Large Epoxy Granite Vertical CNC Mill

[petruscad]

The X and Y axes in a mill need to be adjusted perpendicular to each other. In this case, since I use linear rails, it basically means that the linear rails need to be perpendicular to each other. The Y rails are already bolted to the base which means I need to adjust the saddle such that the X-rails bolted to it are perpendicular to the Y-rails. The only easy way of doing that that I could come up with was to copy the perpendicularity of a master square of known accuracy. I don't have such a master square so... I set to make one!

A square can be thought of as two parallels held fixed at 90 deg from each other. I came up with this clamp that does just that:



Here's how it works. The clamp has two 1/2" wide channels machined into it to locate the parallels. The parallels are held them down magnetically: you can see the four neodymium magnets on the sides. Each parallel is supported on a line and a point making it completely stable. The lines (which need to be parallel to each other) are provided by pressure fitting a precise hardened steel pin through the clamp. The two points are provided by the tips of two screws. Here's how the parallels fit into the clamp. Note that the parallels do not touch the magnets, which are a fraction of a millimeter lower that the supporting pin and screw tip. They are close enough however to provide quite a bit of clamping force.



The screws can be twisted to adjust the angle between the parallels. You can see the adjustment screw holes in this picture.



After squaring the square is mounter to a vise to so it can be secured to the base for the squaring procedure of the mill (I will post on that later).



Now, the all important question: how do you adjust this square to precisely 90 degrees? This is a topic in itself but basically the same technique is used as for testing a master square for squareness. The technique requires that the arms of the square be straight and that the two sides of each arm be parallel to each other (that's why I use two new 6" parallels for this). A surface plate, a 0.0001" dial indicator and a simple jig is also needed. Rather that me explaining how this is done you can just watch this video which is how I learned about it!



I can easily adjust this square to an accuracy of 0.0001"-0.0002" per 6" which is good enough for squaring the mill. The square is also very repeatable. After squaring I can remove the arms and place them back in a different orientation and the square pretty much retains its precision. When I get my hands of some 12" long parallels I can improve the precision even further. One idea which I want to try is to use pieces of linear rail as parallels but I have to check what kind of straightness do they have over 12".

With the master square in hand we can now square the mill...

[dwalsh62]

On the precision mold topic.

Sure, making a very precise mold would be nice. The problem with it is that its difficult, slow and expensive. And only a few surfaces need precision anyway. A simpler way of making precise bases used in industry is replication. They basically cast the base with a cheap, imprecise mold and then add precision to the few surfaces where it's needed: to bolt linear rails for example. But they do not do that by grinding the granite the way I do it here but by molding another layer of epoxy to the base in a precise way (Note though, that grinding of EG bases after casting was used in industry in the 60s and 70s). The way they create precision is by applying a thin layer of epoxy-based paste (e.g. moglice) to the place where precision is needed and then lowering a straight edge or surface plate on top of it in a very precise way. Injecting liquid filler between the rough base and the precise surface is another way. After curing the reference plate/edge is removed revealing a precise surface behind (this is why its called replication). Cincinnati Milacron used this technique on some of their their grinders for example. Of course this still requires a precise reference surface, which is expensive but at least you need a smaller one than a whole mold. In any case for this build I will stick with grinding rather than switch to the replication method because I know how to do it now, it seems to work and I only need to grind a few surfaces.

I'm only talking about one surface with precision in a base mold and two surfaces with precision in the column mold, why do more work than necessary.

[DICKEYBIRD]

I think its a 3L280 belt and it does work pretty well. Not much force is needed so even a 2L belt will work. With the router motor though, make sure you do not exceed the RPM rating of the abrasive wheel (by much!).

Thanks. I was thinking I'd make it with 2 pulley ratios stacked so I'd have 2 speed choices.

Funny, the lower priced 4" CBN wheels I've been looking at list the max at 4700 rpm even though they're just an aluminum disc with a skinny outer band of the CBN stuff. You'd think they would be safe at high speeds but apparently not. It's gonna take a big spindle pulley to tame my 30K trim router down to 4700!

[petruscad]

The alignment of the X and Y axes for perpendicularity was done like this. The 'master' square is secured to the Y axis rails and adjusted so that one of the arms is parallel to the Y-axis. This is done by securing a dial indicator to the saddle and moving it back and forth until the needle stays in one place! Then, a dial indicator secured to one of the X-axis carriages is moved along the other edge of the 'master' square and the saddle is adjusted until the needle of this second dial indicator does not move either. And that's it, its squared! In practice of course its a little harder and more tedious than this and the whole procedure takes several hours to complete, but you get the idea. Here are some photos of the setup. It looks intimidating, doesn't it?







The vise with the 'master' square attached to it is mounted on a large steel straight edge which is clamped to the Y rails. On the saddle the fixed dial indicator (the large one) is clamped to the X-axis rails. This is the one used to make the square parallel to the Y-rails. The little dial test indicator is mounted in a surface gage and rides with an X-axis carriage. This is the one used to slightly rotate the saddle until it becomes perpendicular to the Y-axis. The 'slight' adjustment is done by tapping with the brass weight you can see there. The 16 bolts that clamp the saddle to the Y-axis carriages are slowly tightened up and the squareness checked again and again until they are completely tight.

[phil916]

Hi,

wonderfull construction, any news please ?

[Bongodrummer]

Have been silently following along here. Awesome project! Need more please, am hooked

[mcphill]

Your audience is still interested, if you have made an progress, please post up!

[CaptainVee]

and using a polymer based binding agent (epoxy is just a name brand for a polyester based binding agent)

As far as using epoxy goes, West Shore Marine sells some stuff that isn't polyester based

I may have missed some of the meaning in context, as I sometimes do as a non native English speaker but over here (EU), Polyester and epoxy are very different beasts and by no means can epoxy be called a name brand for polyester.

Polyester is the resin that has been used for pleasure craft since the 1950s in what is generally called fiberglass.
What may cause some or all of the confusion is that epoxy can be used just the same way.

Polyester, as used in fiberglass, is a resin that over time will change from a liquid state in to a solid.
To speed this up a catalyst is added. The more catalyst is used, the shorter the curing time.

Epoxy is a resin that (for out purposes) comes as two components that need to be mixed as accurately as possible. If the ratio is off, one of the ingredients will not be used completely during curing and the end result will be weaker than when the ratio is correct.

Polyester does not bond very well to quite a few materials, but epoxy bonds very well to a lot of materials.

For building quartz concrete, epoxy is the only option, unless you want to be very adventurous...

[ebrewste]

I may have missed ... as a non native English speaker but over here (EU), Polyester and epoxy are very different beasts.

You are correct. Any issues you have with English does not make polyester into epoxy

[petruscad]

Yeah, I did not post for a long time, did I? I've been very busy lately with... moving to a new place.

The column mold is completely finished now with all bells and whistles required (well, it was actually completed before the move). Here's a picture of the bottom side of the mold being glued together.



The holes in the bottom (there are six of them) are where the bolts mounting the column to the base will go. You can see I'm trying hard to make all angles be 90 degrees! And here's the mold almost finished.





You can see the shiny tubes that are going to create the column to base mounting holes. I also added thick metal plates on the top side to distribute the pressure so the EG does not crack while tightening. Here's another view of that.



There are two rebar assemblies that will be embedded in the column and the rebars also hold the bolts to which the Z-axis linear rails will be mounted to as you can see from this picture.



I will post on the detail of the rebar and the mold in later posts. I hope this quenches you curiosity for the time being...

[TOTALLYRC]

Looking good.
In all my spare time I will have to do this as well!!!!!!!!!!!!!!

Mike

[RotarySMP]

Cool build. Like the way you used the rebar to locate the inserts during the casting process. What will it weigh? 800lb+ Column only.

Mark

[mcphill]

I hope this quenches you curiosity for the time being...

Are you kidding me?!?! Now I am thirstier than ever!!!

Impressive attention to detail and thanks for the update. Going to be a great machine!

[petruscad]

I fussed a lot about the best practical way to put the re-bar assembly together. The Z-axis linear rails, like the X and Y ones, are going to be bolted into metal inserts placed inside the EG. The inserts are just large plain bolts. The X and Y axes rails are pushed down by the weight of the saddle and table, towards the base. However, for the Z-axis, the ~100kg weight of the head assembly will be pulling the rails away from the column. Due to this the inserts need to be anchored in a way that distributes the load to the entire column, not just to the surface. The solution I came up with is to attach each insert bolt to a piece of re-bar that spans the whole depth of the column. This has the added benefit that the insert bolts can be suspended at the right distance under the surface of the column (~10mm). The fuss was about how to attach the bolts to the re-bar and the re-bars to each other. I really wanted to weld them so I considered buying a welder but then re-bar is not particularly weldable either. So I went with the thread way. The bolts were drilled and tapped like so.



And the 1/2" re-bars were turned and threaded like so.



And then they get tightened together... really tightened!

And here is the jig used to hold all pieces in place for soldering and tying with wire.



Yes, soldering. It works pretty well if the scales on the re-bar are removed and bare metal exposed. Its not very strong but together with double wire tied at each joint it should work fine. The longitudinal re-bars span the entire length of the column and they are 5/8" in diameter; there are four of them. I think that should do to reduce column bending under the weight of the head and torsion due to lateral cutting forces.

Anyway, the column is not cast yet so any suggestions on how to make it more rigid are welcome.

[louieatienza]

I fussed a lot about the best practical way to put the re-bar assembly together. The Z-axis linear rails, like the X and Y ones, are going to be bolted into metal inserts placed inside the EG. The inserts are just large plain bolts. The X and Y axes rails are pushed down by the weight of the saddle and table, towards the base. However, for the Z-axis, the ~100kg weight of the head assembly will be pulling the rails away from the column. Due to this the inserts need to be anchored in a way that distributes the load to the entire column, not just to the surface. The solution I came up with is to attach each insert bolt to a piece of re-bar that spans the whole depth of the column. This has the added benefit that the insert bolts can be suspended at the right distance under the surface of the column (~10mm). The fuss was about how to attach the bolts to the re-bar and the re-bars to each other. I really wanted to weld them so I considered buying a welder but then re-bar is not particularly weldable either. So I went with the thread way. The bolts were drilled and tapped like so.



And the 1/2" re-bars were turned and threaded like so.



And then they get tightened together... really tightened!

And here is the jig used to hold all pieces in place for soldering and tying with wire.



Yes, soldering. It works pretty well if the scales on the re-bar are removed and bare metal exposed. Its not very strong but together with double wire tied at each joint it should work fine. The longitudinal re-bars span the entire length of the column and they are 5/8" in diameter; there are four of them. I think that should do to reduce column bending under the weight of the head and torsion due to lateral cutting forces.

Anyway, the column is not cast yet so any suggestions on how to make it more rigid are welcome.

Well to me what comes to mind would be a steel bar with holes drilled for the bolts, which would be threaded to the bar. Then in between there would be larger holes for the rebar. You can braze or weld those on. Or even peen the ends using a simple jig with a hole that fits the rebar and a bandsawn slot, and use a clamp to pinch the rebar while peening the end.

[RotarySMP]

That is a pretty deep section. The stiffness of the E/G alone should easy react the forces with minimal deflection. Internal steel load paths are not normally used for E/G machine structures, just use sufficifient cross section.
Mark

[Hezz]

I must admit. Machining the rebar to get it to properly align and hold together was a stroke of brilliance.

[jhovel]

I really think you should incorporate counterweights for your spindle head assembly! That might mean casting in some mounting points for pulley attachments. 100kg is a LOT to have hanging from the Z axis rails.
I have an old horizontal mill and in my case the knee was putting too much strss on the z-axis. The difference my counterweights made is amazing - and only less than a third of the knee+saddle+table weight. I have machined some heavy workpieces (to be added the the load on the Z-axis!) and it is much easier on all moving components.
Something like that or an even bigger proportion will reduce the 'pulling out' force of your head significantly - without affecting the 'backlash' assistance from gravity.
Cheers,
Joe

[RotarySMP]

Here is a VMC being build by some guys in Germany who made a business out of E/G framed machines. The proportions look very similar to yours.

Peters CNCECKE

As you can see, you have no need for internal steel load paths. Just making sure your anchors are well attached (you have) is enough. The steel square tubes you can see are only there as core forms to create voids in the design, and some foot attachment points mount off them.
Peters CNCECKE

Your long rebar's wired to the individual insert location rebars, are going to do about nothing in the finished casting. Even if they would pick up loads, they are not attached to the interface brackets which will be reacting those loads.

I didn't understand the plan with the inserts. In photo 5.jpg above, they look like they are above the surface of the mold, and will therefore be completely buried in E/G? Normally, individual inserts for mounting small bits are simply attached with a screw through from the outside of the mold, so they won't move during casting. For a row of mounting holes, rather than trying to precisely locate a whole row of inserts, rigidly enough to stay in place during the violent vibrating required when casting on a large E/G mass like this, it is easier to simply mold in strips of iron, to be milled/ground flat after casting, and then tap the holes into the iron. The german guys do attach a row of bolts to the back of the steel strips to increase their tooth into the E/G.

Mark

[LeeWay]

Not to be too dumb here, but wouldn't all thread have been just as good?
I have seen it used in anchors for concrete before.