For a while now I've been wishing my Shapeoko 2 was more rigid. I've cut aluminum with it before, but it was not an experience that inspired confidence. The bit chattered a lot, and I had to stand there with WD40 to keep it lubricated. So the time has come to upgrade it. Significantly.
For reference, here is a model of my current machine, which is a 1200x500 Shapeoko 2, which uses Makerslide, 6mm GT2 belts, v-wheels, and Nema17 motors. (I should also note it uses a Makita router and aluminum spindle mounts)
Early version of the upgrade design using non-captive stepper motors on leadscrew:
For a while I thought about sticking with a belt design, using 9mm wide belts. Gantry was going to be 2x4 rectangular aluminum tube, and would have to be machined at a real shop.
Design as of 11/7/2017, switched to 3x6 extrusion for the X-axis gantry and 3x3 for the Y axis, with a pre-made Z slider from CNCnewbie.
Current design as of 12/4/2017, updated the Z axis to my own design using long-version profile linear bearings.
At first this was going to be just a gantry upgrade, but it quickly turned into much more as I thought about each piece. As you can see, this is turning out to be more of a complete rebuild, rather than an "upgrade." I'm not even sure any of the bolts will be original parts when I'm done.
Goal #1: Keep the same overall size
My current machine is 1200mm in X, and 500mm in Y. I know this is backwards from a stiffness point of view, but the long gantry makes it really easy to work with. I don't have to reach over the side rails. Plus, I just finished building an enclosure, so it has to fit inside that.
Goal #2: Eliminate the sources of flex as much as practical. Be able to cut aluminum with ease. Stretch goal: Be able to cut steel on occasion.
Most of the flex I am seeing manifests itself as twist around the X axis, but there are still several sources that contribute to this (in order of significance):
Goal #3: Eliminate sources of inaccuracy and calibration headaches.
- Torsional rigidity of the extrusion itself: At first I was just going to just buy the "wide makerslide" to replace the original double makerslide setup. But with a 1200mm gantry (longer than SO3 XL), I'm not confident that this will give me the stiffness I'm looking for since the SO3 uses a variation of 80x40 extrusion.
- Flex in the carriage: I can clearly see that the carriage is also flexing relative to the extrusion. There are things I could do to stiffen this up, such as mounting each front and back pair of wheels on a single, longer bolt. But there is still a significant amount of flex in the delrin wheels and bearings. Steel wheels would be more rigid, but those cant ride on aluminum makerslide). I could try Openbuilds' solid wheels, but they will never be as rigid as other methods, and it still has the annoyance of running over debris that falls on the extrusion, causing the cutter to walk. So I decided to ditch all that and go with what the big boys use, which is Hiwin-style linear bearings.
- Flex in the Z-axis: Here we have the same v-wheel setup, with the same problems I just mentioned. Let's get rid of that. CNC4Newbie.com sells Z-sliders that use LM12 linear bearings on round rod. Bonus, it includes ACME rod! It comes with a hole pattern that will fit an X-carve, which will fit an SO2. With a slightly different hole pattern (free of charge), it will fit my Makita mounts. Haven't quite decided if this is the unit I want yet. He sells some other units which look promising also.
The main culprit here is the belts. My machine lives in the garage where temperature fluctuations happen a lot. Belts stretch and the controller needs to be calibrated to the specific belt tension. The belt is the thing reacting the tool forces and will stretch against the load, and cutting aluminum well will require something better, more accurate, and free of stretch. And, aside from cost, belts just don't seem like a good idea on a CNC machine. For a 3D printer they are great, but a CNC needs something better. Whenever someone asks me about my Shapeoko, I describe it as a "CNC designed by people who usually build 3D printers."
▼ Build Updates
I ran some FEM analyses using Creo Simulate on a few different extrusion sizes and compared to the double makerslide.
Deflection of 1200mm double makerslide (bolted together) with a 30 in-lb moment applied in the middle. If I do the trigonometry to get the deflection at the cutter tip, I come up with .006"
Here is Misumi's "rigid-type" 80x80mm with the same moment. Deflection at the cutter tip would be reduced to less than .001".
And here is Misumi's 100x50, which is not quite as good as the 80x80:
To be clear, on my current machine I am seeing about .015" of deflection at the endmill bit, when I apply 5 pounds on it, at about 6 inches away from the rail center (5lbs x 6in = 30 in-lbs). The fact that this is more than the .006" calculated by the FEM means there is significant deflection in areas other than the beam twisting, such as V-wheels, carriage, belts, etc. I already knew that, which is why I am replacing the V wheels with linear guide.
Major Update & redesign
- 3"x6" T-slot extrusion for the X (gantry), 48" long (automation4less). It's 28 lbs!
- 3"x3" T-slot extrusion for the Y rails, 20" long (automation4less)
- 1/2" thick aluminum end-plates all around, sourced from a local metals supplier.
- 3/8-8, 4-start lead screw on the Y axis (dual driven) (source TBD)
- 1/2-10, 5 start lead screw on the X axis (source TBD)
- Motors will probably be 269 oz Nema23. (source TBD)
- Linear rails are BLH 20mm rail. X guides are 1155mm, Y guides are 506 (2mm short of 508, which is 20") (automation-overstock).
- Z axis will be a purchased unit, probably from cncnewbie
And now for pictures.
- Approximate cutting area is now 34" in X and 13" in Y. Bottom edge of the carriage plate is 4.1" from the spoilboard. Haven't quite decided if I want to keep that or go slightly taller to accommodate thicker materials.
Leadscrews are held using a fixed-supported arrangement, with the fixed portion being the end opposite the motor with a couple of pillow blocks from servocity. The end closest to the motor is just supported by the motor itself with a flexible coupler. I suspect the bearings that come with the pillow blocks are radial contact. If I get whipping I may switch them out for angular contact.
Limit/homing switches and hardstops on the X axis are independently adjustable.
Cross-section thru bearing support:
The leadnut block on the gantry plates is a 2-inch piece of 1x3 extrusion, with an AB nut from dumpstercnc. Limit/homing switch position is adjustable. Hardstop is just when the linear bearing runs into the Y end-plates, which have a cutout to accommodate the grease fitting.
All end-plates are intentionally simple, so that I can just peck the hole-centers and cut a shallow part outline on my current machine, then cut out the part on my table saw or chop saw, and drill all the holes in my drill press.
(view from the rear of the machine with X rail hidden). X axis AB leadnut is from cncrouterparts.
Update 11/14: Build Phase Started
Now for the exciting part. Hardware is starting to arrive.
Setting up the tool paths. Note that because my current machine lacks stiffness, I'm just pecking the hole-centers with a V bit, and cutting out a .050" deep outline of the parts (using trochoidal paths, which might be unnecessary). The final part outline is to be final-cut using a table saw and miter saw with a metal-cutting blade, and corners filed by hand. I did have to mill out the larger holes for the stepper motor and bearing block. Using a 1/4" 3-flute end mill and trochoidal pockets, it wasn't too bad, although I did notice that the large hole isn't 100% round so it may need some hand-filing if the motor doesn't fit. Used 20 ipm feed, 10,000 rpm, .0625" depth of cut with a 1-degree spiral entry.
One end-plate almost done. Here I'm tapping the motor-mount holes to accept 10-24 helicoils later.
Fit check with the gantry rail.
I ordered some anodizing supplies. Plan is to red-anodize all of the end-plates. Note: if someone is planning to reproduce something like my build, do NOT install the helicoils until AFTER anodize is complete. The steel helicoil will dissolve in the acid.
X and Y extrusions from automation4less.com, and linear guides & bearings have arrived from automation-overstock.com. They appear to be very good quality. One minor annoyance is that the spec posted on the website says the bearings have M6 threaded holes, when in-fact they have M5 holes. (so don't use that spec!) I had been emailing back and forth asking about the long version of these bearings, which are available but not posted, and they sent me the spec sheet for ALL sizes (attached here at the bottom). that is when I noticed that parts of it didn't match the other spec and checked the bearings I received. When I brought this to their attention, I got radio-silence. So if you order these BLH bearings from them, beware! Unfortunately I had already drilled the holes in my gantry plates for M6 screws, so there is more slop than I had intended. (It's a 5mm/.197" screw in a 6.75mm/.266" hole -- pretty loose. Anyway, I bought some tube so I can make "bushings" with it, which will fit over the screws before inserting them, and take up some of the slop. I was sorta hoping that by bringing this discrepancy to their attention that they might offer me a discount on my next order. If & when I get a response from them I'll update here.
Ends of extrusions tapped and helicoils installed.
Bearing arrangement re-design, with BEEFIER bearings. I couldn't find any bearings already mounted in flange blocks that didn't cause problems (most are just way too big), so I will just have to make my own.
Kinda slow progress as I am only really able to work on this when my toddler daughter is sleeping.
Successfully anodized some of the plates! The parts are a brilliant red
Basically my process is:
1. Sand parts to get rid of scratches etc. I went up to 800 grit.
2. Degrease parts. I used brake parts cleaner in a spray can.
3. Wash parts. I used simple green and a scrub brush. Parts need to be "surgically clean". Any oils will interfere. Don't touch parts with bare hands beyond this point.
4. Anodizing: Sulfuric Acid solution needs to be about 15% concentration. Battery acid from the auto parts store is roughly 35% sulfuric acid, so diluting 50-50 with distilled water will get close. I bought a battery hydrometer to check the specific gravity to verify, because according to the MSDS for the battery acid, the tolerance on the 35% is quite large, and I expect some of the water to evaporate over time. (Remember add acid to water, and wear gloves& goggles). For cathodes I have 2 titanium plates (10cm square) suspended on the sides of the tank using hooks made from bent aluminum rod. (Ti is supposed to last longer than Al). To suspend the part I also used titanium rod. Hook up a battery charger, + to the part and - to the cathodes. For current, you need to calculate the surface area of the part. The rule is 720 amp minutes per square foot will produce .001" of oxide. In addition it should be at least 12 amps per square foot -- I'm using about 16 a/sf. For the 2 larger parts (165 sq. in. each) I ended up doing about 20 amps for 30 minutes with "motor drive" mode on my smart charger. Voltage was about 14v but to you should try to use constant current and let to voltage do what it needs to. Cathodes will bubble a lot. I did it in the garage with the door open and a fan blowing because it produces hydrogen. (Actually on the first attempt, the wire I used to hang the part was too thin, and started glowing. Probably not a good thing with explosive gases being produced!)
5. Rinse part. I rinsed in tap water but then spray with distilled water to keep from affecting the PH of the dye bath.
6. Dye. Mix up dye per directions with distilled water. I used actual anodizing dye but you can also use Rit, etc. (I read that Rit will fade I'm sunlight). Dunk the part and leave in there for up to 15 minutes. I did about 10 min and didn't worry about heating it up. You can heat it (but only a little, like 100F) which will reduce the time.
7. Seal. Boil the part for at least 10 minutes. I'm still experimenting with the boiling time. I ended up doing about 20-30 minutes. Some of the dye will leach out at first, so make sure the color starts out slightly darker than you want. Steaming instead of boiling will apparently reduce that, but can also result in splotches.
The beefier bearings arrived, and I was able to press them into the blocks I made using a vise.
Heavily upgrading my 1200x500 Shapeoko 2.
- Build License:
- CC - Attribution Share Alike - CC BY SA
Reason for this BuildBecause I want to cut aluminum and my SO2 is wimpy.
Inspired byShapeoko 3, OX, CNCrouterparts