[Zach_G]

Well after years of inspiration, figured I should give back to the community by documenting my own build. The plan is to end up with a vertical mill that mills aluminum and some steel. Overall work volume should be about 19x10x14 inches X Y Z with future 4th axis. I had been considering gantries and other nonstandard layouts but finally decided on C frame for manufacturing simplicity.

Spindle is a mini mill run by a 1.5 HP treadmill motor. I feel this to be the weak link so far, but it's what I have on hand presently and in the future could be switched out.

Ballscrews are from our favorite Chinese manufacturer of cheap rolled precision ballscrews. For some reason there's alot of drag on them (like hard to rotate the nut and screw by hand) I'm attributing to tight rubber seals that hopefully loosen up since they spin fine without em. Linear guides for Y and Z are 2 each NSK 35mm 4 way contact long blocks with high preload, and 25mm THK on X.

Main structure is 8x8x3/8 and 5x5x1/4 steel tubing with an E/G sandwich between to help dampen vibrations. Total weight should be around 700lbs, lightweight for the work volume but hopefully the choice of materials and FEA simulations will keep things stiff. Using surplus flat ground 3/4 steel waterjetted for various mating plates, and then epoxy/metal powder to level the way surfaces on the steel tube. Still trying to find an answer on how necessary stress relieving is on these A500 steel shapes. Id like to weld em together but again, stress relieving on such an ungainly structure may be costly. The table is a replacement from a 8x28 HF mill, only like $130 to my door, not too shabby. Overall trying to avoid milling parts since access to machinery is limited, at worst there is a pay per personal use shop nearby (I'm in Dayton Ohio).

Controls provided by Mach3 and parallel port (I hear that's the most reliable Mach setup) I spent the last 2 weekends wiring up the enclosure and got my motors to spin! Zeta4 drives will push about 400W through some Nema 34 steppers from Lin Engineering and compumotor (love not having a giant transformer). Kinda a waste having such high quality motors attached to high quality drivers, the anti resonance driver settings seems to overlap the Lin's superb rotor design, didn't need any of the damping settings which were necessary for the old compumotor stepper, but hey they were all cheap ebay buys so it's all good =p

Oh yeah, and I'm pretty commited now since the steel and ballscrews are cut to length, we'll see how fast progress is.

Enough chitchat, here's some eyecandy:

[Eson]

I like it! Are you buying the linear guides new?

[Zach_G]

Its all mostly ebay/surplus so far, except for the ballscrews and structure. The 35's for the Y and Z are new in box, but the X was a pull from a scrapped machine. The whole thing has been designed around components I could scrounge.

Bit of a status update, hit a big setback when I blew up my control computer trying to get mach to control spindle speeds. Managed to connect -60V through the parallel port ground and fried the motherboard and the non-optoisolated breakout board. Thought I was smart enough but guess we all get our daily value of stupid. Hopefully a new $50 motherboard, some replacement chips for the breakout board and Im back to where I started a few weeks ago.

Ill start doing mechanical stuff once I get set up in a machine shop at work later this week.

Here's some pictures of the allmighty pile-o-parts and a nice render I put together. For some reason linearmotionbearings2008 cut the ballscrews a good bit longer than i specified. Guess I get to machine them to length, sucks because i paid for end machining too =\ Theyre also really tight in the bearing blocks which appears to be the rubber seals that will hopefully wear in.

[T.W.O.]

I Like the idea of 2 concentric square tubes.
Do you think that the tubes are straight/flat enough to bolt your linear guides on or are you going to mill/grind them first?

greetings & good luck with the build

[johnohara]

Zach~

May I suggest reading Dr. Eberhard Bamberg's "Principles of Rapid Machine Design" before you go ahead with the E/G sandwich?

Beginning on page 90, he discusses the concept of "Constrained Layer Damping" and addresses segmenting the inner sandwich to improve the dampening characteristics.

Although his discussion involves concrete inside circular tubing, his conclusions are very informative and relevant to your build. (Thank you Dr. Bamberg.)

http://www.mech.utah.edu/~bamberg/re...e%20Design.pdf

Have you decided on an E/G formula?

~John

[TarHeelTom]

Excuse my ignorance. What is an "E/G" sandwich?

[johnohara]

What's a peanut butter and jelly sandwich? Peanut butter and jelly between two pieces of bread.

What's an E/G sandwich? Epoxy-Granite between two pieces of steel.

~John

[TarHeelTom]

What's a peanut butter and jelly sandwich? Peanut butter and jelly between two pieces of bread.

What's an E/G sandwich? Epoxy-Granite between two pieces of steel.

~John

Thanks. Although I used to work for a company that made sinks and bathtubs out of similar material, I'd never heard it called E/G. Actually, I think we used polyester resin and granite.

Tom

[jsheerin]

Looks cool.

I read someone's post recently (I think in the diy router forum) where they mentioned they took apart the bearing blocks from linearmotionbearings2008 and found that the rubber seals were what the screw and nut were hitting (instead of the little metal spacers), that the inner geometry of the block was messed up and no preload was being applied to the bearings, and that there was no lube on the bearings. So they made some new spacers and added some oil, I believe. Anyway, you might want to check that out. I have parts from him as well that I need to take a look at before I put them on my router.

[pete from TN]

This is almost exactly what I am considering building. If I do actually build it I will go for more travels and make the parts larger but other than that I am right there with you on this. What are your thoughts on mounting the linear rails on the tubing, Do you intend to weld the column to the base or make some bracketry to bolt it all together for adjustability once it is working? Do you have any way to weigh those two main tubes you bought for the column and base? I will be following this thread with close interest. I have a friend who has some steel tubing like that who lives nearby and he said I could buy it pretty cheap. Do you have an idea what you are paying for the linear rails and ballscrews? Peace

Pete

[Zach_G]

To answer questions/comments:

Im 99% positive that trying to mount the rails directly to the tube will result in terrible accuracy/binding. I plan to level the surfaces by mixing up some epoxy and steel powder (kinda my own moglice formulation) and coat the tube then press a well waxed surface plate against it to transfer its flatness into the tube. As for joining the Y and Z columns, Ive considered welding them but worry about stresses causing shifting (need to stress relieve the tubes still as well). The option Im leaning toward would be to weld some flanges to the tubes, stress relieve, epoxy level the linear bearing surfaces, then align using leveling screws or maybe a big right angle plate and inject epoxy between the flanges. Tighten with screws once its all set up. That way I have stable, disassembleable fabrication.

Johnohara, I actually read through Bambergs paper and designed my structure to meet his recommended machine tool stiffness. The inner tube is part of the structure, Im banking on the E/G to provide damping and transfer sheer loads like a composite sandwich. Could fill with sand for more damping effect as well. The E/G mix will be play sand, gravel and epoxy.

Hi Jsheerin, hope your design is coming to fruition. I got antsy and started building before going through all the calculations, hope it still pays off.

Pete, Id love larger travels too but this thing is pushing 700lbs which is topping out my weight (has to be moved next year) and budget. The tubes weigh 94 91 40 38 lbs. So far Ive invested about $1800 with probably another $1000 to go. Ballscrews and bearing blocks were $334 and linear bearings usually can be had for ~$130 a pair on ebay.

[johnohara]

Zach~

I recommend adding some Aluminum Oxide abrasive to your E/G mix. 220 grit (or 320 if you can find it) at 10% by weight will work well.

Some reasons:
1. The Al Oxide will fill in nicely with the sand and gravel.
2. The Al Oxide is a little heavier and will add more weight.
3. You'll find the Al Oxide easy to work with vs. sand.

I suggest mixing the Al Oxide with the sand before mixing with the epoxy.

Another suggestion, if I may. Use pea gravel. It's cheap and has a nice
range of small sizes that will work well with the Al Oxide/sand mix.

Some links:
2 Lbs. 220 Grit White Aluminum Oxide Abrasive
Epoxy :*Epoxy Resins and Hardeners with #556 hardener.

~John

[Zach_G]

It’s been awhile time to share some progress!

Things are going slow, but Im also having to build up my "shop" from just a few handheld power tools. Ebay's great for CNC components but when it comes to tools, Craigslist is king! Floor standing drill press - $100, 110V Miller "squirtgun" welder - $200, some shop benches and vise - $150 etc. Granted it takes time to score the good deals, but I have no deadline here, just a bit of impatience.

Decided the mill warranted more spindle hp. Picked up a surplus 2hp 3 phase 220V AC motor from my favorite store. You online folks know it as MECI but I get to dig through their 3rd floor warehouse called Mendelson’s. It sure does dwarf the mini mill spindle! A chinese VFD from ebay should drive this guy under Mach control well. Guess I get to rewire the garage for 220V now…


Anyways, the progress! Ballscrews cut to length and turned down. Ended up a bit rough and 1-2 thous undersized but still gives that nice hydraulic fit with some grease. Made a few rigid steel couplers to connect to the steppers out of some scrap tool steel. What a pain to tap! Cut a sixteenth of a turn, break chips, repeat a hundred times. TS isn’t for Tool Steel, it’s for Tough Sh*t!

V ways needed to be milled off the table to make room for linear bearings. Was using someone else's manual mill with set screw tool holders and the cutter kept pulling out! I continue to live by the mantra that set screws suck! Granted I was hogging a 5/8” end mill through cast iron... Ended up having to take off a couple hundredths from the backside of the table due to a gouge, grrr. Granted I could’ve probably filled it with some epoxy/iron oxide (my own JB weld recipe) like the plan for the frame surfaces but I just wanted a functional part now. The owner of the mill also didn’t have a drill chuck for some reason so ended up spotting the holes with a ball mill and using a drill press to finish them off. Still have quite a bit of that youthful impatience and accidentally drilled the first hole with a through drill rather than tap drill. Oops, get to figure out how to repair that mistake... Other than that everything went together nicely and bolted the rails to within 0.0015" according to the indicator. Started by indicating one rail off the side of the table, slowly tightening the bolts and tapping with a mallet to straighten the rail as I went, then repeat for the other rail, indicating off the 1st.


All the flat ground stock was cut on waterjet. Wish I could've found a place with a multi axis head to compensate for edge taper but no one in the area with that capability was willing to take on such a small job. Oh well, ended up machining the few edges that needed to be square at work on our Minitech mill, which I absolutely detest. Its aluminum frame construction warps all over the place and is the inspiration for building my mill beefy! The edge taper actually was very small, only needed to face a couple thousandths off the bottom edge of the cut. Also had them just spot jet all the holes to save time/cost. The saddle took a while to drill on the drill press since I used end mills to counterbore. Real counterbores probably would have had less chatter but overall it came out well, maybe 1/16 larger than expected due to chatter. Made oversized through holes to give plenty of room for error and adjustability during alignment. I’m pretty good with a drill press, but just barely got everything to line up square.

Screw it all together and with a few reference edges and strategically tightening bolts you can assure everything comes out square. The screw turns smooth without any binding over the length of travel! Attach control box and we have the first CNC controlled motion! I also played with the indicator a bit. Those cheap chinese ball screws from ebay appear to work well as the table faithfully moves 0.001" back and forth without any noticeable backlash. However since they are only lightly preloaded I can move the table 0.002" by pulling on the saddle with about 80lbs.



That’s probably all for a while. It has taken a lot of work and learning to get to this point. Need to spend some time outdoors in this great spring weather!

[jsheerin]

Nice work!

[RotarySMP]

It sure is cool the first time you get motion out of these things.

Have you put your DTI on the end of your ball screw and done that pull test to see if the slop might be in the bearing blocks, motor bearings (you shouldn't use the motor bearings to react the machining forces), or couplings?

[Zach_G]

Mark, good point. This setup is not carrying the machining forces through the motor bearings, theres an angular contact bearing mount on the other side of the screw. The stepper only supports the free end of the screw, and the motor output shaft actually floats in its bearings to compensate for thermal expansion. This setup shouldnt put any more load on the motor bearings than say a tensioned belt drive would. There's no flex in the rigid coupler either like there might be with a flex coupler so as long as everything's well aligned it should work great and be simple.

I indicated between table and end of ballscrew, and got about 0.001" from the angular contact block. Between saddle (where the ballnut is mounted) and end of screw and got another 0.001" from the nut. These both add up to the saddle to table flex of 0.002". Not sure how acceptable that is but it looks like an order of magnitude more flexible than the structure in the FEA simulation which is unfortunate. Wish i couldve incorporated this in the model but without a test setup like I have now there'd be no way of knowing since the supplier couldn't/wouldn't provide any documentation. Ill have to get a tension gauge to come up with the actual stiffness and compare that to my FEA. Be curious to see how much variance is in the other two screws, this thing is turning into a big experiment!

[RotarySMP]

Hi Zach,

"... this thing is turning into a big experiment!"

That is the main thing, you are doing it. Good on you. Sorry I missed seeing the AC bearing block at the other end of the screw.

I like your nice clean design.

the slightest misalignment will cause a 2/rev stress reversal in your motor shaft. There are a few photo on CNCZ of guys whos stepper shafts have broken off due to this. Then again they are so cheap these days, that you could wait and see if it will be an issue.

[Zach_G]

Well it's been close to a month, guess I should do some more updates.

Had the steel tubing stress relieved and blasted. Wish I could've welded the base and column together beforehand but couldn't come up with a method within my means to finish the surfaces flat and square after joining the two. I figure it will be ok though since my 110V welder can only penetrate halfway through the tubing and shouldn't induce too much additional stresses to cause warping. The blasting also cleaned the oxide from heat treating and left a rough surface excellent for paint and epoxy to adhere. Was only $83 total, well worth the peace of mind.

The tubes were especially warped after heat treatment. I had always planned to finish the surfaces. Rather than mill the heck out of it, I mixed up a batch of epoxy and iron oxide (20/80 ratio by weight) and sandwiched it between one of my lightly greased up flat ground steel plates and the tube. Used an old milling table to support the ground plate since its only 3/4" thick. The plan was to use a 0.05" thick layer plus 20% squeeze out. The first go it was hard to coat the whole surface, ended up having to jump up and down as well as rock it back and forth which left a few voids in the final surface. Also, its important to evenly distribute the epoxy with a good bit extra near the corners before sandwiching otherwise youll end up not coating the whole surface. The second go used a 0.07" layer and was much easier to spread, though still didnt quite reach all the corners. To separate the metal just used a chisel and hammer and they popped apart without much fuss. Scored the excess and chiseled it off to leave a clean edge. Note that it only takes a small amount of grease on the surface plate to release, too much gives the epoxy a hazy appearance, though still flat.



To drill hole patterns I like to print out 1:1 surface on a large plotter. They are accurate within a couple hundredths over a foot in the paper feed axis, and a few thousandths in the print head axis according to my measurements. Therefore it’s most important to align the linear rail holes with the print head axis. Can then center punch and drill the holes. The tubes barely fit on my drill press. I had the quill fully extended to allow the bit to wander concentrically in the center mark. Was tedious but got them drilled and tapped. Forgot to take a picture of the test fit but it came out really well, had plenty of wiggle room to adjust the rails into alignment. Would've been much quicker to just throw it on a mill, but I have very limited access to one.



Ive also nearly finished the milling head. Welded the standoff tube to the face with my little Miller welder. The welds look ok, a bit of porosity here and there since Im bit out of practice, but its functional and decent for flux core. I then milled the backside of the standoff to be parallel with the front plate and cut the motor mounting slots. Finally the whole assembly was clamped bolted and judiciously tacked together in preparation for welding the mounting flanges. As the weld progressed, I removed one screw at a time to prevent warping.



Looks good, just need to bore out the spindle pulley and its ready to spin! In case anyone's curious I borrowed the mini mill spindle mod from Mike Aber in http://www.cnczone.com/forums/diy-cn...le_idea-7.html. Eventually I’d like to fabricate a new housing using back to back angular contact bearings but this will get things cutting for now.



That’s the progress so far. I glossed over a lot of details but my fingers are tired of typing. Feel free to ask questions about the methods thus far. Next up is to finish drilling/tapping the Z column and then join it to the base. Will be fun trying to keep things square, but I’ve got a plan to be unveiled, after the break.

[RotarySMP]

It is tough to do this sort of project without having access to decent sized machines. You are doing a really cool job of making this.

[Zach_G]

Spent the past week joining the columns and checking the fit. First pic is how I attempted to keep everything square while I joined the columns. I used straps to pull the column and linear rails tightly together, then bolted the column flanges together before welding them on. I welded half the flanges with the mill sitting on its side then tipped it all up to get access to the other side before removing my makeshift square. I also put a couple guide pins in the flanges before welding it all together to help ensure it could be reassembled.



Well, it didn’t work. After assembling the axis and running a dial indicator along an angle plate, it’s off by about 0.08 deg in both the x and y directions. That’s 0.018” out of plane over 12”! Not sure how this is possible considering I did not detect any gaps between the mill and table while welding.



Guess after removing the straps and table, residual stresses from welding pulled it out of square. You can see how much the metal relaxed after the flange bolts were removed in this picture. Time for plan b, looks like there’s a whole lot of shimming in my near future =\ Might try injecting epoxy/iron oxide into the gap after shimming. That way there’s more surface contact that shimming alone so it’s less likely to shift in the future. That’s what the really big fabricators do.



I decided to assemble the rest of the mill as a test fit and as a motivational endeavor. It’s always nice to see things put together even though there’s plenty to do with it disassembled. Bolting the saddle together was certainly an exercise in patience. I had to fabricate a couple low profile hex keys to tighten the bolts due to the super low clearance. Actually had to shim the saddle up and slowly drop it as bolts were tightened down 60 degrees at a time. It took about 2 hours to get everything tightened down and aligned. That angle plate sure comes in handy often.




Here it is all put together, looks nice just not quite square.



Think my next move is to fabricate a stand to make it easier to work on and incorporate way covers. Also, anyone have recommendations for protecting the bare metal? This hot and humid weather makes rust fighting a daily battle. I considered mixing some of my epoxy and iron oxide (blackening agent) to paint the surfaces.