Thanks for all the suggestions in the comments. I decided to scrape the surfaces and everything looks fine so far. There will be a video about that soon :)
I remember Stefan Gotteswinter describing use of a two-component epoxy for machine alignment, the way he did to tram his vertical milling machine. It's what I'll likely be using to build my own traveling gantry machine which, I have to say, I'm patterning after yours. It's a privilege to watch these - I'm grateful for the insight your videos have given.
Wow, finally someone talking about the leveling problems they are having, and how crazy it actually is to get these things corrected. Thank you for this video, as I am having the same problems with my lathe build, and all of the videos out there doing these builds make them look so simple, but it really isn't when it comes to the fine tuning and leveling... Huge pain in the ass!
Great to hear you're not giving up on the project. There are always setbacks, though this seems like a big one, there is a way through. Looking forward to seeing what you decide to do.
Hi Alex, I really like your machine! You could make your own lapping tool by drilling a hole in the centre of your base between the rails. Get a piece of 5mm steel plate cut to a diameter large enough to cover all your steel mounting pads and place a hole in the centre so you can insert a pivot pin. Cast a concrete slab onto your circular plate maybe 5 or 10 cm thick for some weight and to stabilize the steel disc. Have the disc turned or surface ground flat. Now coat the disk and your mounting pads with lapping compound or valve seating compound and turn the disc a number of times by hand until you see your mounting pads have been flattened across the majority of their surfaces. You could also ask to borrow a granite surface plate from a machine shop. If you know where the surfaces are too low, coat those with a thin layer of slow curing epoxy. Cover the granite surface plate with plastic and the turn upside down and place on top of the mounting pads. The weight of the surface plate should force the epoxy to only fill the low spots. Hope this gives you a workable idea and good luck with the rest of your build!
There is one problem. I have edges in the inlays to press the rails against. So I can not use the lapping method. But I got some ideas here that will work with the edges and are quite easy to try.
Could you CNC the bottom of the linear rail, instead of the top of the steel inlay? Much easier to get that to a machine shop. Or failing that, add a ~3mm spacer bar between rail and inlay on both sides, machined to compensate for the inlay's imperfection?
I'm not sure if I would machine the rails. But the spacer piece sounds like a valid option to try. I already got some recommendations what I could try. This one is also on the list now :)
If you are confident that your left rail is straight then move the indicator base to a left cart, remove the right rail and start with your highest point and start working it down and measuring as you go. You don't have to scrape. I remember an amateur scraping video where he used a Dremel with a sanding drum for roughing in. Maybe just a stone with some WD40 by hand like someone else recomemded.
Thanks for sharing your lessons learn, I plan to build my own machine soon and your video is a good help. The problem with casting with wooden forms like plywood, is that the pressure of fresh concrete can move your parts, wood is compressible and easily warp. The better way as a lesson learn you can use rigid form like steel plate reinforced by steel rods and nuts in betweens. Epoxy grout is also better alternative and can pour slowly by layer to reduce pressure.
@@marson8870 can you elaborate on the steel rods and nuts reinforcement? Are you saying to have the steel rods inside the mold to reinforce the actual object were pouring with holes drilled in the plates for the rods to stick through and then tighten up nuts on the outside of the mold to those rods basically self clamping the mold together. Then when it’s cured you cut I off the excess rods sticking out of the poured object? Or are you saying the rods will be outside the mold. I’m guessing the two side plates will overlap the front and back plate and then you use a rod to pull the two overlapping plates together to hold them pressed against the front and back plates? Sort of like an “H” shape but where the “-“ is in the “H” it’ll be a square/rectangle. I’m probably off on both. Also what’s a minimum thickness for these plates we can use? Wondering if I can pick these up at Home Depot or if I’ll have to special order these.
@@ChristopherJones16 you can simply make a steel skeleton and then pour the resin into the mold. There is also a uhpc with a young's modulus of 80 gpa, a little better than aluminum. With that you need steel only for the rails base. You need a nanodur cement and an artificial gravel called Durigid.
Hello! Serious project! Is the information private? What concrete or mixture did you use to make the machine bed? Have you noticed any deformations or changes in the geometry of the machine over time?
Here is an idea if you can get one of the rails true with shims you could attach a bench grinder on the rail (maybe use the second rail as another axis to attach the bench grinder on) and grind it flat. Then reverse it and grind the first rail flat. Then use a flat piece of granite or iron with a precision level to check it. Just make sure you level your machine before you start using some geometry on the column. This idea is based partially on the jank o grinder from nyc cnc so watching their video may help.
You have a Typical modern day way of thinking and solving problems . You have to many options . Today people have a tool to do anything . You have the answer to your problem. Scrape it you will be surprised it will take you a lot less time and work than hauling that back outside . It is a slow process but you will learn and appreciate it more. Scrape measure scrape measure. Every machine tool of old was hand scraped. I'm retired now but my fist accurate item to make was a one inch cube out of cold rolled steel cut from a 1.25" plate, I was about 16.The only tools we were allowed was a hacksaw and one file. For ever one thou you were out of square from all sides, was one mark off. Good luck with your project it looks awesome.
It's not the scrapping he's is afraid of, its the wandering concrete will continue to wander about. After about 15 years concrete stabilizes an easy mistake as an afterthought ~ we have made clever after the fact!
@@aion2177concrete has water in it, and it takes decades before it's all gone. The method used mostly is epoxy/granite, which has been the choice for over 20 years
Hi Alex, very impressive project! If the error is soo small, like you say, 10microns or less, yo can try to solve with fine sandpaper and trial and error, but its very time consuming.
hey alex, i made a CNC, i had the same problems as you, i grinded both with a big granite surface plate, i glued two 80 grit belt grinder paper and grinded by hand pushing back and forward 1000 times, eventually i had nearly 0.05 on 600 or 700mm length
Have you tried contacting Stephan Gotteswinter? He definitely knows scrapping very well, although that might be too big of a project for him. Maybe he knows someone with a big surface grinder that can help you. Greetings from Portugal. Looking forward for more progress on this beast.
I already got in touch with him. Unfortunately he has no time at the moment for something like that and also not really the right measuring tools for something that big.
Stephan has also talked about and showed how to true machine surfaces with a plastic compound that can be easily worked to get perfect long lasting surfaces. I forget what it's called but he says it's used in the industrial field for bringing precision surfaces back to perfect.
Have you considered using a large thick panel of plate glass, glue a sheet of automotive sandpaper and then use it to rough lap then fine lap the rails?
The problem with that is that edges for the linear guides on both sides. So I would ned a plate that exactly fits between those. An Option would be to remove those. But I'm not sure about that yet.
Hey Alex! You are pioneering a new field with new problems. Don't worry you'll figure this out. :) One suggestion I have: Ever see how woodworkers flatten large slabs of wood that cant fit in a planer? They use a sled to make the surface even. I'd say you have the talent to make a very precise machining sled which attaches to the sides. You'd then be able to bring both surfaces to the same height with a cheap hand router since the depth isn't that much (< 0,5mm)
Hi Alex, hope your doing well. been watching for a long while, since you were building your first mill. Sorry to see that your having this problem. First it should be said that this error is actual quite small about .001 thousand twist total. Matthew here in the US state of Wisconsin, 40 years career machinist. Wish you were closer id jump in the car and come give you a hand. The error looked much worse than it is because as you have your indicator set up it magnifying the error by 8 x., but its probably enough to cause the block to bind up when all 4 blocks are bolted up tight. The best way to fix is to hand scrape, the scraper that i use i made from an old file with a small piece of carbide held in with a small clamp. For a master look for a hardened and ground parallel 1 inch by 2 by 12 long. Should only take a couple of hours to straighten it out, don't know if its just the one on the right or both. Ill check back later to see if you found this, post any questions if you have them. Oh ya celifane off a pack of cigs is .0005. good luck. Matt
Thanks. I already have everything here I need for scraping so that's the way I will go. I did a test on a small steel part and there scraping worked fine and I was able to get a flat surface quite fast. But first I want to finish some other projects before scraping. Best Alex
It would be nice if you explained why you used UHPC rather than epoxy granite. Which is better in terms of accuracy in the long run for a precision CNC machine?
The UHPC has a lower tensile strength but is a lot easier to process. A normal concrete mixer is everything you need. Long term stability should be pretty much the same. Maybe the UHPC is a little bit more stable. I mainly choose UHPC because of the easy processing.
That'd be my first recommendation, if scraping is out. There's a few companies out there that do self-leveling epoxy for built-in surface plates, and the small deviation (in an absolute sense) *should* be manageable with the minimum cast thickness (~2mm?). You can use a less expensive epoxy (or maybe not worth it) to bring the concrete base height up to the bottom of the steel inserts. I've also heard/seen (but never done!) "scraping" with an angle grinder. Maybe not good for hitting super high points per area like hand scraping, but this is to bolt the huge rails down as opposed to creating a nice uniform oil surface. Good luck! I'm embarking on my own router and appreciate all the shared knowledge, good or bad.
Epoxy might also be a thing to try. But for that I would need a proper master surface and I don't have that. I got some other recommendations that might be a solution and don't require anything I don't have. The good thing is I can always try the epoxy if all the other things failed.
I think the suggestion is to use gravity to level everything. It’s popular on a UK CNC forum. Build a channel connecting both rail surfaces, build a wall around the rail surfaces. Then pour in the epoxy to get two surfaces flat and co planar
I plan to start a project similar to this, I am currently collecting information. Could you say which composition you used to make the structure? ("concrete...?") Thanks
Is it Possible to make a temporary grinding tool that mounts to the one good rail to machine the surface of its pair. it would not be exact with such large overhang but would bring the run out much closer for scraping. I wish you luck and hope for the best.
Awesome project. In order to decrease time for scraping you can try to use Zedex or any other competitor's product. It is a polymer. Easy to scrape. And can be glued on top of your inaccurate surfaces.
This is what I did with a surface grinder column ways. I built a template from steel and scraped it flat and parallel on a 2X3(600X900mm) surface plate. Then I used the template to scrape the four sliding surface parallel and perpendicular. I got them square and flat within 5 microns (2 tenths). I checked the straightness with a 4 foot (1.2m) granite straight edge lab grade (AAA) 2 micron flatness full 1.2m. I checked the squareness with a 12" (300mm) cylinder square. It took me 2 months for template scraping and ways scraping. The ways were twisted and tapered. Twist was about 20 thou (0.5mm). Taper was 15 thou (0.37mm). So there was a lot of scraping all done manual
Hi, is concrete dimensionally stable enough for such purpose? I know it shrinks during the curring process, but don't know for how long. Maybe this is the reason of warping? Many use epoxy granite for CNC bases, as it is maybe superior material, but I suppose it's too expensive for such big machine.
It is a special concrete (Nanotec UHPC). I found some researche about that where it was more dimensional stable than epoxy granite. The nice thing about the concrete is, that you can mix it with a normal mixer and it doesn't need any special treatment after poured.
Hi! Thank you for VERY interesting videos!! I did need to see them to trust that tis kind project is possible to do bt one man, thank you again!! Where did you find those linear bars? I cannot find these from anywhere, please help me..!!
Hi Alex, very impressive project! I am building one soon, i am finishing my CAD file. Can you tell me what aggregate you used, like what size of sand and in which mixing ration ? Thanks !
Awesome project. Would you mind to share the link where to buy the linear rails like your or their specs? I can't find them in this size. Thanks a lot.
Hi, the rails are size 35 from Schneeberger and INA. Got them as new old stock quite cheap. You can have a look in their catalogue. But I do not really want to know what the would cost new 😅
Achei uma ideia excelente usar concreto comum para a base da máquina. É barato, relativamente fácil, mas o concreto não é ruim pra absorver vibrações? E o concreto quando curado não contrai, deixando as partes chumbadas com uma brecha? São dúvidas que eu queria tirar, pois tenho interesse em fazer como você fez, mas tenho estás dúvidas. Muito obrigado, e agradeceria mais ainda se vc pudesse me responder, obrigado.
how the hell did that happen?? werent those surfaces milled in place at the exact same time in the exact same fixture? the machine looks incredible soo far, im jealous.. I guess that .2mm would irk me also , seeing as how everything else was going well.. I wonder if you could surface grind the tops of the linear rails and correct the issue , or would the side ball guides still cause it to ride high? subscribed
Unfortunately tops are ground only for the looks and tight dust cover fit. Grooves on the sides are very precise, sometimes even matched to carriage and so shouldn't be touched. The only surface left is bottom of the rail, wich indeed could be ground to correct twist but would require a lot of unbloting and bolting back in to check progress. If the rails have to be removed anyway I think it would be better to attack the real culprit. BTW I'm also wondering how something like this could have happened, especially same error on both parts.
Can you give some information where you bought your UHPC? myuhpc or durcrete? Did you think about shrinkage over time? UHPC tends to do this unless it's been annealed (48h at 90°C) Very nice project!
I have used the nanodur (durcrete) UHPC. I let the parts dry and shrink over one month before machining. And in the end it took about half a year before I finished all the important surfaces. So I do not expect a lot of movement anymore.
Very impressive build! What is the linear rail size on all Y, X and Z-axis? I understood the square profile is 50x50 so the rails must be 35mm? Also what is the brand of the rails.
Hello, awesome project. I would appreciate if you are be able to share with recipe of epoxy granite and which resin did you. I look forward to hearing from you.
You should try to find surface grinding . But you need to do slot 1-2mm for grinding wheel out. In this case you ca made very accurate surface and side support.
Hi Alex! I have question regarding thermal expansion: Is there potential problem with combining UHPC concrete with steel parts (or is it cast iron?) in it? Something like a "bimetal" effect? Thanks! Milan.
Awesome project! Don't despair. Scraping isn't hard to learn. I learned to do it in less than an hour. The only challenge is scraping multiple sides, but your scraping job is among the more simpler challenges. You only need to ensure the bed and column ways are perpendicular. I would get a large flat granite piece. You can get them for cheap from a kitchen counter fabricator. If you still want someone to help you, try contacting Stefan Gotteswinter. Btw, he demonstrates another method where he uses epoxy as a "liquid" shim.
remove the linear rails, have the bottom of the rails machined in a way that will correct for the discrepancy? Or add a block of steel between the body of the machine and the rail and have this machined to correct for the twist instead of having the entire block machined.. this way you don't have to take the whole concrete slab to the big milling machine again. What you think?
I've seen several videos! but I can't know what kind of concrete or epoxy is used because nobody shows them in detail, and what about sand? and cement?
Build two dams or troughs around each of the mates (ply/mdf with a release agent works) and pour about 3~5mm deep and low viscosity epoxy grout on them, make your own grout with epoxy and carborundum, aka silicon carbide, 600 grit, an extremely fine and hard dust. You need to choose a very thin epoxy with a slow set up time and pack it with the grit but not so much that it becomes a mastic paste, it needs to flow still, aim for the consistency of warm black honey. It will flow to perfect flatness, I.E as flat as the curvature of the earth which for the spanned distance is certainly more than good enough. If you connect the two troughs with a channel between them at each end you will also achieve perfect parallelism between the two. This is how it is done with extremely large machines with rail mounts too large to mill, and also, my own. Premade mixtures also exist, look up epoxy bearing grouts like chockfast, nordbak etc. P.S. You can coat your rail mount bolts with melted beeswax and screw them into your mounting holes prior to the pour, they will release after it sets leaving a nice wax coated hole.
wanted to do that, but some people said the epoxy wasn't hard enough compare to steel and will deform under the linear rail. gpa of epoxy is 4 steel 200.
@@m3chanist well yes maybe, but the fact seems real about the stiffness. And the fact I coudn't see a single company doing this. And I searched quite a lot.... If you have more information to help look in to it. I would greatly appreciate. thank you.
@@jesuismika It's not real, it is standard procedure for every large project with an entire industry of epoxy grout manufactures, some of whom I listed, that service the sector, EG mills large enough to cut ship sized parts, National sized dam/generator projects. Rest assured, if it's good enough for that scale of project, it is more than good enough for a home gamer.
@@m3chanist well I guess it depends the expectation, a huge machine won't have the same tolerance as a small one. And maybe if you are talking micron, the epoxy would deform a tiny bit, which is acceptable for some project, and maybe less for other. I'm trying to see the solution to be in the micrometer perimeter with the less equipment possible, if possible. ;-)
You could have used industrial steel section box those sizes as horizontal/vertical beds. Drilled, tapped plugged & machined; then filled with Granite concrete aggregate or concrete. The concrete idea will be constantly moving about like a machine shop floor. In answer to your problem make a x-z axis plate 3" /75mmm thick flash ground of graded steel commercially available - take your guide rail system off x/z and fix to the machined plates 3 mm deep slots as soft mallet fit ~ and drill and tap and locate your linear guide rail/s system. [ reason is that the concrete hasn't the inertia power under the creeping process of settling concrete to warp steel plates. Steel cover plates stronger than concrete. ] I hope this of an idea for you Sir.
Concrete or granite- concretes Etc, will always shift and settle due to the expansion of air and water into it. At the design conception of the machine, the granite or concrete suppliers give an expansion and contraction quota for its [ properties of materials ] Chart. At 10 microns you said adrift at present is good enough for general and precision work. You would have to be in an atmosphere controlled environment on C.M.M Beds to even reliably measure this! Further saying that it's possible outside of this is misleading and confrontational at best. Otherwise, machine builders, who spend millions on inspection equipment and controlled environment to ascertain machine tolerances in manufacture. It best to remember their claims to accuracy at best is as good until the machine is outside the door as the warranty ends right there with them. In part due to the Cold War and nuclear accidents throughout Europe Canada the States Australia china Russia back in the 1970s - mid80s. For Eg, You machine a Flange to your gauges and send it by post to as many states and regions as to spend two days in each area of your country. then see if it fits as transitional fit on its test rig you made for it in your home factory as stabilized. This tells you the time in space and time you exist in your home workshop different to other postal codes in your area. linear vertical and radial warp in the test piece is relative to sight and organic matter called FAR OUT Funny shapes and sizes people heads faces extremities to Cherknoble reacter blow up is relative to all materials is the point of the test.
Impressive build! Have you considered using steel filled machine epoxy such as DWH to set the linear rails on to achieve perfect accuracy? This could be a good solution for you and much faster than scrapping.... All the best.
Another possibility if the two surfaces are close to flat would be to lap them in. You can get large old cast iron laps for not too much money on ebay, and use it across both surfaces at once to even out the twist. Just start with a more aggressive grit lapping fluid to move more material
I have an edge on every side for the linear rails to be pushed against. So the plate must not be wider than 250 mm to fit in between. I have started scraping now and so far it goes like expected :)
I have read most of the comments previously written up until now. Maybe you can shim with something more pliable (easy to deform), like copper or tin. If you are able to fix the issues you can use the same method in conjunction with some adhesive like others have suggested. Just my two cents of thoughts. I'm sure you will figure something out. Good luck!
Hello Alex, I'm currently designing a cnc gantry mill with the sole purpose of machining steel and aluminum. Roughly a 20" x 20" x 10" machining envelope. I will use concrete like you to add mass and rigidity. I really like your cnc machine do you have a design of it? My biggest worry is will my gantry be rigid enough. I am thinking of sharing my design now for collaboration from other builders. Did you collaborate with others to fine tune your design? Great job on the mill and thanks for sharing... Mike - Plus8Precision
Usually I'd mount the carriage in a vice and indicate bottom of the rail to see if groves are parallel to mounting surface first. If they are the error can be caused either by one rail going up or twisting away (and it's probably the latter as you said). Instead of scraping you could just stone the surfaces, applying more pressure and spending more time on the high side. If it really is twist, then indicated 0.1mm means "only" 0.02mm to be removed.
The rails are good. I swapped them and the result is the same. They are also a good brand. I would need to remove about 0,025mm. I shimed 0,03mm at the end and that was to much. But I don't really have the right tools to measure. I need to remove the twist, get the surface flat and parallel (and same height) to the other surface. Nut I have another idea I could try without moving the parts. Need to check if that is a valid option.
@@AlexCNCen I thought some more about your problem. It is possible to calculate how much of what you are measuring is twist and how much is a height difference by measuring on the inside strip and outside one, then comparing results (or even measuring ends of a parallel attached to the carriage to exagarrate the error). Doing this and adjusting distance between rails should be enough to make linear rails happy (not fight against each other all the time). The real problem is that by doing only realative measurments one could accidentally make a smooth, repeatable... sphere. Do you have any straight edges or time to make one? BTW I'm just throwing in some ideas, hope you find a way to fix this and continiue with the build :D
I have some more measuring tools including a 500mm straight edge. I just showed one measurement in the video. Would have taken to long to show everything. I will try to measure more exactly where are the problems are so I know better where i have to remove material.
How about working a calibrated granite surface plate inverted onto the rails with fine grit water paper ? Blue the plate up first just like you would if you were going to scrape it, then just work it back and forth. Then scrape to final flatness. Might just work. ( I read below that the inlays have edges. maybe just grind those off before you start, then you would have to make some other plan for alignment) good luck
The problem with that is that edges for the linear guides on both sides. So I would ned a plate that exactly fits between those. An Option would be to remove those. But I'm not sure about that yet.
@@AlexCNCen maybe just remove the edge on the one side so you still have your reference, then once the one rail is mounted, it should be easy to use a spacer to align the other side to that.
Hi Alex, great project! Can you give us more details about concrete recipe? Is it possible to make/find all ingredients somewhere else or only buy from specific company?
I will make a video about the concrete. I made a test piece where I try to measure stiffness and strength. But not sure when. Free time is expensive at the moment ;)
Nice project Alex. I look forward to seeing it progress. Forgive me if this has already been asked but what concrete have you used? I've been researching how ordinary portland cement corrodes steel and your remarks about how it "flows" tells me this isn't just any old concrete :)
... or you can collaborate with somebody who have some experience in scraping and proper tools. it may be a good variant overall. ( 1. you don't need to move your machine 2. you learn and practice in scraping 3. you get more content for youtube )
Don't worry about the surface in the base, just take the rails off and grind them to match the base. Instead of small shim, surface grind more on one side of the bottom of the rail.
Hm that would be super difficult to exactly get the imperfections on the rail to compensate for the ones on the steel inlay. Especially when I'm not able to measure it accurate ;)
@@AlexCNCen can't you use one rail top, indicate along the other both sides, mark it, remove bottom material, then you are close. Then scrape plates in concrete in situ. Also are you sure you cannot zero this out in software?
I don't think this is a big issue. Although you read approx 0.1 mm deviation on the dial indicator, we're talking about a deviation of approx 0.01 mm on the guides foundation. If you're sure that the origin of the deviation is the foundation twist of one guide (and not an overall flatness error), I think this won't affect too much to machine accuracy.
Hi Alex, are you able to share your CAD files? I’m planning a very similar build and it would be a great starting point! I’ve already checked out your friend Sebastian’s project. Thanks!
Hi Alex. Great build so far, following along on TH-cam and on Instagram. Just a question, I have been thinking. Of doing a cnc with epoxy granite, but after seeing your build, thought that the concrete route might be a cheaper route seeing that epoxy is very expensive. Just wondering how you were able to get the mixture for the concrete to have the right strength for what you are using It for. Also is it a special type of concrete or can I use the 52N concrete for the mix? Anyway. Hope you can help. Cheers.
Hi, it is a special concrete. You can find information about it with Nanodur E45 UHPC. Its noticeable more expensive than normal concrete but easier to handle than epoxy granite. That was the reason why I use concrete instead of epoxy granite.
Thanks for the reply! I went to check out that product and it helps a lot. Looks like here in South Africa we don't have those kinds of products. I have read some papers on how they created Uhpc with local products so might be able to come right with that, otherwise the epoxy route will be the way. One more question. Did you do anything extra with the drying process, or did you just let the mold stand for a few days? WAs. Wondering if it needs to be heated up to get rid of all the moisture content in the concrete to get it more stable.
@@banzaiwill7672 I haven't done anything special. It is important to wrap the parts in foil for some days so it does not dry out. The added water is only that much to finish the curing process. During the first day the parts got quite hot by themselves. Something around 50°C. After that you should wait at least a month before machining the inlays. The concrete shrinks a little bit in that time and is than stable.
Just remove your rails and install a precision ground plate about 25-40mm thick and re-fix your rails to that.You will loose some spindle height but it will be flat.You can use longer screws to fix through your rails through the plate into base rails.You can shim the plate to it's final position and then epoxy it so it becomes rock solid with your base.The epoxy cement will act as grout to fill in the gap between the plate and your base rails.
I just started scraping the ways. Looks like I can get it pretty precise with that and won't have to alter anything in the construction. In case that does not work for in the end I can still try something else.
Hi! Any news yet ? If you lived in sweden I would help you face it. I would probably take it out again and mill it with a large facemill and also check it with a dial indicator before I remove it from the machine. Easiest would probably be a large knee mill.
Maybe Grind the botton of the rails ? Dont mill it ....put it in surface grinder. Or make few mm /cm flat steel / aluminium bar as a shim :) (or just make 1 big few cm thick steel plate) for me second option is the best idea becouse even if u make if wrong on 1 side ....u can remake it :)
I don't know if you've solved this yet, but maybe this could work: Alternate version of scraping it in. But hire someone to create an ultra precise hand scraped (or if good enough, ground) master to match the rail mounting surfaces (as they are meant to be) and have them ship it to you. That way you can expand your geographic radius of who can help you. Then you take over and blue the master, run it along the mounting surfaces (both the flats and walls) and clean them up until they are perfect (or good enough). This would require less metrology tooling on your end, and less effort and skill in scraping. Also things of this mass are flexible, especially on the scale that you wish to achieve precision, so moving it out of your basement and back can reintroduce errors, even if the gridding or milling you do again is better than the first time. So leaving it in its final resting spot is best to then scrape it in. Ignore everyone saying to touch the rails. These rails, no matter how expensive, are not designed to be perfectly straight or flat or whatever. They rely on your mounting references to achieve straightness and parallelism. The only true relationship between surfaces on the rails are the bottoms, one side, and the bearing races at any "2d cross section" (because again, they can be ever so slightly twisted or bowed). Any other surface is not useful as a reference. that is an incredible job so far, i understand the feeling of finding out your reference surfaces are not up to the job. a few months back i "finished" a CNC router only to have the gantry immediately rack and kill the ball nuts on both Y axis'. Of course I was using aluminum extrusion and aiming for a level of precision 1 to 2 orders of magnitude lower than you. But spending a year designing and building only to get so close to done and then catastrophe is heart break. I agree, do it right, and do it once. Good luck
Thanks for your suggestion! I will definitely try to scrape it. I already borrowed a big enough straight edge from a friend. So I have all the metrology equipment here to get the surfaces flat enough. I also tested scraping on some mild steel already and it worked quite well. Now I only need some time to continue the project :) Best Alex
If it's a twist. Buy a reasonably sized surface plate, and some fine grit sandpaper. Tape the sandpaper to the surface plate. Lay the surface plate across both rails. Elbow grease. Not perfect solution, but effective.
I have an edge on every side for the linear rails to be pushed against. So the plate must not be wider than 250 mm to fit in between. I have started scraping now and so far it goes like expected :)
A linear guide is statically over-determined. _Four_ linear guides even more so. The dial-gauge test proves nothing. You should not be using the rZ-position from the linear guide in the first place. Are you _sure_ that the rails are not coplanar (ignoring the rotation-direction)? You could test this by having a laser-pointer on a carriage with three (not four!) balls, two resting on one rail and one on the other. When you slide the carriage on the rails, does the laser-pointer deviate substantially? If so, your recourse is to shim the second rail to make it parallel. In any case, the carriages on the linear rail must be decoupled in all but 5 DoF total (e.g. 2 + 2 + 1 restricting degrees) if you hope to achieve precision at all. Dan Gelbart has a nice video-series on TH-cam, and has designed and built his own ultra-precise lathe. Perhaps some inspiration there?
It appears Alex is using roller guides, Roller guides are very sensitive to binding compared to the ball type they must be very carefully mounted. this requires that the saddle and guide way mounting surfaces are within the tolerances stated in the manufacturers data sheets. Many engineering text books suggest the use of more than three points of attachment is kinetically redundant. Maybe for designing an instrument three points is a good Idea however where heavy dynamic loads are being carried more than 3 is the norm. 3 wheel cars have never taken off because they tend to be quite unstable and unsafe when cornering. Machine tools have similar issues particularly CNC machines with high acceleration. Regards John
@@johnmcnamara3719 Agree both comments. Measuring setup is defenetly not relevant. Magnet base should be placed on the block over straight edge (check nsk rail installation video). I'm pretty sure most of this reading comes from rails plane angular screwing. Same time, roller liners have tight tolerances for inplane. It would be hard to make this two surfaces laying in one plane, since they are protuded and connected to heavy concrete base. No ways to check planes with surface plate. Basing big straight edge over ballscrew mount surfaces is also not best way to fix this issue, because those ballscrew mount surfaces should be scraped after rails have been properly installed.
@@kirillkirillov3809 A small surface plate and blue will reveal the errors, the steel rail supports set about 5mm above the casting. allowing the test to be performed. For the Y axis it would be better to lay it flat then lower the small surface plate onto it.
@@johnmcnamara3719 there are side support surface outside both rails. Will be hard to find surface plate that will fit inside this pocket. Also, not sure if ballscrew support surfaces should lay in same plane.
@@kirillkirillov3809 Unlike the machine built by Mr Sebastian Ent The machine Alex has built appears in the video to have flat milled surfaces. Maybe a friend nearby could lend a small surface plate for the test.
Hi Alex; let me tell how I would tackle this situation hoping that maybe you would find something useful in my explanations . 1) I would look for the ''best'' spot along the inlays where I could find 3 points, and I will explain the reason for them in a short while; these should be 2 points on 1 inlay and another 1 point on the other inlay. Where about these 3 points could be found is debatable. They could be found at either end of the inlays or, why not, right in the middle of the inlays. These 3 points should be chosen so that to be as close to one another as possible. More precisely: if , say, your X axis is 500 mm, I would choose the first point at 240 mm on the LEFT inlay (measured from the Z axis outwards), the second point at 260 mm on the LEFT inlay as well, and the 3-rd point at 250 mm on the RIGHT inlay; 2) Say that your X axis inlays are 300 mm spaced apart from one another; I would make a rectangle out of ordinary steel, about 500 mm /300 mm and I would drill/insert 3 bolts and cap nuts (semi-spherical nuts) at precisely 240 mm and 260 mm on one longer side and at 250 mm on the other longer side of the rectangle; 3) I would attach on this rectangle as many dial indicators as I could afford, perhaps at least 4 dial indicators on either side, spaced apart evenly and I would place the rectangle on the 600/430 granite plate that you already have; 4) I would set all the dial indicators to zero and than I would transfer the rectangle on to the 3 points on the inlays described earlier; this way , the lack of coplanarity becomes measurable and you can start scraping the inlays because you would know where and how much. Here is the thing: it wouldn't make much of a difference if it was in the middle of the inlays or 100 mm in or out, provide that you would place the rectangle every time in the same spots. What matters is for the 3 points to be as close to one another as possible because it is this closeness that makes it a viable solution. Have fun !
Thanks for the advice. I just started scraping and used a machinist level for that. Looks like that works pretty good and with a resolution of 0,02mm/m per division it should be accurate enough to get the surfaces to a flattens of 10μm or better.
Hi, What kind of shirt is that and where can I get one? I love it! Also I feel your pain on the email thing. I feel if I used my company email that I would get a better response, but it seems that companies dont want to deal with people who dont have a company email :(. Its a shame too. I was looking for cast iron and wanted to get quotes.
I figured out how to make it perfect bud! First pick the 2 rails that are PERPENDICULAR TO EACH OTHER...2nd put your indicator on THAT RAIL,,run it all the way down the bad side writing on the rail and a pc of paper the amount it is out INCREMENTALLY and mark front and back etc. After you figure out exactly were it is (hopefully not wavey) you can flip it over and shim exactly what the error is and mill the bottom....make sure you machine the HIGH RAIL !!! HAHA,, i would ether hold it in a few vices with jacks under it or on magnets. Hope this helps! I really wanna build my own as well but cant decide on a size yet...ive done large custom machining on mazaks ,Nicolas correa's, n Cincinnati's for the last 14yrs,,n 12yrs in job shops making little parts before that. Remember measure once,,cut twice! Haha!
Nice machine, why not just strap a grinder on a cross slide, and mount it to the z axis then surface grind the 0.1mm off. Use a paint pen and paint the ways, then grind both. It's like surface grinding on a tool grinder.
Thanks for all the suggestions in the comments. I decided to scrape the surfaces and everything looks fine so far. There will be a video about that soon :)
great progress 👌👍
can I reupload your build playlist in arabic to youtube ??
I remember Stefan Gotteswinter describing use of a two-component epoxy for machine alignment, the way he did to tram his vertical milling machine. It's what I'll likely be using to build my own traveling gantry machine which, I have to say, I'm patterning after yours. It's a privilege to watch these - I'm grateful for the insight your videos have given.
First I will try scraping. I can still use the leveling epoxy if scraping fails.
Wow, finally someone talking about the leveling problems they are having, and how crazy it actually is to get these things corrected.
Thank you for this video, as I am having the same problems with my lathe build, and all of the videos out there doing these builds make them look so simple, but it really isn't when it comes to the fine tuning and leveling...
Huge pain in the ass!
Awesome project Alex wish you luck with the problem solving!
i got some nema 34 1600oz stepper motors that should be plaenty for a lager CNC mill?
Great to hear you're not giving up on the project. There are always setbacks, though this seems like a big one, there is a way through. Looking forward to seeing what you decide to do.
Hi Alex,
I really like your machine! You could make your own lapping tool by drilling a hole in the centre of your base between the rails. Get a piece of 5mm steel plate cut to a diameter large enough to cover all your steel mounting pads and place a hole in the centre so you can insert a pivot pin. Cast a concrete slab onto your circular plate maybe 5 or 10 cm thick for some weight and to stabilize the steel disc. Have the disc turned or surface ground flat. Now coat the disk and your mounting pads with lapping compound or valve seating compound and turn the disc a number of times by hand until you see your mounting pads have been flattened across the majority of their surfaces. You could also ask to borrow a granite surface plate from a machine shop. If you know where the surfaces are too low, coat those with a thin layer of slow curing epoxy. Cover the granite surface plate with plastic and the turn upside down and place on top of the mounting pads. The weight of the surface plate should force the epoxy to only fill the low spots. Hope this gives you a workable idea and good luck with the rest of your build!
There is one problem. I have edges in the inlays to press the rails against. So I can not use the lapping method. But I got some ideas here that will work with the edges and are quite easy to try.
Hope you solve the problem. Best of luck! Looking forward to this great project.
Could you CNC the bottom of the linear rail, instead of the top of the steel inlay? Much easier to get that to a machine shop. Or failing that, add a ~3mm spacer bar between rail and inlay on both sides, machined to compensate for the inlay's imperfection?
I'm not sure if I would machine the rails. But the spacer piece sounds like a valid option to try. I already got some recommendations what I could try. This one is also on the list now :)
Sounds like it would induce wear/slop
I can guess how badly you felt when you started finding faults in your massive project.
Hope you find a good, quick and not overly expensive fix.
The machine looks great! What a suprise! Thanks for sharing. Hope everything further goes well.
If you are confident that your left rail is straight then move the indicator base to a left cart, remove the right rail and start with your highest point and start working it down and measuring as you go. You don't have to scrape. I remember an amateur scraping video where he used a Dremel with a sanding drum for roughing in. Maybe just a stone with some WD40 by hand like someone else recomemded.
The best thing I heard from you in this video is "I won't give up". 👍
Life is a challenge Alex , good luck, Terry from Scotland
Thanks for sharing your lessons learn, I plan to build my own machine soon and your video is a good help.
The problem with casting with wooden forms like plywood, is that the pressure of fresh concrete can move your parts, wood is compressible and easily warp.
The better way as a lesson learn you can use rigid form like steel plate reinforced by steel rods and nuts in betweens. Epoxy grout is also better alternative and can pour slowly by layer to reduce pressure.
You can do the granite epoxy layer by layer? Like let it dry overnight and then pour another layer? That's good to know.
@@ChristopherJones16 there is epoxy grout for big layers like 50 cm. Check mapei planigrout 350, i’m planning to use that for my cnc structure.
@@marson8870 can you elaborate on the steel rods and nuts reinforcement? Are you saying to have the steel rods inside the mold to reinforce the actual object were pouring with holes drilled in the plates for the rods to stick through and then tighten up nuts on the outside of the mold to those rods basically self clamping the mold together. Then when it’s cured you cut I off the excess rods sticking out of the poured object?
Or are you saying the rods will be outside the mold. I’m guessing the two side plates will overlap the front and back plate and then you use a rod to pull the two overlapping plates together to hold them pressed against the front and back plates? Sort of like an “H” shape but where the “-“ is in the “H” it’ll be a square/rectangle.
I’m probably off on both. Also what’s a minimum thickness for these plates we can use? Wondering if I can pick these up at Home Depot or if I’ll have to special order these.
@@ChristopherJones16 you can simply make a steel skeleton and then pour the resin into the mold.
There is also a uhpc with a young's modulus of 80 gpa, a little better than aluminum. With that you need steel only for the rails base. You need a nanodur cement and an artificial gravel called Durigid.
Lovely that you explain how to hoist heavy parts to the downstairs milling mahine.
Hello! Serious project!
Is the information private? What concrete or mixture did you use to make the machine bed?
Have you noticed any deformations or changes in the geometry of the machine over time?
Here is an idea if you can get one of the rails true with shims you could attach a bench grinder on the rail (maybe use the second rail as another axis to attach the bench grinder on) and grind it flat. Then reverse it and grind the first rail flat. Then use a flat piece of granite or iron with a precision level to check it. Just make sure you level your machine before you start using some geometry on the column. This idea is based partially on the jank o grinder from nyc cnc so watching their video may help.
You have a Typical modern day way of thinking and solving problems . You have to many options . Today people have a tool to do anything . You have the answer to your problem. Scrape it you will be surprised it will take you a lot less time and work than hauling that back outside . It is a slow process but you will learn and appreciate it more. Scrape measure scrape measure. Every machine tool of old was hand scraped. I'm retired now but my fist accurate item to make was a one inch cube out of cold rolled steel cut from a 1.25" plate, I was about 16.The only tools we were allowed was a hacksaw and one file. For ever one thou you were out of square from all sides, was one mark off. Good luck with your project it looks awesome.
It's not the scrapping he's is afraid of, its the wandering concrete will continue to wander about. After about 15 years concrete stabilizes an easy mistake as an afterthought ~ we have made clever after the fact!
@@mcadamdavid1 you mean concrete is not a good option? is not clear to me what "wandering concrete" means.
@@aion2177concrete has water in it, and it takes decades before it's all gone. The method used mostly is epoxy/granite, which has been the choice for over 20 years
Scraping is a decent option but i dont get the weird roast towards using modern tools
Hi Alex, very impressive project!
If the error is soo small, like you say, 10microns or less, yo can try to solve with fine sandpaper and trial and error, but its very time consuming.
hey alex, i made a CNC, i had the same problems as you, i grinded both with a big granite surface plate, i glued two 80 grit belt grinder paper and grinded by hand pushing back and forward 1000 times, eventually i had nearly 0.05 on 600 or 700mm length
Have you tried contacting Stephan Gotteswinter? He definitely knows scrapping very well, although that might be too big of a project for him. Maybe he knows someone with a big surface grinder that can help you.
Greetings from Portugal. Looking forward for more progress on this beast.
I already got in touch with him. Unfortunately he has no time at the moment for something like that and also not really the right measuring tools for something that big.
@@AlexCNCen Yeah, that's what I figured. Scrapping is a long process.
I'm sure someone will help you figure that out.
Stephan has also talked about and showed how to true machine surfaces with a plastic compound that can be easily worked to get perfect long lasting surfaces. I forget what it's called but he says it's used in the industrial field for bringing precision surfaces back to perfect.
How did you manage to get good and true holes on drill press without cross table? I could never get serious precision on drill press.
Have you considered using a large thick panel of plate glass, glue a sheet of automotive sandpaper and then use it to rough lap then fine lap the rails?
The problem with that is that edges for the linear guides on both sides. So I would ned a plate that exactly fits between those. An Option would be to remove those. But I'm not sure about that yet.
@@AlexCNCen you may need to accept the inevitable and remove them and have them surface ground
Hey Alex! You are pioneering a new field with new problems. Don't worry you'll figure this out. :) One suggestion I have: Ever see how woodworkers flatten large slabs of wood that cant fit in a planer? They use a sled to make the surface even. I'd say you have the talent to make a very precise machining sled which attaches to the sides. You'd then be able to bring both surfaces to the same height with a cheap hand router since the depth isn't that much (< 0,5mm)
Thanks. I will try scraping first. In theory you can get the flatness down to some μm. If it doesn't work I can still try something else :)
Hi Alex, hope your doing well. been watching for a long while, since you were building your first mill. Sorry to see that your having this problem. First it should be said that this error is actual quite small about .001 thousand twist total. Matthew here in the US state of Wisconsin, 40 years career machinist. Wish you were closer id jump in the car and come give you a hand. The error looked much worse than it is because as you have your indicator set up it magnifying the error by 8 x., but its probably enough to cause the block to bind up when all 4 blocks are bolted up tight. The best way to fix is to hand scrape, the scraper that i use i made from an old file with a small piece of carbide held in with a small clamp. For a master look for a hardened and ground parallel 1 inch by 2 by 12 long. Should only take a couple of hours to straighten it out, don't know if its just the one on the right or both. Ill check back later to see if you found this, post any questions if you have them. Oh ya celifane off a pack of cigs is .0005. good luck. Matt
Thanks. I already have everything here I need for scraping so that's the way I will go. I did a test on a small steel part and there scraping worked fine and I was able to get a flat surface quite fast. But first I want to finish some other projects before scraping.
Best
Alex
It would be nice if you explained why you used UHPC rather than epoxy granite. Which is better in terms of accuracy in the long run for a precision CNC machine?
The UHPC has a lower tensile strength but is a lot easier to process. A normal concrete mixer is everything you need. Long term stability should be pretty much the same. Maybe the UHPC is a little bit more stable. I mainly choose UHPC because of the easy processing.
Perhaps shim with epoxy?
That'd be my first recommendation, if scraping is out. There's a few companies out there that do self-leveling epoxy for built-in surface plates, and the small deviation (in an absolute sense) *should* be manageable with the minimum cast thickness (~2mm?). You can use a less expensive epoxy (or maybe not worth it) to bring the concrete base height up to the bottom of the steel inserts. I've also heard/seen (but never done!) "scraping" with an angle grinder. Maybe not good for hitting super high points per area like hand scraping, but this is to bolt the huge rails down as opposed to creating a nice uniform oil surface. Good luck! I'm embarking on my own router and appreciate all the shared knowledge, good or bad.
Epoxy might also be a thing to try. But for that I would need a proper master surface and I don't have that. I got some other recommendations that might be a solution and don't require anything I don't have. The good thing is I can always try the epoxy if all the other things failed.
I think the suggestion is to use gravity to level everything. It’s popular on a UK CNC forum. Build a channel connecting both rail surfaces, build a wall around the rail surfaces. Then pour in the epoxy to get two surfaces flat and co planar
@@ewildgoose this is exactly what i thought.
how % difference and means to performance of your machine from using linear rail guideway @ 25mm instead of those bigger you use now ?
I plan to start a project similar to this, I am currently collecting information. Could you say which composition you used to make the structure? ("concrete...?") Thanks
Is it Possible to make a temporary grinding tool that mounts to the one good rail to machine the surface of its pair. it would not be exact with such large overhang but would bring the run out much closer for scraping.
I wish you luck and hope for the best.
That is my favourite at the moment. I have to measure a bit more if that could work.
Awesome project.
In order to decrease time for scraping you can try to use Zedex or any other competitor's product.
It is a polymer. Easy to scrape. And can be glued on top of your inaccurate surfaces.
can you post a link? All i can find is a plastic for making generic plastic parts
Yes, scraping! Always the best solution. You will be an expert by the time you get that all square :D
I would use a surface plate and scrape the surfaces for flatness and parallelism
This is what I did with a surface grinder column ways. I built a template from steel and scraped it flat and parallel on a 2X3(600X900mm) surface plate. Then I used the template to scrape the four sliding surface parallel and perpendicular. I got them square and flat within 5 microns (2 tenths). I checked the straightness with a 4 foot (1.2m) granite straight edge lab grade (AAA) 2 micron flatness full 1.2m. I checked the squareness with a 12" (300mm) cylinder square. It took me 2 months for template scraping and ways scraping. The ways were twisted and tapered. Twist was about 20 thou (0.5mm). Taper was 15 thou (0.37mm). So there was a lot of scraping all done manual
Can't wait to see the next video. Great job. Scraping really isn't that difficult, just time consuming.
Hi, is concrete dimensionally stable enough for such purpose? I know it shrinks during the curring process, but don't know for how long. Maybe this is the reason of warping? Many use epoxy granite for CNC bases, as it is maybe superior material, but I suppose it's too expensive for such big machine.
It is a special concrete (Nanotec UHPC). I found some researche about that where it was more dimensional stable than epoxy granite. The nice thing about the concrete is, that you can mix it with a normal mixer and it doesn't need any special treatment after poured.
Hi! Thank you for VERY interesting videos!! I did need to see them to trust that tis kind project is possible to do bt one man, thank you again!! Where did you find those linear bars? I cannot find these from anywhere, please help me..!!
Hi Alex. Can you tell me exactly where you got Nanodur cement from and how much it cost? Same with the additive used in the recipe. Thanks.
Hi, I got the cement and additive from moertelshop.de. The rest like sand I got from a local hardware store.
THIS MACHINE IS INSANE ! ( in a good way )
Is this UHPC or "normal" Concrete ??
and Greetings From Finland
Hello. Congratulations on your work. Can you tell us what brand of linear rails did you put on your machine ?
Where did you buy the rails?
Hi Alex, very impressive project! I am building one soon, i am finishing my CAD file. Can you tell me what aggregate you used, like what size of sand and in which mixing ration ? Thanks !
Awesome project. Would you mind to share the link where to buy the linear rails like your or their specs? I can't find them in this size. Thanks a lot.
Hi, the rails are size 35 from Schneeberger and INA. Got them as new old stock quite cheap. You can have a look in their catalogue. But I do not really want to know what the would cost new 😅
What size are those rails?
Achei uma ideia excelente usar concreto comum para a base da máquina. É barato, relativamente fácil, mas o concreto não é ruim pra absorver vibrações? E o concreto quando curado não contrai, deixando as partes chumbadas com uma brecha? São dúvidas que eu queria tirar, pois tenho interesse em fazer como você fez, mas tenho estás dúvidas. Muito obrigado, e agradeceria mais ainda se vc pudesse me responder, obrigado.
cara, vc obteve alguma resposta sobre a composição dos materiais de enchimento que Alex utiliza em sua máquina?
great video! i've start getting the materials together for my build in australia :)
thanks for the insights
There is shimming stock called onion skin available on ebay. As thin as .0002 inches. CNC4XR7 used it on his linear rail conversion.
Thanks. I will try to get some 5μm shimming stock. That will be also helpful on other parts.
Great work Alex, what type of concrete you use, and what are its technical specs
Thanks, 👍
Uhpc
how the hell did that happen?? werent those surfaces milled in place at the exact same time in the exact same fixture? the machine looks incredible soo far, im jealous.. I guess that .2mm would irk me also , seeing as how everything else was going well.. I wonder if you could surface grind the tops of the linear rails and correct the issue , or would the side ball guides still cause it to ride high? subscribed
Unfortunately tops are ground only for the looks and tight dust cover fit. Grooves on the sides are very precise, sometimes even matched to carriage and so shouldn't be touched. The only surface left is bottom of the rail, wich indeed could be ground to correct twist but would require a lot of unbloting and bolting back in to check progress. If the rails have to be removed anyway I think it would be better to attack the real culprit.
BTW I'm also wondering how something like this could have happened, especially same error on both parts.
Can you give some information where you bought your UHPC? myuhpc or durcrete?
Did you think about shrinkage over time? UHPC tends to do this unless it's been annealed (48h at 90°C)
Very nice project!
I have used the nanodur (durcrete) UHPC. I let the parts dry and shrink over one month before machining. And in the end it took about half a year before I finished all the important surfaces. So I do not expect a lot of movement anymore.
Very impressive build! What is the linear rail size on all Y, X and Z-axis? I understood the square profile is 50x50 so the rails must be 35mm? Also what is the brand of the rails.
Right the rails are size 35 and the square tubing is 50x50 mm. THe rails are from Schneeberger for Y and INA for X and Z.
@@AlexCNCen Hi Alex , the machine is impressive . Can you tell me where did you buy the linear guides ? Thank's
Looks amazing, can you share the recipe of concrete?
Yeah i also want the ratio of materials !
Hello, awesome project. I would appreciate if you are be able to share with recipe of epoxy granite and which resin did you. I look forward to hearing from you.
Hi,
I used the E45 UHPC mixture from nanodur.
You should try to find surface grinding . But you need to do slot 1-2mm for grinding wheel out. In this case you ca made very accurate surface and side support.
Hi Alex! I have question regarding thermal expansion: Is there potential problem with combining UHPC concrete with steel parts (or is it cast iron?) in it? Something like a "bimetal" effect? Thanks! Milan.
Hi the thermal expansion is about the same for this UHPC concrete and steel. So i do not expect any problems here.
Have you thought about lapping the surfaces? If you watch Tom Lipton's videos on them, you can likely build your own laps.
yes but how to lap those surfaces with a small lap surface? They are so big.. can you explain how it works?
Awesome project! Don't despair. Scraping isn't hard to learn. I learned to do it in less than an hour. The only challenge is scraping multiple sides, but your scraping job is among the more simpler challenges. You only need to ensure the bed and column ways are perpendicular. I would get a large flat granite piece. You can get them for cheap from a kitchen counter fabricator. If you still want someone to help you, try contacting Stefan Gotteswinter. Btw, he demonstrates another method where he uses epoxy as a "liquid" shim.
I just started scraping and so far everything looks good. I borrowed a big straight edge from Stefan Gotteswinter and he gave me some tips :)
@@AlexCNCen Excellent! I hope there will be a video on it soon!
remove the linear rails, have the bottom of the rails machined in a way that will correct for the discrepancy? Or add a block of steel between the body of the machine and the rail and have this machined to correct for the twist instead of having the entire block machined.. this way you don't have to take the whole concrete slab to the big milling machine again. What you think?
Very good suggestion 👍
I've seen several videos! but I can't know what kind of concrete or epoxy is used because nobody shows them in detail, and what about sand? and cement?
Hi. The concrete I used is UHPC from Nanodur with the E45 mixture.
Build two dams or troughs around each of the mates (ply/mdf with a release agent works) and pour about 3~5mm deep and low viscosity epoxy grout on them, make your own grout with epoxy and carborundum, aka silicon carbide, 600 grit, an extremely fine and hard dust. You need to choose a very thin epoxy with a slow set up time and pack it with the grit but not so much that it becomes a mastic paste, it needs to flow still, aim for the consistency of warm black honey. It will flow to perfect flatness, I.E as flat as the curvature of the earth which for the spanned distance is certainly more than good enough. If you connect the two troughs with a channel between them at each end you will also achieve perfect parallelism between the two. This is how it is done with extremely large machines with rail mounts too large to mill, and also, my own. Premade mixtures also exist, look up epoxy bearing grouts like chockfast, nordbak etc. P.S. You can coat your rail mount bolts with melted beeswax and screw them into your mounting holes prior to the pour, they will release after it sets leaving a nice wax coated hole.
wanted to do that, but some people said the epoxy wasn't hard enough compare to steel and will deform under the linear rail. gpa of epoxy is 4 steel 200.
@jesuismika Some people don't know what they are talking about, they have an opinion but no experience 😉
@@m3chanist well yes maybe, but the fact seems real about the stiffness. And the fact I coudn't see a single company doing this. And I searched quite a lot.... If you have more information to help look in to it. I would greatly appreciate. thank you.
@@jesuismika It's not real, it is standard procedure for every large project with an entire industry of epoxy grout manufactures, some of whom I listed, that service the sector, EG mills large enough to cut ship sized parts, National sized dam/generator projects. Rest assured, if it's good enough for that scale of project, it is more than good enough for a home gamer.
@@m3chanist well I guess it depends the expectation, a huge machine won't have the same tolerance as a small one. And maybe if you are talking micron, the epoxy would deform a tiny bit, which is acceptable for some project, and maybe less for other. I'm trying to see the solution to be in the micrometer perimeter with the less equipment possible, if possible. ;-)
Hast du es mittlerweile gelöst?
Bin gerade dabei die Flächen zu schaben und es sieht alles gut aus :)
Guess no mineral concrete for these? And not precision ground but milled?
I used UHPC for the two parts. In the following parts you can see, that I scraped all important surfaces to get very good accuracy.
You could have used industrial steel section box those sizes as horizontal/vertical beds. Drilled, tapped plugged & machined; then filled with Granite concrete aggregate or concrete.
The concrete idea will be constantly moving about like a machine shop floor. In answer to your problem make a x-z axis plate 3" /75mmm thick flash ground of graded steel commercially available - take your guide rail system off x/z and fix to the machined plates 3 mm deep slots as soft mallet fit ~ and drill and tap and locate your linear guide rail/s system. [ reason is that the concrete hasn't the inertia power under the creeping process of settling concrete to warp steel plates. Steel cover plates stronger than concrete. ] I hope this of an idea for you Sir.
Concrete or granite- concretes Etc, will always shift and settle due to the expansion of air and water into it. At the design conception of the machine, the granite or concrete suppliers give an expansion and contraction quota for its [ properties of materials ] Chart. At 10 microns you said adrift at present is good enough for general and precision work. You would have to be in an atmosphere controlled environment on C.M.M Beds to even reliably measure this! Further saying that it's possible outside of this is misleading and confrontational at best. Otherwise, machine builders, who spend millions on inspection equipment and controlled environment to ascertain machine tolerances in manufacture. It best to remember their claims to accuracy at best is as good until the machine is outside the door as the warranty ends right there with them. In part due to the Cold War and nuclear accidents throughout Europe Canada the States Australia china Russia back in the 1970s - mid80s. For Eg, You machine a Flange to your gauges and send it by post to as many states and regions as to spend two days in each area of your country. then see if it fits as transitional fit on its test rig you made for it in your home factory as stabilized.
This tells you the time in space and time you exist in your home workshop different to other postal codes in your area. linear vertical and radial warp in the test piece is relative to sight and organic matter called FAR OUT Funny shapes and sizes people heads faces extremities to Cherknoble reacter blow up is relative to all materials is the point of the test.
Impressive build! Have you considered using steel filled machine epoxy such as DWH to set the linear rails on to achieve perfect accuracy? This could be a good solution for you and much faster than scrapping.... All the best.
Why concrete instead of epoxy granite?
Another possibility if the two surfaces are close to flat would be to lap them in. You can get large old cast iron laps for not too much money on ebay, and use it across both surfaces at once to even out the twist. Just start with a more aggressive grit lapping fluid to move more material
I have an edge on every side for the linear rails to be pushed against. So the plate must not be wider than 250 mm to fit in between. I have started scraping now and so far it goes like expected :)
I have read most of the comments previously written up until now. Maybe you can shim with something more pliable (easy to deform), like copper or tin. If you are able to fix the issues you can use the same method in conjunction with some adhesive like others have suggested. Just my two cents of thoughts. I'm sure you will figure something out. Good luck!
Awesome project Alex! Can you please tell me the manufacturer of the concrete you used for the base?
It is UHPC from nanodur with the E45 mixture.
@@AlexCNCen Cool, thanks. I've been looking at that as an alternative to polymer concrete. Looks superior to me.
Hello Alex, I'm currently designing a cnc gantry mill with the sole purpose of machining steel and aluminum. Roughly a 20" x 20" x 10" machining envelope. I will use concrete like you to add mass and rigidity. I really like your cnc machine do you have a design of it?
My biggest worry is will my gantry be rigid enough. I am thinking of sharing my design now for collaboration from other builders. Did you collaborate with others to fine tune your design? Great job on the mill and thanks for sharing...
Mike - Plus8Precision
Usually I'd mount the carriage in a vice and indicate bottom of the rail to see if groves are parallel to mounting surface first. If they are the error can be caused either by one rail going up or twisting away (and it's probably the latter as you said). Instead of scraping you could just stone the surfaces, applying more pressure and spending more time on the high side.
If it really is twist, then indicated 0.1mm means "only" 0.02mm to be removed.
The rails are good. I swapped them and the result is the same. They are also a good brand. I would need to remove about 0,025mm. I shimed 0,03mm at the end and that was to much. But I don't really have the right tools to measure. I need to remove the twist, get the surface flat and parallel (and same height) to the other surface. Nut I have another idea I could try without moving the parts. Need to check if that is a valid option.
@@AlexCNCen I thought some more about your problem. It is possible to calculate how much of what you are measuring is twist and how much is a height difference by measuring on the inside strip and outside one, then comparing results (or even measuring ends of a parallel attached to the carriage to exagarrate the error). Doing this and adjusting distance between rails should be enough to make linear rails happy (not fight against each other all the time). The real problem is that by doing only realative measurments one could accidentally make a smooth, repeatable... sphere. Do you have any straight edges or time to make one?
BTW I'm just throwing in some ideas, hope you find a way to fix this and continiue with the build :D
I have some more measuring tools including a 500mm straight edge. I just showed one measurement in the video. Would have taken to long to show everything. I will try to measure more exactly where are the problems are so I know better where i have to remove material.
How about working a calibrated granite surface plate inverted onto the rails with fine grit water paper ? Blue the plate up first just like you would if you were going to scrape it, then just work it back and forth. Then scrape to final flatness. Might just work. ( I read below that the inlays have edges. maybe just grind those off before you start, then you would have to make some other plan for alignment) good luck
The problem with that is that edges for the linear guides on both sides. So I would ned a plate that exactly fits between those. An Option would be to remove those. But I'm not sure about that yet.
@@AlexCNCen maybe just remove the edge on the one side so you still have your reference, then once the one rail is mounted, it should be easy to use a spacer to align the other side to that.
Hi Allex, would you like to Share the the drawing of this project!! We can try to learn something! Please..
What size linear rails are you using?
Those are size 35.
You can correct the error by CNC program
Hi Alex, great project! Can you give us more details about concrete recipe? Is it possible to make/find all ingredients somewhere else or only buy from specific company?
I will make a video about the concrete. I made a test piece where I try to measure stiffness and strength. But not sure when. Free time is expensive at the moment ;)
@@AlexCNCen Sounds great! Please do it as soon as possible 🙂. Also, is it ok for you to share with us CAD file?
Scrapping steel is a massive undertaking grinding is your only feasible option
Nice project Alex. I look forward to seeing it progress.
Forgive me if this has already been asked but what concrete have you used? I've been researching how ordinary portland cement corrodes steel and your remarks about how it "flows" tells me this isn't just any old concrete :)
I'm not an expert, but from his description it sounds like UHPC.
... or you can collaborate with somebody who have some experience in scraping and proper tools.
it may be a good variant overall. ( 1. you don't need to move your machine 2. you learn and practice in scraping 3. you get more content for youtube )
Unfortunately it is difficult to finde someone with the right skills and enough time. Especially the time part.
You can add some epoxy resin... there are some made just for shimming machine parts, super sturdy, super hard, can solve your problem
I will use such an epoxy to align the column. I just started scraping and so far it looks like I can get a pretty good result with it.
Don't worry about the surface in the base, just take the rails off and grind them to match the base. Instead of small shim, surface grind more on one side of the bottom of the rail.
Hm that would be super difficult to exactly get the imperfections on the rail to compensate for the ones on the steel inlay. Especially when I'm not able to measure it accurate ;)
@@AlexCNCen can't you use one rail top, indicate along the other both sides, mark it, remove bottom material, then you are close. Then scrape plates in concrete in situ.
Also are you sure you cannot zero this out in software?
I don't think this is a big issue. Although you read approx 0.1 mm deviation on the dial indicator, we're talking about a deviation of approx 0.01 mm on the guides foundation. If you're sure that the origin of the deviation is the foundation twist of one guide (and not an overall flatness error), I think this won't affect too much to machine accuracy.
Hi Alex, are you able to share your CAD files? I’m planning a very similar build and it would be a great starting point! I’ve already checked out your friend Sebastian’s project. Thanks!
Hi,
I won't share the plans before they are finished so I can guarantee that everything will work.
Hi Alex. Great build so far, following along on TH-cam and on Instagram.
Just a question, I have been thinking. Of doing a cnc with epoxy granite, but after seeing your build, thought that the concrete route might be a cheaper route seeing that epoxy is very expensive. Just wondering how you were able to get the mixture for the concrete to have the right strength for what you are using It for. Also is it a special type of concrete or can I use the 52N concrete for the mix?
Anyway. Hope you can help.
Cheers.
Hi,
it is a special concrete. You can find information about it with Nanodur E45 UHPC. Its noticeable more expensive than normal concrete but easier to handle than epoxy granite. That was the reason why I use concrete instead of epoxy granite.
Thanks for the reply!
I went to check out that product and it helps a lot. Looks like here in South Africa we don't have those kinds of products. I have read some papers on how they created Uhpc with local products so might be able to come right with that, otherwise the epoxy route will be the way.
One more question. Did you do anything extra with the drying process, or did you just let the mold stand for a few days? WAs. Wondering if it needs to be heated up to get rid of all the moisture content in the concrete to get it more stable.
@@banzaiwill7672 I haven't done anything special. It is important to wrap the parts in foil for some days so it does not dry out. The added water is only that much to finish the curing process. During the first day the parts got quite hot by themselves. Something around 50°C. After that you should wait at least a month before machining the inlays. The concrete shrinks a little bit in that time and is than stable.
Just remove your rails and install a precision ground plate about 25-40mm thick and re-fix your rails to that.You will loose some spindle height but it will be flat.You can use longer screws to fix through your rails through the plate into base rails.You can shim the plate to it's final position and then epoxy it so it becomes rock solid with your base.The epoxy cement will act as grout to fill in the gap between the plate and your base rails.
I just started scraping the ways. Looks like I can get it pretty precise with that and won't have to alter anything in the construction. In case that does not work for in the end I can still try something else.
Hi! Any news yet ? If you lived in sweden I would help you face it. I would probably take it out again and mill it with a large facemill and also check it with a dial indicator before I remove it from the machine. Easiest would probably be a large knee mill.
Maybe Grind the botton of the rails ? Dont mill it ....put it in surface grinder.
Or make few mm /cm flat steel / aluminium bar as a shim :) (or just make 1 big few cm thick steel plate)
for me second option is the best idea becouse even if u make if wrong on 1 side ....u can remake it :)
Hi Alex, Nice project. What was your concrete recipe?
I used the Nanodur E45 recipe. You should be able to find the ingredients online.
@@AlexCNCen Ok, Thanks.
Take the rails off and hand scrape the table below until it’s flat then remount the rail.
I don't know if you've solved this yet, but maybe this could work:
Alternate version of scraping it in. But hire someone to create an ultra precise hand scraped (or if good enough, ground) master to match the rail mounting surfaces (as they are meant to be) and have them ship it to you. That way you can expand your geographic radius of who can help you. Then you take over and blue the master, run it along the mounting surfaces (both the flats and walls) and clean them up until they are perfect (or good enough). This would require less metrology tooling on your end, and less effort and skill in scraping.
Also things of this mass are flexible, especially on the scale that you wish to achieve precision, so moving it out of your basement and back can reintroduce errors, even if the gridding or milling you do again is better than the first time. So leaving it in its final resting spot is best to then scrape it in.
Ignore everyone saying to touch the rails. These rails, no matter how expensive, are not designed to be perfectly straight or flat or whatever. They rely on your mounting references to achieve straightness and parallelism. The only true relationship between surfaces on the rails are the bottoms, one side, and the bearing races at any "2d cross section" (because again, they can be ever so slightly twisted or bowed). Any other surface is not useful as a reference.
that is an incredible job so far, i understand the feeling of finding out your reference surfaces are not up to the job. a few months back i "finished" a CNC router only to have the gantry immediately rack and kill the ball nuts on both Y axis'. Of course I was using aluminum extrusion and aiming for a level of precision 1 to 2 orders of magnitude lower than you. But spending a year designing and building only to get so close to done and then catastrophe is heart break.
I agree, do it right, and do it once.
Good luck
Thanks for your suggestion! I will definitely try to scrape it. I already borrowed a big enough straight edge from a friend. So I have all the metrology equipment here to get the surfaces flat enough. I also tested scraping on some mild steel already and it worked quite well. Now I only need some time to continue the project :)
Best
Alex
If it's a twist. Buy a reasonably sized surface plate, and some fine grit sandpaper. Tape the sandpaper to the surface plate. Lay the surface plate across both rails. Elbow grease. Not perfect solution, but effective.
I have an edge on every side for the linear rails to be pushed against. So the plate must not be wider than 250 mm to fit in between. I have started scraping now and so far it goes like expected :)
A linear guide is statically over-determined. _Four_ linear guides even more so. The dial-gauge test proves nothing. You should not be using the rZ-position from the linear guide in the first place. Are you _sure_ that the rails are not coplanar (ignoring the rotation-direction)? You could test this by having a laser-pointer on a carriage with three (not four!) balls, two resting on one rail and one on the other. When you slide the carriage on the rails, does the laser-pointer deviate substantially? If so, your recourse is to shim the second rail to make it parallel. In any case, the carriages on the linear rail must be decoupled in all but 5 DoF total (e.g. 2 + 2 + 1 restricting degrees) if you hope to achieve precision at all.
Dan Gelbart has a nice video-series on TH-cam, and has designed and built his own ultra-precise lathe. Perhaps some inspiration there?
It appears Alex is using roller guides, Roller guides are very sensitive to binding compared to the ball type they must be very carefully mounted. this requires that the saddle and guide way mounting surfaces are within the tolerances stated in the manufacturers data sheets.
Many engineering text books suggest the use of more than three points of attachment is kinetically redundant. Maybe for designing an instrument three points is a good Idea however where heavy dynamic loads are being carried more than 3 is the norm. 3 wheel cars have never taken off because they tend to be quite unstable and unsafe when cornering. Machine tools have similar issues particularly CNC machines with high acceleration.
Regards
John
@@johnmcnamara3719 Agree both comments.
Measuring setup is defenetly not relevant. Magnet base should be placed on the block over straight edge (check nsk rail installation video). I'm pretty sure most of this reading comes from rails plane angular screwing.
Same time, roller liners have tight tolerances for inplane. It would be hard to make this two surfaces laying in one plane, since they are protuded and connected to heavy concrete base. No ways to check planes with surface plate.
Basing big straight edge over ballscrew mount surfaces is also not best way to fix this issue, because those ballscrew mount surfaces should be scraped after rails have been properly installed.
@@kirillkirillov3809 A small surface plate and blue will reveal the errors, the steel rail supports set about 5mm above the casting. allowing the test to be performed. For the Y axis it would be better to lay it flat then lower the small surface plate onto it.
@@johnmcnamara3719 there are side support surface outside both rails. Will be hard to find surface plate that will fit inside this pocket. Also, not sure if ballscrew support surfaces should lay in same plane.
@@kirillkirillov3809 Unlike the machine built by Mr Sebastian Ent The machine Alex has built appears in the video to have flat milled surfaces. Maybe a friend nearby could lend a small surface plate for the test.
I do hope you found your fix! Even if it consists of taking it back to the machine shop.......next time check the work before you go home! Lol
Hi Alex; let me tell how I would tackle this situation hoping that maybe you would find something useful in my explanations . 1) I would look for the ''best'' spot along the inlays where I could find 3 points, and I will explain the reason for them in a short while; these should be 2 points on 1 inlay and another 1 point on the other inlay. Where about these 3 points could be found is debatable. They could be found at either end of the inlays or, why not, right in the middle of the inlays. These 3 points should be chosen so that to be as close to one another as possible. More precisely: if , say, your X axis is 500 mm, I would choose the first point at 240 mm on the LEFT inlay (measured from the Z axis outwards), the second point at 260 mm on the LEFT inlay as well, and the 3-rd point at 250 mm on the RIGHT inlay; 2) Say that your X axis inlays are 300 mm spaced apart from one another; I would make a rectangle out of ordinary steel, about 500 mm /300 mm and I would drill/insert 3 bolts and cap nuts (semi-spherical nuts) at precisely 240 mm and 260 mm on one longer side and at 250 mm on the other longer side of the rectangle; 3) I would attach on this rectangle as many dial indicators as I could afford, perhaps at least 4 dial indicators on either side, spaced apart evenly and I would place the rectangle on the 600/430 granite plate that you already have; 4) I would set all the dial indicators to zero and than I would transfer the rectangle on to the 3 points on the inlays described earlier; this way , the lack of coplanarity becomes measurable and you can start scraping the inlays because you would know where and how much. Here is the thing: it wouldn't make much of a difference if it was in the middle of the inlays or 100 mm in or out, provide that you would place the rectangle every time in the same spots. What matters is for the 3 points to be as close to one another as possible because it is this closeness that makes it a viable solution. Have fun !
Thanks for the advice. I just started scraping and used a machinist level for that. Looks like that works pretty good and with a resolution of 0,02mm/m per division it should be accurate enough to get the surfaces to a flattens of 10μm or better.
Maybe you could use Diamant Plastik. Stefan has a Video about it for tramming his Z Axis.
WUdhuozouz I like this idea.
@@markamy357 Stefan does it in this video. I think it would work great. th-cam.com/video/U7Qs-J2swIc/w-d-xo.html
Hi, What kind of shirt is that and where can I get one? I love it!
Also I feel your pain on the email thing. I feel if I used my company email that I would get a better response, but it seems that companies dont want to deal with people who dont have a company email :(. Its a shame too. I was looking for cast iron and wanted to get quotes.
Stefan Gotteswinter might be able to help you out but in sure you already asked him
I already got in touch with him. But his schedule is quite full at the moment.
The angle you put under the z travel is not necessary
Why you had this problem when you machined all surfaces in one clamping ? This should not be happening ?
I figured out how to make it perfect bud! First pick the 2 rails that are PERPENDICULAR TO EACH OTHER...2nd put your indicator on THAT RAIL,,run it all the way down the bad side writing on the rail and a pc of paper the amount it is out INCREMENTALLY and mark front and back etc. After you figure out exactly were it is (hopefully not wavey) you can flip it over and shim exactly what the error is and mill the bottom....make sure you machine the HIGH RAIL !!! HAHA,, i would ether hold it in a few vices with jacks under it or on magnets. Hope this helps! I really wanna build my own as well but cant decide on a size yet...ive done large custom machining on mazaks ,Nicolas correa's, n Cincinnati's for the last 14yrs,,n 12yrs in job shops making little parts before that. Remember measure once,,cut twice! Haha!
Just remember cut the high rail not the low! Or shim it with a shim assortment ,,its only .004 thousandths of a inch right?
Nice machine, why not just strap a grinder on a cross slide, and mount it to the z axis then surface grind the 0.1mm off. Use a paint pen and paint the ways, then grind both. It's like surface grinding on a tool grinder.
I already had that idea. But I try scraping first. Started with that some days ago and so far it looks like it will work :)
Why would you have a engineering shop in a basement with such limited access?