[graham_f]

Hey, I've been watching this forum from the shadows for a long time and finally need to ask for some help. I'm feeling kinds stuck with my own build and could use some outside perspective.



I've built (and now rebuilt/upgraded) a CNC router of my own design, and I've tried to take advantage of information gained from other builds to avoid some common issues.
The basic setup is a steel box tube frame, with an aluminum epoxy filled gantry. A boxed z-axis, servo motor, 3.7kw ATC spindle... etc. All built with quality parts.
Measuring things shows that I now have somewhere in the range of 20 thou deflection in the X and Y direction with a full body shove on the spindle when its extended all the way down.

The Problem:
I've recently upgraded from the standard 2.2kw spindle to a nice 3.7kw ATC unit. Now it seems that its not ridged enough and chatters like crazy... even on softer materials like wood and acrylic with light loads.
I never had any issues with the old spindle. The overall design hasn't changed. I had assumed that the z-axis with the new spindle and more mass would, if anything improve vibration resistance, not make things worse.
Am I missing something here? Maybe my design just sucks. ha

Things I have Checked:
-spindle clamp is tight.
-linear blocks and rails are all tight and new (Hiwin 20mm x 4)
-ball screws are tight and have no measurable backlash.
-the gantry carriage has almost no flex ~1 thou with a good push.

Any advice or things to take a second look at? I'm at a loss and would appreciate any input.

[joeavaerage]

Hi,

I had assumed that the z-axis with the new spindle and more mass would, if anything improve vibration resistance, not make things worse.

That is not correct. Basic physics a of a spring/mass system is:

f =0.5*PI* (k/m)^0.5 where k=spring constant in N/m and m is mass in kg

So changing mass will change the frequency of the vibration.....not whether it will vibrate or not. EVERY mass restrained by a springy structure will vibrate......that plain physics.....no avoiding it.
My guess is what has happened is that you have increased the mass of the vibrating system and thereby lowered the natural frequency of vibration and it is now in range of being 'excited'
by the cyclic cutting forces whereas previously those cyclic cutting forces were well below the natural frequency and therefore those large excursion vibrations did not occur.

The solution is to replace the gantry. Filling a hollow section is a waste of time.....it sounds good intuitively but does not give any significant advantage. You'd be better off just making the wall
section thicker..

Steel has a Youngs modulus of 205GPa whereas aluminum is 70GPa. Thus for the same size and thickness section a steel beam will be three times stiffer than aluminum. Bang for your buck
steel far exceeds aluminum in stiffness relative to dollars spent.

Craig

[pippin88]

Show us a photo / video of your deflection test

Are you testing static deflection?
I would not expect changing the spindle (only) to significantly change static deflection

20 thousandths of an inch. That's half a mm.

[Dancnc86]

Hi Guys, completely new here, can you please advise on how I make a new post? Just bought a genmitsu 3020 and having issues with dimension consistency as move along the X axis, many thanks Dan

[graham_f]

Hey Craig. Thanks for the detailed explanation. Forgive me if some of that goes over my head, I'm not a math guy and haven't delt with spring/mass systems before.

that being said, from what i understand even an aluminum hollow box section filled with an epoxy stone mixture would have a better vibration dampening effect rather that using a steel box section alone. At least that what the articles i have read seem to show. My setup is all built out of 1/4" wall tube so its not light stuff... especially after filling with the epoxy mixture. Or maybe i misunderstand and your suggesting i use steel and also fill with the epoxy mixture as well? As for comparing the cost... Its what i had on hand, so that s what i used.


Pippin88... I'll admit my testing procedure is very basic. Just a simple dial indicator on a mag base stuck to the table measuring different directions while man handling the spindle while extended all the way down so i can get some good leverage. I did the same "scientific procedure" ha, in several direction with multiple test locations and the results are pretty consistent.
I did some similar tests while measuring my aluminum gantry for deflection and twist, and to my surprise its pretty solid. At most i get 1-2 thou in any direction. As far as i can tell that seems pretty good.

my only other thought is maybe my boxed z-axis and the 8 linear rail blocks just isn't a rigid as i expected. Maybe its racking just enough to allow the deflection at the tool tip?? Can you see any improvements to be made in that area? I tried messing around with FEA in solidworks, but nothing useful came of it. My knowledge just isn't great here.

anyways, let me know if you have any other ideas and thanks for the help
.
Graham

[peteeng]

Hello Graham - 1) are you running the exact same feeds and speeds and jobs as you did with the old spindle? 2) I suggest you swap back to the old spindle and check it does what it used to do and observe some sort of standard cut. Say a square or a round at set S&F's. 3) then swap back to the ATC and do the same thing. ie you could set up a trial cut now that does wobble then swap back to the old spindle and run the same program to compare... The dynamic stiffness of a machine is complex. Are you using approximately the same spindle height 2.2 vs 3.7kW? is the 3.7kW sticking further down then the 2.2? When you changed the spindle over did you do anything to the Z axis assembly?

As Craig says you may have just lucked out and picked a combo that's live. Shorten the spindle in the clamp as far as possible ie as less stick down as possible and see if that changes the dynamics. Does it wobble in only one direction? Does it start the wobble on corners in one direction? eg in the Y or X direction... more info needed.... can you do a section image of the z axis so I can see the carriage arrangement? But if its worked well before it should work well now...Peter

Does your box Z axis only have 4 carriages? I can't see any bottom set of bearings? I can see the 4 gantry X axis bearings. Does the box only have 4 carriages, one on each corner? if so this is not a good geometric arrangement. Even though it has 4 bearings these being in-plane, its unstable. It needs 8 bearings to be stable, do you have 8? Peter

[graham_f]

Hey Peteeng, Just have a minuet at work here... but i can quickly answer your questions.

1. unfortunately i can not just switch back to the old spindle... its dead but the new one is set at the same height as the old one. At least the cutting tip is... (the overall height of the spindle is longer due to the ATC). I'm also somewhat limited because the spindle has an acceptable clamping zone that i must stay in. The only thing i changed between the old and new spindle is a fresh set of linear rails. same size and brand.

2. the vibration starts each time it changes direction and grows until the direction is changed again. This cycle repeats over and over.

3. Yes, i have 8 linear rail blocks, 4 in the front and 4 in the back. They are spaced about 8" to the outside ends. I'd go wider but I'm limited by the z travel and the rail lengths.

[peteeng]

Hi Graham - 8 bearings is good. Have you tried different tools with different teeth numbers? This changes the dynamics eg use a 1F tool and a 2F or 3F tool on same path and same F&S's? Plus use a new tool or at least a sharp tool. Blunt tools will create more friction and more vibrations.... What preload class are the carriages? What's your typical feed and speed and tool type? no of teeth? Does your phone camera have slow motion or hyper motion? Can learn a bit by using it...Peter

[joeavaerage]

Hi,
when you fill a metallic tube, either aluminum or steel, with a 'squishy' material like a filled epoxy it does not stop vibration.

The metallic tube is very much stiffer than the material which fills it. Even when you fill it up the resultant combination is still only a few percent stiffer than the metallic tube on its own.
A filled epoxy is in the region of 10GPa to 20GPa. It is true that a filled epoxy has a very damped oscillatory character.....but that does not stop a vibration in the structure.

The damping of a material, any material, comes from the internal heat generation by the material being cyclically being strained....ie an energy loss. Filled epoxies absorb vibrational energy
many times better than that of regular metals.

The problem is that given that the metallic 'shell' is so much stiffer than the filled epoxy that there is only very small cyclic strain applied to it...and that assumes that the epoxy is perfectly adhered to the
inside of the shell and therefore very little vibrational energy is absorbed by the filler.

No matter what :ANY elastic structure of finite mass WILL VIBRATE at its natural frequency as I posted above. There is absolutely no way around that. In a well damped structure, that vibration will
decay away and reduce in amplitude quickly whereas a less well damped structure, will carry on oscillating for longer. What does not change is that both filled and unfilled structures will oscillate.

Once you fill the tube it will be heavier and yet its stiffness will be (within a few percent) the same, and therefore the natural frequency will reduce, being inversely proportional the the square root of mass.
This is a mistake because the lower the natural frequency the more likely that vibration will be excited and the machine chatter. As you have found.

In an ideal world you would INCREASE the stiffness of the structure, say by using a stiffer material like steel, and with a thicker wall and maximum geometric dimensions while simultaneously DECREASING
the mass. This has the effect of markedly increasing the natural frequency so that it is NOT excited by the cutting forces of the machine.

The bottom line is that the stiffer the structure the less inclined will be the machine to chatter. You can add as much damping as you like.....but you cannot stop the chatter....reduce it....but not stop it.
The most cost effective way is to forgo the filler but make the wall of the metallic tube thicker. The cost of the epoxy filler is not money well spent, or rather that the performance
gain achieved is less than the cost of doubling the wall thickness.

In industrial practice the most often used material is cast iron. It has a modulus of about 110GPa....about half that of steel, but is quite nicely damped. Despite cast iron being one of the very oldest
of mankind's industrial materials it is still the most widely used in machines. The next most widely used material is steel. The stiffness for a given number of dollars is head and shoulders above ANY other material.

Craig