6-axis Horizontal Machining Center for Education

[handlewanker]

Hi, I believe you when you say that your 6 axis design is better than any 5 axis type......for the reasons you give...…..as my brain is not orientated to the way you want to machine I took the simple approach and thought that any job has to be anchored to the table, rotary or fixed....whatever...…..by one face, purely from the hypothetical aspect as I will not be indulging in a build.

So, as I see it, all it takes is 5 axes to reach all 4 faces of a job and the end face...…...with a rotary table on a trunnion...…. this is easily accomplished and with a simple spindle layout as is used on all vertical mills.

I understand that with the design configuration you are pursuing,...IE, having the spindle in a horizontal mode although it can be any which way possible with the jointed type you are going to use...…….. 6 axis is the only way to accomplish it.

Just for the exercise, I cannot see where a 5 axis conventional vertical mill would not produce a job like the one you showed of the whisky jug in the previous thread. ……..as long as you can rotate the work piece in 5 planes you can machine all of them

In simple terms, what shape can you not machine with a vertical mill having a 5 axis rotary table/trunnion design?

As you can guess, I am a firm believer in the axiom that when all things are equal the simplest way is the best.
Ian.

[ishi]

Hi, I believe you when you say that your 6 axis design is better than any 5 axis type......for the reasons you give...…..as my brain is not orientated to the way you want to machine I took the simple approach and thought that any job has to be anchored to the table, rotary or fixed....whatever...…..by one face, purely from the hypothetical aspect as I will not be indulging in a build.

So, as I see it, all it takes is 5 axes to reach all 4 faces of a job and the end face...…...with a rotary table on a trunnion...…. this is easily accomplished and with a simple spindle layout as is used on all vertical mills.

I understand that with the design configuration you are pursuing,...IE, having the spindle in a horizontal mode although it can be any which way possible with the jointed type you are going to use...…….. 6 axis is the only way to accomplish it.

Just for the exercise, I cannot see where a 5 axis conventional vertical mill would not produce a job like the one you showed of the whisky jug in the previous thread. ……..as long as you can rotate the work piece in 5 planes you can machine all of them

In simple terms, what shape can you not machine with a vertical mill having a 5 axis rotary table/trunnion design?

As you can guess, I am a firm believer in the axiom that when all things are equal the simplest way is the best.
Ian.

Ian,

That's an interesting exercise!

Well, let me start with this: the whiskey tumbler was machined on a 5-axis machine, for the very simple reason that 5 axes are all you need for this piece, and 6-axis machines are very exotic (and tend to be very large). In general, from a purely geometric standpoint, 5 axes are enough for most pieces.

Unfortunately, geometry is only one part of the equation. There are many other aspects to consider. A critical one is envelope. If your part is large and needs to be machined in fairly convoluted ways, a 6-axis machine might be the only option. For example, take a look at this one:

http://www.f-zimmermann.com/fileadmi...Z100_EN_US.pdf

There, you clearly see that removing a single axis from the 3-axis head would prevent many cuts from being made.

Or take the machine that we are currently designing:

https://drive.google.com/open?id=1jz...VtX68r3PZ5QY7M

If you remove the axis of the rotary table, your envelope alongside the Z axis (the table's short dimension) is reduced by half all of a sudden. Not only that, but you won't be able to make cuts with the spindle being horizontal on the face of your part that is looking away from the spindle.

So, what can't you do on a VMC with a trunnion table? Well, first and foremost, you won't be able to cut large pieces, because your trunnion table will be small and won't be able to handle heavy pieces. To convince ourselves of that, let's look at Haas' largest trunnion table, the TR310:

https://www.haascnc.com/machines/rot...els/tr310.html

With this 1,720lb beast, you're limited to a 12.2in platter with a 500lb capacity. That means you're going to need a large VMC, something like a VF-6, but you won't be able to cut pieces larger than 12.2". Let's do some math here: with a VF-6, your envelope is 64" x 32" x 32", which is 61,440 cubic inches. On the trunnion table, you've got 12.2" to the cube, or 1,816 cubic inches. By simply adding a trunnion table, you have wasted 97% of your envelope. You paid for the whole machine, but you're getting to use only 3% of it.

Not only that, but this very large machine is doing a ton of work moving this 1,720lb trunnion table around, just so that you could cut a piece that is likely less than 250lb, or possibly much less. As a result, its accuracy is dramatically diminished compared to the accuracy of a machine that would just move the piece, which is what we are doing with our machine, having moved 2 axes of rotation from the piece (the trunnion table) to the tool (the 2-axis spindle head).

Compare that to what we have: 20" platter (almost twice as large, which means 8 times the volume), and a load capacity that remains to be calculated but is very likely to be in excess of 500lb. How much of the machine's envelope is being "wasted" by the rotary table? Very simple: the table's area is 30" x 20", which is 600 square inches, while the rotary table's area is 10" x 10" x PI, or 314 square inches. Therefore, we've wasted less than half of our envelope. In other words, our waste is 17 times less. From a volumetric standpoint, we're 17 times more efficient that a VMC with a 2-axis trunnion table. And what is our total cutting envelope on the rotary table? 9,425 cubic inches, which is more than 5 times larger than the Haas VF-6 with the TR310. The Haas is 195" x 102" x 130" (2,585,700 cubic inches). Our machine is 69" x 69" x 80" (380,880 cubic inches). Therefore, our machine's volume is 6.8 times smaller than the Haas', which means that our ratio between cutting envelope and machine envelope is 35 times greater than the Haas'. That's how efficient our design is from a volumetric standpoint.

In light of these explanations, I invite you to re-read the post I referenced yesterday. I know it's kinda theoretical and probably boring (no pun intended), but the essence of what we are doing is in there, somewhere to be found.

https://www.cnczone.com/forums/uncat...ml#post2198222

What can we learn from this little exercise? Quite simply: a VMC with a trunnion table is probably one of the worst configurations for a 5-axis machine. It's popular because it comes as an easy add-on to 3-axis machines, it's easy to understand, and it's easy to design. But from a volumetric standpoint, it's massively inefficient. As a result, "native" 5-axis machines never come in this configuration. Instead, the most efficient of them from a volumetric standpoint usually come in the form of a static 2-axis trunnion table and a spindle attached to three linear axes, like this one for example:

https://i.pinimg.com/originals/1f/4e...4eaac15443.jpg

Is that design "simpler" than ours? I really don't think so. For two main reasons:

1. The trunnion table is massive and very difficult to engineer and build.
2. The 3 linear axes are built on top of each other, making it very difficult to engineer properly.

I wish I could design a machine like that. I really do. My benchmark in that respect is the DMG MORI NMV5000 DCG:

http://www.ir4i.it/cgi-bin/images/pa...0-pdf-data.pdf

This is the most beautiful design I know for a 5-axis machine. But if you study the brochure, especially the Octagonal Ram Construction for the Z axis, you will quickly realize that it is anything but simple. It's a very sophisticated design that is positively out of reach for "regular" builders like me. Therefore, I would disagree with the statement that a VMC with a trunnion table is a simpler design. If you want volumetric efficiency, it is not...

Once you dive deeper into all these considerations and put them all together, you will realize that our design is probably one of the simplest that you can imagine in order to get 5 axes with volumetric efficiency, volumetric efficiency being defined as the ratio between part envelope and machine envelope, with the assumption that the part envelope is fully contained within the cutting envelope. And it gives you a 6th axis just for kicks. Like the eleventh position on a guitar amplifier, just because louder is better. ;-)

Best regards
Ismael

[ishi]

As a follow-up to the previous post, I would like to develop the idea of volumetric efficiency for 5-axis/6-axis machines a little bit more.

First, I will define two metrics:

1. Envelope Volumetric Efficiency
This is defined as the ratio between part envelope and linear axes envelope, with the assumptions that the part's envelope is fully contained within the cutting envelope, and that the part is mounted on the machine's rotary table. If the machine does not have any rotary table (Cf. Zimmermann FZ100), the second assumption can be ignored.

For the machine that we are building, the part envelope is ? * (10in)² × 30in, or 9,425in³, and the linear axes envelope is 30in × 30in × 20in, or 18,000in³. This gives us an Envelope Volumetric Efficiency of 9,425 / 18,000, or 52%.

For a Haas VF-6 with a TR310 2-axis trunnion table, the part envelope is (12.2in)³ or 1,816in³, and the linear axes envelope is 64in × 32in × 32in, or 61,440in³. This gives it an Envelope Volumetric Efficiency of 1,816 / 61,440, or 2.9% (18 times less).

2. Machine Volumetric Efficiency
This is defined as the ratio between part envelope and machine envelope, with the same assumptions as the ones used for defining the Envelope Volumetric Efficiency.

For our machine, the machine envelope is 69in × 69in × 80in, or 380,880in³. This gives us a Machine Volumetric Efficiency of 9,425 / 380,880, or 2.5%.

For the Haas, the machine envelope is 195in × 102in × 130in, or 2,585,700in³. This gives it a Machine Volumetric Efficiency of 1,816 / 2,585,700, or 0.07% (36 times less).

Why do these metrics matter? Simply because at the end of the day, all that matters is the piece that was cut. The machine is a means to an end. The part is the end that matters. Therefore, the part's envelope is the critical factor. Now, the larger the machine, the more expensive it is, and the less accurate it gets. In the past, CNC machines used to be sold by the ton, which means that their price would increase in a cubic fashion in relation to their linear size. Today, I think the relation is more quadratic than cubic, but it still follows a power law. And as far as accuracy is concerned, everybody knows that it is usually measured in relation to lengths. For example, the flatness of a surface plate is measured in microns per meter. Therefore, the bigger the machine, the more expensive and the less precise it is.

Considering all this, and assuming that we want to machine the largest possible pieces with the highest accuracy and at the lowest cost, Volumetric Efficiency as defined above (be it of the Envelope or Machine kind) becomes a critical selection factor for a 5-axis/6-axis machine.

Why do I qualify this statement with the 5-axis/6-axis restriction? Simply because 3-axis machines have more consistent volumetric efficiency ratios. There are only so many ways that one can design a 3-axis machine compared to a 5-axis or 6-axis one (the combinatorics are huge there). It's only when you go from 3 axes to 4 or 5 axes that volumetric efficiencies can drop in a rather dramatic fashion and becomes something that you really should pay attention to.

Okay, but which ratio should we consider? Envelope or Machine? Well, I think Envelope Volumetric Efficiency really is the one to pay attention to, because it's more closely related to machine accuracy and machine cost, for the reason that it won't be skewed by peripheral accessories like a large automatic tool changer or a palette changer. Most importantly, the accuracy of a machine is directly influenced by the lengths of its linear axes, not by the linear dimensions of the machine itself. As a result, I will focus on this metric moving forward. In my opinion, if I had to pick a single metric to evaluate a 5/6-axis machine, this would be it.

With that in mind, how does our machine compare to the DMG MORI NMV5000 DCG? Well, its part envelope is ? * (9.85in)² × 17.7in, or 5,395in³, and its linear envelope is 28.7in × 20.1in × 20.1in, or 11,595in³. This gives it an Envelope Volumetric Efficiency of 46%. So, in that respect, we're doing a tiny bit better.

But is our design really better? No it's not, for a very interesting reason: unlike linear axes which accuracy goes down with length, rotary axes tend to be more accurate the larger their diameter is. This is due to the fact that the smaller the diameter, the less torque you have, especially when using torque motors. With a torque motor, the only factor to consider is torque, and torque is linearly correlated both with diameter and height. The larger the diameter, the more torque you get, and the same goes for height. You can see that when looking at the HSD catalog for 2-axis spindle heads:

http://www.hsd.it/bo/allegati/Files/2604_ita_web.pdf

With the relatively small head that we are planning to use (HST310), we get 30 arcsec resolution on both A and C. Move to the 10 times heavier HS810, and accuracy goes up to 2 arcsec on A and C. It's almost as if there is a linear correlation between weight and accuracy, which means a cubic correlation between linear size and accuracy.

This is a very, very important thing to consider: rotary axes do not work like linear axes when it comes to accuracy. For the former, bigger is better. For the latter, smaller is better.

Bottomline: when designing a 5/6-axis machine optimized for volumetric efficiency, try to get your linear axes as short as possible, and your rotary axes as large as possible.

But do you really want the biggest rotary axes possible? Probably not either. You see, the DMG MORI NMV5000 DCG will get better accuracy than our current design, because it uses bigger torque motors. But we could get even better accuracy than the DMG MORI by moving to a bigger 2-axis spindle head. That is because the accuracy of a torque motor in relation to its diameter does not follow a cubic law forever. It looks like that at first, then it reaches its optimum, then goes down from there. And as far as I can tell, for the kind of machine that we want to build, the optimum is somewhere around what the HSD HS810 (single sided or double sided) offers. That probably means a machine footprint of 90in × 90in × 90in. That will be the ISHI OMC 45 Mark I for you...

But for now, we should focus on the OMC 30, because we're still a very long way from being able to cut our first chips with it...

[ishi]

Great news: we just had a call with GATE Technologies, Inc., and they confirmed that the TL105 will work perfectly for what we need.

TL Series | Modular Precision Link Transfer Conveyor | Walking Beam Transfer | High Precision Indexing Conveyors | Gate Technologies Inc

They also really liked our design and confirmed that there is a market for entry-level 6-axis machines, far beyond the education market.

The conveyor isn't cheap ($30k for one, 10% off when ordering 10), but it's not something that we feel we can design on our own at this point. Also, I expect that we could get price below $25k with a bit more negotiation. The price includes the reducer to which the servo motor will be connected. We will be able to use the exact same servo motor that we are using for all our linear axes, which will simplify our supply chain and our connection to the SINUMERIK controller.

Most importantly, they can provide engineering assistance for the design of our tool fork. And if it does not work, they'll help us design a tool holder with a more positive locking mechanism. This would require an extra actuator, but it's good to know that we have this option, just in case.

At this point, we have direct support from the following suppliers:

- RAMPF for the machine base
- SIEMENS for the mechatronics
- HSD for the spindle head
- GATE for the automatic tool changer

The last major supplier that we need to identify is the one for the iron castings. Once we have that one lined up, we'll be able to go to the next phase of the project, which will be to prototype the tool holder in partnership with GATE, because we need to get this part right. If we don't, the entire machine's design would fall apart.

[ishi]

Here is a quick status review for the project.

We've been at it for about a month now (initial post on June 24). At this point, we have the following:

- An original design for a 6-axis machine (I have yet to see another machine with the same configuration)
- A very simple yet effective design for the ATC (single actuator)
- A relatively complete and coherent list of parts
- An almost complete group of suppliers (still looking for the best place to get iron castings)
- Some options for where the original prototype could be built (still working on confirming one)
- Some preliminary market validation (a lot more work remains to be done there)
- A handful of experts telling us that the high-level design makes sense
- Nobody (yet) screaming out of their lungs that it does not

With that in mind, I would like to call this Milestone 1: we have a design that we like. Therefore, we are freezing the following parameters:

- Mineral casting for the base
- 6-axis design in fully-balanced 3+3 configuration
- Approximate machine envelope (69" × 69" × 80")
- Approximate part/cutting envelope (30" × 30" × 20")
- Accuracy of 2 microns on every linear axis
- Repeatability of 1 micron on every linear axis
- Accuracy of 30 arcsec on every rotary axis
- Chip-to-chip tool change below 10s in average spindle and conveyor positions
- Mineral casting supplier (RAMPF)
- Mechatronics supplier (SIEMENS)
- Spindle head supplier (HSD HST310, with three spindle options)
- ATC conveyor supplier (GATE)
- Prototype development costs: $200k
- Cost of parts per unit for a batch of 10: $150k

For better or for worse, this is it. These parameters are critical, because they will prevent us from wandering around, or going off a tangent. Any good artist or engineer needs constraints, be they dimensions on a painter's canvas or technical specifications for a machine.

So, where do we go from there?

Well, Phase 2 will be about refining the design. Most importantly:

- Deciding whether we want a hydraulic counterbalance for the vertical Y axis.
- Deciding whether we want to add support for coolant recycling and chip evacuation.
- Figuring out how to install our linear and rotary encoders.
- Freezing travel lengths for X, Y, and Z.
- Redesigning our base to take advantage of our switch to a mineral casting.
- Performing some basic finite element analysis for our iron castings.
- Selecting our supplier for iron castings.
- Finalizing the design of our electronics.
- Positioning the electrical, hydraulic, and pneumatic enclosures.
- Designing the machine's enclosure.

Doing that would allow us to freeze most of the machine's parameters and ensure the soundness of the end-to-end design, up to a point. We hope to be done with Phase 2 by the end of August. After that, some real prototyping work needs to be done.

By switching to a mineral casting, all technical risks related to the use of granite stone have gone away. At this point, the only major technical risk that we can foresee is related to the way tool holders are attached to the ATC conveyor. At present time, we are considering the use of a plastic fork, but a more positive tool holder retention mechanism might be needed. And if we get that wrong, the whole design falls apart, because the mineral casting will incorporate everything we need to attach our ATC conveyor, and switching to another ATC would translate into a $50k mistake ($40k for the steel mold and $10k for the initial casting). Definitely something that we want to avoid...

Therefore, we better test our tool forks extensively. How will we do that? First, we will buy as many forks as we can:

https://www.hsspindles.com/product/hsk-63f-tool-fork/
HSK 63F Plastic Tool Holder Forks for Automatic Tool Changer | Global Sources
HSK63F cnc tool finger forks for HSK 63F tool holder clamping

Second, we will buy a long HSK 63F tool holder with arbor:

https://www.hsspindles.com/product/h...125-inch-long/

And we will attach a 5kg mass to its end in order to simulate a fairly heavy tool.

Third, we will build a testing rig with a Siemens SIMOTICS 1FK7060 servo motor to take the tool holder in and out of the fork, with a simple rotation. Doing so with the same servo motor that we will use for our linear axes and for the ATC conveyor will give us a feel for the motor. And we will run this rig for a month, non-stop. I estimate that we could get 20 insertions/extractions per minute, which means 28,800 per day, or close to a million in a month. Every day or so, we'll take the tool holder, tool fork, and tool fork mount assembly in our hand, holding the tool fork mount. We'll then shake the whole thing out to ensure that the tool holder does not snap out from its fork. If it stays in place after a month, we'll know that we have a working design. And if it does not, we'll go back to the drawing board.

Using a similar testing rig but replacing the servo motor with a torque wrench, we'll check how much torque is required to take the tool in and out. We'll then make sure that we get more torque on the A axis of our 2-axis spindle head in order to perform the same operation.

These two tests should be enough to validate the overall design of our ATC. But to remain on the safe side, we'll also contract GATE Technologies, Inc. to review (and improve) our design. This will cost us a few hundred dollars, but it should be money well spent (budget: $1,000).

In parallel to this work, we'll have to design our mineral casting. There, we have a pretty steep learning curve, because mineral castings are a lot more complex that granite plates. Why is that? Because you can do all kinds of things with a mineral casting. The file called "CPT Presentation" on the public drive explains that in quite a bit of details:

https://drive.google.com/drive/u/3/f...kwBI-?ogsrc=32

To make a long story short, a mineral casting can include not only thread inserts but other steel mounting elements, a steel frame, polystyrene blocks to reduce the weight of the casting, cables, hoses, pipes, etc. Also, a mineral casting is much more flexible than a granite plate in terms of geometry. Therefore, you can include shoulders, flanges, ribs, or cutouts for many different purposes. For example, having cutouts for the forks of a pallet jack would not be a bad idea. So, in order to take advantage of all these possibilities, we need to make an inventory of everything we'd like to get in there, ensure that it's the best way to do it, figure out how it should be done with a mineral casting, and update our design accordingly. To do so, we will read as much as we can on the subject, getting RAMPF to feed us with all the literature they can get their hands on.

Once we have a good-enough design, we'll run it past RAMPF and pay their engineering department for a formal review. This should cost a few hundred dollars as well (budget: $1,000), but again, this should be money well spent, and this should take us to the end of September.

Once we have a pre-final design for the machine base, we'll focus our attention on the three carriages for the three linear axes. These will be built using cast iron. Once we've selected a supplier for it, we'll refine their designs with their engineers and order a first casting (the one for the X axis, because it will allow us to play with linear rails and ball screws). We'll then have it machined at a local machine shop to ensure that we can get the right level of accuracy in our iron casts (budget: $5,000).

From there, we'll build the entire Z-axis drivetrain around the X-axis carriage (budget: $13,000). This will give us a first introduction to the kind of mechanical parts that we'll have to deal with for the rest of the machine. At this point, we'll have spent about $20k, and if we're not totally scared off by the whole experience, we'll carry on with the rest of the project. I expect that we'll be in December by then.

The next step will be to get the machine base fabricated. Before we order a $50k chunk of gravel and epoxy, we'll pay the manufacturer a visit on the East Coast. We'll study their process as carefully as we can, and look at different castings they've made in the past. Equipped with all that knowledge, we'll head back home and build a scale model of the machine out of cardboard, cables, hoses, and PVC pipes. We'll make sure to have every single cable and hose replicated, as well as every single nut, bolt, and screw. And we'll pay special attention to the order in which parts are being assembled, thereby ensuring that our machine can be built as designed. Once we're satisfied with everything, we'll do another round of validation with RAMF's engineers and order the most expensive part of the build (because of the $40k steel mold).

Once we receive our shiny machine base, we'll then build the X axis (budget: $15,000) and test the accuracy and repeatability we get on the X and Z axes. If we like what we get, we'll carry on. If we don't, we'll try to figure out what's wrong and fix it before we spend another $85K... By that time, Spring should be in full bloom.

At this point, we'll order the iron casts for Y and X as well as the drivetrain for Y ($30K). We'll run another batch of tests and asses our design again. At this point, we'll be $115K into this. If we still have it in us, we'll carry on and order the spindle head, torque motor for the rotary table, and controller. That will be an $85k check, and that will give us everything we need for a complete machine. Hopefully a year from now.

Wish us luck!

[ishi]

In order to learn more about metal casting, I just ordered the following two books:

Principles of Metal Casting by Mahi & Sahu, Sam Sahoo
https://www.amazon.com/Principles-Me...dp/9339218167/

Complete Casting Handbook by John Campbell
https://www.amazon.com/dp/0444635092

This should be a good starting point...

[ishi]

Take a look at this video:



First, it's awesome to see how such a large machine is built.

Second, one thing surprised me when watching the video: even on a machine of that size, they seem to install a hydraulic counterbalance for the vertical Z axis. And that reminded me that we should make a final decision regarding whether or not to install a counterbalance for our vertical Y axis. But before I go there though, I have a question to ask: now that we are calling our machine an Omnidirectional Machining Center, it's neither vertical nor horizontal, it's both. Therefore, there is no reason why we could not label our vertical axis Z instead of Y. So, for the sake of familiarity, should we go for that?

Okay, back to the original point: should we add a counterbalance for the vertical axis?

Here is how the issue was framed in an earlier post:

"There is some debate among advisors to the project regarding the need of a counterbalance for the vertical Y axis (the axis driven up and down the vertical column). An earlier version of the design included a hydraulic counterbalance made of two gas tanks and two hydraulic pistons. Later on, we decided to remove this apparatus, with the assumption that we could achieve similar results by using larger servo motors for the Y axis and a proper motor module (Siemens SINAMICS S120 Smart Line). From a performance standpoint, specifications would be the same or better. From a design standpoint, things would be greatly simplified by using mechatronics instead of hydraulics, and the same would be true regarding maintenance (tanks and pistons need to be replaced on a regular basis). Nevertheless, one advisor raised the issue of safety, which is also critical to us, especially within an academic environment. Therefore, this issue raises the following question: can we get an equivalent (or better) level of safety with mechatronics only?"

From a mechanical standpoint, two servo motors with brakes are all we need. But from a safety standpoint, having a counterbalance would be really nice. Here is why: let say that we don't have a counterbalance and that I am stupid/clumsy enough to let the tool dive through my hand. Now, let say that I am fast enough to press on the emergency stop button before the machine shreds my entire arm to bits. Now, I have about 150kg of machinery sitting on the vertical axis, and unless my name is Hulk, I won't be able to lift it with my other hand, no matter how much adrenaline is flowing through my brain right now. So, what do I do beside screaming out for help? Turn the machine on again and hope for the best? Or wish that I had purchased the counterbalance option in the first place?

Unless I missed something, the $2,000 counterbalance option (1% of the fully assembled machine's total cost) will seem like a rounding error when blood is dripping down on the machine's bed. So, unless someone comes up with a better option, we'll follow the recommendation of one of our advisors and bring that option back into the design.

Now, where do we put the two hydraulic cylinders, and why do we need hydraulic cylinders in the first place? Why could not we use simple pneumatic ones? Well, for a very simple reason: when using hydraulic cylinders, you can adjust the pressure of inert gas (nitrogen) that you put in the cylinders in addition to the oil. As a result, by adjusting this pressure, you can precisely balance the amount of force exerted by the counterbalance in relation to the weight of whatever the vertical axis is moving up and down. Therefore, the design of the counterbalance becomes totally independent from the design of whatever is riding your vertical axis.

Okay, so, where do we put our two hydraulic cylinders?

https://www.bimba.com/Products-and-C.../TRD-HH-Series

My initial thought was to put them inside the vertical column, next to the ball screws. But this won't work with our current design, because we do not have enough space within that column. You can see this clearly on the picture called "Top". There is enough space for the motors and ball screws, but not for the cylinders. And if we try to make space for them, the vertical column will become very fragile, which is not a good idea, or we'll have way too much stuff in there, which will make installation and maintenance a nightmare.

But there is another obvious option: we can put them on the front side of the column, sitting on either side of the vertical axis carriage, right underneath the two flanges used to support the ATC conveyor (see picture called "Machine Front"). This space is empty and nothing comes to mind when trying to imagine what it could be used for. Furthermore, having the cylinders on the sides would make their replacement really easy, because this replacement could be made without having to remove the vertical carriage assembly (a huge benefits compared to other machines).

So, here we go, the parts for the vertical counterbalance have been added back into the BoM, and we know where to put them in the CAD (hopefully sometime tomorrow).

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

Many thanks to K. for his wise words of caution.

[skrubol]

Have you done any FEA simulation yet? Your saddle does not look very rigid, and your base and column, especially now that they're going to be a composite that's less stiff than straight granite (which is 2.5-4x less stiff than gray iron,) that section underneath your ball screws looks pretty thin.
If you want this to perform anywhere near as well as a UMC-750, you will want need some more rigidity. Does RAMPF give material properties for their finished castings?

Those trinkets from Discommon you bring up occasionally, the tumblers you mention requiring 5 axis machining look to be done on a 4 axis HMC. Most of their stuff could be done on a $5k Tormach.

[skrubol]

From a mechanical standpoint, two servo motors with brakes are all we need. But from a safety standpoint, having a counterbalance would be really nice. Here is why: let say that we don't have a counterbalance and that I am stupid/clumsy enough to let the tool dive through my hand. Now, let say that I am fast enough to press on the emergency stop button before the machine shreds my entire arm to bits. Now, I have about 150kg of machinery sitting on the vertical axis, and unless my name is Hulk, I won't be able to lift it with my other hand, no matter how much adrenaline is flowing through my brain right now. So, what do I do beside screaming out for help? Turn the machine on again and hope for the best? Or wish that I had purchased the counterbalance option in the first place?

The machine shouldn't be allowed to run with the enclosure open.
Normally even with a counterbalance on the vertical axis, you still would have a brake on the servo I think, as the difference between no tool in the spindle and a large face mill might be enough to overcome friction. You're probably going to be stuck there either way..

[ishi]

Have you done any FEA simulation yet? Your saddle does not look very rigid, and your base and column, especially now that they're going to be a composite that's less stiff than straight granite (which is 2.5-4x less stiff than gray iron,) that section underneath your ball screws looks pretty thin.
If you want this to perform anywhere near as well as a UMC-750, you will want need some more rigidity. Does RAMPF give material properties for their finished castings?

Those trinkets from Discommon you bring up occasionally, the tumblers you mention requiring 5 axis machining look to be done on a 4 axis HMC. Most of their stuff could be done on a $5k Tormach.

skrubol,

Which "saddle" are you talking about?

No FEA yet, this is part of my plan for August.

Yes, the base must be thicker, for several reasons: rigidity and height of the table (from an ergonomics standpoint). I think we can make it 6" to 8" thicker, while still getting 30" of travel on the vertical axis now that we've streamlined the spindle head's carriage. I was planning to work on that today actually.

Here are the specs for RAMPF's EPUMENT.

Density: 2.4g/cm³
Modulus of Elasticity: 40-45kN/mm²
Poisson's Ratio: 0.3
Compressive Strength: 130-150N/mm²
Flexural Strength: 30-35N/mm²
Coefficient of Thermal Expansion: 15E-6K-1
Logarithmic Decrement: 0.022

The attached screenshot also provides a nice comparison between different materials.

Bottomline: a mineral casting is best as far as corrosion resistance and vibration dampening are concerned. For the rest, cast iron or steel win. But for what we need, I still think that a mineral casting is the way to go.

[ishi]

The machine shouldn't be allowed to run with the enclosure open.
Normally even with a counterbalance on the vertical axis, you still would have a brake on the servo I think, as the difference between no tool in the spindle and a large face mill might be enough to overcome friction. You're probably going to be stuck there either way..

Fair points!

Well, the proposed design for the vertical counterbalance has no side effect on the rest of the design (beside a few inserts on the vertical column), so it's good to know that we have this option should we decide to add it.

[ishi]

Have you done any FEA simulation yet? Your saddle does not look very rigid, and your base and column, especially now that they're going to be a composite that's less stiff than straight granite (which is 2.5-4x less stiff than gray iron,) that section underneath your ball screws looks pretty thin.
If you want this to perform anywhere near as well as a UMC-750, you will want need some more rigidity. Does RAMPF give material properties for their finished castings?

Those trinkets from Discommon you bring up occasionally, the tumblers you mention requiring 5 axis machining look to be done on a 4 axis HMC. Most of their stuff could be done on a $5k Tormach.

skrubol,

If you send me your address, I'll send you one of those tumblers (assuming they can still be ordered). And if you still believe that they can be made on a $5k Tormach, I'll buy one right away. I've used a Tormach, and I cannot fathom how anyone could machine such a piece with such a nice finish on such a machine. If you know, please teach me.

For reference purposes, the tumbler was made on a Kitamura Mycenter-HX300iF.

I don't have the specs for the "F" version, but here is the "G" version:

https://www.kitamura-machinery.com/p...centerhx300ig/

Accuracy: ±0.000079” Full Stroke
Repeatability: ±0.000039”

As a point of comparison, the combined positional accuracy of the best Tormach (1100M) is 0.0013” per foot (Cf. User Manual). That's 16 times less (the Kitamura's full stroke is one foot exactly when using the rotary table). And that's for a $25k machine, not $5k. I'm sorry, but this tumbler cannot be machined with that level of accuracy. Can you machine a tumbler on a Tormach equipped with a rotary table? Absolutely. Can you machine this kind of tumbler? Positively not.

Please don't get me wrong though: Tormach makes great little machines, and the value for the money is probably second to none, but you get what you pay for, and if you need more accuracy than what they can deliver (33 microns), you'll have to pay more.

[ishi]

With the current design, if we put the motors for the vertical axis at the bottom of our vertical column instead of the top, we can have a travel of 30", which is 50% more than what we had designed the machine for originally (20"). With the ATC installed, the machine is 75.4" tall, without vibration dampening feet, which would add about 2.5 inches. The internal height of an intermodal container is 92". So, that leaves us about 14 inches to play with (I would really like to be able to ship the machine with its ATC installed).

Currently, our table's top stands 26.4" above the ground. If we add the vibration dampening feet, that's about 29". Ideally, we'd like it to be 7" higher at 36", which is the standard height for a workbench. Therefore, we could increase the height of our base from 300mm to 475mm (sorry, the machine is designed with metric units). And this would still leave 7" of headroom in the container. The height of a standard wood pallet is about 6.5". So, if we were to remove the vibration dampening feet when transporting the machine (probably a good idea), we would still have 3" of clearance in the container when putting the machine on a pallet.

Going for such a heavy base will give us all the rigidity and stability that we need there. And the 30" of travel on the vertical axis are plenty enough. Conclusion: we'll go for that.

Total weight: 6.2 metric tons.

Many thanks to skrubol for having pointed out that our base was too thin! This problem has been fixed now. Next!

***

High res pictures on the public drive:

https://drive.google.com/drive/u/3/f...3dQrx?ogsrc=32

[ishi]

Here is the machine with its Y-axis drivetrain with motors mounted at the bottom of the vertical column, with some recesses on the base casting in order to provide enough clearance for them while using standard length NSK HSS ball screws. Next: Y-axis counterbalance and linear encoders on all three linear axes.

[ishi]

Designing the vertical counterbalance for a relatively long 30" of vertical travel is turning out to be much more difficult than originally expected. This is due to the fact that a hydraulic cylinder with a stroke of 32" (2" of margin) is 39.5" long when retracted (71.5" extended). As a result, if mounted like on a Haas VMC, it goes almost all the way down our vertical column, with very little room left for extension. Even more problematic: we would need a very long arm (25" or so) to attach it to the vertical carriage.

https://diy.haascnc.com/procedures/v...ce-replacement

Clearly, this does not work.

And we can't really move it upward because it would increase the total machine height, while interfering with our ATC conveyor...

Argh!

I really can't think of any solution to that problem...

Can you?

UPDATE: If the hydraulic cylinder were to be mounted upside down going through the base, it might work. We'd have to move a few things around, but it might work... This would be funky though, which means that it would not pass our "if it does not look right, it's probably wrong" test...

[ishi]

Coming back to the Kitamura Mycenter-HX300iG:

https://www.kitamura-machinery.com/p...centerhx300ig/

Accuracy: ±0.000079” Full Stroke
Repeatability: ±0.000039”

Repeatability is 1 micron, while accuracy is 2. How do you do that knowing that the most precise linear encoders give you 2 micron accuracy only? How do you get repeatability twice as good as the accuracy of your linear encoder? Is that a simple matter of statistics (sometimes you go a bit too long, sometimes a bit too short)? Can anyone explain that?

In any case, this is my goal now: 2 micron accuracy and 1 micron repeatability.

And to do that, we'll switch to the Heidenhain LF 185, which is the only encoder I know that provides ±2 µm accuracy (everybody else does ±3 µm).

https://www.heidenhain.com/en_US/pro...-tools/lf-185/

Besides, I like their mounting bracket a lot better than the one used for the LC 495. Many screws alongside the encoder feels a lot better than a single screw at each end of the encoder...

UPDATE: A cursory review of all 3 linear axes seems to indicate that we won't have any problems mounting these larger encoders.

All is good in the neighborhood...

[ishi]

With our current set of constraints, 30" is as much as we're ever going to get on Y, and 20" is what we get on Z. But what about X? With our current design, we could get a whopping 50", but that would do us no good with a 30" long table, would it? So, how about we extend the table a bit? If we were to do so, we could get a 40" long table with 40" of travel on X, while keeping everything balanced. This would give us a part/cutting envelope of 40" × 30" × 20". These are pretty good looking numbers... Of course, that would degrade our Envelope Volumetric Efficiency by 33%, but it would increase our table area by 33% when using the machine in VMC mode. So, all in all, I think it's a reasonable trade-off.

Now, to do so while still using a standard NSK HSS ball screw, we would need to upgrade to the 1550mm model (1150 thread length), which means that our drivetrain length would go up to about 1,775mm, which is 25 mm longer than our current 1,750mm base width. But an intermodal container gives us 2,352mm, so we have plenty of room to play with. In such a context, we might go for something like 75" × 75", which would also help with the addition of our two hydraulic cylinders for the vertical counterbalance.

And that will be our primary task tomorrow...

[skrubol]

Designing the vertical counterbalance for a relatively long 30" of vertical travel is turning out to be much more difficult than originally expected. This is due to the fact that a hydraulic cylinder with a stroke of 32" (2" of margin) is 39.5" long when retracted (71.5" extended). As a result, if mounted like on a Haas VMC, it goes almost all the way down our vertical column, with very little room left for extension. Even more problematic: we would need a very long arm (25" or so) to attach it to the vertical carriage.

It's fairly common to use roller chain with a sprocket on the ram to get a 2:1 travel ratio (and need double the force on the rams.)

[ishi]

It's fairly common to use roller chain with a sprocket on the ram to get a 2:1 travel ratio (and need double the force on the rams.)

Of course! Clearly, I'm still not thinking like a mechanical engineer... And I have an aversion for chains and sprockets because they're messy and mechanically complex. I really need to find a way to get over this...

Thanks a lot skrubol!

[extent]

Repeatability is just what it says. If you command a 1.0000" move and the machine moves 3.00002" every single time you have easily satisfied your ±0.000039” repeatability spec (and have a horrible accuracy) If your encoder is only 1 count per inch, but you always stop between 1.000039" and 0.999961" you have a repeatability of ±0.000039” and an accuracy of ±1"