New CNC of 2000x600x200mm High Precision

[Zorbit]

I've never seen a composite moulding that has anywhere near the accuracy numbers in this thread. Not even close.

There isn't much point paying a fortune to have a 1000mm mould made to +- 0.005mm if the part moves 0.5mm as soon as it's demoulded.

[Jim Dawson]

Hi GM - airs density is ~1.2kg/m3 so I imagine you mean 120 or 1200kg/m3 Merry Christmas. Your local engineering supplies will have machine tools, find out the reps contact and get in touch with them... Peter

Aerogel

[gmoreno_cnc]

Sorry, you were right about the density of the PU board, I was referring to 1.200 kg/m3

In regards to the high accuracy needed, I still will keep at least +/- 0.01mm due to the fact that the piece to be produced is a high competitive sailplane, in which the wing Airfoils need to be respected as much as possible.

Many thanks

I've never seen a composite moulding that has anywhere near the accuracy numbers in this thread. Not even close.

There isn't much point paying a fortune to have a 1000mm mould made to +- 0.005mm if the part moves 0.5mm as soon as it's demoulded.

[peteeng]

Hi GM - To build anything to better then 0.01mm requires toolmaking precision, which means finding a tool shop that has machines verified at 0.001mm for the entire machine bed. Even though the machinery can make things to 0.01mm does not mean the object will be built to that tolerance. To ascertain your build geometry you will need to find someone with a CMM (co-ordinate measuring machine) or scanner or optical device that can verify your mould geometry before you mould something from it and prefer before the mould is release from the machine so it can be adjusted once the actual geometry is compared to the intended geometry. You need to discuss your needs and intent with a good machine shop to figure out the required path and cost of such precision.

Even your CAD model will need verification. The way CAD models loft surfaces or build surfaces can produce wavy geometry at some geometry scale level. Creating A class surfaces with curvature continuity is not an easy thing. So you need to investigate that further as well. The first kayak I modelled and had machine cut was embarrassing in terms of fairness. Even though I had looked at the surface using zebra displays, curvature analysis and everything I could do with two CAD systems as soon as it came off the machine we could all see the waves in that surface. It had to be hand finished. So do not believe what you see on the CAD screen. The render of the screen is created/designed to look good, it is not the actual mathematically intended surface!! and then you export that and import it again so there are two more levels of possible error to get the CAM...So more research needed in that area as well. Peter

[Zorbit]

I think you'll be disappointed if you are trying for that accuracy on your wing. Don't be surprised if your part moves 100 times more than your tolerance when you demould it, and in use it will move thousands of times more.

[routalot]

I don't like to be the bearer of bad news,but every polyurethane tooling block I have used is subject to movement as soon as you cut it due to the internal stresses from being cast being released. Back in the 80's it was quite normal for a 1.5m board to produce two banana shaped sections if you split it in half with a gap in the middle of 15mm or so. In the early 2000's I used a better board that only had about a quarter as much movement and the current generation is better again. This doesn't mean the situation doesn't occur,it has just become a smaller dimension and the same thing happens if you machine a slab of tooling block to a 3D shape,unless it is very substantial.The reality is that the tool may follow exactly the trajectory you need but he block itself can move and produce a different result and then when you have a mould in the workshop such matters as local reinforcement to prevent it sagging can affect the rate at which it can react locally to temperature variations and further deviation from the theoretical ideal takes place.

The points raised by peteeng illustrate some of the other factors that occur and even if you set the deviation in the CAM system to 0.001mm and accept the vastly longer toolpath generation time that occurs,other errors can creep in. Incidentally ,the best way I have found to view fairness on screen is to use the environment mapping feature with a mirror background.Not totally perfect,but a bit more helpful to my eyes than zebra stripes.One final point,will the finished moulding be used in its ex-mold state or will a finish be applied?

[peteeng]

Hi Routalot - Yes I agree environmental mapping with a mirror or a true sphere is very good. But what I wanted to say is that the actual geometry modelled and the rendering mesh you see are different. What you are actually "seeing" is a mesh superimposed onto the mathematical surface which is an approximation of that surface. If you are using a NURBS modeller or a NURBS surface ensure the tolerance settings of the NURBS is 10x better then the tolerance you need. A NURBS surface is a polynominal that does not exist as a smooth entity, it is a naturally wavy object that exists between tolerance bands. Its not really a single surface hard to explain but to achieve a fair, smooth surface at tight tolerance requires understanding its nature and how it is built and how it is exported/imported to a CAM system then how is it actually descretised by CAM and cut. Don't want to make it sound difficult, its all in the manuals. But having seen surfaces come out wavy, or jittery on some jobs especially with geometry from clients with no depth of knowledge on surfaces shows me its not a trivial thing to get good surfaces especially aerofoils that need to be functionally correct vs appearance correct such as in car panels. Peter