[ishi]

Here is the fourth iteration for the full machine.

The table's length was extended to 35". Pushing it to 40" as suggested last week was not possible. Therefore, its final dimensions are 35" × 20", and it currently stands at 35" above the ground without feet. Knowing that vibration dampening feet will add about 2.5" and that a typical height for a workbench is 36", we'll try to bring it down by 1.5" (that should not be a problem).

The linear encoders for the X and Y axes have yet to be added. The one for X won't be a problem, but we'll have to be creative for the one on Y. Several options are available for it, but none are as straightforward as what we can do for X and Z.

The saddle (Y axis carriage and rotary table) is very beefy now, and some weight reduction is probably in order, but I wanted to focus on dimensions and clearances first.

The height of the X-axis carriage was increased in order to provide more clearance underneath the rotary table's torque motor's stator for the drag chain that will be used for cables and hoses.

Very curious to read what you think if this latest iteration.

***

High-res pictures on public drive:

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

[handlewanker]

I don't believe it.....$4,000 for a mold to make 20 or 30 nylon parts for a prototype????...…...Batman, this thing is getting bigger than both of us.....I know Robin, it's truly unbelievable.

This machine is going to be a one off, a prototype, for your own use and as a guide line for others to make according to your designing expertise...…..without the nylon forks the ATC for you won't happen......and if the nylon fork subsequently requires a mold rework to make it work.....kiss your money goodbye.

By this admission you are now saying that you are going into production to make hundreds.... nay thousands of the nylon forks just to suit this particular ATC device.....I don't think any plastic molding company would set up a molding machine or consider an order for less than a couple of thousand parts.
Ian.

[ishi]

I don't believe it.....$4,000 for a mold to make 20 or 30 nylon parts for a prototype????...…...Batman, this thing is getting bigger than both of us.....I know Robin, it's truly unbelievable.

This machine is going to be a one off, a prototype, for your own use and as a guide line for others to make according to your designing expertise...…..without the nylon forks the ATC for you won't happen......and if the nylon fork subsequently requires a mold rework to make it work.....kiss your money goodbye.

By this admission you are now saying that you are going into production to make hundreds.... nay thousands of the nylon forks just to suit this particular ATC device.....I don't think any plastic molding company would set up a molding machine or consider an order for less than a couple of thousand parts.
Ian.

Batman,

A good-quality HSK 63F fork retails for about $50. I need 36 of these, so they will cost me $1,800. But if I make them myself, it will be $4,072. And they will be made by making a tiny little change to an existing mould design. Therefore, I'll break even on the mould after the second machine. And even if I don't, the extra $2,272 are totally worth the experience of making my very first injection molding part.

Also, don't worry, the injection molding company I work with is perfectly happy with an order of that size.

Not everything should be evaluated from a purely economical standpoint...

[ishi]

Now that all our components are assembled with proper joints, we can easily check for travels and clearances. For example, this view shows that we can't take full advantage of our Z-axis travel because of the hydraulic cylinders. These need to be lowered further into the base casting, or the column casting needs to be raised, by about 5.4". I expect that we'll find a handful of these problems, but I am reasonably confident that none will break the overall design. This week's focus will be to add as many components as possible (linear encoders, rotary encoder for the turn table, way covers, drag chains, etc.) in order to make sure that we have accounted for everything. Ideally, we'll find some time (and some CAD drawings) to add the chip collector as well. And after that, we'll give a first shot at the machine enclosure.

[ishi]

We've added an extrusion to our base casting in order to lift the machine with a heavy duty pallet jack like this one:

https://www.uline.com/BL_1920/Heavy-Duty-Pallet-Truck

The extrusion does not go all the way through. Instead, it stops at 1,250mm in order to prevent any interference with the recesses made for the hydraulic cylinder. Also, the back of the casting is thinner than the front, and we do not want to make it any thinner than it already is (270mm).

[ishi]

We just had a one-hour call with Siemens' top mechatronic engineer for CNC machine tools, and it was pure goodness!

To make a long story short, our basic design was pretty much validated in its entirety, with only a handful of minor suggestions for improvement. Our accuracy target was deemed aggressive but achievable, especially if we pay close attention to thermal control and chip evacuation. Based on our selection of linear encoders, we cannot go beyond 3µm accuracy, and 5µm is probably as good as we'll get, so we'll use that as a target for the time being. But only for the linear axes, because our spindle head's rotary axes are not accurate enough to get even close to that target (we already knew that). If we want a consistent level of accuracy on the rotary axes, we'll have to move to a much more accurate (but much larger) spindle head, going from 30 arcsec accuracy down to 2 arcsec, but going from 75kg up to 750kg... That will be for the machine's big sister...

From a purely mechanical standpoint, we need to beef up our Z-axis carriage by making it a bit thicker and by adding some support ribs on the back (quite easy). We also need to lighten up our Y-axis carriage by removing some material on the underside. This should be fairly straightforward as well. This will help increase the machine's natural frequency. With its current design, it was estimated to be in the 15 to 20Hz range. As any mechanical engineer knows, you want this number to be as high as possible, and you do that by increasing the machine's stiffness while decreasing the weight of its moving parts. The exact equation for this can be found there:

https://en.wikipedia.org/wiki/Fundam...anical_systems

And here is a great presentation on the topic:

http://www.mech.utah.edu/~me7960/lec...Structures.pdf

The use of two motors and ball screws on every linear axis was discussed, but the reasons we gave for this design were confirmed. This lead us to discuss the best way of controlling pairs of motors on linear axes. For larger machines, the gantry approach is the way to go, and it is best supported by using two linear encoders (scales) on every axis (one per ball screw). But for smaller machines like ours that tend to be very rigid, a much better alternative is available: it's called "Torque Master-Slave" or "Load Sharing". The basic idea is that one of the two motors is designated as the master, while the other is designated as the slave. The linear scale is installed as close as possible to the master. Equal acceleration orders are sent to both the master and the slave, but a closed loop is established by measuring torque on both the master and the slave and making sure that the torque differential remains as close to zero as possible, at all times. This form of control provides much better accuracy than the traditional gantry approach and is fully supported by the SINUMERIK platform.

We also talked about different strategies for "Compensation" that can be implemented at the software level, especially with respect to the squaring of linear encoders. These do not need to be precisely squared at a mechanical level. Instead, "beam sag" and "angularity error" compensations can be used. More details on this can be found there:

https://support.industry.siemens.com...5_en_en-US.pdf

In order to help with machine tuning, the Siemens engineer we talked to recommended that we use ballbar testing, as explained there:

Ballbar testing explained

This will require the use of a Renishaw QC20-W ballbar system, which we will bring onsite when the machine is installed:

QC20-W ballbar system

We also discussed various options for thermal management. The cooling of the castings was deemed to be a very good idea, and so was the injection of cold air underneath the way covers in order to control the temperature of the ball screws and linear rails. The use of water-cooled ball screw nut mounts was deemed overkill (as previously mentioned by two other readers on the forum) unless we can keep their price low (we think we can), and the benefits of water-cooled servo motors do not seem to justify the increase in cost (these are seen very rarely on any CNC machine).

In order to achieve the level of accuracy that we are targeting, the primary factor seems to be related to the evacuation of chips. This is especially true for a production machine. In the engineer's experience, most manufacturers that try to achieve this level of accuracy tend to dedicate a significant amount of their R&D resources on addressing this particular problem. This is due to the fact that chips tend to be hot and accumulating chips can radiate a lot of thermal energy toward the machine's base, thereby causing significant material expansion.

In order to deal with this issue, we can think of two complementary options:

1. We need to add the chip conveyor that we have been writing about last week.

2. We should consider adding a thin layer of thermal isolation material on the floor and walls of the base's middle canal, essentially shielding our chip conveyor. This would probably be quite cheap and could have a significant impact. The same material could also be used to line the bottom of our X-axis carriage.

One last thing was discussed: the impact of our hydraulic counterbalance on surface finish. Contrary to what we thought, the counterbalance will not improve accuracy. In fact, it will most likely reduce it. This is due to the fact that for the Z-axis to preserve its full potential accuracy, compensation must be different whether the Z-axis carriage is moving up or down. With the SINUMERIK platform, this was not possible until just two years ago, but directional compensation (my terminology) is now an option that can be purchased. It won't be easy to configure, but the engineer we talked to agreed to help us set the whole thing up in due time.

Overall, this was a really good call, and we now have a few modifications that we need to apply to the design.

[RCaffin]

My only comment is that the Y axis carriage seems to be a long way up in the air, with consequences for the length of the Z axis. I would be looking to lower both the X axis rails and the Y axis rails if I could.

Cheers
Roger

[ishi]

My only comment is that the Y axis carriage seems to be a long way up in the air, with consequences for the length of the Z axis. I would be looking to lower both the X axis rails and the Y axis rails if I could.

Cheers
Roger

Roger,

You're absolutely right. The top of the table is 1.5" too high right now, which we can save by lowering the height of the X-axis rails. We can't really do much for the Y-axis rails though, because we need some clearance within the X-axis carriage for the cable carriers. But we'll drop our hydraulic cylinders and raise our vertical column a bit, which will ensure that we get our full 25" of travel on Z.

[ishi]

Four swivel vibration-damping leveling mounts have been added:

https://www.mcmaster.com/#6221k83/=1dxyr4n

[ishi]

In order to lower our hydraulic cylinders without compromising the integrity of our base's mineral casting, we've decided to mount them on steel plates that will be bolted into the underside of our base. This is a bit funky, but it will lower our cylinders further than any alternative solution we can think of, while preserving the soundness of our casting. If we were to mount them directly onto the casting, the piece of casting onto which this mounting would occur would be very thin and made weaker by the presence of threaded inserts. Each cylinder will only support about 125kg, but we don't want to take any chances with our casting. Also, this arrangement will ensure that no fluid of any kind could be left sitting within the cylinders' recesses, because we do not have recesses anymore, we have through holes, and we can easily machine fluid evacuation holes in the two mounting plates.

[ishi]

A few height adjustments have been made:

1. Column raised by 2.5"
2. Cylinders lowered by close to 3"
3. Top of table lowered to 36" (feet included)
4. Bottom of chip collection canal lowered by 2"
5. Bottom of column raised by 2"

1. and 2. give us back our full 25" of travels alongside the Z axis.

4. and 5. together make the casting for the base a bit more symmetrical alongside the X axis. Good symmetry is critical for castings because it reduces the risk of warping. The base was already almost perfectly symmetrical alongside the Y axis, and now it's more balanced alongside the X axis. And once we replace our ugly flanges at the top of the vertical column by some metal brackets, the column casting will be almost perfectly symmetrical alongside both Y and Z axes.

With these changes, the table does not feel as "high up in the air" as before, and its top standing at 36" is right where it should be from an ergonomics standpoint. If we were to shrink things down any further, we could probably save one inch, but it would likely come at the expense of something else. Therefore, we'll stick to what we have for the time being.

Overall, when looking at the picture called "Profile", things look a lot more balanced than before. Many thanks to Robert for his encouragement to fix our height issues.

***

High-res pictures on public drive:

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

[ishi]

Here is a stiffer Z-Axis carriage. Its 63kg are in line with the 75kg of the spindle head it will move up and down. Eventually, we should be able to bring it below 50kg, but this should be good enough for now.

[ishi]

By raising our column, we've managed to get further clearance for our motors from the shoulder on which the back X-axis drivetrain is installed. In turn, this allowed us to bring the vertical column 40mm closer to the table alongside the Y axis, thereby extending the effective travel in VMC mode on the Y axis by about 1.5". Sweet!

[ishi]

Here is a very important test for the design: can the axis of rotation for the spindle go past the axis of rotation for the rotary table in VMC mode? We need the answer to be positive in order to do any kind of turning with the rotary table. And as we can see on the attached picture, it is positive indeed.

If you were with me in my garage, you could have heard a deep sigh of relief a few minutes ago...

[ishi]

As you can see on this orthographic projection, we have a pretty large space that is wasted in order to provide clearance for the two servo motors of the Y axis. Furthermore, this volume is situated on top of the front side of the front X-axis carriage, which is also wasted as a result. Together, these spaces occupy a volume of 1,900mm × 200mm × 300mm, which is 114,000,000mm³, or more than 0.1m³! Unfortunately, I have no idea how it could be reduced in any significant way, or what it could be used for.

Any suggestions?

UPDATE: This post surfaced a bug with the motors for the Y axis going beyond the machine's footprint. The bug has now been fixed, as evidenced by the second picture. No matter how good your parametric CAD model is, I don't think there is a way to fully prevent this type of bug without over-constraining the model and making it very inflexible...

[ishi]

With our latest design, the travels are:

- X: 1,125mm (44.3", but the table's length is only 35")
- Y: 525mm (20.7", but we'll call it 20" to be safe)
- Z: 650mm (25.6", but we''ll call it 25" to be safe)

We pretty much consider these to be final from now on.

***

Updated high-res pictures on public drive:

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

[handlewanker]

We've added an extrusion to our base casting in order to lift the machine with a heavy duty pallet jack like this one:

https://www.uline.com/BL_1920/Heavy-Duty-Pallet-Truck

The extrusion does not go all the way through. Instead, it stops at 1,250mm in order to prevent any interference with the recesses made for the hydraulic cylinder. Also, the back of the casting is thinner than the front, and we do not want to make it any thinner than it already is (270mm).

GASP...…..a pallet truck with a 5 1/2 ton capacity.... for this epoxy granite mill build????.....Batman, this thing is definitely going to be bigger than all of us…..I'm outta here before it starts to breathe fire...….last man standing turn out the lights..
Ian.

[RCaffin]

Eh, pallet truck or fork lift - whatever. Fork lifts to 50 ton are available.

Just don't drop the damn thing!

Cheers
Roger

[handlewanker]

Th limitations of a pallet truck are in the title of it...…..they are designed to lift a pallet....whatever.....100mm or so off the floor and then trundle it to another location also on the floor......they require a flat hard surface.....no compromise or exceptions on that score.

If the machine is being built up on a pallet or blocks ….on the floor.....then you can trundle it around…..for whatever purpose, but if it's on a table at bench height then plan B will have to apply and that means a crane, fork lift, chain block attached to the roof beams or a gantry.....whatever.

That means some point(s) on the machine MUST be capable of lifting the entire weight of the machine if a rope or chain lift solution is going to be used...….a fork lift is pretty agricultural.
Ian.

[ishi]

Th limitations of a pallet truck are in the title of it...…..they are designed to lift a pallet....whatever.....100mm or so off the floor and then trundle it to another location also on the floor......they require a flat hard surface.....no compromise or exceptions on that score.

If the machine is being built up on a pallet or blocks ….on the floor.....then you can trundle it around…..for whatever purpose, but if it's on a table at bench height then plan B will have to apply and that means a crane, fork lift, chain block attached to the roof beams or a gantry.....whatever.

That means some point(s) on the machine MUST be capable of lifting the entire weight of the machine if a rope or chain lift solution is going to be used...….a fork lift is pretty agricultural.
Ian.

The machine's weight will be in the 7 to 8 metric tons range, and its lifting will usually be done with eye bolts, but adding the space necessary for using a pallet jack could be useful in some instances, mostly when the machine needs to be moved around the shop.