[routalot]

Did I really count 6 axes?I don't know anything about Fusion,but I would be both amazed and impressed if it can create toolpaths for a machine of this complexity.Could I suggest creating,or downloading, a model to use as a trial job?Then create a roughing and a finishing toolpath and perhaps a pencil cut-with a different tool-to see what sort of code comes out.It would be best to establish the capabilities of the software at this early stage before dealing with the challenges of creating the machine.

[dimmaz88]

I'll create the toolpaths with fusion and output to ATP, then IRB CAM will create code for the machine.

Fusion can actually create code for industrial robots, so there's no problems there.

[peteeng]

Hi Dim - Your Z axis has been too small from the start. There is no footprint impact from making it bigger. You really should be concerned about UBER stiffness. Your designs are tall and thin and WILL vibrate, even from machine motion. Sure PS is soft and the tool will hardly know its there but palm trees move in the slightest of breezes, pyramids do not. Plus if you build a machine that is not stiff enough its very hard to fix this without a major redesign. I have been involved in upgrading a PS gantry cutting machine (3x3x2m) that had a wimply Z axis designed by people that should have known better. Keep at it.

The machine structure is its DNA, once set its nearly impossible to rectify easily. I've spent a lot of time evaluating UHPC (years) and I'm not yet convinced its the path for a "small" machine like yours. Aluminium sits right for your machine. AL is twice as stiff as UHPC, you can machine it on your router, no moulds needed and bolted connections are very efficient in AL. Its friction co-efficient is 0.6 so bolts transfer nearly all their preload. A bolted connection has the same damping factors as cast iron. The list goes on.... I would even consider F16 plywood for your first one. I've made 2 ply machines that can cut aluminium easily so I'd expect it to be very damp, easy to make and with some sheet aluminium laminated in correct places for it to be a very good machine...

Hi Routalot - Fusion has 100's of CAM configs including robots. Its very impressive in that dept. Peter

[peteeng]

Hi Dim - You have to take advantage of all available space. Make parts as big as the space is available. The column see attached. Geometry maximisation always wins. Straight long columns suffer from vibration and adding webs like that only creates weirder vibs.. .Peter

[routalot]

Hi Routalot - Fusion has 100's of CAM configs including robots. Its very impressive in that dept. Peter

Interesting to know.I remain concerned about actually generating the Gcode for the machine which is why I suggested working through an example and including a tool change.Different tool lengths normally require the CAM system or the controller to modify the motion to compensate for the difference and I was curious about how this might be accomplished.

The other aspect of the machine that will be challenging is accurately lining up all the axes.In particular,the turntable axis and the vertical axes of the machine.Next to that the co-location of the two axes of the head will be quite simple.

[peteeng]

Hi - There is a bit of a rigmarole in getting a 5 axis lined up. Especially with the CAM centre and the actual part "centre or zero" Dim will have to investigate that his physical configuration has a suitable post and how those two things are coordinated or synchronised. I'm about to go through this experience in Fusion soon as I'm designing a 5 axis machine. Some controllers (if you look up some of HAAS's 5 axis videos) have an automated offset correction. If this is not available, you have to ensure your CAM centre is in fact the part centre or intended centre. There's quite a bit more to setting up the 5 axis vs a 3 axis... as far as I can read at the moment that is. Peter

[dimmaz88]

If I'm honest, working out how to do the setup of the axes in the software is something I'll have to figure out.

What are we thinking about the new version? 500 x 500 mm square footprint to the column, with 200 x 200mm Z. I'm thinking steel for the column and alu for the Z (I'll switch to calling the axes the standard names.)

I think I'd infill the "ladders" with sheet steel do stiffen the whole thing.

Attachment 496214

Attachment 496216

[peteeng]

Hi Dim - A thin infll is not much use. If you use 45deg angled braces in your column vs horizontal braces, then it would be stiff in bending and torsion. If you intend to weld then it will distort. Usually with steel you weld then stress relieve then finish machine. If you make the column using open sections (C sections for instance ) then you can bolt it together. This potentially is the same stiffness as welded if you do it properly. Or use heavy C section aluminium for columns. I'd build it in F17 plywood will be very stiff enough for your application, quick to do and modify. To maximise stiffness you have to maximise geometry ie use the total perimeter boundary as a membrane. So ply suits this. Look up aircraft structures and see how they are rivetted and bolted togther. Peter

[dimmaz88]

I don't like the idea of plywood as we have very damp autumn/winters here. I just assume it will move too much?

Would you say a column of these proportions would be sufficient, if I use diagonal bracing instead of horizontal?

What would be the best linear rails and blocks for a structure of this size? I mean the tracks on the base..

Just to clarify axis labelling from now on...




BC head detail..

[peteeng]

Hi Dim - The column size is fine - Z is debatable. If the ply is coated in epoxy I find it to be very stable. I use formply which has a coating on the faces and I seal the edges with epoxy. I have a few sheets here that went through a flood and they came out fine and I used them on a machine... 20mm linears will be OK... Peter

edit - what size spindle do you want to use?

Re axes - The tilting head on the Z will be called the A axis as it revolves around the X axis and the rotary table will be called the B axis because it rotates around the Y axis. (ie X - A Y-B and Z-C)

[dimmaz88]

I'm going to use a 2.2kw spindle just to keep the weight down on the Z.

Edit: Would a 100 x 100 x 10mm steel section be stiffer than 200 x 200 x 6mm aluminium?

[peteeng]

Hi Dim - I'd use a 800W water cooled spindle, maybe a 1.5kW with Er16 collet so you can use 10mm tools... a 2.2kw is a beast for cutting PS. The 200 al section is twice as stiff as the 100mm steel section... see calc attached. Peter

edit - are you going to use burrs or bits for cutting the PS?? Here's a mostly plywood machine and it cuts aluminium easily... it has an 800W water cooled spindle...

[dimmaz88]

A mixture of burrs and cutters, the reason for the 2.2kw spindle is the ER20 collets. Anything smaller and I think I'll struggle to find big enough tooling.

Earlier you mentioned the trucks jamming if things aren't aligned properly, doesn't the plywood move even slightly to cause alignment issues?

[routalot]

I'm not keen on dismissing somebody's ambitions but I'm going to be less opaque.I think you have very little chance of achieving satisfactory alignment of al the elements of such a complex machine.Overlooking,for the moment arriving at a situation where the X,Y and Z axes are all guaranteed to be flat and perfectly perpendicular throughout the working envelope of the machine even though this alone is quite an undertaking,we have to consider the other three axes.They need to be precisely parallel to the planes and the B and C need to intersect precisely which can best be explained as the tool tip describing a perfect spherical motion.Challenging enough and I have been through this exercise on a more conventional 5 axis machine on more than one occasion.Just imagine how much more care it will take to align the turntable axis so that a rotary motion at a height of two metres is correctly located.Factor in cumulative errors in alignment and build,combined with reactions to stepper torque at full extension and it might become clearer why large machines are quite expensive if you seek accuracy.For something with a large element of artistic interpretation it might not matter so much but it might be a good idea to list the acceptable deviations for each type of motion and then try to engineer them out.Alternatively try for a more conventional 5 axis machine and buy an aftermarket 4the axis with matched tailstock to mount horizontally as it will be an order of magnitude easier.

[dimmaz88]

I agree it's quite the undertaking, and yes as it's for more artistic purposes I'll have a bigger margin for error.
The way I see it, if I can get each axis flat and the rails parallel then alignment should be straightforward enough.

I intend to build the BC head first as it's the most complex part, and work backwards from there.

[peteeng]

Hi Dim - Yes the ER20 is attractive for the collet size. This means you have a serious weight at the end of your Z axis plus the mechanics for the A axis.... Do you have FE? your using SketchUp? I have not found the ply or MDF to move resulting in cars jamming. I have made pressed steel parts that resulted in jamming. I use medium preload or high preload cars. You will need the same to remove hysteresis from the motion. Being an unusual machine config means it will have challenges... maybe that's a good reason o build a ply prototype, sort the issues then build the "real" one. The second one is always 10x better than the first one. Or the ply works out fine and your done. 17mm F17 ply is equivalent to 10mm thick alum plate in bending so its pretty stiff. I've just finished doing some evaluations here's the numbers. I use 2mm aluminium sheet laminated on plywood (both sides) to make very stiff parts. I keep coming back to ply as its easy to work and the numbers always stack up well. And cost is less than metals. I'll stop talking about ply now I've harped on it long enough... I prefer to make carbon fibre parts but that's lots of $$$ Peter

edit - do not build anything until you have resolved the design on paper to your best satisfaction. All parts of the machine are connected and once you make something it can't be changed easily and it affects the entire machine. CAD is the space to do all the development in... You had also better put time into understanding the post processor. If you build this then find out you have to pay someone to write a custom post it will cost much more than the machine! 5 axis posts are not simple things....

[routalot]

I agree with everything peteeng has posted.Except for one thing,you need a six axis post processor to be sure the turntable will follow instructions.It will be a bit harder to generate that a five axis equivalent.

I have been struck by the thought that in essence the concept of this machine is not far removed from a Portal machine that is missing one side and which has been rotated so it's other side is on the ground.

[dimmaz88]

Routalot, this was how I was thinking when coming up with the idea. I just imagined tipping my large format router on it's end and adding the BC head.

Peter, thank you for the maths. However, I don't understand it ??

I assume you're comparing the stiffness of different materials..

I've had the offer of a company to write the post processor for me as a bit of a test for them, so hopefully that side of things is taken care of.

I'm leaning towards steel at the moment, as I'm comfortable working with it and it's fairly cheap for what I want.

I'll start a model soon, although I'm ripping out my kitchen tomorrow and we're expecting a baby in 3 weeks!!!

[peteeng]

Hi Dim - Well from our point of view the machine is more important than kitchens and babies

A sections rigidity is a function of its geometry, the bigger it is the "stiffer" it is. Plus the stiffer the material the more rigid it is. Steel has a stiffness of E=200GPa whereas aluminium is E=70GPa. So steel is 3x stiffer then Al as a material. So the same size beam in st and al means the steel one is 3x stiffer. The sectional inertia of a beam is calculated and its a function of its size. This is the "I" value. When comparing different materials beams the rigidity R is calculated as the material stiffness x the geometric inertia R =ExI so I calculated the inertia (the resistance to bending) of the two sections and multiplied the modulus with that to create a big number for each section. Then since the alum number was twice the steel number the alum beam is twice as stiff as the steel beam in this case.

There are two modules to a post:
1) The actual geometric transformations required that create the mechanical movements. These can be very complex in a 5 axis machine as there are mechanical locking points called inflections. If the machine moves to this position it locks and cannot be unlocked. So the code has to look forward and not go there!! Because you have an A head this may not be a biggy.
2) a post creates code that provides things like air on off, ATC functions, start up and shut down info... It also brings the code into line with the physical position of the machine. If the people think that a post is 2) then all good but you still need 1) if they think 1) is a short exercise then forget it!! find a commecial solution, someone has spent years and lots and lots of $$$ to get there, Look at Fusion NOW and figure out if it can do it. Last time I researched 5 axis software (other than fusion) it all started at around $12000USD for basic capability... This is a very big rabbit hole. 3 axis is a cake walk, 5 axis is technically difficult and costly. Sort it out before you make anything.... the software side can kill the project. Look at linuxcnc (free) I think it has some multiaxis capability. It does have robot capability last time I looked....

even fusion 5 axis is an extra $2000 per year so you have to make it worthwhile to own....

[dimmaz88]

Thanks Peter, I'll try and squeeze in some CAD around babies and DIY.

It's amazing how section size trumps material stiffness in such a big way, I guess that's how engineered wood/OSB I-beams work.

I'm happy to pay for fusion as they've been good enough to give me 3 years worth of trial membership, and it can be paid for monthly which is less of a commitment.