CNC-2116 - Up and running, sort of (2015-03-21)

[Yvan256]

Here's the new Y-axis assembly for the belt drive.

As I mentioned before, the two panels are adjustable so that the bronze bushings align perfectly with the smooth rods.

The first photo shows the assembly sliding "closed", the second one shows the sliding "opened" and the third one shows the bottom with the belt anchor and the slotted holes that allow the alignment of the two sides.

[Yvan256]

The Y-axis assembly is finally in place and it's sliding very smoothly.   

I'm searching for something to use as a temporary idler pulley, installing the belt and then testing the whole thing with TurboCNC and the driver board. If all goes well, I should have a video online "soon".

[Typefighter01]

Looks like your adjustable "Y" panels are going to make aligning them with the rods nice and simple.

I should have a video online "soon".

Hurry up 

[Yvan256]

Looks like your adjustable "Y" panels are going to make aligning them with the rods nice and simple.

I should have a video online "soon".

Hurry up 

Still looking for a suitable "idler pulley" option with the parts I have right here.

[Yvan256]

Here's the underside of the machine, to see details of the temporary belt installation.

And the video of the first movements is .

2014-05-26: update on the F686ZZ flanged bearings, the package has arrived in Montreal, so I should have them tomorrow.
2014-05-27: I have received the F686ZZ flanged bearings.
2014-05-28: the new front plates for the flanged bearings are ready.
2014-05-28: I'm being stopped by a lack of 6mm bolts... I should be able to get them before the week-end.

[Yvan256]

I'm dropping this project and getting a Shapeoko 3 instead. It's much less complicated.

[Rick]

I'm dropping this project and getting a Shapeoko 3 instead. It's much less complicated.

Sorry to hear this! I like the idea behind the Shapeoko, but there is something to be said for the homebrew CNC machine. (I've been trying to determine if my own skills are up to creating a 4'x8' CNC or else, paying someone else for a full-on kit.)

Can we not sway you back?

EDIT: YOU SUCK. Post left for posterity. (Thanks for the PM.)

 

[Yvan256]

Sorry to hear this! ... Can we not sway you back?

I've almost posted a photo of a pile of CNC64 parts and wrote "too late it's already in pieces".

I've re-done and re-cut so many parts that I probably have close to enough parts to build another machine. :p

[Malenko]

can you use your Shapeoko 3 to make me a CNC64?

[Yvan256]

can you use your Shapeoko 3 to make me a CNC64?

I think I'll just attach my Black & Decker "Shapeoko 3" to my CNC64 instead. 

As for making other CNC64's, that's the medium-term plan.

[Nephasth]

 

You had me until the demonstration on various materials. Gotta love infinite axes!

[Yvan256]

I made my custom Web-based CNC software for my Roland MDX-3, so all the coordinates system is off by a factor of 40. I'm really not looking forward to fixing all that to support regular g-code.  Not only that, but every function has to be coded three times: one for the HTML canvas preview, one for the Roland RML-1 code and one for the g-code. I guess I should plan ahead of the g-code rewrite and think about splitting every little cut into a function so that I only code three times once.

In the meantime, I'm currently testing a new function to cut rectangle outlines with tabs (the tabs are in green in the image). I'm also testing new 1mm bits I got on eBay, and the noise reduction from a smaller cutting bit is just amazing. The stepper motors of the MDX-3 are the major source of noise now. 

I also got two bonuses: the spindle motor has a lighter cutting load and I'm now able to cut smaller holes and finer details.

I'm currently cutting extension panels for CNC64, to increase the Y-axis cutting length by 40mm. It's not much but it's the maximum length increase I can do for that axis so that CNC64 still fits inside my PVC box.

of my Roland MDX-3 cutting a rectangle with tabs in 3/8" MDF using a 1mm cutting bit, in order to test the RML-1 code output of the new "rectangle_outside_with_tabs" function of my custom Web-based CAD software.

edit: okay, so I messed up the code a little bit. The "tabs" being cut aren't tabs at all, they're uncut parts of the total material height (imagine the green lines not being there at all, or white). I'm fixing the code to cut the 2nd panel. 

edit 2: first photo below, plate with messed up tabs on the left, plate with proper tabs on the right. It's new to me, so it seems a bit weird to see the right plate "float" like that. 

edit 3: more pointless plates with ugly square corners. It reminds me of plastic models for some reason.

edit 4 (2014-06-15): as you might have guessed, I'm still busy re-writing major parts of my Web-based CNC CAD program. It's going well so far and will allow me to add tabs to any shapes and also output regular G-code in addition to Roland RML-1 code.

[Yvan256]

As I said in my update above, I'm still busy re-writing major parts of my Web-based CAD/CNC program. It was already a mess of code with almost-duplicate code all over the place depending on the shape to be cut. Adding tabs to all of them and adding G-code afterward would have been a maintenance nightmare.

So far the rewrite is going well and it will allow me to add tabs to any shapes and also output regular G-code in addition to Roland RML-1 code.

I've simplified the functions quite a bit and these functions call new "lines" functions that output HTML5 Canvas/RML-1/G-code, so any new function that I write from this point on is automatically compatible with all three output formats. In short, the function that outputs drawing/CNC code is the same for all shapes, no more duplicate code.

I'm planning to have custom functions in my CAD for all the common "2.5D" CNC design requirements . The basic ones being "hole", "pocket", "rectangle", etc. I'm still not sure how I'll design the complex arbitrary shapes because the path usually needs to be relative to the milling bit (inside cut, outside cut).

I've yet to complete the RML-1 and G-code outputs, but here's a screen capture of the new "hole" function which takes a parameter for the number of sides. Since the RML-1 doesn't allow for circles or arcs, my CAD software will also only use lines for everything including circles.

The green lines in the image are tabs. The ends of the tabs are even rounded according to the milling bit diameter, too! 

2014-06-16, edit 1: RML-1 code output already completed? At this rate, adding G-code output should take less than one hour, once I understand which codes to output.

2014-06-16, edit 2: Testing a cut with the pattern shown below, I should be able to add a photo by the end of the day. It's working well so far, but there's quite a number of circles to cut. Middle ones are done, still need to cut the last two big ones.

2014-06-16, edit 3: I had to stop the machine at the last circle, it's 22:30 here. I'll post a photo of the cleaned-up plate tomorrow morning.

2014-06-17: After cleaning up the front of the plate, which came out absolutely perfect, I turned it around to clean up the back. Nothing, no slots, totally flat. Turns out I forgot to select the correct material before outputting the RML-1 code for my MDX-3. But I did watch it cut the first few holes and the Z-axis did go up and down, so I know the ramping tabs are working. Next step, flat tabs. Attached are the cleaned plate (rotated 180 degrees on the MDX-3 so I can cut some more tests on the left part of the plate) and a close-up of my CAD software with my stupid mistake highlighted.

2014-06-18: since I'm now cutting with 1mm end mills, I need to rewrite a number of functions that were hard-coded for 1/8" end mills, such as the "pocket" functions (half-depth holes and cutouts). See the last two images for boring details.

[Yvan256]

I'm still busy with my CAD software rewrite. It's going very well, but a lot of CNC64 parts need to be "re-traced", so to speak, since my software is very basic and is only mill-diameter-aware for shapes like holes and rectangles.

One good thing from using 1mm end mills, apart from the much lower noise, is the fact that I can pack a lot more parts per plate. I'm slowly re-designing a "CNC64 Mark II" with the primary goal being to minimize board waste. My old "corners" were oddly-shaped, had two versions ("left" and "right") and could only fit four corners per board. My new corner is a single version and has been resized so that I can fit eight of them on a single board. My CAD also now has an "offset" function, which took under five minutes to implement. I guess my new code is getting easier to maintain. 

Included below is a screenshot of the CAD screen for the eight new corners. I will cut a plate of those corners tomorrow, I'll update with a photo after they're cut.

edit: the plate is being cut as I'm writing this. I'm also testing a new cutting depth which is three times deeper as usual. The noise increase is barely noticeable. It took about 45 minutes for the first corner, so there's about 5 hours of cutting left. Next time I'm going to increase the cutting depth even more. 

edit 2: the plate is cut, but... I messed up in two ways. First, cutting too deep totally ruined the milling because instead of having 0.5mm notches in the inside corners, I got totally rounded corners that don't even look straight to the naked eye. I guess the new cutting depth was too much. I'm lucky the end mill didn't break after cutting eight of those parts. Second mistake, I know this is a code problem because my 3.5mm slots for the bolts aren't even 3mm wide. All that cutting for nothing. I should have made more tests, taken more measurements and cut a single corner as a test before trying to cut eight of them. 

[Yvan256]

I added a poll, for those who would be interested in purchasing a CNC64 kit, if it becomes available.

Plywood should be a better environmental choice and be lighter for shipping, but would be more susceptible to weather changes (temperature, humidity) and could require varnish/painting (I'm not planning on offering a pre-painted option either).

Expanded PVC and HDPE would be the lesser choices for the environment, however they're less susceptible to weather changes and would come in color, which means that if enough people were to order a kit, I could offer multiple choices (ex: black/red, black/green, etc). HDPE would also probably cost more to ship because of its weight while expanded PVC is a middle ground between plywood and HDPE as far as weight goes.

You'll note that MDF and HDF are not available. 

edit: there's a lot of updates on the parts of the optimized version, but I can't post anything about it yet. 

[PL1]

Any ball-park (+/- $100) cost estimates for your kit and also for the stepper motors/control boards/other needed parts to make it functional?

Might make it easier for people to determine if a CNC build can fit into their budget.


Scott

[Yvan256]

Any ball-park (+/- $100) cost estimates for your kit and also for the stepper motors/control boards/other needed parts to make it functional?

Might make it easier for people to determine if a CNC build can fit into their budget.


Scott

I haven't really priced all the parts for any given volume yet, but my target would be around 200 to 250$USD for a complete kit (CNC64, stepper motors, belt+pulleys/ACME rods+nuts, control board and stepper drivers and all the various hardware such as nuts and bolts), you would only need to supply two things: your own rotary tool (I would probably offer two or three choices for the mount) and your own power supply. I'm trying to use a regular ATX power supply for the setup, since those are easy to get everywhere, even if it's from old PCs.

Because of the popularity of RepRaps, the price of the hardware needed for a small desktop CNC has gone down quite a lot for the last few years, even more since the time I started my CNC project thread.

Even so, I'm guessing volume pricing on the parts needed would probably allow me to go even lower than the prices I mentioned above.

edit: revised the price range.

[Yvan256]

I'm still busy writing new parts for my CAD256 program.

The latest addition is a function to cut alphanumeric text. I've based the font on what is possible with a 16-segment display and retouched some characters a bit.

[Yvan256]

Took quite some time, but they're ready. It's nice to see eight plates on the same board. 

[Yvan256]

Well, adding support for G-code was easy enough. It's not totally optimized yet, but it should work.

I even tested it with SimpleGCoder HTML5 G-code Editor, and it looks the same as the output as CAD256, apart from my program displaying tabs and not showing the above-the-workpiece movements.

CAD256




SimpleGCoder HTML5 G-code Editor




So... since I haven't had time to complete CNC64 yet, is anyone stupid brave enough to volunteer and test the G-code output of CAD256 on his/her CNC machine*?

* 1/8" end mill (Dremel #650) required and limit switches on all ends of all axes is highly recommended. 

[Typefighter01]

Send it to me, I am brave supid enough 

[Yvan256]

This file should cut four J22B and four J22N plates, using a 1/8" end mill, in 9.7mm thick material (cut through 9.950mm, requires a spoilboard underneath your material, I use this setting for 3/8" MDF) with 1mm passes, with plunging cuts (I need to re-add ramping cuts in the new code).

There's only four kinds of G-codes used in my files (G21 to set system to metric, G00 for above-material rapid moves, G01 for all cuts and G28 at the end to send the cutting head home) because my CAD needs to stay compatible with RML-1 which is more limited. All curves are converted into lines. No speed codes are used so anything you set in Mach3 or other should stay set for the whole cutting operations.

Home is assumed to be to the left/front of the material, with the zero Z being the surface of the material.

The breakaway tabs are "ramping cuts" so I'm not sure they're long enough in the current file, be careful.

Please air-cut first, make sure you're wearing eyes protection and please make a video of it. 

[Typefighter01]

Hey Yvan,

Downloaded and ran your file. I dragged it to the desktop and tried opening it inside Mach, but it didn't see it, so I renamed it as a ".TAP" file, and it loaded fine. The aircut was a success and it behaved as expected.

I will pick up a 1/8" endmill this week and cut some material (yes, I will make a video).

It took 24min to cut a plate at normal speed and cranking the feed rate upto 300% it only took 8min (makes sense).

The program actually produces a catchy tune when the steppers are cutting the holes and turnung up the feed rate during this portion of the code was pretty cool.

Good work so far buddy...I will report later on in the week once cuttin bit is purchased  .

[Yvan256]

Hey Yvan,

Downloaded and ran your file. I dragged it to the desktop and tried opening it inside Mach, but it didn't see it, so I renamed it as a ".TAP" file, and it loaded fine. The aircut was a success and it behaved as expected.

I will pick up a 1/8" endmill this week and cut some material (yes, I will make a video).

It took 24min to cut a plate at normal speed and cranking the feed rate upto 300% it only took 8min (makes sense).

The program actually produces a catchy tune when the steppers are cutting the holes and turnung up the feed rate during this portion of the code was pretty cool.

Good work so far buddy...I will report later on in the week once cuttin bit is purchased  .

Thank you for the feedback.

When you say 24 minutes per plate, do you mean it would take around 3h12m to cut all eight plates included in the file? It seems incredibly long, considering my MDX-3 didn't even take three five times that long.

I'll be adding a few things, like add a dead-zone when dropping the cutting head above the material, ramping cuts, etc. They could be ready before you purchase your end mill, so once the optimizations are done I'll make a new post with an updated file.

[Typefighter01]

Hey Yvan,

Downloaded and ran your file. I dragged it to the desktop and tried opening it inside Mach, but it didn't see it, so I renamed it as a ".TAP" file, and it loaded fine. The aircut was a success and it behaved as expected.

I will pick up a 1/8" endmill this week and cut some material (yes, I will make a video).

It took 24min to cut a plate at normal speed and cranking the feed rate upto 300% it only took 8min (makes sense). It seems incredibly long, considering my MDX-3 didn't even take three times that long.

The program actually produces a catchy tune when the steppers are cutting the holes and turnung up the feed rate during this portion of the code was pretty cool.

Good work so far buddy...I will report later on in the week once cuttin bit is purchased  .

Thank you for the feedback.

When you say 24 minutes per plate, do you mean it would take around 3h12m to cut all eight plates included in the file?

I'll be adding a few things, like add a dead-zone when dropping the cutting head above the material, ramping cuts, etc. They could be ready before you purchase your end mill, so once the optimizations are done I'll make a new post with an updated file.

Yes Yvan, if no changes to your file are made, it would take 3h12m to cut all 8 plates. 300% increase to feed rate is as far as Mach goes, but from my own experience with routers and MDF, I would say, I could drop cutting time from 8min by another 2 or 3 minutes and still safely cut all 8 plates. That puts a total cutting time of around 40min. How fast do you want to see them cut?

Question...where (when) is cutting speed generated? I know in Artcam express, when I select different cutting tools, before I generate code, each tool has a feed rate listed beside them (suggested rates anyway as they are user defineable).

I will wait for your optimized code and aircut a potion before cutting real wood.

[Yvan256]

Yes Yvan, if no changes to your file are made, it would take 3h12m to cut all 8 plates. 300% increase to feed rate is as far as Mach goes, but from my own experience with routers and MDF, I would say, I could drop cutting time from 8min by another 2 or 3 minutes and still safely cut all 8 plates. That puts a total cutting time of around 40min. How fast do you want to see them cut?

Just cut them at the speed you are comfortable with. With my MDX-3 and the 1mm end mill, I'm used to something around 8 to 10 hours to cut all eight plates. Taking only 40 minutes to cut all eight plates seems like lightspeed to me.  Just be careful if you do buy the Dremel #650 end mill, I've read that it burns easily.

Question...where (when) is cutting speed generated? I know in Artcam express, when I select different cutting tools, before I generate code, each tool has a feed rate listed beside them (suggested rates anyway as they are user defineable).

My program doesn't generate any specific cutting speeds at the moment for the G-Code output, only paths. I do use codes G00 and G01, so there's "travel speed" and "cutting speed" parameters in the file, but the actual speeds for those codes should be defined by your software. FYI, the travel speed only occurs above the workpiece.

[Yvan256]

Typefighter01, did you have time to pick up a square end 1/8" end mill yet?

Hold on before cutting the plates, I'll make you cut something else instead. 

[Typefighter01]

Yes I did 

Grabbed a 1/8" for your testing and a 1/4" for me.

I just need to come up with a hold down system. I do have double sided tape that I will probably use for now.

Send me some g-code when your ready.

EDIT: These are Up-Cut Spiral bits, not end mills (thanks macattack)...

[Yvan256]

Yes I did 

Grabbed a 1/8" for your testing and a 1/4" for me.

I just need to come up with a hold down system. I do have double sided tape that I will probably use for now.

Send me some g-code when your ready.

EDIT: These are Up-Cut Spiral bits, not end mills (thanks macattack)...

Double-sided tape? It really didn't work on MDF last time I tried. Doesn't your router table have T-slots? Let's switch this conversation to your thread.

[Yvan256]

Small text-only update for whoever is still following this thread: I'm having problems with the 1mm cutting bits. I think they're flexing a little while cutting and it's messing up all my measurements on the parts by 0.2mm, which doesn't sound like much until you try assembling parts together.

I've added 200 milliseconds of dwell time between all lines to see if it helps. It does slow down cutting holes quite a bit though, I'll have to add an on/off parameter for the dwelling time. 

edit: nope, cuts are still messed up. I'll switch back to my 1/8" end mill to see if the problem is with the 1mm bits or with the MDX-3 itself.

2014-08-10: tested cutting a 10x10mm part with the Dremel #650, 1/8" straight router bit, there's still a 0.2mm difference on the cut part. The weird thing is, it's 0.2mm smaller so it's definitely not a spindle runout. The MDX-3 is getting old, maybe there's play in the frame since one side is broken, maybe the belts are under-tensioned, etc.

[Yvan256]

Coming soon to a thread near you: CNC-2116.

And a huge thank you to Typefighter01 for helping me out! 

[Yvan256]

Thanks to my "sandwich" shipping design concept and Typefighter01's packaging skills, the six plywood panels seem to have arrived intact. 

Also, thanks to Canada Post for delivering said package directly to my apartment instead of making me go get it at the post office.

[Yvan256]

Since the parts where shipping in panels, they had to be held in place on the panels by bridges which I then needed to cut once I received them.

A few hours with a Dremel cutting wheel and a router flush trim bit later, I had a pile of parts. I still need to cut out about eight parts but in the meantime I started to assemble everything. It's beginning to look like a CNC.

And this time it's not made of %$#@ MDF.   

Refer to the filename of each photo for a description.

[Typefighter01]

And this time it's not made of %$#@ MDF.   

The cabinet grade birch was really nice to work with and looks good, even without any sort of stain or surface treatment.

It was fun testing and prepping for your final design and it is cool seeing something I cut on my CNC being assembled 100's of kilometers away 

Now hurry up and get this thing together so you can help me design bartops 

[Yvan256]

Now hurry up and get this thing together so you can help me design bartops 

We don't need my CNC to be up and running for your bartops, we can work on them before that. 

[Yvan256]

The idler wheel for the X-axis is in place. The one for the Y-axis is another story, since I made a 1mm error in my plan. 

I'm now hunting for 6mm washers that are 0.5mm thinner. 

[Typefighter01]

I'm now hunting for 6mm washers that are 0.5mm thinner. 

You got a bench mounted belt sander? I have resized washer thickness many times on one. Literally just hold it against the belt for a few seconds should do it, or, relieve the area where washers go with a Forstner bit (you will want to do this on your drill press with wood clamped).

[Yvan256]

You got a bench mounted belt sander? I have resized washer thickness many times on one. Literally just hold it against the belt for a few seconds should do it, or, relieve the area where washers go with a Forstner bit (you will want to do this on your drill press with wood clamped).

I'm just going to trust gravity on this one. I have the same setup on the Y-axis of CNC64 and moving it manually doesn't make the idler bearings go up vertically, so I've put a washer at the bottom only for now.

[Yvan256]

Sorry for the lack of updates on my CNC, but I'm working on Dainamikade with Typefighter01, along with other personal projects.

The CNC is finally up and running, but there's still a lot of things to complete. I didn't notice when they were on CNC64, but the rods for my Z-axis aren't even 8mm (they're 5/16") so there's a tiny bit of play and my cuts aren't 100% precise. I'll have to cut one of the Y-axis rod of CNC64 to complete CNC2116. 

In the meantime, here's a few photos of the machine for those who are still following this old thread! 

The first photo is the "completed" machine without any tool attached.

The second photo is with the Black & Decker RTX rotary tool and a plastic spacer board (to clear the bolts heads of the bed itself, as seen in the first photo) installed on the bed along with a hardboard spoiler board.

The third and last photo is something the machine cut this afternoon (I really need to upgrade the code in my CNC program) which is half the spacer board. Tomorrow I'll be cutting the right half and I'll be able to replace the plastic spacer which I'm afraid is too slippery and could add movement to the part to cut.

[Typefighter01]

Looking good buddy.