But letâs talk about facts (I know, these days, you have to add âI read it in the internetâ to make it 100% bulletproof information, so read this): w=sqrt(D/m)
where w(=omega) is the resonance frequency, D = âstiffnessâ (=spring constant, in case of simplifying the model as a spring) and m = mass. So increasing the mass will reduce the frequency. That does not help your printer from being strangely constructed, but moves the resonance frequency (âEigenfrequenzâ) to a lower value where it usually is not disturbing that much. Of course, the position of the mass plays a role because the equation above takes into account, that the whole mass is swinging⌠And maybe you shouldnât add 25kg to your printhead.
But a nice granite plate should do as a stand, screwed with the printerâs base. I might have overdone, but in the shelf I placed my printers in, i replaced the original ground plate of the shelf with an 8mm steel plate (20kg, granite was too expensive!). and of course, I do not place my printer on wobbely squash balls. So the shelf needs some âcountermassâ to react (=not react) on a swinging system. I screwed the shelf on the wall and it looks good for my engineerâs heart And finally: I am very happy with my prints. Even on my Mega S!
Black arrows: wall screws of the (industrial) shelf
Red arrow: Lead bars (taped with black tape)
Black ground plate: Steel
I use black silicone tubes as the table âspringâ, since I never touch them anymore since I am using Z-Tilt-Compensation. Fan sticker from my supplier of filaments, network switch for the printer server (not visible) and you see almost no original part on either the table and the printhead. I use a MK3 direct extruder (thingiverse), own designed chain holders for head and bed, 5mm Alu-table with PEI sheet on top (mainly because of the nice pattern that the first layer gives to the part, I used FP4 before and was 100% happy with it). Y-axis belt tightener (thingiverse), own turned belt-tightener screw out of brassâŚ
I would caution people to add large masses to their printers and the surfaces they put them on as Iâve seen many cases where they do not fix print problems (ie shadowing). The additional mass will change the natural frequency of the system as well as the resonant frequency which could improve the situation or make it much, much worse.
The first thing that somebody should do if they have a resonance problem is look at the build surface and ensure that it is as rigid as possible. What I would recommend is printing on grade (ie a basement floor) as you literally canât get anything more rigid/massive than the surface of the Earth and then doing the resonance testing. Comparing that to the results of resonance testing on your build surface will tell you where the problems lie.
Usually you get the most bang for your buck running off as rigid a build surface as possible. Right now Iâm running off of an old oak dining room table, which my grandfather (who made it) would be completely scandalized that Iâm putting it to this use.
I do not agree 100%. If we assume the problem to be basically a âswing problemâ, the appropriate model looks like this:
xâ'(t)+k/mxâ(t)+D/mx(t)=0
where
xââ(t) is the acceleration (at time"t")
xâ(t) is the velocity
x(t) is the position
k is the damping factor (some constant value, never mind)
m is the mass (my lead bars)
D is the âelasticityâ/Stiffness or spring constant
you see, that m decreases the velocity and the overall position (=amplitude).
You are right: If you move the natural frequency from a frequency, where it does not harm anything to a frequency where it does, you start to have problems. But my experience is, that adding a lot of mass improves the overall response to the natural frequency at least for printers.
Just give it a try. You may unload the additional weight, if you get in trouble. One example for âadding weight is funâ is Taipei 101⌠Lots of weight added here!
Itâs not clear how you define the system. At a coarse approximation level, there is at least two (three) harmonic oscillators: the moving masses (bed and toolhead) interacting with belts and stepper springiness, and the overall mass of the frame interacting with the rigidity of its support.
A concrete slab will reduce the natural frequency of the printer-support system. However what we are really interested in is the deviation of the toolhead relative to the bed.
Sure, we donât want the whole printer to vibrate: if the natural frequency of the support is somewhat close to the resonance of the moving parts, there will likely be some phase shifts between components that will excite the toolhead-bed system and produce visible artifacts.
This is likely a very bad idea, but it is another way to reduce the natural frequency of the printer-support system. Rubber feet âworkâ on the same principle.
On that setup, the accelerometers will report a very low frequency resonance, but Iâd argue that we donât want to try to cancel it with IS, as long as the frequency is low enough to not interact with the moving parts.
Of course, that will increase the amplitude/gain. What Iâm trying to say is that it is not that bad if the resonance doesnât interact with the harmonic oscillators of the moving parts. This could explain why we see some people reporting that rubber feet âwork for themâ.
For the shake of sanity we donât want to worry about all these interactions, so a solid, weighted down support is the way to go in my opinion. Weighting down the top bar introduces another spring-mass oscillator, which can work if tuned to a node, or fire back.
Beware of the case of a support/frame resonant frequency that is close but not quite equal to the resonant frequency of the moving parts. This is the worst case for input shaping. IS works best with a single sharp peak.
I think youâre starting with a very defendable assertion (that in a 3D printer, the most important thing is the relative movement and position between the nozzle and the print bed) but I donât think youâre following it up with a realistic strategy.
For your three points you seem to make some assumptions about physical systems that I donât think are realistic. The big one seems to be that you seem to think that it is possible to create a system that perfectly dampens forces due to toolhead motion (along with the gantry and any other moving parts) and does not impart any reaction forces against the toohead. If this were possible in the real world, then I would agree with you.
Any unbalanced force against a physical object will result in some type of âringingâ, a (hopefully) dampened movement back and forth at the natural frequency of the object. The more rigid the object and the more rigidly itâs attached to its base, the higher this natural frequency which will minimize any forces being applied against the toolhead and print bed. It should also be noted that the longer the object is away from its base, the lower the natural frequency will be.
For most 3D printers, the movements are relatively slow, which makes it fairly easy to create a rigid frame and support that will have a natural frequency high enough that there will be little movement and what movement and ringing that results will be dampened out very quickly. With a very rigid structure, it now becomes an exercise in understanding the forces (and ringing) of the toolhead and gantry which can be done with accelerators and the smoothing algorithms available.
Therefore, a very rigid frame that is on a very rigid surface will be the best and most practical way of minimizing vibrations that will affect the toolheadâs motion relative the print bed and negatively affect the quality of the resulting model.
I highlighted the word âpracticalâ in the sentence above because nobody here has the resources to accurately model the physical characteristics of a 3D printer and specify the appropriate dampers/shock absorbers to the structure to absorb energy from the toolhead and gantry movement.
I admit, this partially is based on my empirical data.
I have done countless tests with the ADXL on my printer:
Various feet designs from sharply pointed (like used with expensive vinyl record players) to different rubber dampers
Bolted down rigidly
Add weight from above
âŚ
My finding was that soft rubber feet would yield the best results.
Most of the printer constructions use 2020 and (at best) some 2040 profile. This is generally not something I would give the attribute âstiffâ or ârigidâ.
My reasoning and experience:
The whole printer rocking back and forth does not negatively impact quality
On the contrary, allowing these acceleration forces to be drained in this back and forth movement, takes away the need to absorb them in the frame and (so my theory) avoid vibrations in the first place
What is to be avoided is that either the head vibrates but not the frame or vice versa. Any relative movement between frame and head will negatively impact quality
Of course if you would ensure that the entire construction is so stiff / rigid that its natural frequencies are very high, then this would be the best option. But I guess it has a reason that professional equipment already have a foundation of multiple tons and usually go a bit beyond 2040 profiles
If you are moving the printer in a plane with constant velocity, of course you see no difference. But If your hole printer rocks around, you get forces to your printhead (because, you have to change this movement somehow, and this is only done by applying a force =>Newtons Law) I think you are in trouble, more or less. See the picture with the rubber bands. This (in extreme) will not work for printing. I would argue: Movement does hurt! In industry, everyone wants a stiff and strong connection to the ground.
Yes! Else it would start swinging like a undamped spring! The question is: how many damping is needed? See one example: With this argumentation (this is not, what you are saying, it is just an argumentational example!) one could add rubber washers to all connectors, the damping would be âperfectâ but the prints were not because of the implicit movement of the frame in this case (so to speak: Printing at 100,100 is a function of time then but should be constant). As a assumption: If you accept 1/10 Error in your prints, an optimal system damps all energy within this âlengthâ. You now could calculate a âspring constantâ out of this and you will see: it is pretty tough! My argument for rather be âstiffâ.
See (1). Yes, you will damp energy (if it does not slip) but pay the costs. No one does this for a machine tool in opposite: all machines are doweled to the ground. I did (no, I just supervised it) this several times. I do not know a single machine, that is placed on rubber. We placed 25 M16 dowels (12.2 Quality! Stiff shit!) for a milling machine of approx. 3tons.
btw: I also agree with Piezo, that the relative movement between bed and head is relevant, but i doubt (or better: I know it will not work) you can get it with a non-stiff, nox-fixed system.
In German we have a nice saying: âin der Theorie ist die Praxis einfachâ (translate says: âin theory, the practice is simpleâ but it looses its âside kickâ due to the translation I fear), so letâs just try it out.