Inertia matching.... yep, you just threw me down another rabbit hole. I did some quick math. Accounting for the plate and ballscrew (no pulley, linear slides, or "stiction"), the inertia is about 220,000g*cm^2. The 425oz stepper is 800g*cm^2. With a 4:1 reduction, we're in the neighborhood of 18:1. The reason why I brought up tooth count was just for engagement. For vibration and resonance, I planned on filling the entire machine with glass sand (blasting media) as it's a bit denser than regular sand. Although the machine will be in a stable environment, I didn't want to get into possible thermal differences. After more digging, my conclusion remained. If it didn't work, it's removable and would only be $8 down the drain. As for the pulleys, I did not consider using an non-flanged spur. The 60mm limitation would be the available space between the gantry (x) and above the rear frame support (y) if the ballscrews remained in the same place. I'm trying to move the pulleys outside, but that presents other issues. There isn't enough space between the rails and the ballscrew support. (Ignore the oversized plate.) Without pushing the rear 4040 cross member inwards, this would be the only way I can see this working without redesigning the current frame. I will continue to play with it and see what I can come up with. Keep in mind that this is only for the Y axis. The X axis should be a bit easier. I just want to say that I really do appreciate the feedback from everyone. Once everything is up, running,and proven, I'll be posting everything with the build. EDIT: Would shifting the ballscrew to the side be frowned upon?
That sounds a lot better than 275:1 to me! It's certainly a balancing act, especially given that as you get into higher torques, it's usually better to go up a frame size than add length in terms of actual usable power. It could be that a fairly short NEMA 34 motor in the 3-4Nm/425-560oz-in range gives you much better torque and inertia numbers at the expense of a bit of acceleration. But you may gain a lot of that acceleration back from the better inertial matching- and could probably gain some rapid headroom with a lower pulley ratio, say 2:1 or 3:2. I'd play with the numbers a whole bunch in a spreadsheet or something and see what you get before settling on any specific drive arrangement, since your machine seems to be designed and built to a standard that would really benefit from the tuning. Gotcha. Yeah, actual ability to transmit torque certainly helps. As a rule, I stick with XL on any machine with meaningful torque. If it's good enough for the drive on my lathe, I'm ok with it. It's relatively flexible for its size, too, and bore diameters are available in a pretty substantial range. That's a pretty good plan, for sure. That white quartz sand is great for machine-filling by all accounts due to the way it packs. Thermal issues you likely won't have to deal with unless/until you're trying to get below around 3 tenths, or around 7.5 microns, reliably and consistently. If you're just trying to hit 0.002 or something fairly reasonable, definitely not an issue. The aluminum of course doesn't help with that, with its higher coefficient of thermal expansion. What I'd do, if it's feasible, is move the 2040 inward so that the screw bearing plate is on the outside of it, and then put the motor on both a) some standoffs off of that plate, and b) a full strap beam running side to side across your 4040 cap plates, so it's doubly reinforced without actually being directly mounted to extrusion. That should give you tons of space.
The Nema 34 510oz has an inertia of 1000g*cm^2. This would bring the ratio down to 12:1. Speaking of motor size,these pics show the difference between the 76mm 270oz and the 114mm 425oz. These pics are offset by 10mm, but there is no room for adjustment. If this is the route to go, I would still move it over another 5mm. The tooling plate would remain the same. The only part that would change is the spacer between the plate and the nut block. A little more refining. This is with the 15mm offset. a 3/4" Spacer would be needed to mount the FK12 blocks for the ballscrews, but that shouldn't be an issue. One glaring downside with the change is the lesser back plate. I do have the 4060x295 that I can shove between the gantry. The intention was to cut that down for two smaller internal gantry supports, but that's not really needed anymore. One glaring benefit is the overall size reduction, especially after the extra length of the 425oz steppers.
Hmm. Perhaps there are discussions elsewhere online about this kind of marginal decision-making. Obviously the answer is servos, as it usually is, but $1500 to just get a machine working is pretty crazy. There must be effective conclusions on the matter somewhere, though. Looks good. You could also flip that z-axis motor around to the back side to gear that down too. I'm sure it'd appreciate the reduction with all that load on it, but it would also significantly reduce the visual height of the thing. I'm still not 100% convinced about the y-axis screw offset, though with enough rigidity and spread in the rails, it theoretically shouldn't be an issue. Whether that translates to practice remains to be seen.
I was looking into servo setups. That would be about $350 for steppers and drivers. 15mm Offset vs 20mm drop No Offset It didn't even occur to me, but I can just replace the 4040 with 2040/2060. It adds an extra 3mm in axial spacing which is nothing. After removing half of the support, I just combined the ballscrew support with a corner plate. Regarding the Z axis, it will get the same treatment. I just haven't touched it yet.
Three-fifty?! Per axis that seems pretty low for the power required, and for all three seems like magical wonderland pricing. Where'd you find that, because I was thinking DMM DST 0.4kW (might be able to squeeze 0.2kW) and DYN2 driver as the lowest feasible cost option. Clearpaths would probably be a little bit higher. You don't gain anything in terms of high-speed torque or inertia matching by going with closed-loop steppers, just accuracy and reliability (And even then, $120 an axis would be pretty cheap at this size). That looks pretty good, I like it! This is all kinda giving me an itch to go back to some of my back-in-the-day gantry designs and update them, I do have some uses for a Datron-style machine for aluminum. One project at a time, though!
Tree-fiddy for that eBay/Ali special. I didn't even dare look at the Clearpaths. I've always gotten my steppers through OMC-Stepperonline. They have a really wide selection of steppers, drivers, and related components. One project at a time... Tell me about it. I'm waiting for black friday to get some more vslot for a laser.
It's been a while, but everything (minus the plates) is finally here. A few simple mods were made to allow for both belt and direct drive. I'm honestly only doing that to allow for a temporary option. The steppers would only be supported by two bolts/standoffs like many of the OB screw driven setups. Tolerance is another thing. After playing with the acrylic, every little thing shows up during assembly. Any slight misalignment or off angle tap creates a bit of resistance. To help alleviate that, I'm considering drilling the holes a little larger than 5mm; maybe 5.5mm. I'm also considering taking off a little around the perimeter of the plate; maybe 0.2mm. The F360 plate files have been deleted to avoid any confusion. I recreated the newer version in AutoCAD, which is a much easier program for 2D work. That was then transferred back over to F360. I would like to use a 1/4" end mill for time efficiency, but I'll be going with 1/8" for material efficiency. I can fit the 1/4" plates on two 13"x13" sheets which is an odd size, but sure is convenient!
It's been a while since the last update. Shout out to @Chris Laidlaw for making the z axis rail spacers. He'll more than likely be making more All of the belts and pulleys arrived. The belts feel really plastic like. Is that how a GT2 is supposed to feel? I ended up going with 60t 9mm wide pulleys for the ballscrews, and 20t for the steppers. 100 limit switch and 10 3-way relay board pcb's are being made. The X and Y ballscrew nut block spacers are done. I decided to simply print them in PLA+.with 3 perimeters and 40% infill. I'll do the same for the spindle mount, which should be printing tomorrow. I feel like it's overkill, but better over than under in this case. The spindle mount will consume ~450g of plastic, uses 4 clamping bolts with captive nuts, has 10 mounting holes spaced at 20mm (5 on each side), and will clamp over a height of 120mm.
Hi. ☺️ As seen, it weighs 64lbs. It will be right at 100lbs when complete. Then I still want to add the sand filling, but that will wait till it's housing is make otherwise I'll need a few extra bodies to move the thing. I'm working on the third version of the enclosure. To help with vibration/resonance dampening and isolation, I'm strongly considering TPU feet. I'll just use mdf or some thick acrylic as a temporary bed, but I'll get 1/2" mic-6 plate. I've ditched the mechanical limits. I have THESE hall effects on the way. Previously, only the limit cables were shielded. This go round, the stepper cables are shielded. The VFD cable is double shielded (haven't ordered yet). I know it's been a while, but the end is near.
The HY 2.2kW spindle and VFD are on order! I also forgot to link the end stops. Goals for the enclosure: - flood friendly - vibration isolation - sound insulation - easy access to the machine - compact and bit more modular than the previous full cabinet
The wire that I got for the steppers was nice, but would not work for this application. I have a smaller 18/4 version of the same type of cable used for the spindle on order. I did receive the 16/4 spindle cable and omg, this stuff isn't safe for work. When I get the rest of the wire, I'll share some nice pics. You vs the guy she tells you not to worry about...
I purchased two 50x15mm (inner) cable chains, but will only need one. The spindle and VFD are genuine Huanyang products. It's a 220V 2.2kW (3hp) setup with an ER20 collet and four bearings. The first thing I checked was to see if the internal assembly was actually connected to pin 4 for grounding. I'm happy to say that it is, which definitely makes me feel more comfortable about my purchase. I then performed a continuity test. Any two points on the external surface, including the shaft and collet assembly, are connected to pin 4. I'm currently doing some research to see if pin 4 would be safe to use for a tool probe connection. I would prefer to ground it though. Since I was testing continuity, I tested the entire chassis. Any two points all the way up to the spindle mounting plate on the z axis shows continuity. Due to the printed mount, the spindle is completely isolated from the chassis. It's kind of hard to fit the limits to the machine when USPS doesn't want to deliver them.
One decision I absolutely regretted when I built my Sphinx was my "form fitting" wiring. My wiring from the machine was just long enough to get the control box. This was great and all until I had to move things around. For this build, there is approximately 3ft worth of wiring off of the machine, which is way more than enough in this case. The stepper connections are complete, but I'm not 100% satisfied. I may desolder and trim the conductors back about 3mm to get the outer jacket securely in the clamp. It will work just fine for now though. There is a lot going on in the cable chains. spindle cooling in spindle cooling out VFD power X stepper cable Z limit cable flood hose x2 To help keep it tidy, I purchased some Velcro straps. Zipties tend to leave marks on the wire sheathing. I also made quite the disappointing discovery. The regular drop in "hammer" nuts do not work with Misumi rails. That creates quite a dilemma. I will try the roll in nuts, but after that, the only options are to either buy the Misumi drop-ins or take plates off to add tee nuts for anything. The silicone ground wire came in today. USPS seems to have "rediscovered" my endstoos that where supposed to be delivered on the 24th.
Wiring is the worst part of machine-building! Hahaha. This thing's looking great, I've been watching it come together.
I definitely got it tamed though! The only cable missing from the "spine" is the limit cable. The limit cable is a five conductor double shielded cable (just like the spindle and stepper). It will be hooked up to a breakout board mounted to the gantry. Because I switched from mechanical switches to digital switches, all of the boards I made previously are now useless. The first cut with this would be a one off breakout board. I also made some loc-line brackets from acrylic. They do need to vent to final shape. (It was the last thing the Sphinx got to cut. RIP, you will be remembered and eventually rebuilt.)
For perspective, I'm 5'9" (5'10" on Tinder). The hood will have to wait till tomorrow. I'll dump a gallon of polyurethane onto this thing after all of the holes are made. It also probably weighs twice as much as me, but that's not saying much...
IT'S ALIVE!!! With the gear reduction, the steps/mm is 480 at 1/8 microstepping. I dropped it to 240 at 1/4 due to serious step loss, but the issue is still there. I doubt this will be an issue after some acceleration tweaking. EDIT: I had the wrong microstepping numbers above. They are now correct. It seems to be more of a speed limit. At 1400mm/min it gives out, even at 25mm/sec^2. With this in mind, I bumped it back up to 100mm/sec^2 with no issues below that 1400mm/min threshold. I need at least 2000mm/min, but 2500 would be ideal. I wonder if it's a torque issue. I'm reusing the 269oz 2.8A steppers from the sphinx. The drivers are set to 2.37A peak. I'll bump it up to 2.84A.
I don’t know which controller you’re using (and to lazy to scroll back) but you might want to do the math and figure out how many steps per second you are getting. I think grbl can handle about 20k but ymmv.
Phoenix controller with DQ542MA drivers. 10mm pitch ballscrews 1/4 stepping = 800 pulse/rev 1:3 gear ratio --------------------------------------------- 800*3/10=240 steps/mm Of course theoretical and actual are not going to be identical, but it won't be off by anything more than 2%. As for mm/sec, I can't scientifically measure that, but a rough stopwatch gave 6.36sec to travel 100mm at 1000mm/sec. Accounting for human error and acceleration, that makes sense.
So, 1000mm/min (assuming minutes, otherwise it's reaaalllyyy fast!) about 16mm/sec = 4khz. This should be fine w.r.t. step rate. With good drivers I found that it was actually the step rate that was the limiting factor on my Sphinx. By lowering step rate to 1/4 or 1/2 I could hit speeds of 8000mm/min. The whipping on the lead screws weren't that reassuring though.
Sorry, yes I meant mm/min. One thing to keep in mind is the drive reduction. The 1300ish ceiling would be 3900. It's set to 1/4 stepping. Changing the current and pulse length didn't have an effect.
It actually makes sense. At 1000mm/min, the stepper is spinning at 300rpm. We also have 450rpm at 1500mm/min. Taking a look at the torque curve, it looks like the stepper may just be maxed out. https://www.omc-stepperonline.com/download/23HS30-2804S_Torque_Curve.pdf
Did you end up cutting the plates yourself. or did you have them cut for you.. if so can you share where you got them cut?
Yesterday, I sadly took apart the Sphinx. We won't forget you. Moving on, the stand is built and the enclosure is in place. There is a single 240V outlet in the back corner. It is powered by a 12/4 cable hooked up to a dryer plug (14-50P) on one end, and a smaller 3 prong receptacle (6-20R) on the other. I still have to add strain relief to the 12/4 cable, but it is hooked up and functional. I was originally going to split that into a single 240V and two 120V receptacles, but there are already outlets literally right next to the 240V. If I have to move it, I can always add that feature later. I purchased two 5 gal and one 40 gal container. One 5 gal is for the spindle cooling. The other two are for flood stuffs. I'm still trying to figure out the filtration side. The spindle and VFD are fully functioning on their own. Now I'm playing with hooking it up to the controller.
