Reasonably Rigid Router (RRR) - a cautionary tale (likely) in the making.

[luke peter]

Thanks Routalot.

I agree that external triangulation would pretty much solve the problem, at least in the locations that have base support that can be extended. The issue is with my ball screw assembly, it will prevent any bracing from being mounted more than a third up the C channel beam, and most of the deflection is coming from within the channel (the c displacing up in between the bulkheads).

I will try modeling a complete closure plate for the sides (essentially making it a rectangular box section), I think that will stop the localised displacement of the C flanges, and wont cost all that much in 6mm plate.

I think the machine will still be pretty stiff (even at 37N/um in this direction) but it's not in keeping with the rest of the build, and the way spring constants stack, one weak component has a pretty dramatic effect on the entire assembly.

EDIT:


I modeled closure plates on the full length of the x axis beams, and it essentially doubled the stiffness... its not the 5x I wanted, but it will be well worth the money I feel, and it looks much better.

[luke peter]

last bit of simulation nonsense for today.

I increased the model complexity to include all frame elements and the spindle. As the geometry increases in complexity, I imagine the results will be less and less representative of life, but as previously mentioned, this is not what I do all day, so plod along I shall... Load applied (1000N) on the spindle taper.

In X direction (machine long axis) :

Stiffness

In the Y direction (gantry direction):


The Z axis assembly is the greatest contributor to deflection, so it may benefit from a redesign at some point, I will fiddle with a flange on the bottom to counter the plate cupping, or look at a thicker backing plate/ some sort of bracing inside.

Final stiffness in the x direction came to 13.94N/um

Final Stiffness in y direction came to 34.85N/um

Even with the flimsy base and x rails in the y direction, the compliance of the z axis in the x direction makes it the weaker direction by a significant margin.

That being said... even the 13.94N/um is more than stiff enough for Ali milling. I suspect the machine will perform OK

[JonnyFive]

What stiffness are you aiming for? I’m intrigued to know what value you determine to be stiff enough?

[luke peter]

What stiffness are you aiming for? I’m intrigued to know what value you determine to be stiff enough?

Its difficult to give a reasonable answer to that... I have seen a number range of 20000 to 60000lb/in quoted for aluminium, and 80000 to 120000lb/in indicated for steel. These values were not contested.

The 60000lb/in translates to 10.5N/um, so that's my general target, but "as stiff as I can make it while staying within my means" is the ultimate goal.

[luke peter]

Uprights welded and lightly dressed. They need to have their side holes drilled before being welded to the base channels. I have done more welding over the last three weeks than in the rest of my life combined at this point. Its shameful really, how unproductive I have been.



I am going camping with the family from tomorrow until Monday, so there will likely be no movement in the build until then.

Thanks for following along thus far.

Luke

[luke peter]

What stiffness are you aiming for? I’m intrigued to know what value you determine to be stiff enough?

Having a browse through Bamberg's principles of rapid machine design, he indicates machine tools are generally between 10 and 25N/um (proper milling machines). If I can end up in the bottom of that spectrum, for this kind of machine architecture, I would be ecstatic. Joint efficiencies, and my manufacturing ability, will likely diminish the calculated stiffness of Steve by quite a bit in reality. This doesnt hurt my feelings though, it may mean that some of Steves early work will be to make components to stiffen itself

Edit: A pleasant note, I have loaded Steve in worst case positions for FEA analysis, if I perform the work in one of the corners, it will be significantly stiffer and better supported, so its likely I will, in some portions of the workspace, get to the stiffness that I ultimately desire.

[luke peter]

Input request from all you clever folks...

I haven't spent much time looking into damping, and its probably going to be necessary with Steve being primarily steel weldment. I have no issue with just casting the stationary components full of alumina concrete, it will add a further 45ish Kg to the gantry though, and it increases my acceleration torque requirements by about 40%. it will make the machine slower.

What do you all suggest? I have considered expanding foam, but I am uncertain what its damping characteristics will be - possibly an opportunity for experimentation with some square tube test tokens... Concrete vs foam vs various other materials etc. If someone has already done this, I would appreciate a link or some info.

Thanks.

Luke

[Mogal]

How about a resin instead of the concrete for dampening? Much lighter.
No idea if it would be cost prohibitive though...

[peteeng]

Hi LP - What do you want to cut with the machine? The other thing is that you are restraining the feet of the machine but the load loop is from the tool to the part not the feet. This will change the relative stiffness of the axis again.

Re damping - very complex subject. Just because its steel does not mean it will have a vibration issue. Best to make it as stiff as possible and solve vibration issues if they present, rather than trying to second guess them at the start. If you have FEA then do modal analysis this will give you clues to what may vibrate. If you use "fillers" like alumina concrete you will have great difficulty if you want to drill holes in the structure in future. The usual first order culprits are thin free edges like at the end of tubes and cantilevers like the Z axis. High density PU foam (300-400kg/m3) can fill these things and its user friendly. But don't try to fill in one go as it will over fill and its a messy problem to have. Start unfilled and see how it goes concentrate on max stiffness in design then solve vibration if it appears.

I would not swiss cheese the Z axis. You need as much stiffness here as possible and those small ligaments may vibrate, just what you don't want...Peter

Re bamberg - its quite old now and typical modern machining centers are 150N/um but 20N/um is still very stiff for a DIY machine.

[luke peter]

How about a resin instead of the concrete for dampening? Much lighter.
No idea if it would be cost prohibitive though...

Thanks Mogal.

Resin is pricey, much more so than concrete. Hopefully neither will be required. I am resolved to test the machine first and make changes if its inadequate.

[luke peter]

Hi LP - What do you want to cut with the machine? The other thing is that you are restraining the feet of the machine but the load loop is from the tool to the part not the feet. This will change the relative stiffness of the axis again.

Re damping - very complex subject. Just because its steel does not mean it will have a vibration issue. Best to make it as stiff as possible and solve vibration issues if they present, rather than trying to second guess them at the start. If you have FEA then do modal analysis this will give you clues to what may vibrate. If you use "fillers" like alumina concrete you will have great difficulty if you want to drill holes in the structure in future. The usual first order culprits are thin free edges like at the end of tubes and cantilevers like the Z axis. High density PU foam (300-400kg/m3) can fill these things and its user friendly. But don't try to fill in one go as it will over fill and its a messy problem to have. Start unfilled and see how it goes concentrate on max stiffness in design then solve vibration if it appears.

I would not swiss cheese the Z axis. You need as much stiffness here as possible and those small ligaments may vibrate, just what you don't want...Peter

Re bamberg - its quite old now and typical modern machining centers are 150N/um but 20N/um is still very stiff for a DIY machine.

Thanks once again for your input Peter.

How should I constrain the model to get the load loops to be representative? Should I model a vise or fixture plate and put fixed constraints there?

Modal analysis is outside of my skillset, I don't think inventor can do it, I have an oldish version of ansys stashed somewhere, but zero familiarity with it.

That makes sense, I thought 20n/um seemed low for a vmc.

I will cut Ali most frequently, but I hope to be able to cut steel occasionally, and lightly. I have got a second spindle for that purpose (a servo powered 6krpm max Bt30 Chinese cartridge) and a 24krpm spindle for wood and small tool Ali milling.

Edit. Z plates are already cut, I can't un Swiss cheese it

[peteeng]

Hi LP - I add a block of material to the bed that acts like a vice or part. There is no standard method to measure the static or dynamic stiffness of a machine. If it was std it would be helpful. Some tests pull the spindle to an off machine point. So this would be like you have modelled. Some load from spindle to bed. So the machine stiffness stated in lots of docs is a guideline. Inventor should do modal. Its very easy just select the modal solver and it will do the rest. Peter

doesn't the laser cutter have an undo button? one of my FE programs doesn't have an undo and its very frustrating sometimes when you hit the wrong button... Inventor also has a generative design module very interesting stuff that.

[luke peter]

Thanks Peter.

I will model a block for a vise and toss a fixed constraint there.

I have used inventors part generation before when doing conventional (strong enough to not break) design, its a very cool bit of software that would pair awesomely with a laser sintering printer.

I have wished for an undo button for life in many many projects, measure twice is a lesson I persistently fail to learn

It's cool that inventor can do modal analysis, that will give me something to play with when I am forbidden from making garage noise.

[luke peter]

For posterity.

Do you know how many hours, cutoff discs, and strip and clean discs it takes to cut 10m of 65x160mm c channel to length for Steve's purposes, and strip it of paint and rust?

I do...

It took 14 115x1mm cutoff discs (for South African forumites, inco brand is cheaper and more robust than just about all of our offerings,I found this out too late), 5 strip and clean mesh abrasive discs, and approximately 36 hours...

I finished stripping the last of the cut lengths tonight (they will form the cross pieces under the base plate), and I can say with no hesitation that I am pleased to be done with that process.

[luke peter]

My seas shipped parts arrived yesterday, I haven't had a chance to lay them all out and revel like a child on Christmas morning yet, but I will hopefully do so tonight.

In this box:
2x65v PSU's
24krpm spindle (flange mount air cooled) ER20
6krpm waterproof spindle cartridge, bt30
2x bt3 er32 collet holders + spanner
full er32 collet set
full er20 collet set
4x dm860AH stepper drivers
1.8kw 3krpm servo motor and driver
8x500kg leveling castor wheels.

Quality remains to be seen, but the bt 30 holders look exceptional. If the stuff turns out decent, I will share suppliers and purchasing methods here - the price direct from the chinese sellers is very much less than the resale prices I would pay here in South Africa, even with shipping taken into account. the company that I got the 1.8kw servo from have smaller form factor servos (200-700w) that would make great axis motors, and their dedicated 8krpm spindle motors look like a good option for a rigid diy machine.

spindle servo kits : https://item.taobao.com/item.htm?spm...d=626453555454 ($580 for 3.7kw 8000rpm servo and driver)
axis servo kits : https://item.taobao.com/item.htm?spm...d=599371517161 (roughly $150 per kit - servo and driver, 2500line 600w, 1.9Nm)
from what I have researched, the drivers accept step and direction input - not sure what else would be required to make them work well, but for me and my budget, the time invested will probably be worth the result. I will toss a set of these in to my next bulk shipment I think.

[luke peter]

An indication on what has been spent so far, as well as large items that are outstanding (no price indication):



Chinese imports, and using recycled material, have kept the cost down, but its still significant money for hobby time. I believe I will spend approximately twice what is currently indicated before I am done.

[luke peter]

Sigh.

Almost a perfect import experience, almost... The seller sent my servo drive and cables in two packages and only gave me the tracking number for one. I will have to find the second number and see if the forwarder has kept the items, I hope so.

Otherwise, all present and accounted for.



The castors have nice ratchet mechanisms for raising and lowering the foot.


No damage on the ship, and the servo motor looks nice and beefy.

These bits and pieces (including the servo drive) landed for $950.

[luke peter]

Collet holders look very nice, for $10 each - I doubt I could buy the material for that money...

[luke peter]

I got hold of the forwarder and they found my servo drive in the warehouse. Another $70 to air ship it here, and all will be right in the world.

Some more progress on Steve's bits today...
I use printed templates for hole center punching, little 3mm neodymium magnets in 3d printed finger grips hold the paper down while I transfer hole patterns.


I then drilled out the holes to 8,5mm, until I broke my 8,5mm bit, then grudgingly went to 9mm since the threads here are not hugely load bearing:


These holes will have the spacer bars bolted into them, to maintain even spacing between the bed/x rail supports, it also adds a measure of rigidity to the assembly when the bed is not mounted.


Small progress, as other things have taken priority, but I will take small progress over no progress.

[luke peter]

First rail support/bed beam welded. It came out about 1mm short, I believe as a result of weld contraction combined with my jig technique being iffy. This is not a huge issue, I will mark and drill the x rails when they are on top of this assembly of three supports, using the gantry as a spacer for the critical rail-rail dimension.

I extended the jig to give me 1230mm between uprights, all nice and square.


Some grinder work was required to get the end assemblies to fit on either side of the main beam (it was welded upside down compared to this photo). The one plate top ended up being 0.3deg out of alignment with the other, because the welds pulled unevenly. I think that is correctable, either through replication or machining.


I got another 5 kg of electrodes a couple weeks ago, they were very much cheaper than the brand I usually use (store didn't have stock), and its evident why. significantly more spatter, and more work to make nice welds. I will not use them again.