[peteeng]

Hi Cycle Start RE: FEA of carriages. It can be simple or detailed. Depends on the FE system you have and how much time you want to spend on setting it up:
1) simplest is to consider the connection as bonded. Then make the car less stiff say aluminium stiffness. If your only comparing different geometries then its apples to apples and all's good. if you are looking for an absolute answer then this is not good enough
2) The bearing suppliers publish the bearing stiffness. So you can look that up and modify the carriages modulus until it agrees with the quoted rigidity/stiffness. Then you can model the connection as bonded with an accurate connection stiffness. I've done this several times and now I just make the car aluminium close enough....
3) If your FE models sliding then you set the connection up to slide, plus do 2) and you have a very accurate connection detail. Just remember that if the cars can slide you have to restrain the model somehow. In reality the drive system restrains the motion systems... Peter

A "bonded" connection means the connection is elastically perfect. This is achieved either by a contiguous mesh across the two solids or some sort of connection element between the two objects meshes. Depends on the simulation system you have. Seems the term FERA is dieing out and being replaced by "simulation" most FE systems these days have very good connection options and we can do all sorts of things vs old school FEA... Plus meshless systems exist which means there's no elements (nearly).

[bogan]

Long time no update, but the machine is basically finished in it's initial state (limited covers, no tool changer, no flood coolant).

It cuts really well, in the picture is some steel I've put an o-ring groove in, it machined the grid-plate like a champ. One minor thing is I haven't got the axes all that square to each other, XY is about 1mm out over 500mm, with Z both around 0.05mm over 100mm, spindle perpendicularity is about 0.03 over 40mm. Nothing too much to worry about now that I'm aware of it, at some point I'll probably make up some adjustment parts then take it to bits and chuck em in.

Servos are also not that well tuned, AASD servos from Aliexpress, hard to tune with not the best manual and I'm running a much higher inertia ration than I imagine they had intended.

Spindle is great, glad I got the higher torque option.

Very happy with the design, stoked to be finally making chips, and parts! Treated myself to a wet and dry vac to clean up the swarf and keep the wife happy

[peteeng]

Hi Bogan - Looks good. What's the spindle specs? Black Stig looks pleased. Peter

[routalot]

Its hard to describe the exhilaration of seeing the machine behave as it should when you press the button.Getting a high quality finish while doing so is even better and opens up a world of possibilities.The small misalignment of the axes can be dealt with.Its a long way beyond my controller knowledge but there is a way of incorporating a software correction layer between the Gcode and the machine movement.As I understand it,a correction factor is applied to produce a component with the correct characteristics.I wouldn't know where to find the software or exactly how to apply it,but it might be worth a few minutes of googling.

[bogan]

Hi Bogan - Looks good. What's the spindle specs? Black Stig looks pleased. Peter

FM30F from CNC depot, 12krpm 3hp.

[jaguar36]

Looks awesome! Are the poorly tuned servo's causing you any significant issues? What are you using for a controller?

Would love to see some more pictures.

Bold move having the CNC near the laundry machine!

[bogan]

I haven't machined bearing fits yet, but I expect I'll just need to slow the feed rate down a lot to minimise the overshoot. When doing datum holes there is definite non-circulairies most notable at the four 'edges' of the circle, around 0.03mm.

Controller is Mach4 running on a single board computer (udoo x86) with an Ethernet Smooth Stepper plus an MB3 breakout board, it's a great little control setup and I'll have to get pictures in the cabinet at some point.