[Zach_G]

No trouble at all. Im kind of OCD about my CAD model matching what I build. If I ever want to make changes it's nice to see the results in CAD first. Attached is the PDF schematic I sent the sheetmetal shop. It is in no way a complete set of dimensions, just enough to get a quote but they managed to fabricate without anything more. Let me know if you need any dimensions changed. The drawing adapts to the model so it's pretty much just change a number in a box and out pops a new drawing.

[Zach_G]

I've accomplished quite a bit this weekend. Been fretting over the connection of base a column for a long while now, and it finally came together perfectly. Decided to grout with the epoxy and iron oxide mix rather than scraping since my flange surfaces weren't at all conducive to scraping and grouting allowed the machine to be assembled in alignment without having to constantly scrape and check.

3 set screws were drilled into the Z column flange, 2 in front and 1 in back. The column is then jacked up to an initial gap of about 1/8" before making necessary adjustments to tram the column. An indicator run along the side and face of an angle plate measures squareness. Depending on the geometry and direction of tilt, adjusting the set screws in the right direction actually moves the indicator further from 0! Took some CAD modeling to help wrap my head around that one and visualize the geometry.



Once I saw no movement from the indicator, all the edges were sealed up with clear packing tape and modeling clay for the rough patches. The tape needs to be perforated with toothpick sized breather holes. I had hoped I'd be able to just pour some epoxy/iron oxide mix into the Z column based on prior experience of the epoxy slurry oozing from unplugged holes when filling the base. Probably didnt pour enough in so I had to resort to injecting. Mixed up some more epoxy and iron oxide to a viscosity similar to honey and injected it through the breather holes. The mix was pretty thick and only expanded about an inch but also didn't ooze back out. Checked alignment during injection, after curing, and after tightening the flange bolts with no noticeable change whatsoever! Peeling off the tape leaves a perfect seam. Needless to say Im now a big fan of joint replication.



I decided to fill the Z column with E/G for additional support before attaching the head just in case the offset weight deforms the column. I weigh out all the components on a scale and mix the dry stuff in a 5 gallon bucket before dumping in pre mixed epoxy. A few minutes with a paint mixer results in an E/G the consistency of dry chunky peanut butter. The slurry isnt really pourable, more like scraping it out of the bucket. I then compact the mix with a wooden dowel for 15 minutes to force out as much air as possible. Don't know how effective that is compared to using vaccuum or vibrocompaction, but the bottom face of the base only had a couple small air pockets. There was only enough epoxy left for half, guess my estimates were a little off.



No matter, should be enough to stabilize the column so I went ahead and attached the head and table. Had to use shims to tram the head forward/back since there's no means of adjustment in that direction. Also need to shim the table, it's about 0.003" higher on one side.



I played around with the DTI and jogging a bunch, the advertised rigdity spec of these so called "anti-backlash" screws is a joke. There's about 0.008" of backlash in the Y, 0.002" Z, 0.001" X. Guess I'll need to overhaul the ball nuts like I did the bearing ends to squeeze out that backlash =\ Oh well, sure does look pretty on that stand.

[GTRanj]

I've been following your thread for a while now, man this is inspiring. Once i get some time and money, I plan on taking on this type of challenge myself.

Anyways, keep up the great work. Machine is looking good.

[Zach_G]

Thanks for the encouragement! Youll need quite alot of time and a good chunk of money. Im in about $4k now, and dont want to even think about time consumed. Granted alot of that is spent researching and part sourcing. If I had to do it again (and yes, I've already got a list of things I'd do differently...) probably wouldn't take nearly as long.

Right now Im in a sprint to the finish, of course these things are never really completed are they haha. Always improvements to make.

[jsheerin]

Looking good! Thanks for posting your progress. If you want to post what you'd do differently at any point I'd be interested in that. I think that's some of the most helpful information you can share.

[giz]

Loving this build - if I ever build a metal-working CNC from scratch I will definitely use this as a reference.

[Zach_G]

I've achieved a state of functionality! Lost interest for awhile, then got back into it. Did some more shimming to square up the table and tightened down some things to work out some backlash in combination with software backlash compensation to hit within a couple thous. Also put together waycovers for the Y axis. Tried overpriced flat bellows from Mcmaster but the folds were too stiff that it just ballooned out rather than collapsing back down. So I took a tip from the zone http://www.cnczone.com/forums/genera..._made_way.html and made up my own folded denim waycovers.



Took alooot of work but turned out nice. They are fairly self-supporting over the short Y axis. Not too sure about the waterproofing of the spar varnish, it seems to absorb a small amount of water but we'll see how it goes. Still need to fold up the z cover and make mounts but I felt comfortable enough to go ahead and attempt a cut:



Been a couple weeks since then but I believe that was a 3/8" 2 flute carbide at 3krpm and 0.002" per tooth feed, 3/8 depth and 60% stepover. The smallish table felt like it had some noticeable vibrations but everything else was a faint buzz. Did some lighter finish passes and got a pretty decent surface finish!



Threw chips everywhere so I whipped up an enclosure out of PVC pipe and shower curtains. Small neodymium magnets at the bottom help keep the curtain from billowing out. Thank goodness for the leveling feet and casters, made it so much easier to reposition in its final resting point, now I can park my car in the garage again before it starts snowing!




Still got the limit switches to wire up, and hook up the flood coolant. Maybe a better way to lock the spindle rather than using a strap wrench. Then we should really be cooking. Oh yeah, and a belt cover to contain some of the spindle noise. It's geared 1:1 on a 3440 rpm motor and the spindle heats up noticeably at that speed. Could probably overdrive it with the VFD to 5000rpm with better bearings, though I'm kinda wishing I had geared it down for more low end torque. Might play with the voltage/current curves in the VFD but I doubt this arrangement could run a tap since torque drops off substantially below 1000 rpm; to the point where I can stall it by hand.

Funny how all these little details seem to take more effort than the mechanics. Probably had more motivation earlier in the build too.

[TarHeelTom]

Thanks for posting the pictures of the enclosure. I'll consider making something like that for my Bridgeport.

Tom

[LUCKY13]

That cut sounded good. I dont there is a machine around that your not going to feel some vibration in the metal. It sure didnt show up in the cut itself.


Throw you some mild steel on there and cut it and see what she does. The surface finish will tell the story.


Jess

[giz]

Time estimate for those way covers? Do you see these being feasible on a 10' long, 8" wide axis? Perhaps using a thin plastic?

[Zach_G]

Thanks for the advice Lucky, I'll have to give that a try once I finish up the mill. Wonder if there's a standard part to machine that I could do some metrology on and characterize my mill's performance?

Giz, it took me probably 6 solid hours of work to mark up and fold those bellows, I'd hate to imagine what 10' would take... It might work if the bellows rest along horizontal ways for support. The wider your bellow the less folds needed. I'd recommend not exceeding 110 deg unfold angle since the origami tends to flare. If you went with a 3" wide bellow that would take 49 folds to get 10', compressed length depending on what material you use. I suppose plastic could work, but I'd be worried about the folds being too stiff and causing the bellow to billow instead of collapsing properly like my mcmaster bellows which had about 60 folds.

[danande]

Nice build! I'm building a clone of your concept. I believe this to be the simplest way to make a sturdy mill without the access to a mill..
Could you please tell what you would do differently?
Do you think a thick mic 6 plate could be used to transfer off the flatness to the beam? Or maybe it would be better to have it milled?

Do you have some more pics or videos of it in action?

[fatal-exception]

Nice machine! Hope it is working well for you.

Best thing I ever did for my router was to build a fully contained plexiglass enclosure for it. Cost a few bucks, but worth it considering the amount of time it would take to clean up the whole garage after any little job.

I bought a MTW ZAY7045 that will get the conversion over the summer. I can't wait to have a big powerful mill with a mind of it's own!

[Zach_G]

Thanks for the questions and praise guys, its what keeps me posting updates! Danande, according to Alcoa MIC-6’s flatness tolerance is 0.005" for plates 3/4" and over. I'd recommend using something like a granite surface plate which are pretty affordable at the 0.0002" flatness spec and useful for all sorts of precision setup work.

As requested what I would do differently, but first a test cut in steel:



That's a 4 flute 0.5" cobalt end mill in cold rolled steel 1000rpm 4ipm at 0.13" and 0.25" depth of cut .4" wide. You can see some chatter marks on the side of the cut and a wrench would occasionally vibrate around on the mill table. Some nicely colored chips.

Anyway to answer everyone's question:

Skip the E/G fill and just bump up the wall thickness to 0.5" or 0.63". It worked fine but added complexity and time to assemble plus the added stiffness and vibration damping is an order of magnitude greater than the stiffness in the ballscrews.

Speaking of ballscrews, the ones I got from Chai (linearmotionbearings2008 on Ebay) are ok but could be alot more rigid to take full advantage of the structure's stiffness. I think preloading two ballnuts against eachother would help alot assuming the rolled screw accuracy didn't cause them to bind. The fixed end bearing blocks also needed a complete overhaul and some 35deg angular contact bearings (came with 15deg) would help stiffen these up as well.

I'd also have planned on doing the epoxy / iron oxide joint from the beginning instead of trying to weld it square. A pair of steel plates welded to the base and column would save some fabrication time rather than the flange I made from square bar stock. Of course stress relieve after welding which I wasn't able to do. I lucked out and the structure hasn't moved at all over the past 6 months.

Should've designed in more Z travel, its annoying how much space a drill chuck and vise takes up.

Those are the biggies. A vector drive for the spindle might be nice to improve low end torque. Got some other upgrades to accomplish like a coolant system, wire routing, limit switches, spindle speed feedback and spindle bearing upgrade.

[CROSSHATCH]

Great Build.

[wizard]

As to the Z clearance problem do realize it never goes away, buy a Bridgeport and you will have clearance issues, same thing on a Hass. That being said it looks like you could add a riser block under the column without any other modifications. An inch or so might actually help out a lot. Of course this is based on looking at pictures and may not reflect reality.

As to your milling test cuts, I have to say they are very respectable for a machine that size.

As to the epoxy granite I'm not sure I buy your advice. The epoxy is all about damping and would do little to stiffen a machine. Going to heavier walled tubing isn't a bad idea, I'm just not convinced it would be anymore vibration free, though the frequency will change.

In any event I suspect that you are the envy of many here. This is a big project brought to a successful conclusion. Congratulations.

Thanks for the questions and praise guys, its what keeps me posting updates! Danande, according to Alcoa MIC-6’s flatness tolerance is 0.005" for plates 3/4" and over. I'd recommend using something like a granite surface plate which are pretty affordable at the 0.0002" flatness spec and useful for all sorts of precision setup work.

As requested what I would do differently, but first a test cut in steel:



That's a 4 flute 0.5" cobalt end mill in cold rolled steel 1000rpm 4ipm at 0.13" and 0.25" depth of cut .4" wide. You can see some chatter marks on the side of the cut and a wrench would occasionally vibrate around on the mill table. Some nicely colored chips.

Anyway to answer everyone's question:

Skip the E/G fill and just bump up the wall thickness to 0.5" or 0.63". It worked fine but added complexity and time to assemble plus the added stiffness and vibration damping is an order of magnitude greater than the stiffness in the ballscrews.

Speaking of ballscrews, the ones I got from Chai (linearmotionbearings2008 on Ebay) are ok but could be alot more rigid to take full advantage of the structure's stiffness. I think preloading two ballnuts against eachother would help alot assuming the rolled screw accuracy didn't cause them to bind. The fixed end bearing blocks also needed a complete overhaul and some 35deg angular contact bearings (came with 15deg) would help stiffen these up as well.

I'd also have planned on doing the epoxy / iron oxide joint from the beginning instead of trying to weld it square. A pair of steel plates welded to the base and column would save some fabrication time rather than the flange I made from square bar stock. Of course stress relieve after welding which I wasn't able to do. I lucked out and the structure hasn't moved at all over the past 6 months.

Should've designed in more Z travel, its annoying how much space a drill chuck and vise takes up.

Those are the biggies. A vector drive for the spindle might be nice to improve low end torque. Got some other upgrades to accomplish like a coolant system, wire routing, limit switches, spindle speed feedback and spindle bearing upgrade.

[Zach_G]

Thanks wizard. In my application the E/G granite fill increases the stiffness of the structure by around 3 times. However if I bumped the exterior wall thickness from 3/8" to 5/8" and ditched the E/G then stifness would be comparable. Don't take my word for it though, trust the finite element analysis. Doesn't seem intuitive considering E/G has 1/10 the modulus of elasticity of steel but what is happening is the E/G acts to transfer the sheer stress and limit thinwall deformations much like the foam layer of a composite laminate sandwich. Problem is I'm not modeling the stiffness of the bearings and ballscrews properly so while the model shows a deflection of 0.00063" under 80lbs of load, I can push and pull the whole machine around by 0.004" and most of this is coming from the ballscrews.

[wizard]

However is "stiffness" the only thing to be concerned with here? What I'm interested in is how the tube resonants, that is would am Epoxy granite filled Tube be resonating at a higher frequency.

Frankly I'm not sure how this relates to stiffness. All I know is that the epoxy granite structure I've worked on have been very dead.

Thanks wizard. In my application the E/G granite fill increases the stiffness of the structure by around 3 times. However if I bumped the exterior wall thickness from 3/8" to 5/8" and ditched the E/G then stifness would be comparable. Don't take my word for it though, trust the finite element analysis. Doesn't seem intuitive considering E/G has 1/10 the modulus of elasticity of steel but what is happening is the E/G acts to transfer the sheer stress and limit thinwall deformations much like the foam layer of a composite laminate sandwich.

interesting to say the least. I'm not conversant with the finite element method, so I hope I'm not a bother with the questions. It just seems like there is more to be concerned with than the static deformations.

Problem is I'm not modeling the stiffness of the bearings and ballscrews properly so while the model shows a deflection of 0.00063" under 80lbs of load, I can push and pull the whole machine around by 0.004" and most of this is coming from the ballscrews.

Are you sure about that? From the machine maintenance point of view I generally find issues with way fit ups or leadscrew bearings. 0.004" is a lot of movement.

[jsheerin]

Zach, are you saying you have 0.004" of backlash, or you're actually getting that much deflection under load? It would be interesting to measure (measure force applied while also measuring deflection). I used a pull scale on a dowel pin in the spindle and a dti measuring deflection to get an idea of what my router does.

Stiffness and damping are the typical concerns. However you can have all the damping in the world, but if it's not stiff enough, the structure will still deflect too much to be useable. The resonant frequency of a structure is determined by stiffness and mass. Damping determines how it behaves at that resonant frequency. If it's well damped, the resonance will die out quickly. If it's not, it could ring (keep vibrating). Epoxy granite fill would both add stiffness and mass, but I'm not sure in what proportion, so the resonant frequencies (there are more than one) could go up or down. But the epoxy granite will also add damping, so any resonances will be less noticeable. The damping is probably what you're noticing the most when you say a structure is 'dead'.

I have more experience in audio than in machine design, but audio is very concerned with resonances. Typically there are a few ways to go. One is to make things very stiff to push the primary resonance above the frequency range you want to operate in. Then you aren't very likely to excite it, so you're not too concerned with damping. This is how speakers with metal cones tend to be designed. Another way is to damp everything very thoroughly so any resonances are suppressed. This is typically how plastic cone speakers work. This typically also adds mass at the same time, so resonances might go down in frequency as well.

The other thing I look at with FEA (besides deflection under load, ie stiffness) is the modal response. This tells you what frequency the structure resonates at and what the shape of the resonance is (how the frame vibrates). The general goal is to get the stiffness and primary resonance high as well as have some damping, although in the overall scheme of things, cast iron does not have a lot of damping and it's usually touted as having good damping in terms of machinery design. Damping is beyond what I'd try to do with FEA - all the various interfaces in the machine will come into play.

[Zach_G]

It's a combination of both. I need to get a tension gauge to get some quantifiable measurements but I'd say I'm applying in the vicinity of 100 lbs of force. If I pull only on the mill head i can get maybe 0.001" deflection, but if i pull/push between the mill head and table I can hit 0.004" which I'm blaming on the ballscrew.

According to the modal response of the FEA it's around 80Hz. Got a 1D laser vibrometer at work I could use to validate but I don't think they'd appreciate me taking that $150k piece of kit home.