[TMToronto]

I have a CNC router that uses dual 50 mm OD hardened chrome rails as part of its linear motion design for the X and Y axes.
It is a reasonably rigid machine for this class of hobby machine,, and with a few modifications such as a new Z axis assembly that replaces tubes etc...with quality linear rails/guides etc..., my rigidity tests are showing ~ 1N/um at the endmill (not quite a Haas (-: ).

I am also using an ATC spindle, and currently testing its capabilities by machining aluminum. So far I am pleased with the results WRT tolerances and surface finish, but I am looking to maximize its performance, and learn a bit more along the way..

I have tried to mitigate self-excited vibrations through careful choices of tooling, CAM strategies, work holding, and other variables I can control (and perhaps understand (?)). I also want to turn my attention to other sources of vibration, and I think the tubes are a big contributor to that.

I was recently reading forum posts which talked about solutions involving filling tubes with small lead/steel and oil for damping. It is not practical for me to do this, or sand as is sometimes mentioned. Other suggestions of using a solid hard filler did not seem to match what the science I discovered stated.

One idea in a post caught my attention...

"My understanding about how the rubber between two pieces of metal acts to absorb the vibrational energy is that it is sort of like an impedance mismatch. All the materials have an intrinsic natural frequency that is dependent both on density and Young's Modulus but they are all different. The metal that is the source of the vibration does not pass the energy efficiently to the rubber because the rubber does not want to move at the same rate as the metal is pushing. Similarly the rubber does not pass the energy efficiently to the second metal. The net result is the energy is largely lost due to hysteresis in the rubber which gets warm."

It gave me the idea of using a series of equally spaced 1.5" (38 mm) OD cylindrical rubber isolation pads pressed firmly into each tube, and a steel rod being firmly run through their ID hole (typically 1/2"). I could vary the dimensions and number of pads used, as well as the size and mass of internal rod. If it works, even in the slightest, I am happy to try this method as it is relatively inexpensive, quick to install, and easily removed.

I appreciate any insights others with more experience are willing to share.

In the mean time I will continue to try investigate and measure the various, frequencies, resonances, and vibrations I currently have as a baseline for future modifications I may try.

Thank you,

Tom

[joeavaerage]

Hi,
the problem with all those strategies is that they ALL rely on energy being transmitted to a 'damping' material and that damping material dissipates the energy.

The only way energy can be transmitted to such material is for the comparatively rigid structure to flex and that flexure being transmitted to the damping material.
The more rigid the structure the smaller the flexure of it. Thus any given part of the structure may move or flex at most a few um, and that very small linear motion is applied
to the damping material, if and only if, that damping material adheres perfectly to the structure. Thus any elongation/compression of the structure results in the same
elongation/compression in the damping material.

Thus the stiffer the machine is the less likely that any damping material will get energy transmitted to it, and have an insignificant effect on the machine overall.

The bottom line is don't waste your time and money on damping. The correct thing to do is to make the structure more rigid, either with thicker material of the same modulus OR use a material of higher
modulus OR a bigger geometeric section which allows a bigger moment. Some materials like cast iron are favoured because they have high modulus but also fair-to-good damping characteristics.

There are some whom recommend 'envelope shaping', where the servos are driven in such a way to minimise the 'jerk'.

This sounds great but does not really work. It turns out if you reduce the max acceleration of the servo by 10% you improve the machines vibration characteristics by the same as applying 10% 'jerk control'.
You don't actually gain anything.

Another possibility is 'active vibration control'. This relies on the control system applying a corrective force on the machine or machine component to counteract or neutralise the vibration movement.
This is turn relies on a high frequency, high bandwidth measurement system of the machine flexure, and that the similarly high bandwidth controller can apply a corrective force in the correct phase.
Such control systems are known and used in extremely high end applications, like missiles and fighter aircraft, but are not applicable for hobby CNC.

Craig

[TMToronto]

Thank you for taking the time to reply and for the suggestions and information - I appreciate it.

I have learned a lot in this forum and from you over the years, and when more time and funds permit, may try to apply some of that to building my own rigid machine. There is not much more that I can do to my current machine WRT rigidity and stiffness, but as I mentioned at the start I am not disappointed with what I am able to achieve.

Perhaps for now I will try to investigate more thoroughly the self-excited vibrations that are produced when machining the types of materials and projects that currently interest me on my machine, and try to quantify what improvements I can make. I plan to continue reading about natural frequencies and how to optimize cut parameters so that I am not creating more (self-excitation I think some call it) and resonance. It comes up over and over that chatter can be mainly addressed by examining workholding, tooling/holders, and programming. I have been learning first hand about these through trial and error over the two years that I have machining. One new thing I wanted to try was the use of uneven flute and variable helix endmills.

This is definitely a hobby that keeps your brain busy.

Thank you again for the support.

Tom

[joeavaerage]

Hi,

There is not much more that I can do to my current machine WRT rigidity and stiffness,

That is very much the reality for most machines. There maybe one or more specific parts that can be swapped out with more rigid parts at some later date, but the majority of the structure not so.
'It is what it is' from the time you build it. You are fortunate that you can produce good work and be satisfied with it, that is as much as anyone can hope for.

Craig

[peteeng]

Hi Tom - Dynamic stiffness of structures and vibration damping is a very big field. As Craig has said the machines DNA is pretty much set at birth and any dynamic stiffness issues are difficult to surgically remove without major changes to the structure. You mention constrained layer damping. This is usually between two stiff elements (eg concentric tubes) with a thin viscoelastic layer, say 0.5mm thick. Very difficult to do in practice at a hobby level and even some of the work done on this 20 years ago has not appeared in current commercial machines. What has happened in the last 20 years is that improved engineering design (using FEA and simulation) has allowed machine builders to very significantly increase the static stiffness of machines. I think you would be disappointed filling tubes with stuff, it really does not address machines issues (IMHO that may get some pushback) and yes I have been involved with a machine that was filled with foam and other stuff and all it does is make it heavier and remove some of the acoustic vibrations but not the pesky tool vibrations. So publish some images of your machine and the forum may have some ideas for you... there are many people working on this issue.

I laminate plywood (already a very damp material ) and aluminium to make stiff damp parts. I intend to laminate parts entirely of aluminium or steel to take advantage of this effect. Look of GLARE its a metal laminate used by Airbus on most of their wings. If you have flown Airbus recently you are flying on metal laminates.

So what is the state of the art in this area?

1) 3D printers are decades ahead of hobby level machines in this area. They are light and fast and vibration is potentially/reality a big problem. So they are implementing "input shaping". If your up to installing a new controller, particularly a 3D printer controller from Big Tree tech that has input shaping you will do very well to remove the major vibs from your machine. So investigate input shaping., Not new, 30 years old developed by NASA to reduce space vehicle vibs... takes time to filter some of these things down to our level. I can publish some stuff on this if you like. Their machines are well under 1N/um...
Resonance Compensation - Klipper documentation (klipper3d.org)

2) Jerk control - if your machine is using a simple motion planner it is probably making "rough" motion. So you can implement a control such as Dynomotion that has jerk control. It also has feedback and active motion control potential. This site has a dynomotion forum

3) High end machines can use active control. This involves having an accelerometer on the spindle and this provides vibration feedback and the PID controller mitigates that vibration. Probably out of reach of the hobby guys. This is used on high end hypercars and supercars and racecar suspensions if the class allows it

4) tap testing- HowMachine Tool Dynamics Could Become a U.S. Supply Chain Strategy |Modern Machine Shop (mmsonline.com)
This is what input shaping does but is a bit more refined

https://www.malinc.com/products/cutpro/

Lastly people usually slow down their machines when they have chatter issues. But these sort of analyses usually say speed it up! So as long as you have the correct chipload and good constant velocity feeds you may find that speeding up gets you over the vibration hump. Some test cuts will find out. There will be "humps" of agitation and the idea is to cut in the valleys of quietness. Hope this helps. Peter

[peteeng]

Hi Tom - here's some vibration analysis SW that I found the other day. Haven't had a look at it yet but should be helpful. PeterVA - Visual Analyser 2011 for Windows 7/Vista/XP

[TMToronto]

Thank you @peteeng for taking the time to add your insights.

I have been following many build threads in the forum over the years, and always value the experience you and others bring, especially in the area that this discussion focuses on.

The last major modification I made was to design and build a new Z axis assembly - from my rigidity testing it was one of the weakest points that I could address myself - and that resulted in improved rigidity/less flex at the end mill based on my testing.

I have attached screenshots from my new Z axis assembly video. They show examples of some of my testing results. (Below is a link to my YT page where I keep track of my many projects - it will give an idea of my machine and what I do with it)...

https://www.youtube.com/@tmtoronto3741/videos

Here is one of the videos I have archived on the topic of vibration, resonance, and chatter. I found it informative with the concepts well explained. I believe it also talks about the 'humps of agitation' you refer to.



Moving forward, I downloaded a few apps that can make use of my phone's accelerometer. Not exactly an impulse hammer test, but one of the apps can do a FFT analysis from the data. I thought these might help me better understand what is happening at various frequencies.

[peteeng]

Hi Tom,
The vibration analysis should be useful. Its amazing what apps are on phones these days. Your experience with it published here would be excellent. Your machine is based on large round unsupported rails. These are attractive as they combine structure with motion functions. The designer of your machine has made it simple and slick. I liked it as soon as I saw it some time ago. But the twin beam design, unsupported rails etc limit its stiffness.

Unfortunately they also provide extra directions of freedom that aren't helpful for gaining machine stiffness or reducing motion clearances such as square rails do. Maybe its time to update! Use your machine to build a new machine with all the embedded knowledge you now have at least design it, design time is free and maybe better spent time then trying to update a machine.

What I have learnt in designing and building many machines is that the Z axis is probably the most important element to get right and most people design this last (including myself in the past) which means its usually undersize and lacking real estate to size it correctly. So now I start at the Z axis and design it to do what I want it to do then add some. Once I have an UBER Z axis, build outward to get the machine correct. In this way there is no back tracking, pinching geometry here, adding a little there. After all, the spindle and Z axis is the main functional part of the machine, its the pointy end of the business, it does the work. You have found this out. Peter

[TMToronto]

I am always looking at, and out for, innovative CNC build ideas that I can use my machine to make. I know the linear motion will be through quality ballscrews/nuts and linear rails/bearings.
As for the design, I like the idea of tall rigid raised Y rails with a moving gantry and Z axis. I also have seen a fixed z axis that moves side to side on a gantry that is raised by two ballscews, and using a sliding table for the Y axis. I like the laminated solutions you have been settling on and experimenting with. I have no shop per say, so the materials and manufacturing techniques you use are definitely within my skill set.

I recently discovered another method of vibration analysis that makes use of sound recordings vs the phone accelerometer. It also introduced me to Mathematica and Wolfram language. I made an account for Mathematica today for a free 15 day trial, and am excited to learn to use it - I am a bit of a data junky, and love all things math, science, engineering, design,... If all goes as planned, I should then be able to compare the results from two different methods - sound vs accelerometer data. It would add more meaning and credibility to my findings if the resulting analysis from both yield similar frequency/vibration results. The next step of course will be to put that knowledge to practical use to fine tune machining parameters.

Here are the links if interested:

https://blog.wolfram.com/2017/03/02/...fram-language/

https://www.wolfram.com/mathematica/trial/

Finally, since I have a hard time letting go of things, I thought I'd ask again...

Do you think my earlier idea of adding a series of tightly fitting isolation pads inside the 50 mm tubes, and a tightly fitting 1/2" steel rod through their centres, would have any impact on how, or the extent to which the tubes vibrate or resonate? See attached drawing. I believe @Joe average when he says it may be minimal at best, but I thought it might still be interesting to try, especially in light of my planned vibration testing which might actually be able to quantify any differences. Just thought I'd ask before letting the idea go.

[peteeng]

Hi Tom - like the news items these days I have to say that some of this is my opinion and everyone in the forum is welcome to their own opinion So IMHO

Short and curt - No - Why? For viscoelastic mechanisms to work they have to be in the loadpath. The upload path and the back load path have to be different to create differential movement resulting in friction which makes heat and loss of energy. Also called hysteresis. Your proposed arrangement does not place the "rubber" in the loadpath, when the tube bends or torques the rubber is affected in a minor (or secondary ) manner not in the direct loadpath. This is my same argument when you fill a tube with a material. The tube itself is the majority of the geometric inertia resisting the deflection, the internals do nearly nothing especially if they are a material with a lower modulus as strain travels in the most rigid part and in the shortest path. By filling thin hollow tubes with stuff you definitely improve the acoustics but these acoustic noises are in the air not the structure and do not affect the cut, only our ears. You can solve acoustic noises (organ pipe noises) by closing off the ends vs filling the tube with stuff.

If the tube was vibrating with a huge visible displacement so that the rubbers , rubbed on the internal surface maybe, but machines are designed to be very stiff with very small deflections and that's part of the issue with using materials to provide damping. This is how leaf springs work, they deflect and rub against each other and friction dampens the motion. If you use a single steel spring you then need to use a shock absorber (correct name motion damper) to stop the wheel bouncing around.

Making the machine super rigid means the internal strain is very small so material hysteresis is very small. On page 20 of the attached the Professor says that material damping is not enough. We need to use other methods to produce damp machines. Especially as we now make extremely stiff machines so the internal hysteresis is really tiny so the damping is tiny. Most people think we make machines from cast iron because its damp. No, we make CI machines because its an economical casting process for making many parts, and historically we developed CI's that had lubricity & wear resistance properties as the ways where machined into the part. Now that property is superseded by rails. The damping has always been a low a low order feature.

Viscoelastic damping goes against the primary objective of making a super rigid structure. If you made the gantry from rubber it would be super damp but it also would not be stiff enough to function. Chase stiffness and rigidity first and then some. If it's really stiff you won't have dynamic vibration issues or they will be minimalised. All structures and machines vibrate, we need to map them and find out where their sweet spots are and stay there hence vib analysis is important... Keep us informed about the vib analysis I'm really keen to learn & push into this area but currently have no time!! Peter

[TMToronto]

That makes sense. Thank you for the PDF (and other links). I had a quick skim read, and it looks quite interesting.

I will invest my time in the vibration analysis as my next project. I am excited to see what I learn in general, and about my particular machine specifically.

I typically make a video at the end of each project to share with others, particularly those who have the same machine. I can share a link here if interested. Perhaps you and others can add insights into what I discover.

Thank you (and joeaverage) for you support,

Tom

[pippin88]

What are you trying to improve?

Surface finish?
Material removal rate?

Surface finish can be improved by adding damping to an existing setup.
Damping boring bars. Visocoelastic materials adhered to keys parts of the machine. Etc.
Joeaverage goes on at length about not worrying about damping, but there are plenty of examples of where adding damping can help.
Plenty of research papers to back this up.

Adding a bit of damping will not allow you to dramatically increase your material removal rate.

[joeavaerage]

Hi,

Joeaverage goes on at length about not worrying about damping, but there are plenty of examples of where adding damping can help.
Plenty of research papers to back this up.

It plain physics. A damping material can only dissipate energy if motion is transmitted to it.

Better to make the structure stiffer. If its stiff enough its resonant frequencies go so high (audio and higher) that they have little to no effect on cutting.
If, however, the structure is so compliant such that cutting forces excite low frequency resonances then no amount of damping will stop it, reduce it maybe, but never stop it.

Craig

[pippin88]

That's all well and good in theory.

But are companies going to throw out a lathe and to buy a more rigid bigger machine that they can't afford?
Or maybe an antivibration boring bar makes a bit more sense.

Yes, the opening poster would have more success replacing his machine with a 20 ton 1 million dollar machine, but that ain't going to happen, is it?

To the OP: I say give it a go, and document results before and after. You may reduce chatter and improve surface finish. I don't expect you will be able to increase material removal rate significantly, unless you have been really limited by chatter (unlikely given you mention careful attention to feeds/rpm etc)

[joeavaerage]

Hi,
the hugely expensive 20 ton lathe exists because a cheaper 5 ton lathe with an anti-vibration boring bar cannot do the business.

Who would bother with big expensive machines if cheaper ones could do the job?

OP has already stated that while he would like a stiffer machine he can in fact do good work provided he lives within the limitations of his machine as it is.
I think that goes for all of us, I'd sure like my mill to be double as stiff, but it is what it is, and I can do good work within its capacity. There is no magic
material or technique that is going to get me double the stiffness so I box on and try to get the best I can from the machine I've got.

As it turns out I need to make another headstock, the current one despite being steel is too compliant. I have nearly finished the pattern to have a new
headstock cast in grey iron. It is also a part that can be easily swapped out and yet get me quite some stiffness advantage, and indeed that was always my long-term plan.
I did not have the extra money to do it when I first built my machine, but I knew that someday I would, and that I could make a meaningful upgrade to my mill.

Craig