[OneWound]

Hi Y'all,

Currently I am finalizing a CNC gantry design that will be 10'x5'. As I finish this and go on to how it will be assembled, I am running into a worry of aligning everything properly. On the Y gantry, which is 4x8x.5x71", I am worried about twist angles. Is it possible to shim the whole entire rectangular tubing? Or am I better off to machine all faces? I am hoping to hold tolerances within .001 with this machine (Hiwin ballscrews and linears are being used, and a scanning system will be used to verify everything is parallel/perpendicular to itself to accomplish this). I also know how to machine it, I just don't know what option is better. Thoughts?

[DonKes]

I hope April Fools day is not involved in your .001 tolerance!!!

Don

[dmalicky]

It would be very difficult and costly to make a 5'x10' machine that holds a 0.001" tolerance over much distance (assuming you mean accuracy, not resolution). Not even the ballscrews will be that accurate. I'd suggest doing more much research to understand realistic expectations.

[lucky phil]

Hi I have threaded holes and studs out the ends of my gantry section (100mm x 200mm rhs), on the mating surface of the brackets that are on my Y axis rails I have welded flat plates with set screws in from front, back & bottom which allows adjusting the X axis gantry to obtain a very accurate alignment, parallel with bed and rotating for tramming the spindle.
My gantry is 1.5m wide.

[OneWound]

Why aren't ballscrews that accurate? Isn't that what big companies like DMG Mori, Haas, etc are using (along with linear rails, ofc) using on their machines and they are able to achieve .0001 positional accuracy? I understand .001 is a high goal, considering this is a homemade gantry, and I'd be happy with .005 honestly. But considering this CNC will have more than 10k put into it, as well as using actual CNC machines to machine every component to a knats ass accuracy, and CMM machines to verify everything within the gantry is where it should be to an accuracy of .001 or greater, I don't see where I am losing huge amounts (~.002 in this case) of positional accuracy?

[OneWound]

One thing I do wonder though, would aluminum extrusion be rigid enough to machine foam? The extrusion can handle the load just fine, my only concern is that at 500ipm in foam it'll shake like crazy. Thoughts?

[wizard]

Why aren't ballscrews that accurate? Isn't that what big companies like DMG Mori, Haas, etc are using (along with linear rails, ofc) using on their machines and they are able to achieve .0001 positional accuracy?

There are a number of ways to do this with various degrees of success. Linear scale feedback is one possibility as is lead screw mapping. The problem with most router implementations is that the lack of stiffness means that holding position is an issue. That is you might get an axis to "position" to 0.001" but if you are actually doing anything the machine will deflect off position.

I understand .001 is a high goal, considering this is a homemade gantry, and I'd be happy with .005 honestly. But considering this CNC will have more than 10k put into it,

10k is nothing if you want to implement prison ground lead screws and a CNC controller that can work with high accuracy and very long linear scales.

as well as using actual CNC machines to machine every component to a knats ass accuracy, and CMM machines to verify everything within the gantry is where it should be to an accuracy of .001 or greater, I don't see where I am losing huge amounts (~.002 in this case) of positional accuracy?

If you have access to a machine shop and that access doesn't cost you anything you have the potential to machine everything to a high degree of accuracy. Getting that to assemble accurately isn't a piece of cake. For example getting the Y axis gantry square to the X axis base would require access to the metrology equipment to actually measure the squareness of the assembly. Even then 0.001" is pretty demanding over 5 feet.

Honestly I don't even think it is reasonable to talk about 0.001" over 5 feet, maybe one foot. You would need a lot of machin tool building experience and engineering experience. For a moving gantry machine I'd see these sorts of positioning tolerances as a pipe dream.

[wizard]

One thing I do wonder though, would aluminum extrusion be rigid enough to machine foam? The extrusion can handle the load just fine, my only concern is that at 500ipm in foam it'll shake like crazy. Thoughts?

Depends upon the extrusion! If it is a box section extrusion of suitable cross section it can certainly be strong enough. I'm not talking t-slot extrusions here either. As for sakes or vibration that is something that would have to be analyzed beyond simple deflection calculations.

In any event you need to focus your posts a bit on what you intend to do with the machine. I can't see a realistic reason to have 0.001" tolerance, over 5 feet mind you, in a machine built to machine foam.

[OneWound]

Depends upon the extrusion! If it is a box section extrusion of suitable cross section it can certainly be strong enough. I'm not talking t-slot extrusions here either. As for sakes or vibration that is something that would have to be analyzed beyond simple deflection calculations.

In any event you need to focus your posts a bit on what you intend to do with the machine. I can't see a realistic reason to have 0.001" tolerance, over 5 feet mind you, in a machine built to machine foam.

Well, can you tell me what equations would be needed past simple deflection?

As far as everything else, I work in a research lab (at a university) where they have CNCs, and I am qualified to use them. As far as metrology equipment, I also have access to them and experience to use them. To help with positional accuracy, I will be using a lot of dowels (as I have figured, it'll be easier to use aluminum for everything).

Question, however. So pushback from the foam will cause inaccuracies. So how do they get linear feedback? Is it something to do with measuring how much torque the servos see as they move? Or what is it?

[dmalicky]

Great reply from wizard, as usual.

Here is the sticky on ballscrews: http://www.cnczone.com/forums/linear...-software.html
Even a C3 ground ballscrew is very expensive, and only good for 0.00027" per foot. Over 10', that's almost 0.003". As wizard said, they can be mapped, or glass scales in a closed loop system, both with the right software. After that we also need:
- no sag over the 10'
- no whipping due to critical speeds
- a pure guess is that a 10' long ballscrew would need to be on the order of 50mm diameter to not sag or whip at reasonable rpm, and it would still need auxiliary support -- I wonder how much a 10' C3 50mm dia ballscrew would cost?
- a controlled temperature (high end CNCs use ballscrews with internal liquid cooling to control their temp) -- maybe could skip this with glass scale feedback.
- a preloaded double ball nut ($$) and preloaded high precision double angular contact support bearings ($$).
And that's just the mechanical drive for 1 axis!

The most basic challenge with high accuracy is temperature. Aluminum's coef of expansion is ~13 * 10^-6 in/inF. So just a 5F rise in temp of the machine makes the 10' dimension bigger by 0.008". Steel and cast iron move about half as much. So right away, we need a uniformly temp-controlled room, and maybe active cooling for the ballscrews and other parts.

Other challenges with accuracy are:

Room issues: floor rigidity and stability (in addition to temp). A regular concrete floor may move over time. Large CMMs aren't just machines, they are installations -- same for CNCs.

Component quality, which translates as $$, good decisions, and experienced attention during installation.

Alignment: linear (easiest but not easy), 2D (harder but there are techniques), 3D (hard), and how to maintain alignment over time (hard if the machine is not massively built out of proper materials). In 1D, for example, linear rails are not straight when you buy them, so we can't rely on them for linearity. The usual technique is to machine a highly accurate registration ledge the full length (into a very stiff and stable frame), and force the linear rail against the ledge. Next we need to get the other rail extremely parallel to this master rail, mount it there, and have some way to ensure it stays parallel (what if you move the machine?). In 2D and 3D it gets more complex.

Dimensional stability over time: will residual stresses relax and cause parts to shift over time?

Rigidity, vibration, and damping due to cutting forces, inertial loading, and local forces: FEA is good for this, but accurate modelling of a CNC machine is not simple. Some things can be certainly calculated with equations and analytic methods, e.g., how bearing stiffness and slop geometrically propagates to the cutter tip. For the complex geometry, structure, and loading of a CNC router, only experienced use of FEA will do a good job of predicting stiffness at the cutter, or vibrations of the system. The consensus view of experienced machinists is that it takes massive amounts of cast iron and $ to achieve high accuracy over a large distance.

Machine design/configuration: CNC routers are complex machines with many issues. Assuming someone already has a mechanical engineering degree and many years of machining experience, I'd guess it would take about a year of CNC-specific study, *from the right people*, to be qualified to design one that could hold 0.001" 3D accuracy over a long distance, for years. I'm an experienced ME, have done design/build projects for 35 years, have been studying CNC design on and off for 5 years, and I'm not sure that I could do it (nor do I have much interest in it... seems like a bunch of headaches!) This kind of accuracy is just very difficult and complex.

So hopefully you can see why 0.001" accuracy over a long distance will cost an amazing amount of $, and more importantly, a team with lots of knowledge and experience.

For a point of reference, here is Multicam's top router, I'm guessing about a $70k machine: 7000 Series CNC Router - CNC Cutting Machines for Your Application & Budget | MultiCam
They only spec a repeatability of +/- 0.001". No accuracy is stated because it's considerably worse than that. But few people need 0.001" accuracy over a long distance.

Some suggestions:
There is a lot of good info in the archives -- search and learn, follow current threads -- for a few months.
Also search on practical machinist dot com -- lots of experienced machinists there sharing their knowledge, especially the user 'milacron' -- I have learned a lot from his posts.
Spend as much time as you can in the shop at your university, and learn from the smartest machinists there. Ask to be involved in their special projects.
Don't order parts until you've done at least a few months of external study, refined the design through many iterations and analysis, and it's been vetted by others. Exception: ordering some sample parts is often helpful, especially when less experienced.
Design and build a smaller machine first. You'll learn a lot by just completing one, and it's much easier to build a small one than a big one. Then you can make the beginner mistakes on the little one, and have much better skills to build a big one.

[wizard]

Well, can you tell me what equations would be needed past simple deflection?

Actually I can't as I'm not a mechanical engineer familiar with such modeling. I work as an automation technician and basically have worked on machine tools and automation machinery since leaving high school.

As far as everything else, I work in a research lab (at a university) where they have CNCs, and I am qualified to use them.

That is absolutely huge, just having machinery to work with puts you way ahead of the average builder here. For your own benefit you might want to do a quick inventory of the machinery available and the capacity of each. You may already have a good feel for this but the idea is to be prepared to design for the machinery you have available.

As far as metrology equipment, I also have access to them and experience to use them. To help with positional accuracy, I will be using a lot of dowels (as I have figured, it'll be easier to use aluminum for everything).

About that aluminum, I'm not a big fan of aluminum for basic machine structures. One big turn off is that it is expensive. It also has a lot of shortcomings including the ability to hold screws. The only real advantage is that it sometimes leads to lighter structures.

Question, however. So pushback from the foam will cause inaccuracies. So how do they get linear feedback? Is it something to do with measuring how much torque the servos see as they move? Or what is it?

It depends upon the foam how much push back you will get. You might not get enough to worry about.

As for linear feed back the thought here is linear scales of some sort. The common ones are glass scales but there are other options. There are other ways to measure position but you have a budget here. Don't under estimate how much scales and a suitable controller will impact your machines cost. In any event I'm still not sure why you want a machine designed to machine foam to be accurate to 0.001" over five feet. Seems like an excessively expensive requirement of foam.

[wizard]

Great reply from wizard, as usual.

Here is the sticky on ballscrews: http://www.cnczone.com/forums/linear...-software.html
Even a C3 ground ballscrew is very expensive, and only good for 0.00027" per foot. Over 10', that's almost 0.003". As wizard said, they can be mapped, or glass scales in a closed loop system, both with the right software. After that we also need:
- no sag over the 10'
- no whipping due to critical speeds
- a pure guess is that a 10' long ballscrew would need to be on the order of 50mm diameter to not sag or whip at reasonable rpm, and it would still need auxiliary support -- I wonder how much a 10' C3 50mm dia ballscrew would cost?
- a controlled temperature (high end CNCs use ballscrews with internal liquid cooling to control their temp) -- maybe could skip this with glass scale feedback.
- a preloaded double ball nut ($$) and preloaded high precision double angular contact support bearings ($$).
And that's just the mechanical drive for 1 axis!

The most basic challenge with high accuracy is temperature. Aluminum's coef of expansion is ~13 * 10^-6 in/inF. So just a 5F rise in temp of the machine makes the 10' dimension bigger by 0.008". Steel and cast iron move about half as much. So right away, we need a uniformly temp-controlled room, and maybe active cooling for the ballscrews and other parts.

Just to add a bit of color, I spent over a decade working on diamond turning machines in a nice temperature controlled facility. while the room as a whole was kept at a consistent temperature the fact that there was air conditioning vents prevented some machines from working correctly. We literally had to shut off some vents and reposition others. The parts being machined however where less than 15mm in diameter and thicknesses of a few microns. However trying to keep tolerances in the 1-2 microns range required the machines to be thermally stable, the air conditioner kicking in would throw the machines out of spec. Very frustrating.

Now a couple of microns might seem to be excessively small but I don't see keeping an accuracy of 0.001" over 5 feet to be any easier if any thing it would be harder.
we had a key benefit in that the machines ran the same part in long production runs so the machine would find an equilibrium thermally. A router on the other hand is often all over the place.

So in this day and age, on a machine this size and slow price point, one would likely want to try mapping a lower cost leadscrew to the machine. That would give you a machine accurate at a given temperature point. This assumes too that the controller can actually handle the mapping requirements. Mapping can improve things but it doesn't solve thermal issues and may not handle periodic error well.

Other challenges with accuracy are:

Room issues: floor rigidity and stability (in addition to temp). A regular concrete floor may move over time. Large CMMs aren't just machines, they are installations -- same for CNCs.

Component quality, which translates as $$, good decisions, and experienced attention during installation.

Alignment: linear (easiest but not easy), 2D (harder but there are techniques), 3D (hard), and how to maintain alignment over time (hard if the machine is not massively built out of proper materials). In 1D, for example, linear rails are not straight when you buy them, so we can't rely on them for linearity. The usual technique is to machine a highly accurate registration ledge the full length (into a very stiff and stable frame), and force the linear rail against the ledge. Next we need to get the other rail extremely parallel to this master rail, mount it there, and have some way to ensure it stays parallel (what if you move the machine?). In 2D and 3D it gets more complex.

Dimensional stability over time: will residual stresses relax and cause parts to shift over time?

Rigidity, vibration, and damping due to cutting forces, inertial loading, and local forces: FEA is good for this, but accurate modelling of a CNC machine is not simple. Some things can be certainly calculated with equations and analytic methods, e.g., how bearing stiffness and slop geometrically propagates to the cutter tip. For the complex geometry, structure, and loading of a CNC router, only experienced use of FEA will do a good job of predicting stiffness at the cutter, or vibrations of the system. The consensus view of experienced machinists is that it takes massive amounts of cast iron and $ to achieve high accuracy over a large distance.

Machine design/configuration: CNC routers are complex machines with many issues. Assuming someone already has a mechanical engineering degree and many years of machining experience, I'd guess it would take about a year of CNC-specific study, *from the right people*, to be qualified to design one that could hold 0.001" 3D accuracy over a long distance, for years. I'm an experienced ME, have done design/build projects for 35 years, have been studying CNC design on and off for 5 years, and I'm not sure that I could do it (nor do I have much interest in it... seems like a bunch of headaches!) This kind of accuracy is just very difficult and complex.

So hopefully you can see why 0.001" accuracy over a long distance will cost an amazing amount of $, and more importantly, a team with lots of knowledge and experience.

For a point of reference, here is Multicam's top router, I'm guessing about a $70k machine: 7000 Series CNC Router - CNC Cutting Machines for Your Application & Budget | MultiCam
They only spec a repeatability of +/- 0.001". No accuracy is stated because it's considerably worse than that. But few people need 0.001" accuracy over a long distance.

If they do need that sort of accuracy they most likely would end up with a Bridge mill with a moving table design.

Some suggestions:
There is a lot of good info in the archives -- search and learn, follow current threads -- for a few months.
Also search on practical machinist dot com -- lots of experienced machinists there sharing their knowledge, especially the user 'milacron' -- I have learned a lot from his posts.
Spend as much time as you can in the shop at your university, and learn from the smartest machinists there. Ask to be involved in their special projects.
Don't order parts until you've done at least a few months of external study, refined the design through many iterations and analysis, and it's been vetted by others. Exception: ordering some sample parts is often helpful, especially when less experienced.
Design and build a smaller machine first. You'll learn a lot by just completing one, and it's much easier to build a small one than a big one. Then you can make the beginner mistakes on the little one, and have much better skills to build a big one.

[dmalicky]

Just to add a bit of color, I spent over a decade working on diamond turning machines in a nice temperature controlled facility. while the room as a whole was kept at a consistent temperature the fact that there was air conditioning vents prevented some machines from working correctly. We literally had to shut off some vents and reposition others. The parts being machined however where less than 15mm in diameter and thicknesses of a few microns. However trying to keep tolerances in the 1-2 microns range required the machines to be thermally stable, the air conditioner kicking in would throw the machines out of spec. Very frustrating.

Now a couple of microns might seem to be excessively small but I don't see keeping an accuracy of 0.001" over 5 feet to be any easier if any thing it would be harder.
we had a key benefit in that the machines ran the same part in long production runs so the machine would find an equilibrium thermally. A router on the other hand is often all over the place.

Good story, Wizard, thanks for that!

I agree steel and cast iron are ultimately better for a 'real' machine tool, but there are some things I like about aluminum for DIY machines...
- To get steel without internal stresses, we get hot rolled, but it has scale, inclusions, loose tolerances, and isn't straight or flat. In aluminum, we get MIC 6 tooling plate which is ~stress-free, ground flat, and can be cut on a table saw -- great for Y and Z car backplates. Or 6061-T561 bar stock, which is mostly stress-relieved and stays pretty flat after machining.
- Hand drilling thick steel is fairly challenging; less so through aluminum.
- Unless one has high $ motors and drives, an aluminum gantry does help acceleration a lot.

[OneWound]

Good story, Wizard, thanks for that!

I agree steel and cast iron are ultimately better for a 'real' machine tool, but there are some things I like about aluminum for DIY machines...
- To get steel without internal stresses, we get hot rolled, but it has scale, inclusions, loose tolerances, and isn't straight or flat. In aluminum, we get MIC 6 tooling plate which is ~stress-free, ground flat, and can be cut on a table saw -- great for Y and Z car backplates. Or 6061-T561 bar stock, which is mostly stress-relieved and stays pretty flat after machining.
- Hand drilling thick steel is fairly challenging; less so through aluminum.
- Unless one has high $ motors and drives, an aluminum gantry does help acceleration a lot.

I concur, for a 3x8 thick piece of aluminum running across 5' weighs just south of 200 pounds. To help with tapped holes, helicoils will be used.
All, I'll try to reply to everything the best I can. If I miss something, let me know.

@wizard, this gantry will machine high-density foam, ranging from 15lbs to 40lbs. We can't use a moving table design due to the sheets being in 4x8 form. The reason I want .001 so much is because I'm a perfectionist. While I understand that .001 is unobtainable, I'll do everything I can to get as close to .001 tolerance as possible. I'll be happy with .01-.005, honestly.
FYI - This machine will be used to make carbon fiber molds for 2-3 student organizations, and while the molds may be sanded, if I can ensure accuracy and use small enough stepdowns I can greatly reduce sanding time.

@dmalicky, I understand what you are saying. An active A/C system for this maybe may be viable, but for now I want to get this machine design done at least. I will be posting a design review on here, and I do hope that you are willing to look at it. There will also be an assembly review, as I understand this thing will be a MAJOR pain to install correctly

[louieatienza]

Why aren't ballscrews that accurate? Isn't that what big companies like DMG Mori, Haas, etc are using (along with linear rails, ofc) using on their machines and they are able to achieve .0001 positional accuracy? I understand .001 is a high goal, considering this is a homemade gantry, and I'd be happy with .005 honestly. But considering this CNC will have more than 10k put into it, as well as using actual CNC machines to machine every component to a knats ass accuracy, and CMM machines to verify everything within the gantry is where it should be to an accuracy of .001 or greater, I don't see where I am losing huge amounts (~.002 in this case) of positional accuracy?

Well, go and find out how much a Mori Seiki VMC costs, and you'll know your answer. You couldn't buy one ballscrew for your budget, and their ballscrews and support bearings on their VMCs have through-coolant. Even then they use glass scale linear encoders. You'd need a ballscrew with a relatively large lead to achieve the feedrates you want without the screw sagging and whipping. And the larger the lead, usually, the lower the precision, even with a ground screw.

Heck, look at Haas' gantry style router, about $120k. It is great to strive for perfection, but impossible when you have to make concessions based on budget.

[OneWound]

Well, go and find out how much a Mori Seiki VMC costs, and you'll know your answer. You couldn't buy one ballscrew for your budget, and their ballscrews and support bearings on their VMCs have through-coolant. Even then they use glass scale linear encoders. You'd need a ballscrew with a relatively large lead to achieve the feedrates you want without the screw sagging and whipping. And the larger the lead, usually, the lower the precision, even with a ground screw.

Heck, look at Haas' gantry style router, about $120k. It is great to strive for perfection, but impossible when you have to make concessions based on budget.

Understood, but here are a couple of things that drive that number as well:
1) Rigidity, I only need to cut high density foam. Not aluminum.
2) Labor. I work for free.
3) Controllers. Any VMC or Haas router can outdo whatever I end up using.
4) Speed. A Haas gantry is has more than double the speed that I theoretically have.
5) Spindle. I am not using the very nice spindles that Haas, DMG Mori, etc has. I am using a router head.
6) Motors. I am not using high-torque motors, as I am not machining metal

[dmalicky]

Good advice from Louie. And thanks, that's good to know they use both thru-coolant and glass scales.


Thanks for sharing more of the machine's purpose -- that's very helpful to know.

It's really great you have high standards. But with that in mind, I'd invite you to take a broader view about defining "perfection". In engineering, an ideal design and product is one that best meets the customer requirements, within the project's resource constraints, under reasonable assumptions.

For this kind of foam cutter, it sounds like the customer requirements include:
- High speed and acceleration so it can produce small stepovers with a reasonable job time (24 hours?)
- Reasonable rigidity for cutting 40 lb/cubicfoot foam. That's about the same density as wood, so around 5000 lb/in stiffness-at-the-cutter would be good. Less is probably fine.
- Reasonable accuracy considering the form will be sanded after. I'd say 0.030" is plenty accurate for that purpose, and the size of the form. Or 0.060".
- What other customer requirements have you defined?

Your project also has important resource constraints, and we need to know this context to give good advice:
1. How many people are on the project, and what is their experience level with machining/fabrication and design/build/develop projects? Some examples of their past design/build projects would be great to know.
2. How long has the team already worked on this project?
3. How many total hours/week can the key team members dedicate to it in the future?
4. How many weeks does the team have to finish it?

All these issues are important because customer requirements usually compete with each other and the resource constraints, as Louie mentioned. If 1 requirement (like accuracy) is way over-spec'd, it will inevitably come at the expense of other requirements, or consume resources so the project can't even be finished. I teach ME senior design and have advised about 60 such projects. I love projects, especially CNC builds, and I'm willing to help yours. I ask these questions because I've learned how student projects can fail, and what a team needs to consider to have a successful outcome.

[louieatienza]

Understood, but here are a couple of things that drive that number as well:
1) Rigidity, I only need to cut high density foam. Not aluminum.
2) Labor. I work for free.
3) Controllers. Any VMC or Haas router can outdo whatever I end up using.
4) Speed. A Haas gantry is has more than double the speed that I theoretically have.
5) Spindle. I am not using the very nice spindles that Haas, DMG Mori, etc has. I am using a router head.
6) Motors. I am not using high-torque motors, as I am not machining metal

I'm talking about the rigidity in the machine itself that helps it keep its tolerances. I can build a machine out of closet rods, and with the right equipment can line it up to .001". That doesn't mean it will stay that way, and it probably is unrealistic to set such goals.

What I'm talking about is, you'll be blowing most your budget on precision ground screws and linear bearings, and likely servos. They're only as good as the machine they're attached to. As massive as the Haas is, it is not designed to cut steel, and capabilities include only "cutting" aluminum, not necessarily precision milling.

If you set more realistic goals on precision, it would free up the remaining resources to possibly get heavier frame members, having potential welded parts stress relieved and machined (I don't know if your shop can machine a 10' long structure), better spindle, and better controls/drives. I still think you'd want more HP than what a wood router could give, and having a water-cooled spindle is much more palatable than a wood router sucking up foam dust all day. It won't last as long, you have less selection of collets, the bearings aren't designed to use longer reach tools that are commonly used in mold-making... Plus, you'll be able to take deeper cuts, which would save you time.

I think you could realistically achieve over 1000ipm rapids incorporating a rack-and-pinion system, and the accuracy can be pretty darn good. A ballscrew for the Z would be all you'd need, and there's always a good selection to find, even NOS, on eBay. A helical rack, and spring-preload pinion attached to a planetary servo reducer would be one way. Belt reduction would be another.

Check this YouTube channel out. I wish the guy didn't take down the videos of the foam forms he made, but his screen avatar shows the net result. I once downloaded the video but couldn't find it. I think a scaled-up version would be right at your budget, if you scrounge for stuff on eBay. I keep this build in mind because I'd like to build one using these concepts, someday when I have the space!

https://www.youtube.com/user/ProjektSS7/videos

Build video:
https://www.youtube.com/watch?v=fIVL83AwXrE

[OneWound]

Good advice from Louie. And thanks, that's good to know they use both thru-coolant and glass scales.


Thanks for sharing more of the machine's purpose -- that's very helpful to know.

It's really great you have high standards. But with that in mind, I'd invite you to take a broader view about defining "perfection". In engineering, an ideal design and product is one that best meets the customer requirements, within the project's resource constraints, under reasonable assumptions.

For this kind of foam cutter, it sounds like the customer requirements include:
- High speed and acceleration so it can produce small stepovers with a reasonable job time (24 hours?)
- Reasonable rigidity for cutting 40 lb/cubicfoot foam. That's about the same density as wood, so around 5000 lb/in stiffness-at-the-cutter would be good. Less is probably fine.
- Reasonable accuracy considering the form will be sanded after. I'd say 0.030" is plenty accurate for that purpose, and the size of the form. Or 0.060".
- What other customer requirements have you defined?

Your project also has important resource constraints, and we need to know this context to give good advice:
1. How many people are on the project, and what is their experience level with machining/fabrication and design/build/develop projects? Some examples of their past design/build projects would be great to know.
2. How long has the team already worked on this project?
3. How many total hours/week can the key team members dedicate to it in the future?
4. How many weeks does the team have to finish it?

All these issues are important because customer requirements usually compete with each other and the resource constraints, as Louie mentioned. If 1 requirement (like accuracy) is way over-spec'd, it will inevitably come at the expense of other requirements, or consume resources so the project can't even be finished. I teach ME senior design and have advised about 60 such projects. I love projects, especially CNC builds, and I'm willing to help yours. I ask these questions because I've learned how student projects can fail, and what a team needs to consider to have a successful outcome.

I understand that reasonably accuracy is probably fine at .030, but I'm going to holder better tolerances. Otherwise there is no reason to build this CNC (what I am designing is replacing another CNC).

1. How many people are on the project, and what is their experience level with machining/fabrication and design/build/develop projects? Some examples of their past design/build projects would be great to know. There are two people on this project. Me and the person handling servos, controllers etc. The other person is a CE major. I am designing the CNC. There is already an existing frame that this will mount to, therefore covering the 10' requirement. A lot will be answered once I finish my revamp of the current design. I am also the person responsible for the assembly specs. That being said, everything that has been designed is able to be machined by a Haas VF2. It may be a headache on some pieces, but it is doable. To note, this will be reviewd by lots of engineering students. As well as this forum. And maybe a couple professors on campus.
2. How long has the team already worked on this project? I've worked on it...20-30 hours of actual work time? I don't keep count.
3. How many total hours/week can the key team members dedicate to it in the future? As much as possible
4. How many weeks does the team have to finish it? Probably a month
Hopefully I'll be posting the final design at the end of this week.

FYI: Making all machined components out of steel is not out of the question, but I'd rather avoid it for now.

[OneWound]

I'm talking about the rigidity in the machine itself that helps it keep its tolerances. I can build a machine out of closet rods, and with the right equipment can line it up to .001". That doesn't mean it will stay that way, and it probably is unrealistic to set such goals.

What I'm talking about is, you'll be blowing most your budget on precision ground screws and linear bearings, and likely servos. They're only as good as the machine they're attached to. As massive as the Haas is, it is not designed to cut steel, and capabilities include only "cutting" aluminum, not necessarily precision milling.

If you set more realistic goals on precision, it would free up the remaining resources to possibly get heavier frame members, having potential welded parts stress relieved and machined (I don't know if your shop can machine a 10' long structure), better spindle, and better controls/drives. I still think you'd want more HP than what a wood router could give, and having a water-cooled spindle is much more palatable than a wood router sucking up foam dust all day. It won't last as long, you have less selection of collets, the bearings aren't designed to use longer reach tools that are commonly used in mold-making... Plus, you'll be able to take deeper cuts, which would save you time.

I think you could realistically achieve over 1000ipm rapids incorporating a rack-and-pinion system, and the accuracy can be pretty darn good. A ballscrew for the Z would be all you'd need, and there's always a good selection to find, even NOS, on eBay. A helical rack, and spring-preload pinion attached to a planetary servo reducer would be one way. Belt reduction would be another.

Check this YouTube channel out. I wish the guy didn't take down the videos of the foam forms he made, but his screen avatar shows the net result. I once downloaded the video but couldn't find it. I think a scaled-up version would be right at your budget, if you scrounge for stuff on eBay. I keep this build in mind because I'd like to build one using these concepts, someday when I have the space!

https://www.youtube.com/user/ProjektSS7/videos

Build video:
https://www.youtube.com/watch?v=fIVL83AwXrE

The longest section that will be machined will be ~6'. This is doable if you open up the side doors on our CNC and do incremental moves. Since there will be holes, you can use the position of the last hole to set a new coordinate system each time. Than it is rinse and repeat. Machining components is not my concern. I have enough experience, with 1.5 years of tool and die school, as well as as 2-3 years experience with all CNCs and CAM. My only concern is rigidity of the machine. I am starting to think though if I have a 175 lb slab of aluminum going 6' across on the Y, it should be stable enough to handle machining high density foam.

That being said, I'll be posting my whole design. There is a lot I don't want to type out, and I'll be posting pictures once I finish my current revamp.

As far as the router head is concerned, we will be using Bosch MR23EVS (as that is what is current CNC). It has an 18k spindle with a vaccum and will work just fine.

FYI: The design in the video is very similar to mine, with various tweaks.