6-axis Horizontal Machining Center for Education

[ishi]

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"

extent,

This makes perfect sense, but still: what gives you this repeatability, besides a sound mechanical design? How do you explain that repeatability is exactly half of accuracy? Is it just a matter of statistical average? In other words, if I have a 2 micron absolute linear encoder and a low-backlash mechanical design, should I reasonably expect a 1 micron repeatability?

[extent]

it isn't related to accuracy at all. You could have no linear encoder at all, and as long as the machine stops in the same place every time you hit the stop button you still have perfect repeatability. You can't even count on high accuracy to make up for poor repeatability. Just like if you had a rotary encoder on a servo it has no way of knowing how much backlash there is in the drive screw, super high CPR on the encoder doesn't make up for a lack of preload in the ballscrew. A linear encoder lets you see the actual movement of the table, so it can see that same backlash, but if moving the table over causes the column to shift you're back with the same problem. Your super high resolution linear encoder has seen the movement of the table, and the backlash in the ballscrew, but can't see the movement of the column

[ishi]

it isn't related to accuracy at all. You could have no linear encoder at all, and as long as the machine stops in the same place every time you hit the stop button you still have perfect repeatability. You can't even count on high accuracy to make up for poor repeatability. Just like if you had a rotary encoder on a servo it has no way of knowing how much backlash there is in the drive screw, super high CPR on the encoder doesn't make up for a lack of preload in the ballscrew. A linear encoder lets you see the actual movement of the table, so it can see that same backlash, but if moving the table over causes the column to shift you're back with the same problem. Your super high resolution linear encoder has seen the movement of the table, and the backlash in the ballscrew, but can't see the movement of the column

extent,

Okay, that makes sense, but the fact that the Kitamura's repeatability is exactly half its full stroke accuracy still feels like an odd coincidence to me. And I still do not understand how they can achieve 1 micron repeatability, especially with the super high 2,362ipm feed rates they claim. With our 5mm pitch ball screw and 3,000rpm servos, we'll get 709ipm. As far as I can tell, 5mm pitch is the lowest you can get on a standard ball screw, and this gives you the highest precision. I understand that accuracy is mostly a matter of pre-loading, but still. They're getting 1 micron repeatability with 4 times faster feeds. How do they manage to get that?

[handlewanker]

One micron???…….anybody ever hear of expansion and contraction due to thermal differences?

No matter how accurate or preloaded your screws are they will expand and contract and the longer they are the more they will re-act......this also does not take into consideration the thickness or viscosity of an oil film...…...one micron has to be tested to be seen and if you aren't getting one micron several months down the track who will ever know......and if your workplace is not at a constant 68 deg F 24/7....forget the one micron bit.

If your work piece is not to tolerance then the common solution is to blame the tool due to deflection either in the machine or the tool itself......no two tool suppliers agree on the best tool for the job and speeds and feeds affect any tool parameter.

Did I heart someone say a cleaning pass.....then you do have tool or machine deflection.
Ian.

[mactec54]

extent,

Okay, that makes sense, but the fact that the Kitamura's repeatability is exactly half its full stroke accuracy still feels like an odd coincidence to me. And I still do not understand how they can achieve 1 micron repeatability, especially with the super high 2,362ipm feed rates they claim. With our 5mm pitch ball screw and 3,000rpm servos, we'll get 709ipm. As far as I can tell, 5mm pitch is the lowest you can get on a standard ball screw, and this gives you the highest precision. I understand that accuracy is mostly a matter of pre-loading, but still. They're getting 1 micron repeatability with 4 times faster feeds. How do they manage to get that?

The Servos they are using most likely have 24Bit Encoders this is where the difference is, 1 micron it's just a number, and all conditions have to be meet to achieve this

This is why they can get there high accuracy, is how they build them, and they will be using dual feed back to the control with Linear Encoders and servo Motor Encoder, there construction is also very good they have had years of experience to achieve this, they are using box ways on the X and Y Axes, and dual contact spindle, and linear rails only on the Z axes



Affordable, reliable, durable - Kitamura’s Mycenter-3XD is the newest addition to Kitamura’s VMC line up, designed with space saving flexibility and ease of use for the operator in mind. Truly compact the Mycenter-3XD harnesses an ample table size of 500 x 800mm (19.7” x 33.9”) along with 760 x 510 x 510mm (30” x 20.1” x 20.1”) X,Y,Z travels, making it ideal for small to medium sized workpieces. High speed rapid feed rates of 48m/min (1,889ipm) X&Y, 42m/min (1,654ipm) Z on heavy duty cross roller linear guide ways combined with stiff cast construction offer you the benefits and reliability of Kitamura quality features and components throughout.

Mycenter-3XD Features
Meehanite cast C-frame construction ensures unparalleled rigidity durability and stability. All contact surfaces are hand scraped for precise alignment, fit and performance ensuring high quality and attention to detail in the manufacturing process. Heavy duty cross roller linear guide ways offer the strength necessary to carry heavier loads and power through heavier cuts.

Mycenter-HX500G Features
Manufactured in Japan with hand scraping production techniques, the Mycenter-HX500G offers up a host of Kitamura stand-out features such as solid box way construction with blazing 2,362ipm rapid feedrates, twin ballscrews and motors in the X & Y axes and linear scale feedback on all axes for optimum stability and ultra-high accuracies of ±0.000079” full stroke, ±0.000039” repeatability.

[ishi]

One micron???…….anybody ever hear of expansion and contraction due to thermal differences?

No matter how accurate or preloaded your screws are they will expand and contract and the longer they are the more they will re-act......this also does not take into consideration the thickness or viscosity of an oil film...…...one micron has to be tested to be seen and if you aren't getting one micron several months down the track who will ever know......and if your workplace is not at a constant 68 deg F 24/7....forget the one micron bit.

If your work piece is not to tolerance then the common solution is to blame the tool due to deflection either in the machine or the tool itself......no two tool suppliers agree on the best tool for the job and speeds and feeds affect any tool parameter.

Did I heart someone say a cleaning pass.....then you do have tool or machine deflection.
Ian.

Ian,

1 µm = 0.000039”

This is the repeatability that Kitamura is claiming:

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

They use ball screw cooling though.

[ishi]

The Servos they are using most likely have 24Bit Encoders this is where the difference is, 1 micron it's just a number, and all conditions have to be meet to achieve this

This is why they can get there high accuracy, is how they build them, and they will be using dual feed back to the control with Linear Encoders and servo Motor Encoder, there construction is also very good they have had years of experience to achieve this, they are using box ways on the X and Y Axes, and dual contact spindle, and linear rails only on the Z axes



Affordable, reliable, durable - Kitamura’s Mycenter-3XD is the newest addition to Kitamura’s VMC line up, designed with space saving flexibility and ease of use for the operator in mind. Truly compact the Mycenter-3XD harnesses an ample table size of 500 x 800mm (19.7” x 33.9”) along with 760 x 510 x 510mm (30” x 20.1” x 20.1”) X,Y,Z travels, making it ideal for small to medium sized workpieces. High speed rapid feed rates of 48m/min (1,889ipm) X&Y, 42m/min (1,654ipm) Z on heavy duty cross roller linear guide ways combined with stiff cast construction offer you the benefits and reliability of Kitamura quality features and components throughout.

Mycenter-3XD Features
Meehanite cast C-frame construction ensures unparalleled rigidity durability and stability. All contact surfaces are hand scraped for precise alignment, fit and performance ensuring high quality and attention to detail in the manufacturing process. Heavy duty cross roller linear guide ways offer the strength necessary to carry heavier loads and power through heavier cuts.

Mycenter-HX500G Features
Manufactured in Japan with hand scraping production techniques, the Mycenter-HX500G offers up a host of Kitamura stand-out features such as solid box way construction with blazing 2,362ipm rapid feedrates, twin ballscrews and motors in the X & Y axes and linear scale feedback on all axes for optimum stability and ultra-high accuracies of ±0.000079” full stroke, ±0.000039” repeatability.

mactec54,

That makes sense. Well, we certainly don't have the experience, and we won't be able to do box ways, but we do use 24-bit encoders on the servos, we do use twin ball screws on every linear axis, and we could add water cooling for the ball screws. Should we add the water cooling?

[ishi]

I have a question regarding the vertical counterbalance and the ball screws used for the vertical axis on the vertical column: does adding a counterbalance improve accuracy and reduce wear on the ball screws and ball screw nuts? Here is what makes me think that it might:

Unlike the X and Z axes, the vertical Y axis has to fight against gravity, which means that more energy has to be transferred from the ball screws to the ball screw nuts in order to move the Y-axis carriage (150kg) up and down. This extra energy leads to extra friction, which itself leads to extra heat dissipation and wear on parts (ball screws and ball screw nuts). In turn, this extra heat dissipation leads to lower accuracy because of thermal expansion, unless the ball screws are equipped with water cooling. And the wear on parts leads to increased backlash over time, which leads to an additional loss of accuracy.

Bottomline: the vertical counterbalance might have benefits beyond safety and is worth considering seriously.

[ishi]

I just found our that NSK offers the option of cooling ball screw nuts:

https://www.nskamericas.com/ball-scr...oling-2336.htm

Up until now, I thought that the only option was to cool the ball screw itself, which is really complicated, because you need sliding seals or rotary joints. Since we already need water cooling for the spindle, adding it for ball screw nuts does not seem like a very complex addition. And because our spindle isn't that powerful, our existing water chiller might be enough for both the spindle and the ball screw nuts.

One problem with that option is that the smallest shaft diameter available for cooled nuts (HMD series) is 40mm, instead of the 32mm that we were planning to use. And the smallest pitch for it is 16mm instead of 5mm. That would reduce precision but it would increase rapids from 591ipm to 1,890ipm, and I don't think that it would reduce accuracy. Also, these ball screws do not seem to be standard products. I'll get in touch with NSK and try to get more information from them.

[mactec54]

mactec54,

That makes sense. Well, we certainly don't have the experience, and we won't be able to do box ways, but we do use 24-bit encoders on the servos, we do use twin ball screws on every linear axis, and we could add water cooling for the ball screws. Should we add the water cooling?

For what you are building, you are already way over the top for a first build, adding coolant circulation for the Ballscrews is a none starter for me, and would be a waste on an unknown, not all machines need this, as your rotary and Z axes Axis's will be doing most of the work, X and Y axes won't work hard enough to heat them up to cause any accuracy problems

[ishi]

For what you are building, you are already way over the top for a first build, adding coolant circulation for the Ballscrews is a none starter for me, and would be a waste on an unknown, not all machines need this, as your rotary and Z axes Axis's will be doing most of the work, X and Y axes won't work hard enough to heat them up to cause any accuracy problems

mactec54,

You are probably right. But we could probably still design the machine with HMD ball screws without cooled nuts and upgrade if we establish a need for it. You can get HMD nuts with or without cooling, without having to change the ball screw itself.

[ishi]

I am in the process of reviewing all our dimensions, and I still can't figure out how much vertical travel we need. I used to think that more is always better, but I'm not so sure anymore. What really matters is a balanced design. With a 20" diameter rotary table, the 30" of vertical travel that we currently have seems too much. If I compare our envelope to the one of the DMG MORI NMV5000 DCG, they only offer 20" of travel on the vertical axis for a similar rotary table. And if we reduce our travel on the vertical axis, we can improve accuracy (the shorter the ball screw, the better), while installing a vertical counterbalance without any problem. So, should we go for 25", or even just 20"?

UPDATE: 25" seems like a good middle ground, because going lower than that would bring the effective travel in VMC mode too low (it's about 5" less than the full travel we get in HMC mode). Also, 25" would allow us to use a hydraulic cylinder that is about 840mm, which is just about perfect, because it could be attached directly to the Y axis carriage without any extension bracket.

Conclusion: 30" on X, 25" on Y, and 20" on Z.

[ishi]

Now that we have decided to switch to a mineral casting, it is time to start designing a new version all over again. This will be our third, and we will start with the tool changer. When we talked with Gate TI, we learned that tool indexing is much more precise when one pallet at each end of the conveyor is perfectly perpendicular to the conveyor's longitudinal axis. This is due to the fact that the conveyor uses a cam system, and there is a lot less backlash when indexing is done through incremental steps instead of continuous motion.

This creates an interesting problem though: when you have a 30 pallet conveyor, you have 3 pallets on each end and 12 pallets on each side of the conveyor's longitudinal axis. Because 12 is divisible by 2, there is no pallet in the middle, perfectly aligned with the axis of our spindle. Therefore, in order to make things work, we would need to offset the conveyor by half of 105mm (the offset between each pair of contiguous pallets), or that plus a multiple of 105mm. This creates all kinds of problems, some related to the mounting of the conveyor on top of the vertical column, and others purely cosmetic.

In order to solve this problem, we will move from the TL105 to the TL100:

TL Series

This solves the problem, because this conveyor comes with four types of modules instead of three:

- Head (5 pallets)
- Body with 4 pallets
- Body with 6 pallets
- Foot (9 pallets)

As a result, if we use 3 bodies with 6 pallets, we get a total of 32 pallets (32 slots for tools), with 13 pallets on each side of the conveyor's longitudinal axis. 13 is not divisible by 2, therefore we'll have a pallet perfectly aligned with the spindle's axis once the conveyor is centered on the vertical column.

In doing so, we lose 5mm of length on every pallet, but this should not be a problem with HSK F63 tool holders. Also, we save about 5% of longitudinal space, which means that moving from 30 to 32 tools does not increase the total length of the conveyor by much. Most importantly, the entire ATC fully populated with tools will remain within the machine's footprint.

Last but not least: now that we are calling our machine an Omnidirectional Machining Center, we will label the vertical axis Z again. This will make things a bit less confusing for people who are not used to the labeling of axes on Horizontal Machining Centers.

[ishi]

In our previous design, we added flanges to our mineral casting in order to support our ATC conveyor. This was a poor design, because it made the casting a lot more complex than it should be, while a pair of standard Misumi precision angle plates would be perfectly suited to the task. And at $219.38 for the pair, they'll be a lot cheaper than the extra complexity added to our mold. Also, they will make installation a lot easier. In other words, it's a better design on all fronts.

https://us.misumi-ec.com/vona2/detai...7d&Tab=preview

[Eldon_Joh]

some ball screws are gun drilled and the cooling is through the middle. this is necessary when the ball screw is tensioned with fixed-fixed bearings on both ends, otherwise it would heat up and tension would be reduced or even lost.

I don't see a problem with epoxying the ballscrew nuts into an aluminum tube with a helical channel cut in the aluminum for the machine coolant to passively cool them.

[ishi]

some ball screws are gun drilled and the cooling is through the middle. this is necessary when the ball screw is tensioned with fixed-fixed bearings on both ends, otherwise it would heat up and tension would be reduced or even lost.

I don't see a problem with epoxying the ballscrew nuts into an aluminum tube with a helical channel cut in the aluminum for the machine coolant to passively cool them.

Indeed, but we'll go for fixed/supported. And flowing coolant through the ball screw is much more complicated and expensive. I think nut cooling is the way to go, but it's not even clear that it would benefit us: we probably have many more problems to solve before this one could make a difference.

[ishi]

As I look for solutions to various mechanical problems, I keep noticing a recurring pattern: once I find a working solution, it usually ends up looking like something that can be found on an existing machine. And this should not come as a surprise: most (all) of the problems that we are trying to solve have already been solved before, and the solutions can be found by simply looking at existing machines.

Tonight, the problems that I was trying to solve were to get more spacing between the ball screws for the X axis, find a proper way to mount the shoulder plates for our rails, and make it easier to mount the two servo motors for the Z axis (the one alongside the vertical column). All three problems could be solved with a single change to the design: raising the rails for the X axis by adding some ribs in the mineral casting, then embed a couple of steel plates at the top of these ribs in order to properly machine the shoulders and threaded holes for installing our rails with shoulder plates, without being limited by the fact that threaded inserts on mineral castings can't be close to an edge.

As a result, we can move our vertical column about 5" down, which means that we have more space for mounting our two servo motors for the Z axis at the bottom of it, without having to make recesses for them in the base casting (that was way too funky and would have been a nightmare during installation and maintenance). Also, this new configuration allows us to move the servo motors for the X axis on the outside of the rails, thereby improving rigidity and squareness. And we found a way to move the two servo motors for the Y axis within the envelope of the Y-axis carriage, which will allow us to reduce the depth of the machine or to get more space for our electrical cabinets on the back of the vertical column, while improving the machine's ergonomics by bringing the table 5" to 7" closer to the operator.

Once again, the lesson is this: if it does not look right, it's probably wrong. Keep hammering on it until it looks right!

Okay, enough for today...

[handlewanker]

This build is going above and beyond the realms of credibility....too many variables...…….one micron?....try .02mm at best.

If Kitamura have claimed to get repeatability of that figure it's because they've been in the game for years to achieve that claim.....you are not even near to getting anything off the ground this century.

I did say you.....this is regarding the one off prototype build you are anticipating as any production models would no doubt be built by a fully competent workplace that had the equipment and expertise to get the specs right.

Just cooling a ball nut does not achieve anything...…..cooling the ball screw itself would if you have a problem, but pushing a machine so hard to overheat the ball screws does not give you room to achieve 1 micron accuracy.

You forget that the ambient temperature all year round fluctuates and that will make any dissimilar metal on your build also fluctuate differentially one to another...…..this will also affect the epoxy granite mix as it's not a good heat resistant material.

BTW.....are you also going to temperature control the linear ways...…..being tied down at 60mm intervals will make them exert pressures longitudinally like a railway track in the sun.

I realise you have high ambitions and expectations, even if not immediately achievable, given the very fullness of time no doubt you will get to where you are thinking of, so with my very best wishes.....go forth where not many men have been before and do your best but at the same time accept your failures too.

BTW....I like the idea of the wood panel mock up to see if the design is actually feasible...….my exact mode d'employ…..though I don't aspire to the notion that if it looks good it is good......total BS......if it looks good it's a toy.
Ian.

[RCaffin]

A thought for you if you are going with a cast base and frame:
Include some copper tubing for cooling water for temperature stabilisation.
You will need water cooling for the spindle anyhow, so a good INDUSTRIAL pump will be needed. Do NOT buy a domestic water pump: they don't last.

Cheers
Roger

[jkkmobile]

There are no way covers? If you are locking the final travel now you need to add room for way covers specially between bed and upright colum..