-
1 Attachment(s)
Re: WillMill - HSM Benchtop Composite Mill
Hi William and others - I was watching a video of the Taiwan Machine show from 2015 and saw this, a CF gantry on a Hiwin machine. This particular gantry is a cantilever. I'm sure we shall see more of these around. Peter
https://www.youtube.com/watch?v=03L-ayrKEYw
I do like air bearings and intend to make a machine with these soon.
-
1 Attachment(s)
Re: WillMill - HSM Benchtop Composite Mill
Hi William - another article for you, but settle the material!! . One papers machine is about 30N/um the other is about 150N/um measured on VMC sorry that papers too big to upload.... Peter
-
Re: WillMill - HSM Benchtop Composite Mill
Hi Peter - Things have been moving a bit slow here, lot's of things going on right now. Yes regarding materials, the plan is to use woven carbon fiber for the skin of the components, and then use carbon fiber swarf next to the skin to maximize the potential stiffness. Aluminium swarf will be used for the core as a filler. The amount of carbon fiber used will depend on how much I can get my hands on. I'll focus on getting my supply of materials sorted today.
Update: The revised z-axis is finished. It's a composite clamp for the spindle that mount's to a cast iron plate with a bolted connection. In this way the head can be trammed with shim stock or other means of "adjustments". I have though of a second configuration for the z-axis. I will post some pictures and do a little comparison of the two configurations.
I received my 4.8 CFM vacuum pump with the mail today. I miss being in the workshop so I have decided to make my own vacuum chamber as a little excuse to getting my hands dirty again. I'll just mount some vacuum hose fittings on a thick acrylic lid and put on some silicone sealing paste. Then vacuum lid can the go on just about every pot. I'll do the same for a resin trap. This also saves me a bit of money ;-)
The Mill Stiffness PDF was a great read!
Regards William
-
Re: WillMill - HSM Benchtop Composite Mill
Fwiw..
I have actually built a mill, really a VMC, for milling steel, somewhat efficiently.
2.4 x 2 x 1.5 meters in size, 1600 mm table, 1200 mm movement, 32 mm ballscrews.
When I stood on top of the spindle, 85 kgf, originally it deflected 0.28 mm.
After 5 revisions and 1000 kg more in tool steel and 400 bolts, the deflection was around 0.12 mm (v3,v4) before the latest revisions.
At the moment I needed to take off the front beam, since the auto toolchanger was too large, and I need to put a spacer in to accomodate it.
Once the beam is back in place, I can measure current deflection.
Anything around 0.05 mm will be great.
-
Re: WillMill - HSM Benchtop Composite Mill
Hi Hanermo - That sounds like an beefy machine. 0.12 mm deflection is more than what I would have expected for a machine that can mill still "somewhat efficiently" as you state. I'd love to see a clip of the machine milling some steel.
On another note, that means that my machine is way more stiff than it has to be. This is of course not bad news, it means I can slack a little with the composite the machine will be made of. It doesn't has to be as stiff, which potentially means I don't have to use any carbon in my composite mix... I'll stick to the plan for now: Aluminium swarf, carbon strands and woven carbon fiber skin.
Thank you for the numbers, good to know.
-
Re: WillMill - HSM Benchtop Composite Mill
Hi William
1) Resin trap - You need to select a suitable container to go inside your vacuum pot. eg a icecream container, a std 10L bucket or something that is readily available and a throw away. I use std 10L buckets in mine. I made my trap by laminating fibreglass around a large plastic bucket. If resin flows into your trap (and being a learner it will) its a pain to clean it so by having the resin flow into a container you just throw that away. Plus make it big enough to get your hand in there. You need to wax it very well to make resin removal easy. Do not seal the pot with silicon. You need to be able to get the lid off and on easily. Use some foam or plastic sheet and cut a seal. Test your containers, vacuum will collapse thin or non circular objects.
2) You have not designed the entire machine yet. The entire machine will deflect quite a bit more then you think once it is fully modelled. A machine can never be too stiff it always works out less stiff in reality then on paper for various reasons.
3) You may plan to use CF cloth for the outside but do you have a source yet? and why go to all this trouble yet use a cast iron plate for the Z Axis? What is CF swarf? If you have a suitable supply of CF then you can make a slab of CF for the Z axis. This is then in-line with your objectives....Regards Peter
Re: woven cloth vs stitched cloth. Woven cloth has non-straight fibres (kinked) and therefore is less stiff (at least half the stiffness) then stitched cloth which has straight fibres. So try to get stitched cloth offcuts not woven. If you get woven then place this internally not near the surface. Plain weave is often used for its appearance on the surface as this is the "look" but structurally weaves are very inefficient.
-
1 Attachment(s)
Re: WillMill - HSM Benchtop Composite Mill
Hi Peter
1) Good idea, I intended selecting some kind of pot that would be bigger than a throw away container. Waxing the mixing cups or container for easy clean up, that will definitely make my life a little easier. Great tip Peter.
The silicone will not make a permanent seal between the lid and the pot, it will act as a gasket. There are special types of silicone for this purpose.
2) You're absolutely right, the machine will deflect more than what it does right now, and even more in reality. And yes, the stiffer the better, but if the end result is overkill, then I don't necessarily need to add carbon strands into the composite (unless I get a free source to these)...
3) The cast iron plate is not what gives the z-axis most of it's stiffness. That would be the clamp. The cast iron plate will be precision milled with a 0.01 mm flatness and the linear rails will be mounted to it. I'm not sure just how precise my mill can mill parts flat, since I don't have a surface plate to verify it. I'll buy a straight edge and test this to see weather or not I could mill the plate flat myself, but until then I'll use a flat milled cast iron plate to mount the linear rails of the z-axis to. By carbon fiber swarf I mean carbon fiber strands. Some fiber strands mixed with cutoffs just looks a bit like swarf to me.
Thank you for explaining the difference/properties between woven and stitched cloth. I will use stitched cloth.
I've contacted every company in Denmark somewhat near me, that delivers or uses carbon fiber, for carbon cutoffs.
(the z-axis will also have two screws for the clamp at the top)
-
Re: WillMill - HSM Benchtop Composite Mill
Hi Will- The composite industry talks about short fibres say <50mm milled fibres <5mm long fibres all the rest. Woven cloth, stitched cloth, multiaxial cloth. I still don't see why you want to use cast iron but so be it. Peter
-
Re: WillMill - HSM Benchtop Composite Mill
Hi Peter I have not found any source of flat milled composite slabs. But you right, and it annoys me a little too. I'll make the saddle and z-axis "backplate" of composites too then. I hope my mill will make them flat enough, but if they end up with a flatness of 0.03 mm, then so be it.
-
Re: WillMill - HSM Benchtop Composite Mill
Hi Will - Given your resources you can only do what you can do with your mill. So probably time to give your mill an overhaul in prep for the new mill build. Sharpen your tools before you do a job is always a good plan. In your objectives it says "experience casting composites" so CI is in conflict with that objective. Plus this is a design exercise. There is no value in making a compromise in this project. You do the best you can on paper and at the end you may have to make compromises due to practical or economic reasons. But right now there is no point in compromising. Design the best composite mill you can then figure out how to realise it.
The practical side of infusion of parts is still a huge learning curve and your months away from that unless you start parallel projects... such as making a slab of G10... who will be your composite supplier of wax, fabric, vac bags, tacky tape, breather cloth etc etc??
Peter
https://www.professionalplastics.com/contact they have a european distributor. G10 or FR4 should be readily available but you can make it yourself now you have a vac pump.
-
2 Attachment(s)
Re: WillMill - HSM Benchtop Composite Mill
Hi Will and others - I have been designing a small mill and have come back to looking at gantries. As usual the issue is square gantries lozenge (change to a diamond shape ie shear deflection). In the Maximus thread I determined that the corners needed thickening or to use a diagonal brace. Others have filled their tubes with EG, placed internal webs etc. This has pointed to the fact that gantries are shear dominant and are not strictly beams in flexure. This is called short beam behavior. Short beam theory is very different from long beam theory. FE takes this into account mostly but large shear deflection is best looked at via non linear solvers not linear solvers. But that's getting technical.
So I was looking at a 10kW laser video yesterday and noticed the gantry was triangular (flat bottomed isosceles, point upward) and this brought me back to the brace I had used in Maximus. But what if I just left off the back square bit so the loads went directly to the bearing corners vs around the back of the square section? (called shear flow or shear lag) So I modelled a simple gantry this morning SQ vs TRi and tri wins hands down. SQ 0.62mm TRI 0.45mm nearly 30% stiffer. The question is how do I build this shape in aluminium plate???? Always a hurdle to new things perhaps its time to make a mould...
The gantry modelled is aluminium 6mm thick 1100mm wide. The tool plate is 200mm wide and 25mm thick with a 200mm cantilver. The SHS gantry is 150x150x6mm The tri is 150mm high and 150mm wide and 6mm thick. The tri is lighter and stiffer, win win...
Peter
Will maybe do something similar in F360 and confirm the result?
-
3 Attachment(s)
Re: WillMill - HSM Benchtop Composite Mill
Hi Peter - Well my resources doesn't have to be a limiting factor... The idea of buying a cast iron plate was to find a supplier that could also grind it or make it flat with a precision mill. Your're right Peter, I need to remember that this is a design exercise, and that building the machine will a project for another time. Designing the machine already requires a few projects, like measuring the stiffness of the composites I'll be able to build in my garage (I'm really exited about testing my own composite one day!). I'll design the dream machine, and make compromises when the time comes to build the machine. I've spent most of the day learning about some of the imperial threads used for sealing connections to connect fittings and hoses to. I've also made a little list of things to buy and contacted some sellers on Alibaba to buy a whole bag of barb tee fittings... I've mostly been looking at consumer suppliers, but when i took a look at professionalplastics.com I was blown away. What a wide selection of materials. I just contacted them regarding potential suppliers or warehouses in EU.
Triangles, of course! Why have I never considered a triangle? Great idea Peter. I can confirm the results. I made a quick search for carbon triangles, and found this interesting trapezoidal tube which left me wondering how strong it would be compared to a square. It turns out this shape has about 30% less deflection compared to a square. http://www.dodge.com.tw/rw_products_6e6086ab.htm
Profile: 50x80mm - 400mm long.
Load: 700N in the middle of the front face 40 mm below the lowest point of the gantry.
Weight: Both profiles weight the same +- 100 g.
Square deflection: 0.004931 mm
trapezoidal deflection: 0.003433 mm
Interesting... If I increase the beam length from 400mm to 700mm the trapezoidal profile is only around 20% stronger instead of being 30% stronger. When I think about the beding forces it makes sense. The trapezoidal shape that I stimulated is not perfected for it's job, so it could potentially be stronger... Trying to think of any scenarios where a trapezoidal tube would be practical or useful compared to other shapes, but I can only think of the "cool" aesthetic of the angled faces.
A right triangle will be more practical, but I was just wondering how much of a difference the trapezoidal shape would have compared to a triangle.
Regards William
-
1 Attachment(s)
Re: WillMill - HSM Benchtop Composite Mill
Hello everyone - As mentioned the other day I contacted Jingjiang City Jianken High-Speed Electricmotor Co., Ltd. about a CAD model for their JGD-85 2.2 kW ER-20 spindle. After I few messages I had gotten my hands on a drawing of the spindle with some information that I thought someone could benefit from - so I'll post a picture of the drawing here.
The spindle has a housing that protects and holds everything together. I was wondering what the wall thickness of the housing was for FEA purposes, and the drawing states that the wall of the housing has a thickness of 2.5 mm, but this doesn't seem right to me?? Does typical "consumer" HF spindles only have a wall thickness of 2.5 mm? - or has something gone wrong in their engineering department? With cutting loads and vibrations I would have though that the spindle housing would be at least be 4 mm thick.
- William
-
Re: WillMill - HSM Benchtop Composite Mill
Hi William - a triangle is a fixed geometry. It can't change shape. So the more the trap moves to being a rombus the more compliant it becomes. Even the trap is more compliant then a tri. Even in FE we use square mesh to model various things because it can lozenge and twist. If we use triangular mesh it can "lock", say if we are doing a plastic flow problem and we can't get to a solution. Using square mesh allows the material to flow and allows the solver to keep solving... Peter
-
Re: WillMill - HSM Benchtop Composite Mill
Hi Will - You have reverted to saying "stronger" vs correctly saying stiffer or more rigid. Peter
-
Re: WillMill - HSM Benchtop Composite Mill
Hi Will - When you mould something it has to have "draft" this allows the part to be removed from the mould. So th draft will make traps natural. Usually use 2-3degs on a mould but the more the better. Peter.
-
Re: WillMill - HSM Benchtop Composite Mill
Hi Will - A little bit on infusion for you. Peter
https://www.youtube.com/watch?v=3uhR4jdFiS4
-
1 Attachment(s)
Re: WillMill - HSM Benchtop Composite Mill
Hi Peter, thank you for the replies. I'm aware of draft angles, all the relevant surfaces will have draft angles on the final part. I've been researching again and reading some threads on the forum, I've learned a lot more and gained more confidence in the art of composites, although I still don't have any real world experience (that will come soon). Peter you're right I have reverted to saying strength instead of stiffness without even knowing it. Thank you for correcting me.
Great video Peter, I saw it just a couple days ago haha. I like Easy Composites, they make som great content and they have a nice web shop. I currently have a few items in a shopping cart on their website.
This is a good read for anyone learning about building machine parts with carbon fiber.
https://www.cnczone.com/forums/diy-c...40246-cnc.html
CompoTech makes a variety of components from carbon fiber and composites. They have also designed a carbon fiber machine and written a few whitepapers, read about it here:
https://compotech.com/about-us/downloads/
Design update:
Gantry have been redesigned to a height of 180 mm.
I'm currently working on a simple way to simulate carriages for linear rails - I might just have come up with something genius, I'll share it soon. When the linear rails responds realistically to the forces on the machine, we can do a full gantry simulation with the z-axis and get a better idea of the machines stiffness. From here we can decide whether or not the the height and z-axis travel can be increased.
- William
-
Re: WillMill - HSM Benchtop Composite Mill
stiffness measurements cnc mk3
router stiffness measurement. built from extrusions 1.25N/um.. Peter
-
1 Attachment(s)
Re: WillMill - HSM Benchtop Composite Mill
Hi Will and Others - I have been looking at the placement of bearings on square gantries. Seems the diagonal approach is the stiffest. This is 150x150x6mm sq section with 1000N at 250mm. Not by much but it is stiffer and it also has advantages with bearing & drive placement. Cheers Peter
-
Re: WillMill - HSM Benchtop Composite Mill
Quote:
Originally Posted by
peteeng
Hi William and others - I was watching a video of the Taiwan Machine show from 2015 and saw this, a CF gantry on a Hiwin machine. This particular gantry is a cantilever. I'm sure we shall see more of these around. Peter
I do like air bearings and intend to make a machine with these soon.
Carbon fiber has been quite common for laser machines parts for some years, light weight high speed low inertia gantry's carbon fiber is a perfect fit for machines like this
-
2 Attachment(s)
Re: WillMill - HSM Benchtop Composite Mill
Thank you for sending a link to the thread on stiffness measurements Peter, that was insightful.
Regarding the arrangement of the guideways Peter, could the increased stiffness potentially be due to the extra material? - and not the placement of guideways. Some bearings/cars are stiffer in certain directions, I guess this would be a scenario where one could benefit from certain placements of the guideways.
*I will refer to the sliding block on linear guideways as a bearing.
Regarding the machines design there is not much to update you guys on, I have however found a great way to simulate linear guideways. In order to simulate a bearing so that it deflects realistically it's easiest to get your hands on a data sheet for deflection values of specific loads for the type and model you intend to use. At the moment the choice of guideway and linear rail will be a 20 mm RG bearing from Hiwin. The RG series has roller bearings instead of ball bearings. I have chosen the RG series, because It seems like it will have less vibrations as it's in contact with a bigger area of the guideway compared to ball bearings. The RG-series from Hiwin also has other properties like increased rigidity. I have contacted Hiwin and distributors with no luck, I couldn't get my hands on a data sheet with deflection values for the RG series. Peter has sent me a document from Hiwin with some deflection values for different bearings in the EG series, so my target deflection values will be based on numbers from this document. I'm not sure if I'm allowed to share this document, as it says it's confidential with a red square around it. I'll ask Hiwin if I'm allowed to share it here before uploading, as I'd like to stay out of trouble.
I can imagine four ways to simulate bearings for linear guideways.
1. Full simulation with every part and element of the bearing with correctly assigned material properties.
2. Constrained simulation of simplified bearing (use simulation tools like contacts and other constraints in the used FEA-package to achieve deflections according to the manufactures data)
3. Simulation of an simplified bearing geometry that is modeled to behave according to the manufactures deflection values.
4. Split a simplified bearing into two different bodies and edit their individual material properties in a way that it will deflect according to the manufactures data. (Peters idea)
* Solution 2-4 could be mixed together
Now these options give me a few different ways to achieve the desired results. I'm going to rule out option 1, as I'd like to keep things simple, and because it's unnecessarily complex. Option 2. however will depend on the FEA software available. In Fusion 360 a stiffness factor can be applied to different contacts. There's different contact types for different purposes... A contact tells the computer how two surfaces should interact. I made a "bond" test in Fusion 360's simulation workspace simulating a contact type with a stiffness factor of 0,01 and 100. I found that the stiffness factor only affects the "rough" type contact. This means the the rough contact is the only type of contact of which the stiffness can edited. I made a simple bearing the can be found in the PDF document below. The top surface of the guideway had a bonded contact to the bearing, and the sides of the rail had a rough contact to the beaning so that the deflection from the front load could be controlled by the stiffness factor. I found stiffness factor that had equal deflection for front and top loads, but I couldn't figure out a way to control the deflection for loads applied to the side of the bearing, so this solution was neglected. - By editing the geometry that is in contact with the top surface of the guideway, one might be able to find a solution that has equal deflection in all three directions.
I have tested multiple designs and mixed different materials, stiffness factors for rough contacts while simultaneously editing the geometry in an attempt to make the deflection more controllable. In total there has been a total 368 iterations and even more simulations in an pursuit to come op with the simplest way to simulate bearings for guideways. The solution is 100% geometrical and deflects equally in all directions. The scale of the deflection can be controlled by simply editing the stiffness of the bearing material. The ratio between the deflection of each of the forces can also be controlled relatively easily (not all cars deflect equally in all directions). The beauty of having a geometrically controlled bearing deflection is the fact that you don't have to spend time applying different types of contacts and stiffness factors for each bearing in the simulation - This could be time consuming and have potential for human errors, as there would be a lot of contact's to control and stiffness factors to worry about. Without the use of contacts (all contacts bonded - by default) the bearing can simply be inserted into the assembly and will deflect the way it should.
The solution, the finished bearing design, can be seen in the picture below. It consists of "springs" that basically are thin and scew walls that can deflect. I've put two pairs of these springs on the bearing and oriented them 90 degrees apart, so that the radial and lateral deflection can be somewhat controlled. The thinner the scew walls are, the more the bearing will deflect under a load from the top surface. The scale of the deflection can be adjusted by editing the stiffness of the material applied to the bearing being simulated. The dimension of the bearing that I have used are based on a 20mm RG bearing from Hiwin without the dust caps. The simulations of this design that can be seen below does not have the correct deflection values, they should be closer to 0.03 mm for the type of bearing that I'll be getting. More on this in next weeks update. The simulations in the picture below all use a force of 3000 N applied to the whole surface.
The design will not reflect the deflections of the data sheet from Hiwin 100% but it's close. If you'd like to try and find a better solution than what I have found, I encourage you to do so and post the solution here. Simulate -> results -> edit stiffness factor/dimensions/material properties --> repeat.
I'll upload the file next week along with other details. I'm working on a website that that I'd like the file to be uploaded to, that's why I'm not uploading it here.
-
Re: WillMill - HSM Benchtop Composite Mill
Hi William - The extra material should decrease the stiffness not increase it. I think its because the local load is less as the bearings are further apart (1.4x apart if the tube is square) and the loads are applied to the section more symmetrically leading to less shear deflection. You can publish the hiwin data I have in other spots in the forum and the material was supplied not under a non disclosure agreement so I expect its public knowledge.
In your carriage model can the car still slide? allowing it to slide is more important then getting the local stiffness correct. The sliding produces less stiffness (zero stiffness in the pitch dirn) and if so if the load is "up" how does it stay connected to the rail? or is this a spring connection? Regards Peter
-
3 Attachment(s)
Re: WillMill - HSM Benchtop Composite Mill
Morning William and lurkers - William (and others of course) the axial stiffness of an element is FL/EA. See attached. So you can quickly calculate the required dummy material stiffness for the car. In this case I took the EG15SA car stiffness. It deflects 0.015mm at 2000N in the heave direction. So I made a 40x15x1mm thick pad at the car rail interface. The manual calc says the material has to be 222MPa to do this. Then I made an FE model and it predicted 0.012mm. This is because the FE model includes the poisson's ratio effect so it gets a bit stiffer as it stretches as it's contained. But you can back the E off or consider it close enough. You then make patches in the transverse dirn and your done. In this way you can define sliding surfaces as well and the bearing will behave nicely. Regards Peter
I have attached the bearing data so you don't have to fret about lawyers turning up on your doorstep...:) time for you to reveal a mill model..
Also remember that in the early rounds of the design process this condition is one of the very minor contributions to deflection, sliding is more contributory for instance. So you maybe over complicating things early on. Once you have honed down the design to near final then it maybe worth doing this to see how much it contributes. For instance you have bonded connections vs bolted and fasteners will be very compliant vs the bearings.
-
Re: WillMill - HSM Benchtop Composite Mill
Hi Will - This is the stiffest mill I have found. https://www.mmsonline.com/articles/a...e-for-titanium its 350N/um.... Peter