[peteeng]

Yes the screw being of different length has a different frequency. Peter

[hanermo]

My counterpoint was/is that cutting plywood efficiently, needs a great deal of rigidity.

You are building a tool to cut 2.4 m sheets, of presumably some thickness like 12-15 mm++.
There is no doubt your design will work - to some extent.

But if your gantry bends 0.3 mm, theoretical, I think it might not handle carbide router bits efficiently in sizes 8 mm+ at efficient speeds.
So the cutting will progress at 4x slower speeds using 6 mm cutters, again 2x slower in MRR / work done.
== 8x slower work than it should be, for a semi-work advanced hobby design of some cost.

E:
Look at cutting times for 16 passes (4 shapes, back-forth), 2.4 M long sheets, say 4 m cut length (not rectangles nor artwork).
A single pass might take 12-15 minutes, say 192 minutes total.

This might be ok commercially, if the value you get from the sheet is high, you only need a few sheets per project, assy/design/finishing is the critical path etc.

But if You could increase the frame rigidity (linears),
drive rigidity (ballscrews),
spindle quality,
gantry rigidity (much bigger sections),
the 192 minutes of cut time should go down to about maybe 45 minutes.
(Still 2-5x slower than modern commercial routers at 700 ipm cut speeds at 40k rpm+--16 mins).

Biz (imho):
The router(cnc) commercial-guys sales pitch of cost/part in faster cycle times is mostly wrong.
Your goal here is right - so is the approach.
Do "something", do "better", for "your" market.

biz-case.
Even if you could cut the sheet in 1 minute using a uber-router (500k),
automated sheet loading/unloading,
unlimited power,
no issues with vast quantities of waste and dust,
unlimited space,
free materials racks (40k$),
and logistics equipment like stackers (4k$), cranes (10k$),
loading docks, etc.,
free financing for 200k in sheet goods,
free financing for the cnc uber tool,
You could/would probably sell more/less the same number of your products to your marketplace.

And selling other products to other customers, competing with those who already have all above, mostly paid off, is a fast way to ruin.

But..
If it costs You 40$ to make 1 sheet in 3.2 hours, just in your living allowance at half minimum wage,
then your bos or balance-system should be well profitable to make it reasonable, and very much depend on your needing to have total control of the process.
Otherwise, for 40$ you get your stuff cut in 30 mins at the other place.
And there may be a big value in clients seeing "your" industrial-sized cnc machines.

And slow may not matter if you do other stuff meanwhile.

[Zorbit]

Some good points by Hanermo, there are many occasions when I wish I had a small router spare. Running a 10hp vac pump, 10hp dust extraction and 15hp spindle is overkill most of the time, not forgetting the noise and dust produced by 20mm tools. Sadly I only have room for one router at present, but plans are being made.

[ger21]

I presume you mean the small distance between the bearings?

No, I think he's referring the the Z axis and spindle mount hanging so far below the bearings.

[davida1234]

I presume you mean the small distance between the bearings?

No, I think he's referring the the Z axis and spindle mount hanging so far below the bearings.

Both of them. They, IMO, have to be considered together.

The small spacing between the bearings AND the spindle being so far
below the bearings. It looks like you have a cantilever four or five times the
spacing between the bearings by the time you add the bit to the spindle.

IMO, this will result in quite a bit of deflection of the bit regardless of the
preload of bearings or any other reasonable measures.

[peteeng]

Hi All -Thats a design issue and needs to be resolved in Round 2. Hopefully Hanermo can clarify a target stiffness. A cantilever has the highest deflection so that's why we do FEA to understand and achieve the design requirements. Deflection is a function of the support structures not the bearings per se. So the support structures need to be dimensioned correctly. Peter

[peteeng]

So I reread Mr Bambergs thesis and he quotes some machine stiffness requirements. So I've picked a round figure target of 10N/um (or 0.001mm deflection per 10N force) as a start. I then changed the loadcase in my current model to a load of 100N which should deflect less then 0.01mm and we are at 0.02mm. So now I can start working through the parts again and get to the target stiffness. Getting closer to starting Round 2. I reran the model with the Z axis plate set to rigid ie its infinitely stiff. This reduced the deflection to 0.01mm. This infers that half the deflection is the Z plate and half the deflection is the rest. So I need to make the Z Plate much stiffer. Plus I need to include the bearing compliance soon.

We are in what's called the "Fuzzy Front End" or FFE of the project. But we are nearly ready to nail down a few things so R2 will be better specified and targeted.

Peter

[peteeng]

Hi All - Now we have a stiffness spec to chase (10N/um) I looked back at Scoot. Scoot was not designed to a stiffness spec. It was designed as a MVP (minimum viable product) to test various functional things and commercial arrangements. The philosophy was to build it to a minimum function (vs stiffness) and work it then solve the issues as they came up. Scoot is now at its development end.

I reran the solver with 100N longitudinal and 100N transverse loads. Its long stiffness was 0.25N/um and its transverse is 0.5N/um. Scoot cuts plywood, plastic and soft materials quite well holds better than 0.1mm as far as I can see. But now it's time to up the stiffness. So Maximus will be 40x stiffer than Scoot maybe more if we can. I may measure Scoots actual stiffness next week to see how it goes against the model. Peter

[peteeng]

Hi All - I have been doing some part sensitivity analysis on the current model. Whats showing up is a relationship between the tool plate stiffness and the gantry stiffness. I call the tool plate the part that carries the machine tool, its usually a flat thick piece of aluminium or steel. If the tool plate is not stiff and bends a bit this distorts the bearings and local area on the gantry and the gantry twists and bends. If the tool plate is over a certain stiffness the load pushes the toolplate directly backward and does not twist the gantry it only bends. Which is ideal as the tool stays vertical. So now I have to design a suitable tool plate that is Uber stiff. Currently I'm looking at a tubular sort of design or a deep U shape.

This process is also called a "form finding" process or exercise. We are finding the best form or shape for the task the part has to do. Peter

[peteeng]

Hi All - So after some tooing and froing from CAD to simulation, back and forth here's where we are. I've got to the target of 0.01mm for 100N load. There are two areas to address in the structure to make it better. See image attached I have balanced the gantry vs tool plate deflection and I'm happy with the general design trajectory of the structure. To get the inertia up on the tool plate it will have a bolt on front section. The next model will include bolts and compliant bearings so our machine stiffness will drop and we will figure out how to get it back. Peter

[peteeng]

Hi All - Getting ready to start Round 2. I have rebuilt the bearing into 3 pieces. I have modified the manufacturers STEP model of the bearing. The red ends will be set to a plastic or zero stiffness. They are there only as a geometric reference so I can check clearances to things like bolt holes and stops. The grey top will be steel like the real part and the maroon part will be tuned to deflect like the bearings. This will be a figure of around 5000MPa. I'll look at the bearing catalogue and run a test model to check this out next. Peter

[peteeng]

Ok I have built and tested the bearing model. I looked through the PMI and WON catalogues, couldn't find a stiffness figure. So I've picked some numbers from the Bamburg thesis and Star brand. 600N/um (0,001mm) is about right. So I built the model and used ABS (tried other plastics but too stiff or too soft, can set up a custom material) plastic data for the guide. I applied 60,000N which should compress 0.1mm and it compressed 0.07mm close enough for now. Easy to change in the future now its set up.

So Round 2
Cutting Envelope to be 2500-3000mm x700xZ200mm plus
Target tool stiffness 10N/um
Bearing stiffness 600N/um
Parts to be laser cut and bent
High Rail design, timber bed
Anything else I've missed? maybe target acceleration 0.1g (9800/10=980mm/s/s)

Round 2 shall start by rebuilding assemblies from scratch and adding motion parts. Moving right along... Peter

[ger21]

How often will you need 300mm of Z axis travel?

The long lever of the Z axis require considerably more stiffness than if the Z axis was only 150mm.

[peteeng]

Hi Ger - The last post specs Z200mm plus. I build moulds and cavities by the time I have vacuum clamps and bolsters and thick blanks I'll need 200mm. If I want to make Carbon Fibre gantry parts or Z axis parts I'll need more. Of course a long cantilever needs more support then a short one. If I was doing car body panels which is on the list I'll need 1000mm or 1500mm down the track and 5 axes. Now we have a 10N/um spec its an engineering exercise to achieve that whatever the Z axis length is. Everytime I open CNCZONE I get a blue elephant ad. That machine has much more than 150mm... This is a concept design exercise there's no use in achieving the usual or aiming at the ordinary. It will take a couple of hundred hours to work through this and unless I learn lots or achieve something different I may as well go and learn to play golf.

My machines also have an apron so people can do edge work like dowelling and dovetails and boxwork. This means ideally the tool has to be able to go under the bed height. Attached is a picture of Scoot fixing a bee frame using the apron.

Peter

[peteeng]

Hi All - Been looking at the z axis assembly and deciding whether the gutter design is the go. It's really stiff so is a good benchmark. The tube type design is complicated but should be much much stiffer yet it's only about 30% stiffer for double the weight. Seems the bottom edge of the plate is deflecting more then the global deflection. So its a local issue. So looks like the gutter is in and I'll add a cap to it as well to make it tubular. I'd like to spend a little time sorting the gantry foundations before I formally start Round 2. I'v up'ed the nominal load to 10000N so I get the difference in 0.01mm's Being a linear model this doesn't mean much. I do run non linear occasionally to check no NL effects are present. Peter

[peteeng]

Hi All - Been looking at the gantry foundations and the Z Axis bearings. I've sent off a request about some crossed linears that will save some parts (maybe). I have also consolidated the 3 parts for the foundations into one part which I'm happy about. I'll get this from my sketch book into CAD asap. If I use the crossed linears I still have to have a bearing plate to attach the Z axis drive to. But it will be more compact then using the current arrangement. Plus the load path is direct from the Z axis to the gantry vs through some other parts. Comes down to cost and availability. Peter

[peteeng]

Hi All - Did a little work on the gantry foundations or columns. I consolidated the 3 original parts into one. Still close to 0.1mm at 1000N deflection so all good. Would be good to use the bottom flange of the gantry to tie into the bearings somehow but no inspiration yet. The intent is to have the fwd bearing in line with the tool so there is no tool overhang. But the FE shows that the foundation plate bends a little in the middle so I should bring the bearing back a little. Then it overhangs.... Hmmm maybe a small overhang is the compromise. Peter

Hi Gerry - to revisit the Question on 300mm Z. If I wanted to make this gantry as a CF part I'd need 300mm plus Z to cut the cavity. I'd need a tool+collet 157mm long. Plus the mould would be 20mm thick so perhaps even more Z is needed. Or I have to split the mould and do it in two halves on 45degs or do them as flats, assemble hand detail and finish which is not the game. Peter

I have also asked the forum admin if they can allow MP4 movies to be uploaded. Videos are much better for this sort of thing.

[Zorbit]

I've also made plugs for dagger boards and rudders, from MDF, and I'm intending to do something similar again, hence my interest in this thread and your informative blog. The biggest problem I found when making the plugs ( on a 3 axis Multicam stepper motor machine at the time ) was that the job took many hours because of the small stepover and tight radius surfaces around the perimeter.

As most plug / mould making materials are quite easy to cut have you considered sacrificing some stiffness for speed and better access ?

[peteeng]

Hi Zorbit - Scoot was designed for speed and making foil plugs. It runs at 20m/min and can run faster but I usually don't. I don't see the need to relate stiffness with speed as such. You design & build what is needed. People ask me two things a lot 1) can I get a bigger Z please than 150mm which is sort of a std and 2) can it cut aluminium? Maximus is trying to answer those questions at a kit machine level. First thing is to design a structure that has the required stiffness then we move onto the drive system to get the required speed and accelerations. Thanks for watching. I think it's time to draw up a bench so the Round 1 structural picture will be complete.

In regard to foils a 4 axis machine (prefer 5 axis on the ToDo list) and use a flat bottomed bit is the next step vs a small stepover and say a 1/2" veining tool which is a good choice. I've attached an image of MDF wing plugs made on Scoot. Sting 600 is a 22ft flying trimaran being built by a friend. Cheers Peter

[Zorbit]

There are many things I'd like to ask, but I don't want to clutter up this thread which is so useful as an insight into the design process. Would you consider starting a more "general purpose" thread to cover broader design issues and some of the topics on your website ? If your time allows it of course, I quite understand if one thread is enough to keep up with.