[magno_grail]

Hi All,
Started gathering the parts to build my own CNC. Most designs that I see out there have too limited Z axis travel. Some work will be in aluminum, some acrylic. I mostly need it for prototyping as I have a couple of CNC shops make my parts. The acrylic is for making wax models for investment casting. A casting house wanted $7800 for a simple sand cast mold that could be made out of wood for a clutch housing (11"x11"x3"). Since I have several projects that will need castings it is worth the time and effort to build my own CNC.
The design is a gantry type with round bar rails and linear bearing trucks sourced from VXB. I got 16mm C7 ball screws and a 1.5kW spindle ER20 collet and inverter (Huanyang) from China. The spindle is rugged enough for mold making or so advertised. I have not selected the steppers yet because I needed the weights of the moving parts but I intend to use NEMA 34 size. The frame is made from 2.5" x 1/4" wall aluminum box tube bolted (rather than welding to keep down distortion). There are only a few connection pieces that need to be made. The work area is 350mm cube.
I will start out using LINUX and the parallel port and go from there. I might eventually go to a PCI card. I have GibbsCam, AutoCAD and Solidworks.
Here is a picture of the setup:

[CaptainVee]

Please do not go with this!
Unsupported shaft like this gives a flimsy router! It will flex like crazy requiring very low feeds, making the spindle pointless...

my 2cents.

[magno_grail]

What unsupported shaft? The rails are 20mm steel rods. The X axis has three supports on each rail, the Y and Z are supported on each end. I am not building a router.

[CaptainVee]

Supported shafts are supported along the whole length, the shafts in your design are only supported at the ends.

Unsupported shafts will flex where the same size supported shafts do not. Supported shafts allow higher accuracy and/or higher speeds. Higher speeds require stronger motors, obviously.

[awerby]

Sven's right - there's no way you're cutting aluminum on that thing, and anything harder than foam would be iffy at best. It's easy to swap your unsupported table rails for supported ones, but you'd need to add some structure for the Y and Z axis supports to attach to.

Andrew Werby
www.computersculpture.com

[magno_grail]

Can I see your calculations for deflection versus load?
I considered T rails however, bolting down a rail every 10cm makes it difficult to align the rails and the trucks are lower capacity.

[magno_grail]

I made a change to the Y axis, moving the bars to either side of the motor. This eliminates one side of the clamp (using two truck halves) and the bending moment on the bars due to the offset weight of the motor. The Z axis ball screws are in line with the motor with linear springs to offset the motor weight.
I used the Kennemetal site to calculate cutting loads in 40,000 psi aluminum. For a two flute 1/2" end mill cutting 0.05" depth, full width, 0.003" per tooth at 68.8 in/min the cutter load is 6.6 lb.
The weight of the motor and trucks is 16.98 lb. The Y axis bars deflect 0.000175" with the motor in the middle. A 2" cutter below the end of the motor puts 8 lb. at 24.4º from vertical deflecting the bars another 0.00017" in opposite directions. The end of the cutter deflects 0.00038".
The table is 1" aluminum tooling plate with pockets underneath. Still weighs 35.5 lb though. The X axis bars bend 0.00055" with the table in the middle.
Considering a finish pass is used and the C7 ball screws have 0.0005" backlash the structure is stiff enough for my needs.

[awerby]

You're only kidding yourself, Magno. Anybody who's ever operated a CNC machine can tell you that this design, with all those unsupported rods holding a spindle out in space, is going to kick like a bucking bronco. Those calculations of yours tell you how much force you need to spin the cutter, but not what sort of vibrations are going to happen when you do it using a long endmill into a piece of metal. The static weight of your components isn't what's going to deflect your rails, it's cutting forces, which are many times greater. And the way you've mounted them, on those weak little triangular plates, is only going to make a shaky situation even worse.

If this sort of design really worked, why do you think manufacturers would take the trouble to make massive castings in iron for the frames of metal-cutting mills? But it doesn't seem like you're willing to take anyone's advice on this, so go ahead, build it, and prove us all wrong. Be sure to post a YouTube video of this thing in action, smoothly carving away massive amounts of metal with a 2" cutter or whatever. Please post it back on this thread, so I can be sure to admit I was wrong about this.

Andrew Werby
www.computersculpture.com

[magno_grail]

"smoothly carving away massive amounts of metal with a 2" cutter" - you must be joking. First off, the spindle only has an ER20 collet which takes at most a 1/2" shank. Even with my 10"x54" knee mill I do not take "massive amounts" of metal with any cutter. If you read my post, the calculations are the cutter forces, not the force needed to spin the cutter (which is torque). If you do not believe Kennametal then check out the Machinery Handbook.
The end support plates are not final. I changed the Y axis support because this gives a better support for the spindle. I am not building a Haas VF-1. I have run a Haas VF-1 and have two mills so I know what it takes to hold a piece of metal and cut it. Since there is only air cooling on this I would not do any heavy cutting anyway. Perhaps you have not seen the many router designs that use plates supporting each end of the Y axis?
If you look at the pictures, the ball screw takes the lateral load from the spindle and it is mounted close to the Z axis supports so there is very little bending load applied to the plate. The Y axis bars only load the plate in the vertical/horizontal planes which puts the plate in tension/compression.
What is your calculation of the natural frequencies for the support bars? I previously asked for force deflection calculations and got no reply. I hear advice but no proof. I will accept any graphs, charts or equations that you have.

[007CNC]

I think these gentlemen are just trying to save you some headache down the road. You dont have to take their advice if you dont want to.
I am not a pro, and I dont have calculations for you, but from the little I know from building CNC in the past (from horribly flimsy to very rigid machines), I have a hard time seeing how you will cut metal with that design.
from the structure to the unsupported rails to the Z height... Dont see it.

[derek]

I agree with the advise your getting but from the sounds of things you should probably build it like it is and report back.

I will say though that there are some things I don't need to see a graph on to know it won't work.

Derek

[magno_grail]

Thanks for the input. As you are probably aware, on the internet, everyone is an expert which is why I asked for calculations or some other data showing why this will not work for my intended use. Ages ago I had a Unimat mill/lathe which also used steel bars for the carriage. It was not the most robust design but it worked. If the design is not as stiff as I would like it I can go back and replace the bars with linear tracks. I already have the bars, trucks and supports so I will proceed with this arrangement. If cost was no object then I would have bought a commercial machine and not build a DIY.
The Z rails are supported in the same manner as the Y. The X has an extra set of supports in the middle because of the longer span of the table.
So far, no one has defined what "will not work" means. Will not hold a cutter? Will not move in all three axis? Will not support its own weight? This appears to be a definition of terms.
I am not new to machining. This is a BSA A65 motor with a Yamaha XS650 crank. The engine now has a three bearing crank re-phased to 90 degrees (formerly two bearing 360 phase). I need to investment cast a new cylinder and head to go with it which is one of the first tasks for the CNC, making a mold for the wax patterns.

[TarHeelTom]

.

[steve123]

Well, maybe you will believe this:
http://www.cnczone.com/forums/diy-cn...le_idea-5.html

A nice long thread. Near the end you start hearing stuff like this:

My Z axis has 12" long linear rods and the Z carriage was mid span. I could actually watch the linear rods flex. I could minimize the action by moving the Z axis carriage toward either end of It's travel. .035" was a worst case situation.

Moral of the story: Don't use unsupported linear rod in a machine you plan to use working with aluminum.

Then again, you could just ignore advice and just do it yourself and see.

[magno_grail]

You also find statements such as this:
"I used a .250” 4 flute HSS end mill to mill in 6061 aluminum .100” deep .050” step circular pockets with 50% stepover in .5”, .75”, 1”, 1.25” and 1.5” sizes at 20 IPM and 3000 rpm. All the pockets measured spot on and the spindle handled the job easily. My other spindles can not accomplish this level of accuracy in aluminum."
So what changed? He stated later that the motor was changed to a much larger HP. The Z axis 1/2" rods are fairly small for a machine of that size. Stiffness of the rod goes up with diameter ^4. He also stated that he likes to take heavy cuts. Putting in his numbers into Kennametal's calculator it comes up with 14.8 lb load on the cutter. The K2 web site offers this advice:
"proper technique required if used to cut aluminum"
They move the gantry instead of the table for the X axis, have only end supports on the X axis and the further up the motor is moved on the Z axis the further away the support (Y axis) bars are.
I put one 20mm support bar in the lathe chuck with the far end supported by the tail stock. With the motor set in the middle the bar deflects 0.001". The motor weighs about 16 lb., is supported by two rails and I am not taking great swaths of metal off.

[steve123]

Take a look at this interesting article about chatter:
Cutting Tool Engineering | Article Archives |
A machine must be both statically and dynamically stiff. Are round rails good at dampening vibration? I don't think so.

[RCaffin]

The difference between theory and practice is that in theory there is none.

Cheers

[LeeWay]

The design is very similar to the Prusa Mendell 3D printer that I built recently.
It functions well as a printer that is laying down material, but would be no good at all in material removal. Hit a little hard bump of plastic with the print head and it just bounces right over it almost effortlessly. It isn't the plastic bump that gave away, it is the head that deflected. Granted your rods will be a little larger than the 5/16" rods I used on the printer, but the design remains the same. Flex is the issue on the Prusa and will be on your design too.
When you add speed to the mix on the Prusa, it shows in the product too. It can do prints without showing direction change if you keep the speed relatively low, but ramp it up and printing becomes very poor.
Again this is flex induced by rapid direction change.
You can work around these short comings on a 3D printer. Like a plasma cutter, there is no drag on the bottom of the Z.
Induce a drag there with any kind of cutter and even running it slowly will not yield very accurate cuts.

I suggest that since you have the stuff to build it with, go for it.
That will let you see first hand what it can and can't do. It will give you a basis to work from when you redesign and build your second machine.

Don't despair if it doesn't meet your needs with a cutter on the Z. Put a print head on it and you have a 3D printer.

[magno_grail]

5/16"? 20mm rods have over 40 times higher moment of inertia. I do not know what the span is between your supports or how they are fixed so I cannot compare the stiffness. Looking at the pictures it is held together with plastic pieces and uses plastic bushings. Threaded rods for the axis positioning. You think there is any real stiffness in that structure?
I get the impression that nobody here has ever measured or calculated the forces on a cutting tool.
Cross section has little to do with damping. Damping converts the strain energy into some form of energy loss, usually heat. Steel and aluminum have very small internal dampening. There is a paper on the mechanics of it in 1020 steel here:
Evaluation of the damping capacity of a cold-rolled SAE 1020 steel.
Chatter comes from exiting the vibration between the tool and part. Increasing the stiffness of the structure increases the frequency, increasing the mass decreases the frequency. By changing the tool speed the input excitation frequency is changed. Changing the feed changes the amplitude of the input. If the forced excitation is far enough above or below the natural frequency it is not a problem.
Is the end supported rod support as stiff as one supported every 10cm? No, but is it stiff enough? It can be depending upon the intended use. I could build the bed from one ton of chilled nodular iron and it would be very stiff but it would be overkill.
There is a YouTube video of someone cutting 0.008" cuts in steel sheet with a KD CNC with end supported rods. Their main problem was holding the sheet and reduced the cut to 0.001". I doubt that I would ever try cutting steel with this, that is not the intent.

[ger21]

You mentioned twice that the 1.5Kw spindle has an ER20 collet. All the ones I've ever seen have an ER11 collet.

And I agree with everyone else. I've seen a wood router with the Z axis attached to 1/2" round shafts (admittedly longer than yours). With pressure from 1 finger at the tool, it deflected almost 1 inch.
Typically, guys building wood routers like to use 30-50lbs of cutting force when designing. 8lbs of force when cutting aluminum seems ridiculously low.

As everyone says, build it and prove us wrong.