KRE Mill V3 - Concept

I don't like to post concepts or ideas as there is almost a 100% chance of them not going anywhere, but this is one that I want to get moving very early into the next year.



Disclaimers:

• This is not based on Openbuilds components.
• I am in no way talking bad about any other machine, just stating that everything has it's place.

The project started as a true mill with an XY moving table. However I couldn't quite get the work area I wanted, without having to go crazy on the frame. I wanted at least 8" (200mm) in the Y. I switched to a moving Y table, and a moving XZ spindle. Because that opened up so much space, I went ahead and just gave it a 12" x 12" (300x300mm) build plate.

The revision of the XY table mill turned into this. It seemed like it would work just fine, but everything just felt wrong. The Z was admittedly always sketchy. After that was revised, it really highlighted the rest. (Old Z has the extrusion.) The upright support also didn't inspire confidence. The next issue was the X axis. It was based on two 4040 sections separated, then mounted to the uprights. That's only creating a greater leverage. Plus the 4040 may not be enough for what I want. I did add a back plate with an extrusion mounted to it, which should have reduced any fore/aft flex, but that brings us to the final issue. It was butt ugly. I believe in function over form, but you should still be able to extract form from function.



That brings us to the V3. It is built from four 4080 extrusions at the base instead of two 4080, with some 2040 support. The uprights also share the same side face as the base extrusion which allows for an easy integration of a support plate instead of just the corner brackets.



The X axis is now built from two separated 4080 extrusions, but with a slightly reduced gap. Due to the fact that the uprights are no longer attached to both X extrusions, the possibility for swaying is now a thing. To counter that, I simply used 80mm wide rear mounting plates. There are also short sections of 4060 inside to allow for ball screw mounting hardware, and to a provide a little more "squareness stability." Adding a web would be incredibly easy, but may not be a single bit necessary.



The entire build is based on 20 series extrusion, "HGR20" assemblies, 16mm ballscrews, and a mix of 1/4" and 3/8" aluminum plate. 20 series uses M5 hardware and had a manageable slot gap. Lead screws can be used, but finding decent metric anti-backlash nuts that don't cost a fortune is a chore in itself. That would also affect the serviceability. The work surface, ballscrew mounts, and both z plates are 3/8". Although the Sphinx can machine the aluminum for this, I'd rather build this with acrylic first for two reasons: to find weak points and to buffer the upfront cost. Even with acrylic, I'm sure it would still be above the level of the v wheel system. A good test would be for it to machine its own plates while in its acrylic form.


Goals

• Efficiently mill aluminum
• Use all of what the router/spindle has to offer.
  • Although not by a long margin, the Bosche Colt router was actually able to over power the wheels. Then a chain reaction would start. I'll be placing an order for a 1.5kW spindle in the next two weeks. That will go on my Sphinx, but is intended for this mill.
    
  • This isn't a machine fault though.
• Easier maintenance and disassembly
  • C beam with inner wheels almost requires a complete breakdown.
  • Every axis on this mill can slide off after decoupling the nut.
  • Every axis can also be independently separated from the other axis as well as the frame without compromising alignment or integrity.
• Tighter tolerances
  • Less flex to keep the end mill straight is a big deal in the finishing and time departments.
• To have fun and learn more
  • Although it's for side business/hobby purposes, we're all still builders here.
• Flood safe and easy cleaning
  • This was not an issue before, but I'd use 18-8 SS hardware.
  • I'm going to cover as many slots as possible. What cant be cover by plates would be covered by a printed part.

I'm still working on the details, but I was just curious to hear what others make suggest.

Kev

 

Regarding the screws, I'm still on the fence. 1604, 1605, and 1610 all share the same nut dimensions. 1616 is larger by a tap, but I don't care about this variant. I was thinking about going with 1605 all around I haven't done the math yet on rpm/torque/feed yet, but the Z will have the 1605 for sure. I don't see myself wanting more than 1200mm/min during operations, and 2000mm/min free travel. Maybe it's just resonance, but 2000mm/min begins to sound scary on the 500mm 8*8 lead screws.

Regarding the height, I made it that high to account for different vises. The space between the bottom of the X assembly and the build plate is a safe 110mm. That's accounting for up to a 1/2" plate. I've been looking at various vises and just came across these little guys. They shouldn't be too hard to replicate, and consume very little height. With that said, I can definitely drop the gantry. I just need to be able to work with up to two inch high material (plus vise height).

All of the following pics are with a 1/2" (12.7mm) plate instead of the 3/8" (9.525mm) plate seen above. The end mill is just a round representation of a 1.25"-2.5" fly cutter set to 1.5". There is enough side clearance to allow entry from the left and right to surface the plate. I would keep the minimum 1.25" though.

220mm - current


220mm view without plates


220mm with larger side plate - I did not change bolt locations for the pics, but you get the idea.


195mm with larger side plate - Notice the change in spindle mounting height. You can also take a look at the distance of the stepper from the top of the picture frame. This is also showing min and max Z.


After the 25mm drop, It still has a safe clearance of 85mm. The ironic part is that I make the "do you really need that height" argument all of the time. I do agree that the upright is more than likely the weak point. It's 4080, but then again, so is everything else. I can fill is with something. I should be able to use these brackets instead of the L's for the inside. I dropped the gantry another 5mm in these two pics.



I'll definitely be posting whatever I can. I'm more than likely going to order the spindle in the next week or two. The itch needs to be scratched. I would like to get a water cooled ER16 220V 1.5kW spindle. The issue is that I don't have 220V. Tapping 220V isn't exactly rocket science, but it's not my space, and I don't want to pay an electrician to do something I know I can do. I'll probably just grab a 2kW transformer for the time being. Whatever happens, I do have 220V where I live and can play around with it to confirm it's working. If I don't order the spindle, I'll be ordering the extrusion instead.

Speaking of spindles, another concern was the mount. I absolutely despise the generic cast mounts.

• Weight is good for dampening, but I don't see THAT weight being beneficial. I'd rather just fill the frame with sand.
• I wouldn't be able to precisely add mounting holes where I want. I don't have access to a press.
• That are ugly.
• They could help with heat, but heat isn't an issue unless something is catastrophically wrong.

I'm kind of against the printed route, but is there actually an issue with this? Would a printed mount fatigue? My design is 120mm tall with a 15mm thick border (except the back center). The mounting holes are outside for squaring purposes. I also gave it five holes on each side with a 20mm spacing. The idea is to maybe use some for accessories such as coolant brackets or lights or a roast beef sandwich maker. The clamps are spaced 30mm apart. Everything is M5. I would definitely use washers for the mounting bolts.




Further info:

• 300mm ball screw and guides on Z
• 500mm ball screw and guides on X and Y
• all M5 hardware, which was another reason for keeping the 20 series
• 267oz-in NEMA 23 steppers in the CAD model

 

What do you consider shallow? I was cutting 900mm/min at 2mm DOC with 2mm stepover with a 1/4" diamond o-flute under flood, and 900mm/min at 9mm DOC with a 0.5mm stepover on a 1/4" variable flute ZRN 3 flute under flood as well. 8000mm/min... I'm honestly scared to even try that. I think I capped mine at 2500mm/min. I will definitely try bumping that up slowly. I admit that it's a mental block more than anything. I haven't messed with ball screws yet, but they are much lower in the friction department.

I completely agree with you on the torque-to-speed statement.

Keep in mind that the OB mounts are made for routers with larger tops. The back is only 5mm thick, pretty much there just to keep it together. The thing for me is pressure. I'd rather have very little pressure over a larger area than a high pressure over a smaller area.

 

Someone else running more than 250mm/min at 0.2mm DOC! Did we just become forum buddies? Your post made me go back and check "ideal" feeds and speeds. One critical factor missing from both of our numbers is RPM. Plugging everything back in, the 36ipm (900mm/min - switching back to imperial since that's the more used system for chip loading) works out to be 18k rpm for each tooth at 1200 SFM with a 0.002 chip load. The 3 flute should be at 108ipm (2700mm/min), but I was running it as a finisher. The Bosch did begin to bog so I had to set it to 22k rpm.

I ultimately went with the 2mm DOC and 2mm optimal load for balance the tool use through that specific operation, but I do want to move to full depth from the get go. The goal here is to get that depth (possibly greater than 12mm) and a decent step over/optimal load.

Since I should be able to better control RPM, I'll be properly recreating my F360 tool library.

I was actually going to message Chris to see if he could make the Z axis rail spacers. I can do it on mine, but it's just one of those things I'd rather leave to a big machine.

 

Chip thinning is something I've thought about, but I've never actually put it together with optimal load/stepover/RDOC. In order to compensate, you must advance the cutting edge further along than normal. That 2mm doesn't sound absurd anymore, but it hurt my DOC for sure.

This is a pretty good formula for adjustment. How To Combat Chip Thinning - In The Loupe - Machinist Blog

Following that, dropping from 50% to 25% optimal load increases the desired chip load by 4.

Ultimately, the greatest indicator is the chip itself. If you are throwing good size chips, there shouldn't be any heat issues, nor will thinning be an issue. The only thing that wouldn't be optimized is tool life.

All of this isn't worth much when trying to use wet noodles to force end mills through materials. I'll let it sit for a couple of days (actually hours) before going for the extrusion. Maybe the little router can survive with new machining methods.

 

After replicating the Mod-Vise, I released that the build plate didn't have a dense enough pattern. So here's the updated table and vise.



I've been struggling with limit switches. I want the machine to home the rear right (+x, +y). To that, the bed has to slide forward. I don't want any protruding from the front. These are the larger lever limit switches. It's either this or a proximity switch.



Then there is cable management. I would like to run the cable chain to the -y side of the z stepper. That should save a bit of machine height. The only potential issue is how close the exit may be to the top of the spindle. It would have to fit three water lines, the spindle power cord, and the z limit wire. I left more than enough accommodation for whatever.




I also did try higher travel rates. I didn't realize it, but my machine is hitting a resonance around 1500mm/min, but quiets back down at 2000mm/min. It did 3000mm/min just fine, but either slipped or stalled at 5000mm/min. I didn't really sit down to diagnose. Speaking of resonance, I plan to fill the extrusion on this new machine with sand. No epoxy granite, just sand.

 

I didn't even think about that. I'm pretty sure it's set to 1/16. I'll give it a shot.

 

The extrusion is on order! I also started recreated all of the plates in Fusion.
A big issue I was having was nesting the plates. It took some time, but I finally got a decent layout.

 

I pretty much have everything completed in Fusion. The acrylic also arrived, but this week will be busy from another big project. I'm hoping to have it all cut out this weekend though.

I've been debating how I want to set up the probing operation. I initially thought of using an optical sensor as a rough Z, then follow up with a slow plunge for the final Z. This would all have to be compared to the WCS. I've created a z probe code already that worked fine, but this time, I would like for the probe to only provide a variable that can be used in a different code based around the WCS zero instead of a table top zero. It doesn't help that my explanation sounds confusing.

EDIT: Actually, I can just let the machine operate in absolute and keep all 6 WCS

Here is the original code:
G28 G91 Z0
G28 G91 X0 Y0
G38.3 F100 Z-80 (probe decent and trigger)
G0 Z20 (move 20mm Z+)
G92 Z33.35 (set tool height, 20mm plus 13.35mm touch plate height difference)
G92 Y323.7 (set x)
G92 X325.2 (set y)
G28 G91 Z0 (this should raise it to home zero, while retaining the correct change in height)

This is also why I want this new machine...

 

So um... It's official. I won't start an actual build thread until everything is together though.

I got to unpack the extrusions and let me tell you, the cuts are perfect! I had concerns that it would be relatively light duty, but this stuff is sturdy. I want to say it's thicker than the OB V slot, but I'll have to wait till tomorrow to confirm that.

As I mentioned earlier, I'll be filling the extrusion with sand, "which distributes the energy of the strike over a longer period of time and reduces rebound" according to Wiki. To help with the filling, I have designed some inserts. Some silicone around the perimeter should be enough to seal and keep them locked in.

I didn't see myself getting more 500mm length extrusions. I thought my next step would be 1500mm.

 

I gave my limit switch breakout board a revamp. The completed one (second picture) is what I use on my Sphinx. It accepts two switches per axis. Due to the fact that it is normally open, it does not need two switches per axis. The main output connector was not installed in the first picture. It is filtered and has a dedicated wire shield pin.



The revision only accepts five switches. Z negative is pretty useless if you've already done major damage from burying the end mill into the bed of the machine. The main difference is that the revision is normally closed. The filter and shield were carried over. This one also uses SMD components instead of the through hole components. This one would also be made as a real pcb instead of as a mock up.



As the build progresses, I'll see what circuits would be a good addition. The only other thing I can think of would be a relay panel. Those are already available, but why not.



While we're on the topic of electronics, I plan to simply move everything over from the Sphinx as a temporary solution. I was thinking of going with an Arduino Uno, but there is only one spare pin after connecting everything. The step up would be the Mega, but you can get the 32bit Due for the same price.

Boards & Modules - Arduino

 

The only reason I went No before was because I didn't have access to pcb making. I also didn't have flip operations figured out yet. Having a pcb with a bunch of wires kind of defeats the purpose of having a pcb.

The printer guys already hit that 8 bit barrier. I did some research on whether the Due is supported or not. It wasn't originally, but I believe it is now.


It would appear that I'm having a very bad night. The first extrusion went just fine. The first tap broke in my second extrusion. The second tap broke in my third extrusion. They were both combination drill taps. I'll just try again with my plug taps tomorrow. The issue now is getting them out. The only thing I can think of is to freeze the extrusion with tap, then pour hot water (maybe boiling) around the tapped holes. I can then reach in with a notched flat head and try to screw them out. I have no way of securing them to drill them out.

I did get some other stuff completed,but more on that later.

 

I haven't made an update in a while.

I was never able to get the taps out, but I didn't really try either. The other holes tapped just fine with the plug tap.

The acrylic test plates were technically a failure in two ways. Some of them came out correct, but most had misplaced holes. This was caused when Fusion did not update the CAM positioning after making adjustments in the models settings even after resetting the drilling operation. The second error came from the manufacturer. The acrylic was actually closer to 5.5mm than 6.35mm (1/4"). That means that 12mm long bolts will not work without 1mm worth of washers... 96 of them. I'll own the Fusion error, but this one is irritating. This isn't the first time they've sent incorrect thicknesses. I still have enough to re-cut everything, but

I made plugs for the extrusion. I used silicone to secure them, but the gap wasn't completely closed. I'll just pour in a little epoxy to seal it for good. Then they can be filled with sand. I'll be going with glass blasting sand as it has a higher density than regular sand.

Almost everything is on order. All that is left is the aluminum for the frame: 1/2" table, 1/4" for the frame, and the Z spacer blocks. I already have 3/8" for the Z axis.

I did bump up from the 1605 to the 1610 for X and Y.

It was never an issue before, but after having to move my Sphinx into a far less convenient setup, I've realized just how back and forth jogging is. It would be sweet if I could add jog functions to aa contro panel (along with e-stop, hold, resume, etc).

 

Here is the third iteration of an enclosure.

• Flood proof
  • more than likely going to be made from melamine
  • has room to house flood hardware (pump, tanks, filtration)
  • accounts for plumbing.
• Easy to assemble
  • based around torsion boxes (The bottom panel is just a shelf, and is not part of the structure.)
• Compact but easy access all around
  • easy to clean
  • no tight/hidden corners
  • 800mm wide, 750mm depth plus door, 1550mm height
  • The machine bottom is 800mm off of the ground.

I do not want an all-in-one type setup with electronics housing, tool storage, or material storage. The idea is that if things go well, I can just build an identical unit and treat these as individual modules. However, there will be provisions for attaching all of those.

It won't really change the pictures below, but I will extend the width to 850mm. As of now, the inner/under space is only 600mm (23.622") wide. A standard 5 gallon bucket is a little under 12" in diameter, and it would be nice to fit two side by side without having to pull one forward.

 

The linear motion showed up! These things are pretty big. Seeing them in person vs in the model really provides a different perspective. Two of these slides, maybe even one alone, may be more than what an entire machine worth of wheels can handle...and there are four per axis.

What's left:

• 1/2" for the work surface
• 3/8" for the entire Z assembly and main X/Y ballscrew supports
• 1/4" for everything else
• Z axis rail spacers
• Wiring
• Limit switches
• Proper spindle - My second Bosche doesn't have much more time in this world.
• Spindle mount

The corner brackets, nut blocks, and ballscrew floating supports are in transit

 

Everything is all cleaned up from the protective fluid. Here's an unmounted picture for size reference.

The linear components kind of has me thinking. Why not just beef the plates up to 3/8" from 1/4" and go with a 2.2kW spindle? Blue is what most people seem to get. Green is the ER16 80mm variant that I was looking into. But if the weight of the 80mm 1.5kW isn't a concern, then why not just bump up to the 2.2kW ER20? In this case, the concern wouldn't be weight bearing on the machine, but the weight on the stepper. Then we get into the screw pitch, 5mm vs 8mm, and friction levels, ballscrew vs lead screw. As for the price, the difference is negligible.

I wouldn't use this chart as the holy grail, but it's an idea.



It's a rabbit hole. Can the steppers handle it? I doubt the little 269oz nema 23's that I already have can. The strongest nema 23 is 425oz with a length of 114mm. The nema 24 maxes out at 566oz at a size of 60x60x100. They both maintain the same mounting holes.

 

I did consider belt reduction for the purpose of torque and space saving, but I hadn't considered the "smoothing" benefit as much.

I've spent a good chunk of time trying to find an applicable belt setup, but nothing has satisfied yet. The issue is the 10mm shaft on the ballscrew. The standard pulley maxes out at 8mm. You can get larger bores, but so far I'm really only finding 6mm wide belt slots. The wider 9-15mm belts are all GT2-2m instead of the desired 3m.

The other limiting factor is the max size of the spur pulley. It can not exceed a 60mm diameter. I still haven't played with the design too much though. I may be able to completely get rid of this limitation.

Also 4:1 may not be possible, but 3:1 is very doable. I wouldn't go smaller than a 20t.

A vibration analysis would be cool. It looks like there are arduino accelerometers available.