Designing a new router called Brevis-HD

[peteeng]

Hi Mactec - I don't plan on machining rail lands as the buyer will not pay for it. This machine is low cost aimed at Makers, Hobbyists and DIYers. Its at the same level as an aluminium extrusion machine and they are not machined. I use flex grade cable and catenaries no cable carrier but that is up to the Maker. Its a kit, I supply the structural and mechanical parts plus a shopping list and specs for the rest. I don't disagree with much of what you say in principle, its just that I'm not in the Rolls Royce part of the industry. Brevis-HD is made to a $$$ value and the design tries to squeeze out as much user value as possible from those cost $$$. Perhaps the next machine I design on the forum will be a carbon fibre Rolls Royce one just for the fun of it....cheers Peter

[peteeng]

Hi All - Due to the C-19 virus the place I usually go to today is shut down so I'm beavering on Brevis again. I hit a speed bump with Z axis width. The combination of flaring allowance and flanged bearings made the z axis quite wide. So I had to rejig the design to square bearings. I usually use flanged as they provide a bigger base for bolting to. The other element in the mix was using a std nut housing vs making one which is cheaper. So now after a bit of rework here it is again. I found downloading the Hiwin models very easy. They certainly provide great info and support. Now I can get back to the saddle. Cheers Peter

[peteeng]

Hi MD - About internal stress and "messing" with the build. The first time I built a machine this way some of the parts "sprung" as assembled they needed tweaking and some I had to take back to be reset. This was due to little inaccuracies in bends, angles and fits. Now I am familiar with how the press operators work I have lots of notes in the dwgs about what's important in each part. Last one went together very sweetly so I'm getting to know the medium.

The current modelling process is using the development model as a reference, I'm building the production model from the development parts and they become production parts. Each part is adjusted for the hole clearances and flange allowances. I've hit major space issues with the saddle so will silver solder that assembly but that's good as it will be very stiff as it needs to be.. Now I'm up to the gantry so over the main humps. Going to square bearings has saved a bit of room so all good. Cheers Peter

[machinedude]

same thing happens when you build on the fly like i do. you think you got things nailed down but then you have to make minor adjustments and rework parts for the build. doing what your doing in CAD is good to get the obvious sorted out in the beginning. but i have found things present themselves during the build process too.( for me at least) so i don't spend near the time you do on that side of things. the whole process is time consuming so you have to have interest in the matter to go down this path for me it keeps my mind active and i enjoy doing everything myself and dealing with the issues that present themselves and finding solutions to the problems. building something like this is probably 10 times better in the learning department that could be bought with an education from a school in my opinion

[peteeng]

Hi All - closing shop for the day. Got the Z axis and saddle sorted, now onto the gantry details. Very happy with saddle stiffest one yet, the middle is 15mm thick of steel. It will be a laminated steel construction. The global stiffness dropped from 10N/um to about 4N/um. Most of this seems to be in the gantry torquing. As I'm about to dig through that I'll sort that as well. It needs shear webs at its ends firstly. But onward. Do the 1610 ballscrew drive detail dwgs and all motion bits can be sent out to quote. Cheers Peter

[peteeng]

Hi All - Did some structural loop testing and the Z axis to its bearings is 7.7N/um then the Z Axis plus saddle to the saddle bearings is 3.5N/um and then all the way to the end of the gantry is 1.7N/um. So in % terms the Z axis is 21% of the stiffness, up to the saddle bearings is 47% and the gantry is 100%. So the gantry is 53% of the machine deflection so this needs to be attended to as I detail it. Seems it has to be thicker....Plus I pushed it and pulled it and the deflection was the same so its symmetrical, all good. Peter

[Sterob]

Hi All - Did some structural loop testing and the Z axis to its bearings is 7.7N/um then the Z Axis plus saddle to the saddle bearings is 3.5N/um and then all the way to the end of the gantry is 1.7N/um. So in % terms the Z axis is 21% of the stiffness, up to the saddle bearings is 47% and the gantry is 100%. So the gantry is 53% of the machine deflection so this needs to be attended to as I detail it. Seems it has to be thicker....Plus I pushed it and pulled it and the deflection was the same so its symmetrical, all good. Peter

Excuse my ignorance...For the z axis, I assume that means you are gettting 1um of deflection for every 7.7Newtons of force?

[mactec54]

Hi Mactec - I don't plan on machining rail lands as the buyer will not pay for it. This machine is low cost aimed at Makers, Hobbyists and DIYers. Its at the same level as an aluminium extrusion machine and they are not machined. I use flex grade cable and catenaries no cable carrier but that is up to the Maker. Its a kit, I supply the structural and mechanical parts plus a shopping list and specs for the rest. I don't disagree with much of what you say in principle, its just that I'm not in the Rolls Royce part of the industry. Brevis-HD is made to a $$$ value and the design tries to squeeze out as much user value as possible from those cost $$$. Perhaps the next machine I design on the forum will be a carbon fibre Rolls Royce one just for the fun of it....cheers Peter

The extrusion machines where the rails are mounted are machined on all the good machines and the extrusion can be brought with it already machined

I thought this machine was the Heavy Duty model does not seem that way, I do agree with what you say about low cost everyone wants low cost, but don't realize that there can be hidden costs in buying low-cost kits like this, kit machines very rarely ever get finished so can be good for the seller or a pain with customer service requests

[peteeng]

Hi Mactec - Its HD in the sense of the model. "Brevis" the base model is light duty. I'm thinking of another name for this one as its character has changed from its starting point. I want to get away from HD or light etc as these mean different things to different people. Same issue with accuracy I say high accuracy and get jumped on because its not to 0.001mm but I meant 0.1mm relative to competitors machines...etc .

Hi Sterob -Yes this number is called the static stiffness. Its the spring stiffness of the element. Could be in lbf/in or lbf/thou imperial. So if I support the Z Axis rigidly at its bearings and apply 7.7N to it, it deflects 1um. 7.7N is 0.78kgf. The design load I use is 1000N (101kgf) so its deflecting 0.001mm/7.7N*1000N= 0.13mm at the tool if the tool load is 101kgf. Make sense?

This is also directional, the Z axis is much stiffer in the long direction of the gantry then transverse to it. When designing a machine I place the Z axis in the middle of the gantry and use max Z so it is worse case.The Z in this case is 210mm cantilever.

Commercial VMC's have a static stiffness of 100N/um plus. I've seen machines quoted at 750N/um but these are monster huge stiff aerospace type mega munchers.

Cheers Peter S

[peteeng]

Hi All - Its time for a bit of a summary of the project and to describe a couple of things. The auto industry has a term called the "Marriage" This is the point in the assembly line where the body is connected to the drive train. This occurred last night with the Brevis-HD, the gantry and rails were landed on its base. Now HD means different things to different people and the design has moved away from the precepts of the Brevis model line so its time for suggestions on a new name. I'm thinking YAG600. YAG is the laser that cuts the parts and 600mm is the depth of the machine. There will be a half sheet version as well the YAG1200 maybe a 2400... or maybe call it Atelier or BOB

Another thing the auto industry is into is "digital twins" and I suppose my development of YAG has been exactly that. So:
1) YAG's development has used digital technologies in terms of CAD and simulation
2) It started small and grew to a 1/4 sheet transverse configuration as a large benchtop machine. The wide configuration means a ripped std sheet width can be used or a full sheet can be stepped thru the machine.
3) Its aimed at Makers, Hobbyists and DIY type timber and plastic maker people
4) It has an 18mm plywood machine base with high rails. The high rails are to create a sawdust trap, keep the ballscrews out of the dust and make the machine stiffer by not having columns
5) Parts are made from laser cut stainless steel and are press bent. Stainless steel is used to remove issues with paint/rust and steel is stiff.


Issues along the way:
1) It was decided to move from belts to ballscrews. Cost for asian C7 ballscrews is similiar to building precision belt drives in australia
2) The Z axis had round rails to take advantage of their extra height. There is always a discrepancy between the carriage height and the ball nut height to deal with somehow. But in the detailing phase this was changed to square rails to decrease the Z axis width due to the round rails wide mount and the need for flaring allowance on the part. Made the part some 60mm wider then desired.
3) Packaging everything based on flaring allowance has been interesting. Especially with the saddle dimensions getting very large. So I decided to make the saddle parts flat and to improve stiffness laminate steel parts together via silver solder

Design workflow
1) from experience designing machines in the past I started rough designing the Z axis first. Then the saddle then the gantry. This then allowed rough sizing of the machine base
2) Once the base and envelope dimensions had been resolved, I started at the tool again and designed to detail level. This showed up sheet bending issues and bolting issues to be resolved
3) Also at this stage assembly logic was checked. I have made bits that have been difficult to assemble or access in the past. The horizontal gantry rails are one attempt at streamlining assemble and making the saddle less congested
4) The assemblies are broken into Z axis, saddle, gantry and machine base. Each structural loop is associated by its bearings. eg the Z axis bearings are where the Z axis loop finishes and the rails and cars stay associated
5) At the marriage, the base will be fully detailed then I shall work back through the structures to the tool.
6) This phase is the drawing phase. Each part will be drawn to production std, holes and fits checked and any special instructions included
7) At this point the design can be rested for a while then reviewed with a clear head....

Overall
Cost will determine a few things in the wash-up. I have to decide how stiff this needs to be so I use the thinnest sheet for each part. 3, 4 and 6mm are good thicknesses to work with due to the tooling allowing tight bends. Once the drawing stack is done and a couple more simulations run, the overall picture of part thicknesses can be tuned. Then its build the first machine.... The design has a good "handle" at the moment very encouraging...Peter

PS look up "handle" its the technical name for the feel of toilet paper. It even has an international standard. TP is very topical here at the moment....

https://en.wikipedia.org/wiki/Handle-o-Meter

[peteeng]

Hi All - I have reviewed some of the parts and have decided to use MBA type motor mounts vs the current flange designs. This is about cost neutral but then I don't have to braze the webs and paint the parts. This takes it to a cost down so that's good. But BST automation can't supply MBA size 10 so I have to upsize the Z axis from 1204 to 1605. So I'm waiting on some other costs from other suppliers then I'll redo the model properly. Now working on the machine bed construction dwgs. Seemed to be close but the design will be better. Peter

[peteeng]

The All - The change to the MBA motor mount flowed quite a bit further then I expected!! Had to use 16mm ballscrews (with 12mm bearings vs the 10mm ones) which moved the nut up quite a bit. So today I have been reworking the gantry dealing with this issue. Then I got into trouble in the Z axis. What fun But I think I have it re-re-re-sorted now. I dislike having heaps of spacers prefer none. Modify a couple of the brackets and I'm back on the home stretch. Peter

[machinedude]

The All - The change to the MBA motor mount flowed quite a bit further then I expected!! Had to use 16mm ballscrews (with 12mm bearings vs the 10mm ones) which moved the nut up quite a bit. So today I have been reworking the gantry dealing with this issue. Then I got into trouble in the Z axis. What fun But I think I have it re-re-re-sorted now. I dislike having heaps of spacers prefer none. Modify a couple of the brackets and I'm back on the home stretch. Peter

your preaching to the choir my friend i have reworked just about every aspect of my build look at it this way it keeps you busy and out of trouble. that's what i keep telling myself at least

[peteeng]

Hi All - Wouldn't it be fantastic if the support bearing people talked to the linear bearing people and made the stack heights the same!! At least for a given series.... Peter

I have a couple of reasons for the rails on top of the gantry. 1) It makes the saddle more accessible so the bearing bolts can be got at for levelling and maintenance without pulling the machine to bits 2) it makes the bearings easier to deal with as they are on different parts 3) with my two piece gantry design it makes the connection to the "columns" or bearings easier. When anything is done to make something easier you can confidently bet that it makes something more difficult. So compromises have to be made.

Right now I'm considering the position that the horizontal saddle plate connects to the vertical tool plate. The red bearings are the closest bearings to the load so they do 90% of the work. These are the ones that will wear fastest or come loose first so they need to be accessible. Ideally they need to be moved down a bit (to lessen the cantilever) but then I can't brace them and then they become cantilevers as well.

Where they are you have to use a ring spanner and a fiddle to check them or tighten them (and use hex bolts not capscrews). I'd like to move them down, to also improve the relationship between the Z movement and the gantry. There's a little wasted action in it at the moment. One way to fix it would be to move it up slightly so all bearing bolts were accessible from the saddle (hairs standing up on my neck for structural negative reasons) Or move it alot down and put up with the ring spanner and the extra flex. After all, this is a timber machine and it won't be too bad. But the search for perfection is hard to throw off....

Seems it's time for more FEA work... Peter

[pippin88]

Peter,

I think the rails being on top of the gantry (rather than being on the face) is a poor choice.

You have a big lever there. You normal would have the lower Z bearing level with the bottom of the gantry. With your design doing this doesn't gain much because it's still all cantilevered.

If you want the bearing bolts easily accessible, put the rails on the face of the gantry and just have the carriage/saddle wider than the Z axis.

[peteeng]

Hi Pippin - The basic lever or moment is the same if the rails are on top or on the face of the gantry. If a standard sort of arrangement is used then you have two times the stack height of the bearings/rails out the front of the gantry so the local moment could be more with a std arrangement. If I have the saddle wider then the Z axis then I loss gantry cross travel. All decisions are a compromise and lead you down different rabbit holes ... I'm happy with the current arrangement I'm looking for fatal errors and trying for refinement (called polishing) . can't see any fatals yet...

The image shows the deflection as 0.58mm with 1000N applied. This includes the plywood base being included and compliant bearings. Still probably 50% efficient. But happy with it at the mo.Will try a full contact model soon. this image is a bonded model. Peter

[davida1234]

Hi Peter, Just out of curiosity and not being as much involved in modeling as you are:

1. Are deflections in the model linear, in other words, if applied 1/10th of the load (say 100N), is the deflection 1/10th of the max. deflection?

2. With that cantilever of the saddle and the close arrangement of the vertical linear trucks, are you plugging in zero tolerance in the bearings?

As you know, nothing can have zero tolerance if it has to move, roll or slide. High quality linear motion manufacturers achieve very tight tolerances but also have their pricing accordingly. Less expensive ones have certainly a lot more tolerances to accommodate less precise manufacturing. If you are ruling out machining of rail mounting surfaces due to cost, you probably have to rule out top notch linear rails as well. In that case, calculations/models with large cantilevers/tightly spaced trucks should incorporate a certain amount of tolerance into models. Particularly not so much in max. deflection but in low/moderate loads while changing direction. Something like cutting a circle with moderate loads where the bearing tolerances + cantilever forces will show up as wobble/backlash.

Thank you and keep up the nice work. There is something to learn for all of us in your endeavors.

[peteeng]

Hi David:
1) Yes the model is linear and stiff machines re usually linear. I do a lot of work in the non linear area as well. Occasionally I'll run the NL solver to check it is in fact linear which it generally is. We don't want non linear machinery deflections. I use 1000N as its a round figure and gives a reasonable deflection. Being 100kgf it also represents the top value in cutting that I would expect for this type of machine.
2) All connections in this model are "bonded" you can't model a "gap" (you can but its complicated for this level of work) but you can use low modulus values for the bearing to simulate the bearing stiffness. I model the bearings at about the same stiffness as aluminium to capture this effect. I've attached Hiwins bearing stiffnesses if that's a help. To use the values you model a discrete bearing and adjust its material stiffness until it matches the documented stiffness. Then the model will include the bearing stiffness correctly
3) I try to use high preload bearings in my builds, they have zero clearance and if you look at the table they are linear in behavior and will eliminate the "wobble" you mention when changing direction. You have to be careful when getting bearings/rails from ebay etc as these will be clearance or zero clearance spec and they will have hysteresis when changing dirn.
4) In the FE I use I can set up sliding contacts to simulate the freedom in the rail direction or a rolling bearing or any sliding type surface
5) You use the word "tolerance" but maybe start using "clearance" for bearings. Look uo the bearing specs they come in different tolerance and clearance for different applications
6) at this level I assume the bonded model is about twice as stiff as it should be vs bolted and friction connections. We will find out soon when I run it fully contact... takes 15mins to set it up this way takes about 4 hours to set this up fully contact...

Yes its a two way street in the forum with information and ideas that's why I'm here..

Cheers Peter S

[peteeng]

Hi All - After designing several machines in sheet metal (and building 4) I've decided the next machine I design will be full composite. Struggling against the flaring allowances, bending allowances and metal shape limits I think has shown me all the possibilities/limitations its has to offer. So stay tuned for a full composite machine. But I have to get Brevis-HD aka YaG600 built first... I think the biggest issue with sheet metal work is getting holes close enough to bends without flaring. I continually design things that get bigger then what I want due to this detail. So onward to a material I have nearly full control over....Cheers Peter

[peteeng]

Hi All and Sundry - I've been working on the gantry drive connection. I did not like the bracket & spacers I had. I prefer to connect to a major structural part than through a bracket. But the screw nut and gantry have different stack heights and it became difficult as metal thicknesses did not add up correctly and I hate a stack of odd plates to get to the "right" height. So I decided to make the spacer from Tertrium-S (if you do a search for it you will find it) a composite material that I make. In this way it can be machined or cast to correct thickness (15mm) . Also being composite its damp so will isolate some vibration. The last area to polish is the saddle. Also I have been looking at material costs and have decided to use 3mm zinc coated steel vs S304. Its just under 1/2 the cost of stainless so will make a difference to the final cost. I had a quote of $84AUD for one of the Z axis parts in 6mm SS and $38AUD for plain steel. These things will add up. One thing I wanted to do was have a play with laminated metal. Aircraft use it quite a bit. But I think I'll jump straight to composites, so I do the lot. Peter