[ger21]

About 12 years ago, after I finally got my first CNC running, I started designing a new one.
I had a couple main goals:
1) Dual spindles, with independent Z axis.
2) Very high performance. (fast and strong)
3) Constructed mostly from wood.


Over the years, I've made a lot of small parts (prototypes?), and the design has been constantly evolving. I've already thrown away two complete Z axis, and will son get rid of all the parts for the Y carriage that carried them, as I'm now on the 3rd version.


As of now, here's a list of most components.

X and Y axis linear rails, Hiwin 20mm. 3000mm X axis, 1700mm Y axis.
Z axis linear rails, 2 sets of Hiwin 15mm, I think 300mm long (They're in a box somewhere.)

The two Z axis will be using 570oz Nema 23 steppers, with Leadshine AM882 drives, and an Antek 60V toroidal power supply. These will drive 1/2-8 4 start acme leadscrews with custom anti backlash plastic nuts. In order to keep the two Z axis as compact as possible, there's no room for ballscrews.

X and Y axis will be DMM 400w AC servos, driving helical rack and pinions through 10:1 gearboxes. Dual drive X axis.


Eventually, there will be a rotary 4th axis (axis #6, actually) with another 400w servo, through a 21:1 zero backlash gearbox (I have two of these gearboxes).

Control will be UCCNC, with their UC300ETH controller, plugged into a UB1 breakout board with UD1 daughterboard.

For now, I have two air cooled 2.2Kw round spindles, and a Huanyang VFD. I'm using one VFD for both spindles, using relays and contactors to switch spindles. They may be upgraded to something larger in the future.


Here are a couple Fusion 360 renders to get started. Not that the Y carriage will be completely redesigned from what is shown.

 

[ger21]

This might take a while.....

Today I cut and assembled the first major part. A baltic birch torsion box. It's roughly 42-1/2" x 95-1/2", and about 7" thick.

All parts were cut on a 5x12 CNC router, and the box was assembled on the machines vacuum table, with the vacuum holding it down flat, upside down.
The result is a perfectly square, perfectly flat box, that will form the base of the machine.

Wooden legs will be bolted to the bottom. The X axis linear rails will mount to two 10'ong plywood rails. Each rail is made up of 3 layers of 3/4" plywood, which were laminated together on the same vacuum table, to insure flatness. More on those later.

 

[difalkner]

Can't wait to see this progress through the steps, Gerry. Why the two spindles? Is one for drilling and one for machining? Or is it to keep from having to go add an ATC?

David

 

[RainMan 2.0]

Very impressive. You’ve been talking about this for a while,so I’ll bet your excited to get the ball rolling .
Great having access to a cnc to build a cnc , especially getting those tight tolerances. The torsion box looks awesome , and I’m sure liking how the top and bottom has dado’d lines to fit .
Looking forward to seeing the progress. Looks fantastic so far Gerry

 

[ger21]

Two spindles = fewer tool changes, and a lot more speed if you are only using two tools.
It probably cost me about $750 for the second Z axis, including the second spindle. An ATC would be $2000-$3000 minimum, for a chinese water cooled ATC spindle and toolholders.
And if a chinese ATC goes bad, it's another $2000 to replace it. It's $200 to replace the air cooled spindle.


I've already purchased most of the components for this machine. The main remaining items are two servos and drives, and I think two more lengths of gear rack.

 

[difalkner]

So if you're going from one spindle to the next I assume you know exactly the distance between the two so changing the X axis zero can be handled in your controller software without having to re-zero X manually. Is that how you'll do that? I like it!

David

 

[BalloonEngineer]

I’ve followed your discussions about this machine for the last several years over on CNCzone, hope all that planning leads to a much quicker build! Let’s see if you or Rick finishes building first!

Rick - goes to show that planning eventually can lead to building! One more excuse gone.

 

[UglySign]

This needs to be a Sticky - Popcorn ready

 

[ger21]

So if you're going from one spindle to the next I assume you know exactly the distance between the two so changing the X axis zero can be handled in your controller software without having to re-zero X manually. Is that how you'll do that? I like it!

David

They are supposed to be 5.44375" apart, but I'll need to do some measuring at some point. :grin:
That distance came from an earlier design, where the Z carriages were asymmetrical, and a combination of working in both inches and using metric linear rails.


My M6 tool change macro will automatically apply G52 offsets, and also automatically swap Z axis depending on tool number. (It also will do the VFD switching)


The downside is that I need to come up with a system that assigns specific tool numbers to each spindle, and make sure I stick with it. My current plan is to use tools 0-50 in one spindle, and 51 and higher in the other. (UCCNC currently allows 96 tools).
Or I may just assign all straight bits to one spindle, and everything else to the second spindle. Since almost all projects involve straight bits, this makes sense. And if I need to use two different straight bits, I can always change the tool numbers during programming.

 

[ger21]

Wooden Machines

A few things about building machines from wood.

There are some serious issues that need to be addressed, whenever you need to attach metal parts to wood. Because wood is soft, any bolts holding parts in place will pull themselves into the wood. The more you tighten them, the more they crush the wood. Even the parts themselves can crush and dent the wood, causing misalignment, or parts coming loose over time.
The main way I solve this, is by using phenolic plates and inserts bonded to the wood. All the linear rails are mounted on phenolic surfaces that have been epoxied to the wood, then CNC machined flat, with straight mounting ledges. This gives me straight, flat surfaces that won't crush when screws are tightened, and automatic alignment of linear rails. How to actually bolt the rails to the wood is something I've given a great deal of thought to over the years, with a lot of different ideas.

I've thought about bonding steel plugs in from the back, and drilling and tapping them. But that's very labor intensive. Note that this build is very labor intensive. I'm not averse to lots of extra work, but I really don't want to drill and tap 200 holes, that need to be accurate to much better than 0.5mm.

My other thought was to bond some type of steel nut into a pocket. My concern there is with the nuts not having a firm enough surface to hold them still. The other day, I had an idea that I'm going to try. from the back side, I'm going to bore 3/4" holes to within about 1/2" of the phenolic plates. Then I'm going to epoxy in 1/2" lengths of maple dowels. This way, the nuts will tighten againsdt the maple endgrain, to resist crushing. I'll pour some thin epoxy over the dowels, to soak into the end grain and make it even stronger, then go back and rebore the holes to give me a clean flat surface for the nuts.

The rails use 5mm bolts. To locate the holes, I found a drill bushing that's very close to the counterbore size in the rail, that accepts a #8 drill bit. This gives me a small amount of clearance.
I'll clamp the rails in place, and drill the end holes, and bolt them in place. Then go back and do all the other holes.
I'm using nylon locks nuts on the back. When all the bolts are in, I'll flip the part over, and pour thickened epoxy over the nuts, to hold them in place. The nylon should seal the epoxy, and as long as the screws are oiled, they'll come out fairly easily.
This will be a fairly long process, so I hope to get started soon. Just need to get the garage cleaned up a bit to make more room to work.

Here are some pics of my gantry parts to show a little of what I'm talking about. Looking at the dates on the pics, it looks like I cut them 7 years ago. :surprise:

The gantry is roughly 8" square, and about 67" long. It's a 4 sided box, with each side being 1" thick laminations of 1/4" MDF. The MDF was laminated in my vacuum frame press, with plastic resin glue. This makes them very stiff and rigid. I did some calculation years ago, and cam up with a max deflection of about .001" with a 20lb load. I believe that in reality, it will actually be even stiffer.
If you're wondering why I used MDF, it's because it was free. :smile:
The beam will be epoxied together, with bonded in fasteners. It will be mounted with studs that are bonded into the beam, all the way into the internal ribs.

 

[sreilly]

Great project and detail Gerry but wayyyy over my skill set. Love to see the finished project....

 

[MEBCWD]

Looks good so far Gerry. Looks like this should be a good tight machine with all the bugs worked out beforehand. It does help to have your present CNC to cut the parts and the vacuum table to make sure your glue-ups and laminations are perfectly flat.

Looking forward to following the rest of your build.

 

[ger21]

It's actually the $150,000 router at work, not my old machine. :grin:

 

[mark greenbaum]

Those Thomson rails aren't cheap. I have some of the linear bearings, but the rails got very expensive, so I stopped construction.

 

[ger21]

Total cost for all the linear rails and bearings was about $1600. They are new surplus Hiwins from Automation Overstock.
I bought them 8 years ago. The worst part was the $170 shipping charge for the 3000mm rails.

Two 20mm x 3000mm rails
Two 20mm x 1700mm rails
Four 15mm x 300mm rails

Eight 20mm flange bearings
Eight 15mm flange bearings

The total cost for this machine will probably be in the $7000-$8000 range.

 

[difalkner]

The total cost for this machine will probably be in the $7000-$8000 range.

For what you're going to end up with that's not bad at all. Matter of fact, it's pretty low considering the level of detail and design that's going into this machine. Good job all the way around, Gerry!

David

 

[ger21]

A CNC Router Parts Pro 4x8 is about $10,500.
I expect to have considerably better performance, for less money.
Almost every component I'm using is an upgrade over CNCRP components.

 

[marecat3]

WOW is all I can say

 

[OCEdesigns]

WOW! This is very impressive Gerry! For me it's kind of like reading Chinese because trying to wrap my head around having 2 spindles is crazy! I cant wait to see the finished product and hopefully a video of it inaction.

 

[ger21]

As usual with this project, another year has gone by with no progress.

I just moved into a new house, and am in the process of moving everything from my old garage into the new one.

Once the old house get's sold, I'm going to start getting the new garage shop setup, so that I can hopefully spend all winter building the CNC.