6-axis Horizontal Machining Center for Education

[ishi]

but it's mechanically quite complex, and I am concerned about possible maintenance and adjustment issues.
A bit of string going over a couple of pulleys - complex?
Adjustment would be set-and-forget.
I have done this recently on my hydraulic press to lift the cross-arm, with two lengths of 'string' over pulleys at the top and down on each side. Works dead easy.

Cheers
Roger

Roger,

I think I could make it work, but I don't see how it would address our requirements, because we need telescoping actuators, since the rising front panel is higher than the bottom section that will contain it.

[RCaffin]

I am not sure I fully understand what you are asking for, but could you not lift the bottom panel and have it lift the upper panel?
Failing that I would need a simple diagram.

Cheers
Roger

[trb100]

As Roger wrote, bowdens or string and some pulley could be super easy. One actutor could move the panel easily. You dont have to have telescopic actuator. First, you can mount the actuator horizontaly, second, if you route the strings in the right way, you only need the half of the desired travel for the actuator. It's simple, not a complex system.

Here is some simple solution:

[ishi]

As Roger wrote, bowdens or string and some pulley could be super easy. One actutor could move the panel easily. You dont have to have telescopic actuator. First, you can mount the actuator horizontaly, second, if you route the strings in the right way, you only need the half of the desired travel for the actuator. It's simple, not a complex system.

Here is some simple solution:

Wow, this is clever! I get it now.

Yes, I think that would work, even though it might be tricky to get the right level of rigidity knowing that we don't have fixed pillars and the motion will be vertical instead of horizontal, but it's definitely worth a try.

Thanks a lot Roger and trb100!

[ishi]

Here is a totally redesigned back panel for the roof. It was quite a challenge, because we had to deal with a wide set of constraints:

- Ensure that this panel does not collide with the spindle head
- Ensure that it does not collide with a vertical tombstone installed on the rotary table
- Ensure that it does not interfere with the automatic tool changer
- Find a proper way to motorize it
- Ensure that this motorization does not interfere with the side panel

After considering over half a dozen configurations with one or two sections, articulated sections, and telescoping sections, we settled on a relatively simple design with a single section and an offset hinge.

The panel will rise up whenever a heavy part must be loaded on to the table using some kind of crane.

The panel's actuation will be enabled by a pair of small servo motors using bevel gearboxes. We're using two synchronized motors in order to get enough torque in a balanced way.

https://product.item24.de/en/product...al-9006-68848/
https://product.item24.de/en/product...es-1001374077/

Our light bar is now mounted on the underside of this roof panel.

[ishi]

Here is the new roof panel installed on our enclosure.

As can be seen on the last picture, 100% of the top opening in front of the enclosure's vertical profiles is now covered. Of course, we need to add some additional static panels inside the machine in order to prevent all chips from flying out, but this part should be much easier...

[handlewanker]

87K just for the spindle head!!!!...…...it just goes to show how much you have to pay when you want to deviate from the beaten track and tread unfamiliar paths...……...industry will not have a bar of it that's for sure, otherwise they would have.....my opinion so just label me Mr Skeptical.

I get the distinct impression that the final page of the saga will end with...…" and if you follow precisely and without compromise the words and music of this build to the letter you will no doubt have a machine like no other."

I would hate to think that a revision would occur during the actual build that would ensure a return to the starting line a possibility and make a complete re-design a happening of the grossest proportions.

By that time the design would be obsolete so it would not get further than a querky forum discussion for those that like to muse on the improbable on how a different format for a machine can be achieved using a different approach to the norm and a friendly stand by bank manager with lots of spare funds to lend.
Ian.

[ishi]

87K just for the spindle head!!!!...…...it just goes to show how much you have to pay when you want to deviate from the beaten track and tread unfamiliar paths...……...industry will not have a bar of it that's for sure, otherwise they would have.....my opinion so just label me Mr Skeptical.

I get the distinct impression that the final page of the saga will end with...…" and if you follow precisely and without compromise the words and music of this build to the letter you will no doubt have a machine like no other."

I would hate to think that a revision would occur during the actual build that would ensure a return to the starting line a possibility and make a complete re-design a happening of the grossest proportions.

By that time the design would be obsolete so it would not get further than a querky forum discussion for those that like to muse on the improbable on how a different format for a machine can be achieved using a different approach to the norm and a friendly stand by bank manager with lots of spare funds to lend.
Ian.

Ian,

I won't explain again what the purpose of this project is. I have covered this topic at nauseam in previous posts.

And yes indeed, this thread might just be that, a quirky thread. But I doubt that I could find many other threads where a newbie could have learned so much on so many topics, and that makes it worthwhile on its own. That being said, I would not suggest that anyone follows it unless they're interested to go through the same learning curve, or they're curious about the process itself. This approach is very much an acquired taste...

As far as obsolescence is concerned, I cannot see for the life of me why this should worry anyone. Compared to other industries (like software, which I am familiar with), the CNC machine tool industry moves slowly, very slowly. And obsolescence is rarely an issue, which is why the market for used machines is so active.

The machine that we are designing is not really concerned by the only major source of disruption to significantly affect the CNC machine tool industry in the coming years: additive manufacturing. Beyond that, here is what I see happening in the coming 10 to 15 years:

* Regular motors and torque motors won't change that much. They will just become cheaper as products coming out of China get better and better.

* Linear drives will remain used on larger machines only.

* Spindles will remain pretty much the same, just cheaper. 2-axis spindle heads will become more and more popular.

* Linear guides will remain the same, just cheaper. And you'll be able to source stock ground ball screws from many more places, especially Taiwan.

* Linear scales will remain the same, just cheaper.

* More and more frames will be done with mineral castings because companies won't be able to hire enough skilled welders.

* High-accuracy machines won't use scrapped guideways as much as they do today because companies won't be able to hire and train enough skilled workers.

* More and more machines will be designed in a modular fashion with interchangeable tables in order to provide more automation.

* ATC libraries will be redesigned in order to facilitate the robotic movement of tools from machine to machine, either with ground robots similar to the ones used in warehouses today, or roof robots moving tools from the tops of machines.

* High-flow coolant will be replaced by Minimum Quantity Lubrication for most applications.

* Vibration sensors and Artificial Intelligence will be used to monitor the wear of cutting tools.

* LIDAR will be used to facilitate the indexing of parts.

When you consider all these predictions, you realize that our machine design is highly tolerant to these evolutions. In fact, we've designed the machine with all these assumptions in mind. We want it to serve as foundation for future experimentations. And you should keep in mind that the most critical innovations won't apply at the hardware level, they will apply at the software level. This is due to the fact that the industry has pushed the hardware side pretty far, and opportunities for radical innovation are few and far between. In contrast, the software side of the equation remains totally primitive if you were to compare it to other industries (video games first and foremost).

If I had to pick two area for disruption, I would chose hybrid manufacturing and linear drives.

On the hybrid side, you want to do something like that:

Hybrid Manufacturing Technologies - Home

On the linear drive side, you want to downscale them so that they could be used for smaller machines. Linear drives replace ball screws, which are the primary source of heat when doing high-speed machining. Ideally, someone should come up with assemblies that combine linear drives and air bearings. That way, you would replace two components (ball screws and linear guides) with a single one, and you would remove all sources of heat and wear. This mechatronic innovation could totally change the game, but we're at least 10 to 20 years from such a system being available on the market at scale. In the meantime, everything will be very incremental.

Cheers!

Ismael

[ishi]

As a follow-up to my previous post, here is another area of innovation that I can foresee: modular tool heads.

Currently, machines are designed with interchangeable tools, but for a single type of tool. And the connection between the head and the tool only serves three functions: hold, rotation (when the head is a spindle), and coolant (when using through-tool coolant). With standard machines, there is no way to support additional functions such as power (electrical or pneumatic) and signal. This is why tools like touch probes need IR or bluetooth to communicate with the controller.

In the future, I believe that someone will come up with a set of standard formats for connecting interchangeable tools onto standard head plates. Heads could be static or dynamic (with one, two, or three axes of rotation), and the tools could be of any kind:

- Static tools for turning operations
- Rotating tools for milling or grinding operations
- Laser tools for engraving or cutting
- Laser sintering tools for additive manufacturing
- Mechanical probes
- Optical probes (with static lasers or LIDAR subsystems)

The standard would specify, for different sizes of interconnection plates, the formats and positions of connectors for the following:

- Electrical power (most likely with multiple voltages and phases)
- Signal (most likely an industrial version of USB or Ethernet)
- Pneumatics (for tool holding and/or tool rotation)
- Hydraulics (for tool holding on larger tools)
- Water Cooling (for spindles)
- Oil Cooling (for Minimum Quantity Lubrication)
- Fluid Cooling (for traditional cooling)

Smaller plate formats would have less powerful electrical power ports, no hydraulics, and no oil cooling.

This approach would bring many benefits:

1. The same machine could perform a wide range of operations on the part, in a single setup.

2. The same machine could perform different operations on different parts in a single setup.

3. The tool (especially the spindle) could be easily replaced when broken, thereby reducing downtime.

4. Tools could be shared across machines more easily, because more machines would share the same tooling platform.

5. Companies like Hybrid Manufacturing could innovate more easily.

Having signal going right to the tool would allow the tool to become intelligent. For example, a spindle could embed a computer that would monitor tool vibration and use machine learning algorithms to predict tool wear. Or it could measure tool deflection and dynamically adjust RPM or instruct the controller to adjust the tool path. Currently, this type of application requires custom connectivity between the tool and the controller, which dramatically reduces its applicability to existing machines. In other words, machines are very inflexible: the only modularity they provide is at the ATC level, allowing you to swap one milling tool for another.

When you think about such a platform, you quickly realize that our balanced 3+3 design is ideally suited for it. In comparison, it would be very difficult to retrofit a 5-axis Horizontal Machining Center with 2-axis turntable to accommodate such a modular tool head, because the tool (the spindle) is deeply embedded within the vertical column of the HMC. In contrast, a two-axis spindle head is an ideal platform for supporting interchangeable tools. In fact, I am willing to bet that if/when modular tool heads become commonplace, most machines will have 3+3 designs similar to the one that we have come up with.

And with that in mind, I believe that I have found my next project: working with industry leaders like Siemens and HSD to design such a standard.

[ishi]

Following our previous post, we need to take it to its next logical step: we do not need an ATC.

Why is that? Because if you have interchangeable tool heads (spindle, laser, probe, etc.), it is likely (and highly desirable) that your collection of heads and tools (like end mills) will be managed outside of the machine in such a way that they can be shared across multiple machines.

To do so, you will need a robot that can move back and forth between tool library and machines. There are two main approaches for building such a robot: on the floor, or on the roof. Historically, rooftop robots have been the preferred option, because they are safer and more reliable. But with the advent of floor robots in warehouses used for e-commerce logistics, a floor robot is probably a better option today (cheaper, more flexible, more versatile). For more on this topic, you can look at these links:

https://www.amazonrobotics.com/
https://www.exotecsolutions.com/
https://www.inviarobotics.com/

So, let's assume that we have a robot that can carry a modular tool head (like a spindle) or a tool (like an end mill) on a tool head holder or a tool holder. This robot will move around the shop floor, doing its best to avoid obstacles (machines and humans). And it will have a 6-axis arm to move the tool head or the tool from its tool holder to the machine. For that, the best option is to have two tool head holders and two tool holders mounted on the moving table, within reach of what we will call the tool head fork (the big red part on our machine). And that's pretty much it.

This might not be obvious at first, but such a redesign has massively positive implications:

First, you do not need an expensive ATC on each and every machine anymore. Instead, you just need a robot (or a few robots), and a central tool head and tool library.

Second, the static tool library can be much simpler than an ATC. In fact, it can be totally static: just a bunch of holes on shelves, with shelves probably at a slight angle, with the top of tools pointing away from the robots. By doing so, I am willing to believe that 10 to 20% of a machine's cost could be saved, on each and every machine. And you could build such a library for thousands of tools, at a ridiculously low cost.

Third, when you do not need to design your machine for it to include an ATC, you can design your machine in a totally different way. For example, the roof panel that gave us so much trouble yesterday becomes trivial. And the vertical column can be 200 to 300mm shorter, saving a solid 10 to 15% of its cost when you take into consideration the cost of rails and ball screws.

Of course, what I am describing can be developed for tools only, leaving the modular tool heads for later. And this is what we will do with our machine: we will develop it as a proof of concept for this next generation of machine shop, with centralized tool management first, then upgrade the machine with a modular tool head.

This is exciting...

[ishi]

Here is a short presentation for the ideas outlined in the past couple of posts:

https://docs.google.com/presentation...05ca24a66_0_50

[ishi]

When we started this project, we did not know precisely what we wanted to do. We gave ourselves a certain set of objectives, tried to build a machine that would satisfy them, learned a ton along the way, replaced some objectives by others, and kept pushing until we would stumble upon a really interesting idea. We did not know what such an idea could be, nor whether we would find one at all, but we knew that happy accidents happen, and we did everything we could to create the circumstances upon which one could occur.

Thanks to handlewanker's near-constant negativity, I think we've found our idea: a robot-friendly machine.

This is the kind of disruptive idea that has the potential to create real opportunities: it's inevitable, yet it's hard to embrace for established players. It's inevitable, because automation will happen, whether we like it or not. It's hard to embrace for established players because they're much more familiar with hardware that software, which is the reason why very few machine vendors develop their own controllers. Instead, they use standard controllers from Siemens, Fanuc, Heidenhain, or Num.

Therefore, our next step will be to redesign our machine with two priorities in mind:

1. Make it compatible with future modular tool heads.
2. Make it as robot-friendly as possible.

These two priorities are totally independent from each other. Each on its own can bring tremendous benefits, but both combined can really change the game. We'll see if we can put an industry consortium together for #1, but we'll focus on #2 as far as this particular thread is concerned.

So let's focus on #2. What does it really mean?

Well, first and foremost, as we've written before, the ATC is gone. Therefore, we can simplify the roof of our enclosure. And we need to add tool holders on the moving table. There, we'll probably have room for 5 tools on the short side of the table. This will allow the machine to change between 5 tools without the help of a robot. We will use the short side of the table because we have a lot more clearance alongside the X axis than the Y axis. All we need to do is design a simple cover that will protect these tools from chips. And that will give us the fastest chip-to-chip change time of any machine on the market...

Then, we need to realize that our primary operator will be a robot, not a human. As a result, we probably don't need large windows on our enclosure. In fact, if we want to use laser sintering, we really do not want any window at all, and we probably want the inside of our machine to be all black instead of white. Therefore, we should probably get rid of our windows and allow ourselves to be really creative about it. What do I mean by that? Well, why don't we replace windows by cameras and flat panel displays? These are really cheap now, and they could provide much better visibility into the machine, through multiple angles at the same time. Also, having cameras would allow for the remote monitoring of operations. Later on, having these video feeds would allow machine vision to be used for some radical applications, such as in-situ preview of toolpaths for collision detection. Conclusion: we will remove all windows, add a whole bunch of cameras (some static and some attached to the tool fork), and add a large flat panel display on the front panel.

Next, we should reduce the height of our vertical column. We can do that because we do not need to reach any top-mounted ATC anymore. And by doing so, we should be able to replace our telescoping rising front panel by a single section, thereby allowing us to use two simple pneumatic pistons for actuating it. This will in turn cut the width of our front enclosure pocket in half, because we just need to house one section instead of two. And we'll mount a large flat panel display there.

This display will be used to show the inside of the machine, through multiple angles. Furthermore, some of the cameras that we will use will work in the infrared frequency range, allowing hot spots to be identified, both on the machine and on the part(s). This will be useful both to the operator and to some machine vision algorithms that could be added in the future.

Then, there is the topic of probes for parts and tools. For indexing parts, we'll use a wide range of probes that can be shared across machines. And for tools, we'll put the best laser probes that money can buy in the tool library. Therefore, no machine will include these, which will further reduce costs. Everything that could be shared across machines should! This is a critical design strategy that we must follow diligently.

As far as chips are concerned, we can keep our current design, but we should plan on using our large pallet changing robots for changing chip bins. Therefore, we will have to design our own chip bins, the goal being to be able to run the machine in lights-out mode for extended periods of time.

Talking about the pallet, we will need to redesign our table in order to support the changing of pallets. I do not expect this to be very difficult, and we will design our pallet changing robot at the same time.

Also, in order to make the machine as "universal" as possible, we should give some thoughts about high-speed turning operations. For that purpose, we might want to add a 7th axis on the table, with a chuck and a bar feeder. This might require that we beef up our table quite a bit, but this should be relatively straightforward. With such a setup, the part would be mounted either on the rotary table (B axis) or on the chuck (A2 axis).

Now that we do not have an ATC, I estimate the cost of parts for this 7-axis machine to be in the range of $250k. And for that money, you would really have a machine like no other on the market...

Many thanks to handlewanker for the constant encouragements to push the envelope ever further!

[ishi]

Our little project will move offline for a while. If you want regular updates, please contact me at ishi at ishi dot io.

[handlewanker]

Ishi...….I am now convinced you are actually a robot and are hell bent on getting rid of the Human race by undermining the wage earning capacity of the common workshop environment and replacing it with metal " people" that don't need to frequently pee and take lunch breaks and holidays......that also can function 24/7 as opposed to the Union wage week of 40 hours.

This is a fact, and I am also convinced that each machine produced will have a robot to go with it...…….robots using machines to make robots that make machines to make more robots only means you have been found out and the Human race had better take up arms to combat your army of invasion...…..Orwell got the dates wrong by the way......probably nearer to 2084.

Being a robot I'm sure you won't take Umbridge at the skepticism exhibited so far.

BTW.....you have re-invented tomorrow's world in as little as 2 months...…..I don't think the World or the Human race is quite ready for the Nuovo way to make money without getting your hands dirty yet, maybe in a century or two after we've been to Mars.
Ian.

[handlewanker]

Hi all....just watched a video on UTUBE about the 5 axis POCKETNC-V2...….a very small bench top 5 axis model with a horizontal spindle that you could fit in your pocket......metaphorically speaking..... well it isn't in the same frame that Ishi is working on but it does give you a 5 axis capability and not for megabuck at that.

It appears to my CNC ignorant mind that it seems to be a mill with a horizontal spindle and a type of 4th axis and trunnion table layout.....what that aspect is I don't know......everything moves.....I'll have to take another looksee to see what moves and in what order.

No price was mentioned but he did say a couple of grand.....pant, pant, gasp...…..there goes the family jewels and the kid's inheritance again....these toys are completely insidious to the bank balance.

Edit....just had a Google and the price is around $5k for the basic machine plus a lot of extras available..
Ian.

[pippin88]

Chip to chip time with a decent ATC is <10 seconds. What is the chip to chip time going to be with your robot idea?

Will it really be cheaper to have robots etc rather than multiple ATCs and tools? I doubt it.

What do you do when you need two machines running the same tool at the same time?

[ishi]

Chip to chip time with a decent ATC is <10 seconds. What is the chip to chip time going to be with your robot idea?

Will it really be cheaper to have robots etc rather than multiple ATCs and tools? I doubt it.

What do you do when you need two machines running the same tool at the same time?

pippin88,

The fastest machines currently available on the market will give you close to 2 sec. chip-to-chip time. With our latest design, we should get in the 4 to 5 sec. range. The longer time is due to the fact that we'll have to move a fairly large internal protection panel, which we can't move as fast as a small trap door.

Our design has taken a radical new direction: a universal machine optimized for a robotic farm environment, meaning a large number of machines (10 or more) tended by robots. To get the savings that we expect on the tooling front, you would need to use a fairly sophisticated piece of planning software that would optimize the distribution of tools across machines, with a priori knowledge of which parts will be made by which machine in the upcoming few hours.

Interestingly-enough, a 6-axis robot with 10 kg payload is cheaper than a 40-tools ATC... Therefore, our machine will include such a robotic arm, inside the machine. And we'll make sure to provide internal storage for 20 to 40 tools in the machine, using the outside robots only when more tools are necessary. Think of the internal tool library as a buffer between the spindle and the outside library that is shared across multiple machines.

Having a 6-axis arm in the machine, we can be quite creative now. For example, adding a 7th axis for high-speed turning operations, using the arm to move materials and parts between an internal storage carousel, the lathe chuck, and the rotary table. With that, you would get a truly universal machine combining the benefits of an HMC, VMC, and lathe.

You might want to watch this video:

https://www.youtube.com/watch?time_c...&v=6QZO8ZJTjqo

Now imagine all that packaged into a 2m wide cube. This is what we are designing...

It will be unlike any machine that is available today, for better or for worse.

When you design a machine for robots, it ends up being very different than a machine design for humans. The same happens when you design an electric car: it ends up being very different from a car with a combustion engine, and the differences can be found everything, not just in the power plant or drive train. The BMW i3 is a great example for that. And when you design a driverless electric car, it ends up being totally different from a traditional car.

We're applying the exact same thinking to CNC machines.

[handlewanker]

Sounds like a good plot for a horror movie where some kids break into a factory of the future and get chased around in the dark by busy robots...….on the theme of "Oh no, not another teenage horror movie". the outcome is all blood and guts as they get trapped and loaded one by one into various machines and machined to a pulp......they had a good finish though.....LOL.

Now I quite fancy the idea of a pick and place robotic arm to pick tools out of a rack and stick them up the spindle of my mill.....move the tool not the toolbox as in a carousel etc......but a robot per se doing a walk about in a factory is not my cuppa any day......I get the feeling that it's a case of diminishing returns to produce a few parts with such a huge infrastructure outlay.

That it could happen is on the cards when someone with infinite wisdom decides it should be the way.
Ian.

[skrubol]

Have you seen the DMG Mori Lasertec machines?
I think you need to do a lot more research on your new market. Nothing you are proposing sounds novel to me. There are machines with pallet changing robots, and machines with shared tooling cribs that use robots to transfer tools between the small 40 to 100+ tool magazine in the machine and a 1000+ central tool crib with tool wear management, etc. I don't know of many machines that use the same robot for workpiece and tool movement (I have seen cases where a tool too large for the ATC will have a holder on the pallet or tombstone of its job,) as a robot meant for moving a 1000+lb pallet from the plane of the table of the mill is not likely to be good at dealing with <40 lb tools needing to be at the top of the travel of the machine.

[handlewanker]

The more I read over the past posts the more I come to the conclusion that something is missing...…….how can this machine train you to do 6 axis work when it is unique in that it does not exist in industry and so if you trained to work this machine all you would experience is that a 6 axis mode can do wonders.

Going out into the workforce you would have to retrain all over again as nowhere would there be a machine like this that you are familiar with.

True there are other machines that do 5 or 6 axis work but not as you would be familiar with......you might as well only get familiar with a cheap Ebay mill with a 4th axis and trunnion table to do 5 axis work and then once you know how that works do the real thing in the work place as a newby.

One thing is for sure, if you were in a work place where robots were tending to machines you would not be anywhere near one, and just knowing what a 6 axis machine can do does not qualify you to work on one.

It's an interesting conceptual design....the machine itself I mean, but mainly as a novel approach to making difficult workpieces in a different way...…..if the price were scaled down dramatically perhaps it could get on EBAY and sell like hot cakes......just my opinion.....but I can't see that ever happening with that 2 axis spindle design which is required to make this machine design a possibility..
Ian.