Building a CNC Router

[Grant H.]

It seems to never fail...

Monday evening, My wife caught me headed to the garage and told me we had plans for dinner with some friends... Apparently they called her out of the blue and wanted to get together, so the only progress I got done was Dykem and scratches...

Major Gantry Support Plates laid out:
20210104_192511 by ARNEWB, on Flickr

Starting small and working up. It makes it a lot easier to get the holes where you want them.
20210105_190700 by ARNEWB, on Flickr

5.5mm Holes drilled for the bearing block bolts.
20210105_192142 by ARNEWB, on Flickr

Counter bored to accept the socket head m5 "bolts"... These are set deeper than necessary to provide plenty of thread engagement in the bearing blocks.
20210105_192924 by ARNEWB, on Flickr

8.5mm holes drilled for the plate to Gantry Riser bolts.
20210105_193904 by ARNEWB, on Flickr

Counterbored. Some are actually kind of rough, some are just dirty still. The big counterbore bit is an interesting study in chip load and cutter performance. The holes where I pushed "hard" on the drill press were the cleanest and smoothest, and the cutter seemed happier. The ones where I didn't push as hard made more noise, chattered, and left a worse finish. More thoughts on this in a minute...
20210105_194803 by ARNEWB, on Flickr

Apparently I missed a picture of the countersinking bits... You can see the shank of the larger one that I have been using for the m8 hole and counterbore and the m5 counterbore. Cheapy set of 4 for $9 dollars on amazon, and they work great.

Finished plates "front".
20210105_204015 by ARNEWB, on Flickr

Back.
20210105_204028 by ARNEWB, on Flickr

All four plates on the one side of the router for now. Just verifying bearing alignment, fitment, and motion.
20210105_205416 by ARNEWB, on Flickr

Next up is grinding more t-slot nuts and getting the rails on the other side of the router. The I can move on to cutting my gantry risers. I am going to use the same 80x160 extrusion for the risers and gantry. Then I have to make the interface plates that allow the gantry to bolt to the risers. Hopefully Australia will pipe the EF down tomorrow and I can have a little more time in the garage.

More thoughts on chip load, router performance, etc...:
I stumbled across this video the other day (okay, like 8 months ago...), and the chip load vs finish vs speed surprised me. This video is filmed on a CNCRouterParts/AvidCNC machine that is VERY similar in construction to what I am building, with the biggest notable exception being that they use Rack/Pinion instead of ball screws on X/Y. In reference to the chip load thoughts on the counterbore bit, it seemed a lot happier when I was forcing it to take a bigger bite than if it was taking small bits. Obviously rigidity and mass weigh in as well, but the spindle on a bench top router is doing a pretty damn fine job in steel in this video, once he starts getting a little further into his testing.


Also of interest... After the discussion of epoxy granite earlier in this thread, I was browsing through YT early this morning before I went to bed, specifically CNC router video's, and found this video. Based on the earlier discussion, when this machine comes back apart and gets rebuilt, I think I might follow suit and stuff the frame, at least, with epoxy granite. I want to figure out how much my gantry and Z axis weigh before I really contemplate filling that as well, however I'm not sure about the necessity given the results of the above video cutting steel on a similar machine.
https://www.youtube.com/watch?v=OCKU0K-E0-c&t (Since the board doesn't like more than one video per post)...

Not quite as much progress as I hoped for, but still moving in the right direction, and I suppose dinner with friends isn't a horrible excuse.

[Grant H.]

Gantry interface plate layout is done now too... Apparently the Alreco saw had a rough day when these plate remnants were made... The outside top corners have "R" scratched into them (by me) to denote the reference corner. Those corners are square and a great match on each plate. So those two perpendicular edges are where all of the measurements come from. The inside bottom corners have a nice "swoop" to them. The plates were obviously cut by the same saw at the same time, as they are nearly perfect matches on all edges.

The idea is that the gantry extrusion will sit flush to that "R" corner, and have bolts attaching it to the risers underneath and behind it.

20210105_225607 by ARNEWB, on Flickr



Hurray for more drilling, counterboring, and countersinking...

[Grant H.]

Progress should be reasonably steady on this,

Why... Why do I write/say stuff like this?

Work blew up when Corp IT informed us that their ef-up had caused exposure on the gear that we maintain... Lots of hours. Planning to use the crappy weather as an excuse to hide in the garage and make some serious progress. Best laid plans and all that... Ended up working on O/G customer stuff with my brother (still sort of involved with my brothers company... I'm apparently still useful...)

I managed to get a few things accomplished, but not much picture worthy.
1. The gantry/riser interface plates are all drilled, counter bored, and counter sunk.
2. Ground down more t-slot nuts
3. Cut the gantry extrusion to the right length
4. Cut the gantry risers to the right height

And...

5... Ended up making some changes to the plan...

Despite the tails of the linear rails hanging off of the frame, I am not going to cut them down to use in the Z-axis. I'm going to leave them long so that I can expand the base in the future. It takes up a little more room that way, but it's not a huge deal. So, I ordered up some more rails to build the Z-axis. This will delay things a little bit, but not too bad.

Oh well, you win some days, and you lose some weekends...

More actual updates soon enough.

[Tim K]

I remember when we first discovered chip loading. The theory is easy to understand once it's explained, but we found the recommendations to be counter intuitive. The size of chip required to carry away the heat is always a surprise. I would say that most hobby woodworkers run their routers at feed rates that are half or less of what's required to be at the correct chip load. Even 15 years into it, I'm constantly surprised at just how fast our CNC runs when we have the chip load calculated correctly (which is 100% of the time since it's programmed into the software). The only time we deviate from recommended chip loading is when we're unable to hold the workpiece solidly enough to withstand the lateral cutting forces. Invariably when we do that we get burning on the cuts.

[Grant H.]

I remember when we first discovered chip loading. The theory is easy to understand once it's explained, but we found the recommendations to be counter intuitive. The size of chip required to carry away the heat is always a surprise. I would say that most hobby woodworkers run their routers at feed rates that are half or less of what's required to be at the correct chip load. Even 15 years into it, I'm constantly surprised at just how fast our CNC runs when we have the chip load calculated correctly (which is 100% of the time since it's programmed into the software). The only time we deviate from recommended chip loading is when we're unable to hold the workpiece solidly enough to withstand the lateral cutting forces. Invariably when we do that we get burning on the cuts.

It is interesting watching CNC wood video's and seeing how many seem to burn the edges when they are doing profile cuts.



More progress!

Gantry to Riser Plate bolted onto the end of the Gantry:
20210106_221355 by ARNEWB, on Flickr

Quick test fit of one of the risers:
20210107_155552 by ARNEWB, on Flickr

Gantry with the rails and plates installed:
20210112_205054 by ARNEWB, on Flickr

Since the gantry is going to be supported by t-slot nuts and bolts, I decided to go a little over board...
20210112_205753 by ARNEWB, on Flickr

And.... It's starting to look like a CNC router. The gantry is actually surprisingly heavy (more than I was expecting).
20210112_213051 by ARNEWB, on Flickr


I can grab one riser and slide the entire gantry back and forth without any noticeable binding, or drag, so I am not concerned about the ability to move it with smaller steppers to start with.

Some may notice that the upper riser-rail block has changed. I was originally shooting to have bolts on both sides of the block, but given the ball screw mounts I have, there isn't enough clearance with the bolts in the middle.

There are some squareness issues that I have to chase down, so that should be fun...

Given that I ordered some other rails for the Z-axis build, while I am waiting on those, I am going to continue working on getting the ball screws on the x/y axes, motors mounted, and then I'll start on the power supply, stepper drivers, wiring, and motion. I assume the rails will be here before I make it through that whole list, but at least I have enough to keep me busy until they show up.

Ball screws for the X axis (gantry support - we can argue about axis designation if you really want, but that's how mine is designated) are going to be a little goofy to start with. The ball screws that I have from my original parts buying are only 800mm long, and I likely need 1000mm or 1050mm for the dimensions that I have decided to build this for now. So, they are a touch short for the moment, but I am going to make them work and ensure that the design is functional, and then use it to rebuild some parts for the final version.

[Great-Kazoo]

It is interesting watching CNC wood video's and seeing how many seem to burn the edges when they are doing profile cuts.



More progress!

Gantry to Riser Plate bolted onto the end of the Gantry:
20210106_221355 by ARNEWB, on Flickr

Quick test fit of one of the risers:
20210107_155552 by ARNEWB, on Flickr

Gantry with the rails and plates installed:
20210112_205054 by ARNEWB, on Flickr

Since the gantry is going to be supported by t-slot nuts and bolts, I decided to go a little over board...
20210112_205753 by ARNEWB, on Flickr

And.... It's starting to look like a CNC router. The gantry is actually surprisingly heavy (more than I was expecting).
20210112_213051 by ARNEWB, on Flickr


I can grab one riser and slide the entire gantry back and forth without any noticeable binding, or drag, so I am not concerned about the ability to move it with smaller steppers to start with.

Some may notice that the upper riser-rail block has changed. I was originally shooting to have bolts on both sides of the block, but given the ball screw mounts I have, there isn't enough clearance with the bolts in the middle.

There are some squareness issues that I have to chase down, so that should be fun...

Given that I ordered some other rails for the Z-axis build, while I am waiting on those, I am going to continue working on getting the ball screws on the x/y axes, motors mounted, and then I'll start on the power supply, stepper drivers, wiring, and motion. I assume the rails will be here before I make it through that whole list, but at least I have enough to keep me busy until they show up.

Ball screws for the X axis (gantry support - we can argue about axis designation if you really want, but that's how mine is designated) are going to be a little goofy to start with. The ball screws that I have from my original parts buying are only 800mm long, and I likely need 1000mm or 1050mm for the dimensions that I have decided to build this for now. So, they are a touch short for the moment, but I am going to make them work and ensure that the design is functional, and then use it to rebuild some parts for the final version.

Igf (key word) IF there's issues with the stepper motors ability to move the gantry, smoothly. Have you thought of a Plan B now, if that does occur?

[Grant H.]

Igf (key word) IF there's issues with the stepper motors ability to move the gantry, smoothly. Have you thought of a Plan B now, if that does occur?

I have. It's noticeably more expensive, but it solves a number of issues in one shot.

Clear Path Servo's. I can get nearly double the torque values compared to my steppers in the same body size (Nema 23), and I eliminate the possibility of missed steps, as they have built in encoders to validate the motion. Beyond that they also simplify my control cabinet, as I don't have to have stepper drivers to run the servos. I can feed the step/direction signals directly from my control board to the servos, along with ~75VDC and it's all done.

If that's not enough, something is likely wrong, but I can always step up to Nema 34 frames. Going to better gearing in the ball screw is also noticeably expensive, but it doesn't solve any of the other potential issues, so it isn't really a great solution either.

[Grant H.]

Next up was the table support pieces.
20210122_213930 by ARNEWB, on Flickr
20210122_221850 by ARNEWB, on Flickr

While I am still waiting on a few things to show up, I have decided to start on the cabinet, power supply, and other such accoutrements.

Gutting the cabinet. I have the advantage of having access to used O/G cabinets for a bargain, when they are being decommissioned from the field.
20210122_223940 by ARNEWB, on Flickr

Unregulated 45/25/18VDC power supply. This is just one of the 45VDC legs wired up to test it.
20210122_233641 by ARNEWB, on Flickr

Main cabinet/system power contactor and momentary push button's. Between the N/O contacts and the option to wire this for latching power, it prevents the system from powering back on without manual intervention in case of power loss.
20210122_181911 by ARNEWB, on Flickr

Supposedly some of the parts that I need for moving on with hardware on the machine will be here tomorrow. However, that's USPS tracking, which means it'll probably be 3-20+ days from now. If the box actually shows up it'll mean that I can start adding the ball screws and steppers. My Z-axis rails showed up today, so I can also start working on building my Z-axis as well...

[Grant H.]

So... yeah... almost 3 years later...

I lost all motivation for the photo documentary part of this when my personal life exploded shortly after my last post, I did continue on and make it functional, although calling if finished would be entirely false.

I am ramping up to "rebuild" the router now, and figured there would be some interest in where it ended up, the process to upgrade/rebuild, and likely some of the projects that I want to use it for.

Looking forward - Upgrades:
Increase usable cutting area to at least 4'x4'
Replace steppers with ClearPath Servos
Replace no-name chinese 110v 2.2kw spindle with Jianken 220v 2.5KW spindle with Automatic Tool Change capability

Likely part of the rebuild:
Epoxy granite bed
Epoxy granite gantry supports
Epoxy granite replacements for Y extrusions

Some of the history

Control Cabinet Build:
First cut of the power supply after putting it together on my desk. Used a scrap of plywood to mount it for testing.
20220214_222506 by ARNEWB, on Flickr

Rats nest of wiring for testing motion.
20220214_222459 by ARNEWB, on Flickr

With motion proven, it got built onto the back plane of the cabinet.
20220308_134832 by ARNEWB, on Flickr

Closer shot of the Mesa Electronics board to the Stepper drives. I put the effort in to make the cables nice and clean, and then had to undo it later when I swapped a few things around. Lesson learned. Wait to make things pretty until it's a fully baked product...
20220312_130134 by ARNEWB, on Flickr

Despite the progress on assembling it, that's about as far as it got. I never did get it put into the cabinet, as the machine got put to work almost immediately.

Z-Axis Build:
Z to X axis interface plate
20220207_201937 by ARNEWB, on Flickr

One of the best things about having built this router, and it was one of my original plans, is that I can use it to machine accurate parts to use during the rebuild. I did a decent job getting things laid out and manufactured by hand, but there are some holes that had to be oversized to add enough slop to get it to work.

So much dykem, time with calipers, scribbled dimensions, and drilling...

Mounting the spindle on the Z-axis plate.
20220209_192552 by ARNEWB, on Flickr

I switched the Z-axis around from "traditional" builds and mounted the bearing blocks to the interface plate and the rails to the longer piece. The advantage to this is that I have a much smaller intrusion into the working envelope when the Z-axis is retracted. The "downside" is this makes the machine noticeably taller, but I'm okay with it.

Z-axis plate mounted on the gantry with the X axis ball screw tied in.
20220210_200025 by ARNEWB, on Flickr

Z-axis mounted on Gantry:
20220211_095322 by ARNEWB, on Flickr

Y-axis Motion:
Tandem Y-axis motors and ball screws, mirrored on the left side.
20220211_212716 by ARNEWB, on Flickr

Motor mount and ball screw fixed block.
20220211_212729 by ARNEWB, on Flickr

Long term, the offset between the Y extrusions and the X extrusions is intended to have an end cap with integrated Nema23 mount. The plan is to use 3/4" aluminum and have a nice, clean, integrated plate/motor mount. Now that I am going to swap those extrusions and use the full 1500mm of the rails, this may get changed around.

Proximity limit switches, also used for homing on each axis. I wanted to avoid the physical contact micro switch solution that lots of entry units and first time build use. These are cheap ebay special ones, but thus far, they have been extremely consistent and reliable. Just a simple 3d printed mount. While these will trigger off of aluminum, the reliability and consistency is much better with steel, hence the bolt threaded into the side of the Z-X interface plate for homing the X axis.
20220625_230759 by ARNEWB, on Flickr

Not a ton of other photos before I dove into running the machine.

First test cuts on the 2x4 and then cutting 14ga steel in prep for cutting holes into automation cabinets for the O/G industry.
20220217_182831 by ARNEWB, on Flickr

Cutting parts to adapt an older dust collector to an Oneida cyclone.
20220227_115023 by ARNEWB, on Flickr

Machining aluminum parts for a custom Pan/Tilt mount for $100k IR cameras being used for AI based gas leak detection. Before getting serious about this, I added mist coolant with control from the linuxcnc side.
20220528_192854 by ARNEWB, on Flickr
20220528_192917 by ARNEWB, on Flickr

There are STILL aluminum shavings in most of the garage. I don't run dust collection when I am using coolant.

New ATC 2.5KW spindle next to 2.2kw spindle. The old one weighs ~12lbs and the new one is ~34lbs.
20220226_131942 by ARNEWB, on Flickr

[rfenster]

Thank you--had been following with great interest back in Covid times...very impressive!