Epoxy granite vertical mill build log

Right so I've started on an large cast iron/EG mill build. Plan is to use several cast iron surface plates I have bought on ebay and embeded in epoxy granite, the hope is this will give me a flat surface to mount the linear rails to.

This is the current design in fusion

Attachment 24628Attachment 24630Attachment 24631

Travel: X 400mm, Y 250mm, Z 400mm
Total weight 600-800kg
Rails: 20mm hiwin clone rails
Ball screws: Generic rolled Chinese ball screws, with a plan to upgrade in the future.
Spindle: Initially: 2.2kw water cooled spindle. The plan would be to upgrade to a BT30 and 1.8kw servo from AliExpress if I can get this made and then integrate ATC
Electronics: UC400 ETH, Initially steppers, with upgrade to 400W AC servos.

Epoxy granite: Silimix 282 + west systems epoxy 105/206 10% by weight, 0.25% black iron oxide


Cast iron:
I’ve got one 36 x 16 inch plate

Attachment 24629

Along with four of these 18x12x2in plates

Attachment 24627

The large surface plate is reasonably flat, however the smaller plates are out by about 0.05-0.15mm. I am planning on trying to hand scrape them flat.

Issues:
1. I have got to cut one of the cast iron plates into four for the column mounting blocks, I’m not sure how I will do this as they're 2 inch's thick, I think I will try with a cutting disc
2. The column and table will consist of more than one cast iron piece, I’m not quite sure what the best way to keep alignment is, I think I will have to use a flat surface, possibly the big surface plate to cast against and an angle plate.


I've got the epoxy and Silimix. I have made the mould for the base and I think I will cast this first to check that the epoxy granite doesn't warp the surface as it sets. I'll post back with the results!

Hi Nick
Thanks for posting, will be following with interest.
Mike

Right so I’ve casted the base. I started by preparing the surface plate, I cleaned the underside with a wire wheel and then drilled and tapped several holes for M6 and M10 bolts, so the epoxy would anchor.

The surface plate had a hole drilled in the top when I got it, I thought this might be helpful to run wiring/pipes through, so I routed some PVC pipe to the back of the surface plate. Before I decided to cast the surface plate in epoxy, I did briefly try and cut it, I sealed this cut with silicone. I also drilled two pairs of holes in each side and passed pvc pipe through, this will let me pass some chains through to lift it.

The mould was made of melamine "furniture" board I got from wickes, sealed with silicone and then three applications of wax were applied (three was necessary in testing, it's a huge pain when it sticks). The surface plate was lowered with an engine crane face down into the mould.

Attachment 24642

Mixing the epoxy granite was the hardest part, it took ages. We did it in 25kg batches, with one person packing the mould while the other used a hand held mixer. We also had a bench grinder attached to the mould, with an eccentric weight, but it quickly ripped itself free of its base. It took just under two hours in total and I managed to add some steel plate to act as mounting points later on and a small section of steel as a future motor mount.

Attachment 24643

I let it cure for about 48 hours before removing the melamine mould.

Attachment 24644Attachment 24645

There are some areas that haven't fully combined and some air inclusion, this is worse on the side away from the bench grinder and I think related to poor vibration. But overall i'm fairly happy with how it turned out. As a very crude test, I've put a tenth indicator on a 30cm extension and jumped up and down over the base, I can't really get it to deflect. It weighs about 220kg now.

I was also concerned that the curing epoxy would warp the surface, so I tried to map out the base with a indicator, there is very little movement (apart from the end I cut which shows about 2 thou of deviation). I also used a 2 foot straight edge, I can't pass a 0.02 mm feeler gauge under it. I don't think its warped significantly.

Next step is to scrape all of the cast iron inserts for the column, I've started to do this, its slow work, but I think will be possible. I'm going to have to think of a way to level the end of the base, my surface plate isn't big enough so i'm not entirely sure how to do this?

Nick

Following.

How big is the area you need to get flat? I had similar challenges so had to be inventive with my own concrete build.

Interesting project. I was thinking about building a dedicated milling machine the other day but priced up the off the shelf bits (motor spindle rails screws etc ) and talked myself out of it for now.
You have a nice set of parts lined up so should make a great machine. Looking forward to watching your progress.

The area to flatten is 900 x 400mm and my surface plate is 600x600.

Yeah I talked myself out of it a few times. I'm not sure its going to be very economical, the BT30 spindle itself is a substantial cost.

Chaz, I found your concrete build very interesting and I've thought about a moglice/DWH approach. I'm lacking a big enough reference surface though, any ideas on how I could overcome this?

I'm starting to wonder if I will have to leave the base as it is. If I could be confident in the way I was measuring it and I was sure it was flat to say 0.02-0.05mm I would probably be happy and just shim/scrape it later. Does anyone a good way to measure how flat a surface is that's bigger than your reference surface?

Quote:

---

Originally Posted by Nick

The area to flatten is 900 x 400mm and my surface plate is 600x600.

Yeah I talked myself out of it a few times. I'm not sure its going to be very economical, the BT30 spindle itself is a substantial cost.

Chaz, I found your concrete build very interesting and I've thought about a moglice/DWH approach. I'm lacking a big enough reference surface though, any ideas on how I could overcome this?

I'm starting to wonder if I will have to leave the base as it is. If I could be confident in the way I was measuring it and I was sure it was flat to say 0.02-0.05mm I would probably be happy and just shim/scrape it later. Does anyone a good way to measure how flat a surface is that's bigger than your reference surface?

---

I bought a hollowed out 'reference' block from someone. Its around 900 x 600 x 600 and might be good enough for what you need. It has slots in it and few holes / marks but its what I used to lay my Y rails, then later X and Z from that.

Where are you based? Might be worth a chat, PM your number if you think its worth discussion via phone.

Was just trawling through some older "subscriptions" and think this might be your "reference box" Chaz :

Attachment 24649

From http://www.mycncuk.com/threads/10480...growing-nicely

Quote:

---

Originally Posted by AndyGuid

Was just trawling through some older "subscriptions" and think this might be your "reference box" Chaz :

Attachment 24649

From http://www.mycncuk.com/threads/10480...growing-nicely

---

Yep.

Chaz, that might work, how much does it weigh? I am based in Yorkshire, where about are you based?

Just changed jobs, making very slow progress on the build at the moment.

Quote:

---

Originally Posted by Nick

Chaz, that might work, how much does it weigh? I am based in Yorkshire, where about are you based?

Just changed jobs, making very slow progress on the build at the moment.

---

Hi, its fairly heavy (estimated 300-400kg). Pity you arent closer, I'm in Uxbridge / Ickenham, West London, Near the M40 / M25.

Sure, happy to help if I can.

This is a bit delayed but I owe a big thanks to Chaz who has let me borrow his reference surface. It has been really helpful and I am not sure how I would of progressed without it. I tried to pick it up in a Fiesta, which unsurprisingly didn't work, thankfully it came on a pallet.

I've been making slow progress, I settled for an angle grinder for the cutting of the cast iron blocks, it took forever... I have managed to rough scrape all the surfaces for the column.

Attachment 24898

These blocks of cast iron are the ones highlighted below.

Attachment 24900

I have also scraped the main cast iron block.

Attachment 24899

They're not perfect but within 0.03 mm. I am making the mold and getting the inserts ready, I plan to cast the column next week. Just trying to work out how to drill 18mm holes through the 2 inch cast iron blocks, my drill press isn't up to it and I really don't want to have to buy a mag drill, I don't know if anyone has any other ideas?

Nick

Quote:

---

Originally Posted by Nick

This is a bit delayed but I owe a big thanks to Chaz who has let me borrow his reference surface. It has been really helpful and I am not sure how I would of progressed without it. I tried to pick it up in a Fiesta, which unsurprisingly didn't work, thankfully it came on a pallet.

I've been making slow progress, I settled for an angle grinder for the cutting of the cast iron blocks, it took forever... I have managed to rough scrape all the surfaces for the column.

Attachment 24898

These blocks of cast iron are the ones highlighted below.

Attachment 24900

I have also scraped the main cast iron block.

Attachment 24899

They're not perfect but within 0.03 mm. I am making the mold and getting the inserts ready, I plan to cast the column next week. Just trying to work out how to drill 18mm holes through the 2 inch cast iron blocks, my drill press isn't up to it and I really don't want to have to buy a mag drill, I don't know if anyone has any other ideas?

Nick

---

Scrape Scrape Scrape ....

I was going to ask how you were getting on.

Would the blocks fit on a lathe cross slide to drill the holes using a bit or boring head in the chuck?
Or the other way around with the block rotating on a lathe face place and bit or boring head in the tailstock?

Quick update

That was a great idea with the lathe cross slide although I couldn't quite get it to work, having issues with its size. I ended up getting a mag drill... Drilled 60 holes and tapped for the insert bolts.

Attachment 24938

Mould made and epoxy poured, all 150kg of it. Got it done in about 90 minutes with two of us. Had some slight issues, no vibration, we ran out of pigment towards the end and the mould blew out on one side. But overall went ok, if anyone wants more details on how to go about this I could write a more detailed post / video.

Attachment 24939

Will probably de-mould tomorrow.

It looks bigger out the mold

Attachment 24952

Quote:

---

Originally Posted by Nick

It looks bigger out the mold

Attachment 24952

---

Lovely. Keep at it ....

Small update

I've made a temporary stand and put the column on the base

Attachment 25159

I don't have a square big enough to see how perpendicular it is. Has anyone bought a granite square ? the only one I can find is this one
https://www.zoro.co.uk/shop/measurin...MaAg9CEALw_wcB

I have been making slow progress with the table, I have reduced the table width to 240 mm and I've got two ground 25x50x1000 mm bars to use as the rail mounts.

The table "offcut" from reducing the table size width has been cut into four and scraped flat and parallel to act as spacers. I was starting to scrape the table flat but the carbide came off my scraper...

Attachment 25158

I am slightly concerned that there isn't enough support under the steel bars and they will deflect. You can see this in the renders below, with the overhanging steel bars and the small amount supporting epoxy granite. I'm thinking about adding some threaded rod or increasing the thickness of the epoxy, i'm not sure if this is necessary though, thoughts on this?

Attachment 25153Attachment 25157Attachment 25156

Hi Nick,

Very very nice project! I'm planning the exact same machine but first option is to use steel tubes filled with EG. Second is directly casting EG like you did. Anyway, will follow with great interest.

Would you mind sharing your model for study, or maybe just give us dimensions of base and column?

Thanks for sharing your experience, this surely prove helpful at some point...

Cheers

Envoyé de mon ONEPLUS A5010 en utilisant Tapatalk

Sure, it's not quite finished and I have had some difficulties with exporting it. All the joints have gone, does the iges file below work?

https://www.dropbox.com/s/6w12be2b4w...0v19.iges?dl=0

The rails are from the Hiwin website and I've used these models off grabcad

https://grabcad.com/library/lichuan-servo-60st-m01...
https://grabcad.com/library/ac-servo-110sy-04030-1
https://grabcad.com/library/bt30-spindle-2

The design is based around the cast iron surface plate and blocks that I bought, I think this has resulted in the base being too narrow and long.

I think with hindsight starting with a welded steel filled epoxy structure would of been a good idea.

I look forward to your build, let me know if I can help

Some progress today, managed to cast the table.

Attachment 26076Attachment 26077

Good to see that you are still at it.

Some progress:

It started to look a bit like a mill:

Attachment 26698

Next step was to flatten the base casting so I could get the rails on. It was to big an area to scrape and the reference surface and base were to heavy to invert. So I blocked off two rail sized areas with wood and then used an angle grinder to lower the rest of the surface, a bit crude but worked well.

Attachment 26694

I then flipped my surface plate and lowered it onto the base. It wasn't that far off, you could see damage from the previous owners of the surface plate.

Attachment 26692Attachment 26695

After several rounds of scraping, it is flat to about 0.02 - 0.03 mm

Attachment 26697

Y axis and saddle installed:

Attachment 26691

Next step is to flatten the Z-axis rails, im using the same process as with the base, flipping the surface plate, it's really slow going.

Attachment 26693Attachment 26696

More progress, rails installed, steppers installed on temp 3D printed mounts, temporary head and spindle.

First axis movements (Y axis not working):

https://www.youtube.com/watch?v=rAVX...ature=youtu.be

Attachment 26993

Next steps:

Alignment, use mill to mill better mounts, mill permanent head BT30, servos, way covers.

Does anyone have any experience with the cheap AC servo's on aliexpress? In particular the 60ST-M01930? Not found much on these.

Has anyone bought way covers from aliexpress? I was thinking about getting some "Armoured bellow cover"?

Nick

Ive looked at the servos, Ill be getting some once we get this house buying marlarky done.

Congrats, commented on your video.

Quote:

---

Originally Posted by Nick

Does anyone have any experience with the cheap AC servo's on aliexpress? In particular the 60ST-M01930? Not found much on these.

---

Where you located Nick.? Says E yorks.? Anywhere near Goole.

I've got some 400W waiting to go on a machine which are exactly the same drives and motors just lower power. So if you want to have a play with them on the bench your welcome to come over. For the money, they can't be beaten.

Edit: Actually just had thought.? They could actually be 600w I had set of 400w and 600w but fitted one set to a machine and I think they where the 400w. I'll Check when I get back to work on Monday.

Chaz - First movements have been very rewarding, hoping the pace picks up now.

I'm near Hull about an hour from Goole, that's a very kind offer, it might be a while till I could get over. That's reassuring that you've got them to work. I'm having a hard time making sense of the Chinese-English hybrid manual. I would be using them in the "position mode" initially, have you had any issues with overshoot? Are you using "open collector input mode" or using the suggested "AM26LS31 similar line drive"? Or 0-10v? I found the below video, which was bit concerning having to play around with the internal PID settings...

https://www.youtube.com/watch?v=qYgnIKDoVPY

Nick

Quote:

---

Originally Posted by Nick

I'm having a hard time making sense of the Chinese-English hybrid manual. I would be using them in the "position mode" initially, have you had any issues with overshoot? Are you using "open collector input mode" or using the suggested "AM26LS31 similar line drive"? Or 0-10v? .

---

I've used both Analog and Differential inputs on these drives, both work ok but Analog does require more work with tuning if closing the loop back to the controller
If you're using a decent controller with Step & dir then you won't need to mess around with line drivers as it will provide differential outputs.

Quote:

---

Originally Posted by Nick

I found the below video, which was bit concerning having to play around with the internal PID settings...

---

Don't be concerned because it's quite normal to have to tweak the PID settings. Servos are tuned according to the load they have to move so it should be no surprise that some tweaking is needed as it's impossible for them to be tuned correctly at the factory or when sat on a bench.

To how much tweaking needed depends on many factors including Input Mode and how or where the loop is closed. If using full closed-loop back to the controller then affectively 2 Pids have to be dealt with. First, the motor needs to be tuned to the drive using the drives PID. Then the position loop for the Controller needs to be tuned using it's PID.

Don't be afraid of PID tuning it's not rocket science but does involve a bit of black magic to perfect. Mainly it's just tweaking few settings until it behaves how you'd like. Little patience is all that's needed.

Edit: Checked today and the motors are the 600W version.

Jazz thank you for your post, this makes a lot of sense, you've reassured me re the servos. I've decided on using the analog interface with Linux CNC, mesa 7i92 and 7i77 with the controller closing the outer loop. I might need some help with the tuning. I've bought 600w for the X and Y, I have also committed to the BT30 spindle with a 1.8 kw servo.

Attachment 27310

Waiting for the servo for the spindle, pneumatic cylinder and mesa cards.

Made a start on the pneumatics, I've gone with a "mist cooling" system using the water filter approach I found on the here, works well.

Got a lot of time consuming jobs done, filled the spindle head with epoxy granite, got the y-axis movement working. Made a few chips with jogging.

Next major problem is protecting the ways from chips, I've seen several designs. Would anyone be interested on collaborating on a way cover design?

I've seen a few designs:

Chris DePrisco made some with sheet metal: https://www.youtube.com/watch?v=3GSEayp8Ioo
This looks very labour intensive

This looks less labour intensive and less skilled, could potentially 3D print the sides: https://www.youtube.com/watch?v=G2DJ8RQy4D4

This project (which is a bigger, much more polished version of mine) uses way covers from aliexpress (I found a link and I think they cost about £400), this could be emulated by putting sheet metal on cheaper bellow way covers: https://www.youtube.com/watch?v=nTC5gWVmt0o

Then the simplest option is using rubber or a fixed cover with acrylic. Thoughts on this?

Don't overthink the way covers they don't need to have scissor mechanisms. Just let them slide so push-pull each other in and out. Many industrial machines use this method and it works fine, the Fadal machine I have here just uses the push-pull method. Your welcome to look at or copy these covers which are off the machine.

I'll take some pictures for you to see how they are made up.

Quote:

---

Originally Posted by Nick

I've decided on using the analog interface with Linux CNC, mesa 7i92 and 7i77 with the controller closing the outer loop. I might need some help with the tuning..

---

Do you have any experience of Linux CNC.? If not then get ready for a massive learning curve because with servos and closed-loop it's a different ballgame to Steppers. I'm very experienced with lots of controllers and I'm struggling big time with Linux CNC. Luckily I've got a Linux Guru in my camp working on it, hey Clive S . . :whistle::hysterical:

Quote:

---

I'll take some pictures for you to see how they are made up.

---

That would be great. I've put some rubber sheeting in for the moment.

Quote:

---

Do you have any experience of Linux CNC.? If not then get ready for a massive learning curve because with servos and closed-loop it's a different ballgame to Steppers. I'm very experienced with lots of controllers and I'm struggling big time with Linux CNC. Luckily I've got a Linux Guru in my camp working on it, hey Clive S . . :whistle::hysterical:

---

Nope, It's been a massive learning cure. Very slow progress, hours of scanning the Linux CNC forum. You get stuck on silly things like setting up the home switches for what seems like hours. I gave up with stepconf and pcnconf wizard and just played around with the "hal" and "ini" files. I have however got Linux CNC working with steppers via a computer parallel port.

Attachment 27447

I have made my first chips and broken my first end mill on the servo mount for the X-axis. There were lots of issues milling this the X-axis pulley keep slipping. It looks a bit rough, I left 0.5mm of axial so I had to file it down, no finish pass but I think it will be functional. Waiting for a new belt to try it out.

Attachment 27446

5mm 2 flute end mill: RPM: 8000-10000 WOC 1mm, DOC 4mm, chip load per tooth around 0.04mm, feed 600mm/min
8mm 2 flute: RPM 8000-10000, WOC 1.6mm, DOC 8mm, chip load 0.05?, feed around 1m/min

I have got the china servo motors (analog) working with Linux CNC on the desk, closing the loop in control, it wasn't too painful to figure out. The mesa 7i77 card is really nicely made, nice manual, lots of IO.

If anyone is interested I will write a tutorial on getting the china servos working with Linux CNC, it might help with small things like I got stuck for a a good few hours before I realised I needed to set Pn146 to 0 on the servo.

Very brief video of first'ish chips

https://www.youtube.com/watch?v=WzXHVFuu-qo

Quote:

---

Originally Posted by Nick

I have got the china servo motors (analog) working with Linux CNC on the desk, closing the loop in control, it wasn't too painful to figure out. The mesa 7i77 card is really nicely made, nice manual, lots of IO.

If anyone is interested I will write a tutorial on getting the china servos working with Linux CNC, it might help with small things like I got stuck for a a good few hours before I realised I needed to set Pn146 to 0 on the servo.

Very brief video of first'ish chips

---

Bloody hell not joking when said video was brief I blinked and was done. . . :hysterical:


Regards Linux and Servo's how's the following error, have you checked it in hal scope.? . . . We are struggling (well not me but my guru) with the Step/dir to get full speed without throwing a following error.

Hi Nick,

Here's those way covers I was talking about, pictures didn't do it justice so a little video shows it better. Hope it helps

https://youtu.be/aoHXJ9wZT-E

Quote:

---

Originally Posted by JAZZCNC

Hi Nick,

Here's those way covers I was talking about, pictures didn't do it justice so a little video shows it better. Hope it helps

https://youtu.be/aoHXJ9wZT-E

---

Ah thank you very much, I see what you mean now, those look great, really simple design, just sheet metal and a few welds. I will give making some a go.

Following error wise. What speed are you aiming for? How have you found tuning the internal PI loops? I'm struggling to tune the servos. I'm just about at less than 10 microns f-error now on the X and Y, I am cheating slightly acceleration dropped a bit, speed reduced to 1500mm/m. I think my main issue is the servo speed and torque loops aren't tuned. The error got a lot better when I put in the inertia load into Pn257 after I ran the Fn018 mode to get the inertia ratio. I'm using gain adjustment mode 1 (Pn258) and then adjusting the "rigid grade selection" until I get oscillations / vibration. I found the linuxCNC tune very sensitive to FF1

I haven't worked out how to properly tune the internal PI's because I can't work out how to bring linuxCNC into open loop and then command a specific speed in LinuxCNC. I am thinking about using the internal speed command function and toggling this with push buttons. Say set it to 250, -250rpm and then I will be able to jog it back and fourth while looking at the HAL scope.

Attachment 27471

Quote:

---

I haven't worked out how to properly tune the internal PI's because I can't work out how to bring linuxCNC into open loop and then command a specific speed in LinuxCNC. I am thinking about using the internal speed command function and toggling this with push buttons. Say set it to 250, -250rpm and then I will be able to jog it back and fourth while looking at the HAL scope.

---

Nick. What mode are you using ie position or torque etc. Have you seen or read this:-

https://forum.linuxcnc.org/10-advanc...etailed-how-to

I am in the same boat as you and struggling with a following error when using closed loop. Changing closed to open loop is just one change in the hal file.
This is also my first time with halscope :whistle:

Quote:

---

Originally Posted by Clive S

Nick. What mode are you using ie position or torque etc. Have you seen or read this:-

https://forum.linuxcnc.org/10-advanc...etailed-how-to

I am in the same boat as you and struggling with a following error when using closed loop. Changing closed to open loop is just one change in the hal file.
This is also my first time with halscope :whistle:

---

Yeah, that's what i've been following, really useful. I also found the John Thornton guide's. Using speed mode, via a mesa 7i77, so linuxCNC is running the position loop.

I will try torque at some point I think. I'm sure the issue is with the drive speed and torque PI loops. How have you gotten on with setting the drive PI settings? I'f i put it in open loop I think the axis will drift and I don't think I could command an analog signal? or maybe I could?

Nick, Clive is using Step & Dir Mesa card not Analog +/-10v using Position mode in the drive. He's using the same AASD Drive as you but 30A with 1.8Kw servo.

Quote:

---

Originally Posted by Nick

Yeah, that's what i've been following, really useful. I also found the John Thornton guide's. Using speed mode, via a mesa 7i77, so linuxCNC is running the position loop.

I will try torque at some point I think. I'm sure the issue is with the drive speed and torque PI loops. How have you gotten on with setting the drive PI settings? I'f i put it in open loop I think the axis will drift and I don't think I could command an analog signal? or maybe I could?

---

I believe that torque mode is more difficult . I am using step/dir not analogue which should be easier.

I will be away for a few days but will catch up on my return.

Quick update:

Control box nearly done, MPG installed one of the cheap aliexpress ones, finally got an E-Stop. Its got space for 5 servos and 1 servo for a spindle, 3 installed. The 7i77 and one cheap parallel breakout board to drive 1 stepper (currently the Z axis, will hopefully rotate the tool changer)

Attachment 27662Attachment 27661

I tried to keep AC separated from low voltage, the 7i77 to servo connections are using twisted pairs, I ended up hand soldering these. I have tried to ground everything in one place. Some thoughts:

Those cheap ebay "crimp tools" are really effective. The "DB25 solderless connectors" on ebay were invaluable for checking the wiring to the AC servos and the MPG. I got a 800x600x210mm box, I wish I had gone for a 250mm depth the door is catching on the AC servo encoder connectors when closed. I will have to swap the cable trunking out for a deeper version I used 40mm, think I will go for 60mm or deeper.

Z-axis counterweight system next

Nice Nick.

I'm keen to hear your thoughts on those drivers / servos. Been looking them myself for whatever I build next.