[ishi]

Because it gives me a square footprint and it makes the table more accessible when working on fixturing. Also, I figure that the X axis for the table will move a higher mass than the Y axis on the bridge, especially if I decide to add a two-axis trunnion table down the road. Therefore, I'd rather have less travel on the X axis than on the Y axis.

[ishi]

Also, I'd go for a higher pitch screw than 5mm, and 10-12Nm steppers? I'd go with a 1-1.5kW servo. Medium inertia servo (3000 RPM max speed) with 10mm pitch screw should get 20m/min feeds and 30m/min rapids. That's what I'm planning for my router (which will eventually have a ~500lb table and 16 HP spindle.)

I might be wrong, but I'd rather have a spindle with more speed and less power. This is the whole point of the HSK-E25 form factor. I know it's not very popular (yet), but I think it has a lot going for it.

[skrubol]

More RPMs just gets you faster results with small tools most of the times. More torque lets you use larger diameter tools, which are stiffer (as are larger tool holders,) which will get you better surface finishes when you need long tooling. When you need tiny tooling and long reach, you don't have much option, but if you can fit a bigger tool, you're almost always better off.

[ishi]

More RPMs just gets you faster results with small tools most of the times. More torque lets you use larger diameter tools, which are stiffer (as are larger tool holders,) which will get you better surface finishes when you need long tooling. When you need tiny tooling and long reach, you don't have much option, but if you can fit a bigger tool, you're almost always better off.

That makes a lot of sense. I might need to rethink the way I'm approaching it. Or find a more powerful spindle in HSK-E25.

[pippin88]

Trying not to be too negative, but this all seems crazy.

An untrained person is wasting money trying to design / build a 25-50k machine.

Building a 5-10k machine is different, as commercial machines in this price range are poor quality / value often, and the diy designer / builder can do quite well.

For your budget, you should buy a properly designed and engineered machine.

Your Z axis design appears about as stiff as a cooked noodle.

[ger21]

I have learned a lot in less than 48 hours...

If someone was building a $2000 machine, I'd tell them to spend at least 6 months researching, to avoid throwing money away on the wrong items. I've seen it happen countless times.
If you are spending $50K, I'd recommend 2 years of research.
You could very easily end up with a $50K machine that performs like a $10K machine.

[ishi]

pippin88,

You are absolutely right: I am being quite ambitious (foolish?) with this build, and there is a price point beyond which an off-the-shelf system would make a lot more sense. There are a few things to consider though:

First, I am not doing this as a way to get a cheaper machine for production purposes. Instead, I am doing this mostly as a learning exercise. And clearly, I would not learn nearly as much if I were buying an off-the-shelf machine.

Second, I am trying to do something that has not been done very often, especially in a publicly-documented way. For a machine in the 5-10k range, one can learn a lot by studying your forum carefully. But for the next level up, it's a lot more difficult. This is what I am interested in.

The Z axis will be very similar to Wade'O's design (with the two rails being coplanar). I understand that it's not the most rigid one, but it's very simple. If you can point me to better design alternatives, I would really appreciate it.

https://www.wadeodesign.com/cnc-router-build.html

Thanks a lot for your feedback, I really appreciate it.

[ishi]

If someone was building a $2000 machine, I'd tell them to spend at least 6 months researching, to avoid throwing money away on the wrong items. I've seen it happen countless times.
If you are spending $50K, I'd recommend 2 years of research.
You could very easily end up with a $50K machine that performs like a $10K machine.

ger21,

That is very sound advice indeed, and I might end up doing exactly that, or even giving up entirely (I hope not). But I look at it this way:

First and foremost, a third of the budget will go toward a proven control kit that I could reuse in many different ways, no matter what happens with the build:

https://machinetoolproducts.com/siem...rofit-kit-6nm/

And this kit is itself built on top of a really good platform (SINUMERIK 828D), which I consider to be a great platform to learn (ShopMill looks awesome).

Second, 20% will go toward the granite base, which has a tried-and-tested design (fixed portal, canals for ball screws, etc.). If I design the layout for inserts in a generic-enough way and add a few inserts for future modifications, I should be able to retrofit the base fairly easily. Therefore, this investment won't be lost either.

Third, 15% (or more) will go toward the spindle. This is where I see the largest risk, and I won't rush to buy one. In fact, it might take me 3 to 6 months to select it after everything else has been put together. With that in mind, I will try to design what I now call the X axis (the longest one which rails are mounted on the portal) and the Z axis in such a way that they could properly support a spindle of up to 20kg (the one I currently selected is about 8kg). That way, I have a lot many more options to play with.

Bottomline: the only expensive components that might be wasted are the rails, carriages, and ball screws. Altogether, we're looking at less than $6K, which is about 12% of the total build. If this is the amount that ends up being wasted, I'll be happy, because what I'll learn in the process will be priceless.

That being said, I realize that it is a fairly aggressive approach, and I really do not want to be disrespectful of the community. I am here to learn, and I really appreciate your time and feedback: this is priceless as well.

[skrubol]

First, I am not doing this as a way to get a cheaper machine for production purposes. Instead, I am doing this mostly as a learning exercise. And clearly, I would not learn nearly as much if I were buying an off-the-shelf machine.

Why are you considering an off-the-shelf controller then? A lot to be learned in Linux CNC or the like.

For the Z axis rails being coplanar, I don't know of any other common designs. The gantry rails are sometimes not coplanar (one on the front near the bottom of the gantry, and one on top sometimes,) but it's usually quite a bit more work to get such an axis squared up.

[ishi]

Why are you considering an off-the-shelf controller then? A lot to be learned in Linux CNC or the like.

For the Z axis rails being coplanar, I don't know of any other common designs. The gantry rails are sometimes not coplanar (one on the front near the bottom of the gantry, and one on top sometimes,) but it's usually quite a bit more work to get such an axis squared up.

Great question! I initially went down the path of building the whole control system myself, but the learning curve is simply too steep. One thing I want to avoid at all cost is to spend 25-50k and end up with a bunch of parts on my workbench... So, I have opted for a "divide and conquer" strategy, focusing on the integration of two core pieces that are sourced from two reliable suppliers who can provide a lot of the engineering that I am not capable of myself (yet): the granite supplier for the frame and the controller supplier. That way, I can focus on the higher-level selection of parts for the spindle and the linear motion components, and their integration into a working system. This is already quite a bit to handle, especially when trying to get relatively high-precision components... If I do nothing but that, I'll be a happy camper...

If the machine works, I'll tackle more challenges, but developing a control system from scratch is likely to be fairly low on my priority list. Before that, I would rather spend some time and resources building a two-axes trunnion table with torque motors. That is not something that very many hobbyists have done, and for it to work, I will need a solid controller with proper 5-axis interpolation. And if I am successful with that, I'll try to replace my motors and ball screws with linear motors. The design of my granite base will try to take this possible retrofit into account from day one.

Regarding the rails on the portal, I agree with you: coplanar is the way to go, because it will make squaring a lot simpler. The only change that I am considering right now is to use linear carriages with rollers instead of balls, but Misumi does not provide them. HIWIN does, but I would rather work with Misumi for all linear motion components, because their components are easier to order than HIWIN's, especially for things like ground ball screws. But all this is very much a work in progress: I expect to spend a fair amount of time during the next two months doing all the calculations for selecting the final components. I just hope that I can settle on a layout for the granite base that would be largely independent from component variations. The only parameter that I am not totally sure about is the depth of the canals on the baseplate and on the portal beam: depending on the size of the motors, I might need deeper or shallower canals if I want to mount the motor brackets flush on the granite. But I could also use mounting plates to get more flexibility. In order to take that into account, I am likely to add some extra inserts in the granite and make the canals deeper right away. That way, I can leave more doors open.

Thanks again for your feedback: as mentioned earlier, this is priceless.

[ishi]

Thanks to the great folks at DamenCNC, a better alternative for the spindle has been identified: HSD ES351 - H6161H1036.

ES351 - H6161H1036

Power S1/S6 6,5Kw/7,5Kw
Tool Blocking Automatic with pneumatic piston
HSK E32
Ratered Voltage 380V
Poles 4
Bearings Front Ceramic
Bearings Rear Ceramic
Bearings Long Life lubricated
Encoder LB 1Vpp
Max 40.000rpm
Spindle Body Alluminium Alloy
Cooling Liquid

This should give me twice as much torque, which should be enough for rigid tapping (hopefully). It's also 3 times as heavy (20kg instead of 7.8kg), which might require slightly larger motors, but I can figure that out later.

The DamenCNC folks also suggested to add stiffeners to the carriage and head plates, which will improve rigidity without adding too much weight.

[skrubol]

What's the plan for dealing with the 380V rated voltage to that spindle? You'll need a 1333+ Hz VFD as well (4-pole.)
You'll need a counterweight of some sort for your Z axis most likely. For routers most use a gas spring of some sort. Pneumatic actuator connected to a reservoir seems to work best (longer life, easy adjustment and more linear force.)

[ishi]

One of the main goals for the design is to keep a footprint as small as possible (48" × 48"). Ideally, this would include the enclosure for all the electrical components, as well as the minimum quantity lubrication (MQL) system and the spindle chiller. For that purpose, the controller enclosure will be mounted horizontally on rails within the table, underneath the base. That way, it will work as a drawer. This will require an extra drag chain, but this should not be a problem. The only modification to the enclosure will be to move the front vent to the side, so that no vent radiates hot air toward the top of the table. The enclosure will be a standard NEMA 12/13 40" × 24" × 12", which will leave plenty of space for future extensions (a 24" × 30" × 12" would have been sufficient). This controller drawer will open from the left side of the machine, while the MQL system and water chiller will be accessible from the right side.

This will leave about 25" × 25" of footprint that can be used for an optional Kennedy Compact HDS 7-Drawer Worktop High Cabinet:

7122 - Buy Kennedy

These drawers will open from the right of the machine and be placed toward the front so that they are directly accessible next to the HMI unit. The arm for the HMI unit shown in the attached picture will be moved from the controller enclosure to the front right side of the granite baseplate. For that purpose, a 2" through hole will be machined in the granite baseplate (Pyramid just confirmed that this can be done). This will allow the HMI unit to be flipped back within the footprint of the machine whenever the machine is not in use, facing the right side of the machine (same as the drawers).

Because the Kennedy cabinet is twice as tall as the controller drawer, the drawer will be mounted on top of the fork lift tubes, which will extend only by about 24" (half of the width of the table), which means that the table will be lifted from the left. With that design, the back of the table can be placed directly against a wall, thereby saving some valuable floor space in the workshop. The only drawback of that design is that the cabinet, MQL system, and chiller will have to be removed whenever the table is moved, but with a 2 metric ton machine, this should not happen very often...

Also, the baseplate will have four swivel vibration-damping leveling mounts 6221K83:

https://www.mcmaster.com/#catalog/124/1471/=1dggdyd

And on the top, on all four corners, large 1" threaded inserts will be added for mounting removable eye-bolts that can be used for lifting the granite baseplate (straps should only be used by an experienced operator because the stone surface is too smooth to provide a good gripping surface.

[ishi]

What's the plan for dealing with the 380V rated voltage to that spindle? You'll need a 1333+ Hz VFD as well (4-pole.)
You'll need a counterweight of some sort for your Z axis most likely. For routers most use a gas spring of some sort. Pneumatic actuator connected to a reservoir seems to work best (longer life, easy adjustment and more linear force.)

No plan yet, I am still working on that. Thank you for the reminder (so many "details" to work out ;-)

I had forgotten about the pneumatic actuator! Thank you so much for the reminder as well. I studied those a year ago, but I totally forgot about them. That's a cool project: I'll work on that next.

Thank you so much skrubol, very much appreciated.

[ishi]

The bill of materials has been split into two sections:

# Components:
https://docs.google.com/spreadsheets...SwA/edit#gid=0

# Options:
https://docs.google.com/spreadsheets...#gid=356239105

This will provide a fairer estimate of costs for the basic build.

[wizard]

One of the main goals for the design is to keep a footprint as small as possible (48" × 48"). Ideally, this would include the enclosure for all the electrical components, as well as the minimum quantity lubrication (MQL) system and the spindle chiller. For that purpose, the controller enclosure will be mounted horizontally on rails within the table, underneath the base. That way, it will work as a drawer.

That might sound good and I do realize there are issues almost everywhere with space, but putting a control panel under a machine on slides leaves a lot to be desired. Even locating the control panel to a fixed position between the frames legs has its issues but is far more desirable than a sliding arrangement. For me any ways the ideal solution is a free standing console/control cabinet attached to the machine via flexible cord or conduit.

This will require an extra drag chain, but this should not be a problem. The only modification to the enclosure will be to move the front vent to the side, so that no vent radiates hot air toward the top of the table. The enclosure will be a standard NEMA 12/13 40" × 24" × 12", which will leave plenty of space for future extensions (a 24" × 30" × 12" would have been sufficient). This controller drawer will open from the left side of the machine, while the MQL system and water chiller will be accessible from the right side.

You are right to leave space for future developments and an initial layout that is not cramped. However is you do want the control panel physically in the machine, the ideal solution is to make sure the panel fits between any frame members of concern. You will want easy access to the panel for startup, calibration / tuning and other reasons. By the way that means plenty of clearance in front of the panel box, there are standards for clearances which I forget off the top of my head. I only bring this up because you seem to be under space constraints. One option you have to improve internal space in a control panel is to buy power transformers that are totally enclosed and as such can be wired up to the control panel after being mounted under (inside) the frame. Yes the transformers are more expensive but they also take up a surprising amount of space inside a control panel.

This will leave about 25" × 25" of footprint that can be used for an optional Kennedy Compact HDS 7-Drawer Worktop High Cabinet:

7122 - Buy Kennedy

While such cabinets are nice I'd be reluctant to waste space under the machine for such a cabinet. You will need space for any pneumatics associated with the various controls you will want to implement.

These drawers will open from the right of the machine and be placed toward the front so that they are directly accessible next to the HMI unit. The arm for the HMI unit shown in the attached picture will be moved from the controller enclosure to the front right side of the granite baseplate. For that purpose, a 2" through hole will be machined in the granite baseplate (Pyramid just confirmed that this can be done). This will allow the HMI unit to be flipped back within the footprint of the machine whenever the machine is not in use, facing the right side of the machine (same as the drawers).

What can be done and what is actually a good idea is two different things. I would not rely upon the structural strength of epoxy granites. This especially for a cantilevered mass. In any event you may find it to be far simpler and more cost effective to build a steel frame for most to the mechanical issues in which the epoxy granite base sits. In any event I'm not even sure why we are discussing epoxy granite here. To put it bluntly you are not equipped to build a machine to the precision that is justified by epoxy granite. In other words I see you wasting a lot of money unless you are willing to equip yourself with a bunch of precision tooling and metrology hardware.

Because the Kennedy cabinet is twice as tall as the controller drawer, the drawer will be mounted on top of the fork lift tubes, which will extend only by about 24" (half of the width of the table), which means that the table will be lifted from the left. With that design, the back of the table can be placed directly against a wall, thereby saving some valuable floor space in the workshop. The only drawback of that design is that the cabinet, MQL system, and chiller will have to be removed whenever the table is moved, but with a 2 metric ton machine, this should not happen very often...

I'm not even sure I'm following what this machine will ultimately look like. I might mention though that if you are space constrained it often makes sense to leverage vertical space.

Also, the baseplate will have four swivel vibration-damping leveling mounts 6221K83:

https://www.mcmaster.com/#catalog/124/1471/=1dggdyd

And on the top, on all four corners, large 1" threaded inserts will be added for mounting removable eye-bolts that can be used for lifting the granite baseplate (straps should only be used by an experienced operator because the stone surface is too smooth to provide a good gripping surface.

Not only do you need experience you need significant equipment to handle the fragile epoxy granite structures. I'm still trying to figure out how epoxy granite will be worthwhile in a router type machine. This especially on a compact moving gantry machine.

[skrubol]

I'd recommend going with something a bit more substantial for a table. 3/4" thick aluminum in that large of a span is going to deflect a lot. Plus aluminum isn't a good match for granite thermally. That ~.5m span between the rails on the table will give you about a 6 micron per deg C difference between the table and the granite frame. I'd try to avoid aluminum altogether on a build like this.
Any idea of a plan for workholding?

[ishi]

Wizard,

Thank you for all your comments. Let me clarify a few things:

First, I am *not* using epoxy granite. I am using real granite, as in, the real stone:

Precision Granite Machine Bases - Pyramid Granite & Metals

This is the same kind of granite used for making surface plates. Here is more on that particular material:

Frequently Asked Questions - Pyramid Granite & Metals

With real granite, I would not worry at all about rigidity. I'll do some FEM simulation to prove that point before passing the order for the granite base, but I'm confident thi will work. And if the risers and bridge have to be made wider than 8", I could go as far as 24" in the current design, but that would add a metric ton of granite to be supported by the table, with no benefits (according to the founder of the company providing the granite base). That being said, the table could easily support ten metric tons or more...

Second, this machine does not use a moving gantry. Instead, it has a static portal and a moving table. I have added a crude sketch to illustrate this. As you can see, this is a really simple design. I strongly believe that simplicity is the shortest path toward rigidity.

As far as precision and metrology are concerned, several things to consider.

First, the granite supplier takes care of making things flat and square. Overall flatness tolerance of .000300" for the base plate, 000050"/6" for squareness and perpendicularity of the riser. I don't think you could get anywhere near that with welded steel. The only way to get better is by using cast iron, then scrapping it, but the cost would be about 5 times as much (if not more), and you would not get as stable a base from a thermal standpoint. Nothing beats real granite if your design is kept simple.

Second, I will use a local machine shop for machining the table and mounting plates, because I do not have the right tooling at home, and I am a newbie machinist. I did spent two months at a community college last year to get a solid introduction to manual mills and lathes, but that's not enough to make the parts I need.

Third, the squaring of the linear guides and ball screws will indeed require a lot of metrology equipment that I do not own. Fortunately, by community college teacher has offered to help and has access to all the tooling that we could dream of. This should take care of that part nicely.

Last but not least: the controller enclosure. You are absolutely correct, my design makes no sense. For one thing, it's too complex. For another, the Siemens drivers must be mounted standing right up. Conclusion: I'll just attach the enclosure vertically in between the two horizontal steel beams of one side of the table, as you recommended. And the idea of including a Kennedy cabinet within the table is plain stupid. It would be great, but it makes no sense. I got carried away, sorry... And thank you for bringing me back on track. Very much appreciated.

[ishi]

skrubol,

I think you're absolutely right. This 3/4" thick aluminum plate was (maybe) good enough for earlier iterations of the design that did not use as good components and such a rigid base. Now, I probably should go for steel. And it is quite likely that I will need a more powerful motor for the Y axis (the table's axis). For reference, please check this sketch to see which axis I'm talking about, because I've switched the X and Y axis nomenclature from my earlier posts (sorry for the confusion).

https://www.cnczone.com/forums/attac...8&d=1530129265

I am not spending a lot of time on these components right now, because they will impact the height of the granite risers, but this height is easy to change. Granite provides such a rigid frame that I could go from 12" tall risers to 18" or even 24" without affecting the overall rigidity of the machine. All that might be required is to make the risers and bridge deeper (8" right now).

The one thing I really need to decide soon is the size of the motor for the Y axis so that I can properly set the depth of the canal alongside the Y axis on the baseplate. The rest can be adjusted down the road.

So, for the time being, I'm going to plan for a 1" or even 1.5" thick steel plate with T slots on it. This should be very easy to machine for any decent metal shop.

As far as workholding is concerned, this will be optional. So far, the best options I have found are offered by DamenCNC (again, these guys rock):

https://www.damencnc.com/products/cn...04_w_242__GB_1

[TTalma]

I'm really confused about what you are building and why you feel you need to spend so much. Your first post states that you are planing to mill wood and aluminum.

If that's what you are planning to do your design is so much overkill. That is what I mill with my machine. I will have less than 5K in it with an ATC.

You are worried about surface finish, the tooling you use will have more of an impact than the rigidity of the machine. Especially since you said time is not a factor. A .0001 cleaning pass can make a beautiful surface.

Ply-wood will expand/contract a few thousands overnight. Sold wood is much worse. Cutting wood with small bits can deflect the bit enough that you can measure the wavey-ness with a CMM.

The Siemens control you are looking at is a huge mistake. You will regret buying it instantly. I work with those all day as part of my job. In fact I have the same model 3 feet away from me as I type this. You are buying a unit designed for a commercial machine that will be running at least 8 hours a day, wants to monitor 100's of digital and analog inputs, interact with different machines, and control a safety system. It's like buying a table-saw to cut a piece of bread, it'll do the job, but it's not the right tool. If your not planning to add something along the lines of a robotic loader, you will have nothing but wasted power, and most likely a lot of frustration. You would be much better off with something like a smooth stepper and mach 4. Since you admit you are not very familiar, you can get a lot of online support for those products. You'll get very little with the Siemens if any, and it will only come from the company who sold it to you. And forget about parts. We spend millions with Siemens each year, and the best we can hope for is a 6 week lead time.

I think the best decision you made is to use servos. I would revisit every other aspect of your design.

You really need to look at what you are going to be making and design the machine to do that. There is a reason wood cutting machines are not as rigid as machines that cut steel.

If you can't wait and you can't find a commercial machine that meets your requirements contact Gary Campbell on via this site or camheads.org (he's more active there) He designs cnc machines for commercial sales. He could design and build you a machine that would work for you.

At the end of the day it's not the machine but the parts that come off of it. If you spend $50k on your machine I don't believe they would come out any better than the parts that come off my $5k machine.