit even a FEA expert.

Now that being said i think we can get them on track to a successful design. We do need to cut through the BS though.

I'm not so sure... I've seen a few FEA done on YouTube usi g Fusion360 and it put out results that surprised me.

[wizard]

Hey CNC junkies,

Let me first start off by saying that there is no build that is impossible. That being said, I've heard a lot of people state that this build is impossible,

You can build anything the question is will it meet your expectations. Im going to say right off this machine will have trouble with anything beyond foam.

but CAD models and FEA suggest that it may be feasible within our desired tolerances (+/- 0.003"). I plan on using this thread as a method of documenting the design for others to collaborate, and ask questions about the design as we run into problems.

How much experience does your FEA guy have? I ask because i highly doubt the Z woild hold postion that you want being pushed by my pinky. 30" of Z is a seriously long lever arm, as designed it is going to deflect mire than thr 0.003" you are asking for under load. Not to mention issues with vibration.

This is based on industtial experience with all sorts of machinery not FEA. In any event to clarify communications here, im going to refer to the acises as X, Y, Z from bottom to top. (It is a pet peeve to be sure)

In any event your lever arm is long enough that i suspect you will get deflection out of the X axis frame members, especially in the center. In other words your frame is pretty weak for the tolerances you are trying to achieve.

Our shop is looking to design and fabricate a 4'x4'x30", 5-axis CNC router that will do 70% foam, wood, composite work and 30% light aluminum work. The proposed budget is currently at ~$10K-$15K.

Well at least your budget is some what realistic. Given reasonable budgetting i would strongly suggest beefing up with respect to your steel members. You might want to run FEA but id suggest at a minimal 3" square tubing for the frame and even that might be thin. The good thing with the frame is that you are free to reinforce it as you see fit. Id certainly add more bracing but then you need a way to load the machine. Honestly with a 4' span id seriously consider 4" square tubing

We are in the aerospace industry, so we work with aircraft tolerances, which means this machine has to be very rigid. We are using 4"x4"x0.25" ASTM A500 Grade B Steel for the frame, and 2"x2"x0.120" steel tubing for the structural members.

.
That is one very confusing statement. The frame is a structural member, so is the frame 2" tubing or 4". Also use tubing that is as thick as you can source. The thickness may not impact your machines rigidity that much but it will do wonders for screw holding which can eliminate the need to weld pads for bolted on components.

In any event your gantry beam needs to be far larger. There is an excellent thread in the stickies that covers gantry beam design. Your problem here is the long lever arm of the Z axis which will twist the beam.

The bed will be made out of a dense wood or composite material, and will include a lattice structure style torsion box.

If you are buying steel buy a piece of blanchard ground steel and suppirt it with steel matching the rest of your frame members. The thing here is that the minute you start machining aluminum you will want to use coolants/lubes. This will raise hell with wood based structures.

We will be implementing a dual-driven gantry, also made out of steel, and we plan on using rack and pinion drives (the kit from cncrouterparts.com) for the X, Y axes that drive the gantry and Z-axis assembly.

Why rack and pinion? For a machine this size and the tolerances wanted anti backlash screws are in order.

The Z-axis itself will be driven using a 48" ballscrew. The gantry and X-axis will traverse along Hiwin Profile Rail guides, and the Z-axis will traverse along continuously supported shafting.

You have a realtively robust budget so why cheap out on linear bearings? The Z is probably the last place you would want to use round rails.

The 4th and 5th axis capabilities will come from the Doughty Drive B/C rotary head. I am confident in our abilities to make this system rigid, and for the mechanical and structural systems to be sound.

What I am less confident in, and is the problem we are currently facing, is the electronics and wiring system. We plan on using NEMA34 motors to dual drive the gantry along the Y-axis, to drive the Z-axis assembly (cutting arm) along the X-axis, and the Z-axis ballscrew itself. We will use NEMA23 motors to drive the B/C axes on the head itself in which the spindle sits. That gives us 6 axes to control, and I'm unsure of what control board we will need. I have done some research, and found a 6 axis control board, but I'm worried that this won't allow for the 3 axes + the slaved axes for the gantry, as well as the 2 rotary axes. Does anyone have any experience with this, or know if this will work? We are going to use Mach 3.

You may want to completely reconsider this. Frankly it is better to put off electronics untill you have a good idea of what the final mechanical design is.

In summary, I have a few questions for those of you who've done this before:
- Will the 6 axis control board I've posted a picture below handle what we are trying to achieve?
- Suggestions for spindle head (air cooled, water cooled, 2.2kW?, etc.). Completely newbie to this part of the design.
- Software for generating the toolpath and gcode for 5-axis capabilites (I've heard cnc-toolkit, AutoDesk Fusion)?
- Other tips for the electronics side of things; again, complete newbie with this aspect of the design

This is a tough one because if you are serious about precision you will likely need linear encoder feed back. This puts you into a different class of controller than is commonly discussed here. You need to nail down your specs a bit more. If you want to hold position to 0,003" over 4' you will need linear scales as the most economical way to achieve that. Im not sure what your real needs are here.

Any constructive criticism or suggestions are welcome and appreciated. This is turning out to be a huge endeavor, but I'm enjoying it and I believe that the final product will pay off.

Such a project isn't impossible however dont underestimate the extra efforts required to do something that isn't mainstream. DIY 5 axis machines aren't mainstream yet. This means some trail blazing on your own.

You need to consider how the machine will go together and what local machining resources you have. To be perfectly honest you could get away with foam machinig with only a few enhancements. Once beyond that though i suspect you will have trouble. But back to the machine itselfhoe the machine is assembled becomes an issue in getting goid results.

First you will want to consider having the frame stress relieved. Second you will want to find a shop with a mill large enough to mill the mounting pads for the linear rails. You may even need to have pads welded to the frame to have enough material to mill the pads in place.

This is all based on your rather tight precision requirments. It would be rather easy for a welded frame to distort sometime after assembly causing you accuracy problems.

You also have the issue of getting the table parallel to the gantry and X axis. This might require a two piece frame.

I've posted a few visuals from our current design iteration to hopefully help clear things up.

Im not getting a warm and fuzzy feeling here. The gantry just doesnt look robust for work in aluminum.

Note: the models are very preliminary, most of the mates and geometry are there, but some components like the rack and pinion drive are not in their final places.

Please help if you can!

- Scott

I hope that i have helped some. My biggest fear is that the machine wont meet your needs in aluminum and maybe even wood. This mainly due to 0.003" being tight on most routers. When you combine a long lever arm with rack and pinion drive and some other design choices i dont see happyness.

By the way a decent fifth axis head could be an issue here. This especially if you really want to swing around a 2kw spindle motor. You can blow your budget right here. You may also find yourself having to DIY the fifth assembly to get the results you want. 2kwof power is a lot in alow end fifth axis solution.

Which brings up a final considerstion do you really need the large spindle? Im assuming that you want that large of a spindle because you expect to use all of that power. Only you know how you expect to use the machine but the fact is if you are using all of that power you will be causing reaction forces that will deflect your Z. Not to mention the forces on the Fith extension.

One other thing here now that we have considered the spindle. Your frame has to be large enough to clear the spindle in all orientations which becomes worse with the larger and longer spindles. If you don't factor this in then you loose a significant amount of work area. Not to mention the damage each crash will cause.

Somebody else mentioned a moving table design. This may be a good idea and might actually save you space that would be otherwise used up by an oversized frame. With your design your frame has to be oversized enough to alliw the spindke to clear all four cumns. On a moving table design you could design in clearance around the suppirt columns. Even if you found yourself spreading the two columns out you still wouldn't lose width for the entire machine. Even lengthwise a moving table wouldnt be that bad because your saddles the gantry beam rides on need to be fairly wide to resist the forces involved. This all adds up such that a moving table design might make sense even at this larger scale.

A moving table design may also solve the prblem of liading and unliading the machine and also address ease of tool changes. Tool changes will be intererting with you design as you will likely have to reach a considerable distance.

Hope this helps.

[louieatienza]

Your reply might have been harsh but i think the BS detector was right on. The reference to doing FEA and the indication that the machine would be rigid enough had me shaking my head and im nit even a FEA expert.

Now that being said i think we can get them on track to a successful design. We do need to cut through the BS though.

If there's one thing I've learned in 20 years of participation in online forums, is to never, ever assume or make judgment upon, the intentions of the OP. It's so easy, as it turns out, to fall into this trap, and I've admittedly done so on occasion; happens to the best of us. The term "aerospace" can be pretty broad and it doesn't always mean milling a bulkhead out of a gigantic billet of 7050. I t could be taking a stamping or formed sheet metal and cutting holes or ports into it. He could be a startup. And like I said, I've seen some FEA results with Fusion360 posted online that gets my head itchy. One for example recently where the designer is using relatively small rectangular steel tubing, with 16 or 20mm supported round rod and open round bearing blocks, and claims all the bending on the machine is at the carbide endmill that was modeled, with absolutely no deformation anywhere on the machine. But I think not, something seems to be amiss - his scenario may be true if he was trying to push that endmill though the material - without it spinning (and therefore cutting.)

[louieatienza]

In any event to clarify communications here, im going to refer to the acises as X, Y, Z from bottom to top. (It is a pet peeve to be sure)

A pet peeve of mine is the use of the plural of axis - which is axes (pronounced ax - sieze)

As for nomenclature of the naming of each axis, I default to the standard Cartesian naming used since antiquity - X referring to left and right, from the viewpoint of the operator, Y going away and toward, and Z going up and down. End of controversy!