Hi Josh. Solid looking design!

As for the Z axis, I definitely vote for the single motor. I spent May and June testing various dual motor designs against single motor designs running dry prints involving thousands of Z lifts throughout 24 hour, 8 inch high, prints. After dozens of these tests, and consulting Prusa's evidence for NEMA motor microstepping errors, I've concluded that it is, by design, impossible to keep 2 NEMA class motors in sync over time. And worse, auto-bed-leveling schemes actually increase the error rate. Measuring the two sides of the X axis with digital depth gauges it was not uncommon to find the two sides off by more than a layer height at end of print using two motors, regardless of driver (RAMPS, Melzi, Azteeg, Printrboard), whereas two belt driven single motor designs have logged more than 50,000 Z lifts with perfect accuracy. You can see the resulting printer here http://www.openbuilds.com/builds/onez.2278/.

The main problem was getting closed loop belts the right size for running the two screws. RobotDigg has a wide selection at great prices and their delivery time is now usually 9 days to USA. RobotDigg also has 8mm bore pulleys for the screws. A NEMA 17 can run two belted acme screws but at high voltage and temperature. Stick with your NEMA 23. As for ratio, don't over-think it. Use the same size pulley on the screws that you do on the motor and you have the same ratio as a direct drive. In the firmware simply use the same setting for steps-per-mm that you'd use for direct drive. Your NEMA 23 will have plenty of power without need for down gearing.

If you are going to use a closed loop belt for the single motor Z axis, consider doing the same on the Y axis. Mount the Y axis motor parallel with the screw, fix a pulley to the screw with a matching pulley on the Y axis motor and run the screw with a short 208mm or 132mm closed loop belt. That would make for a compact, under the bed, design that would not increase the height or depth of the printer footprint. And, it will look really cool - justifying the use of an acme screw instead of a belt. If you already have some closed loop belts for the Z axis and would like some 208mm closed loop belts to try out on the Y axis, let me know. I have 6 of them from an abandoned design I'd gladly pass on to you.

 

I do not use acme screws for printers, in spite of the fact that they look really awesome. Consult Josef Prusa's http://prusaprinters.org/calculator. (His calculator for lead screws is mislabeled - for manual input his field labeled "Leadscrew pitch" should read "Leadscrew lead" since "lead" is actually the mm/revolution (lead = pitch * starts)).

On a Z axis using a 1/16 stepper, an 8 lead Tr8*8-4 acme screw can not use full 0.1 layer increments since each step is 0.04mm. Layers 0.2, 0.4, 0.6, etc, are OK but 0.1, 0.3, 0.5 etc land on half steps causing the motor to jump to the closest full step. Since I have no control over the layers that user of my printers may want, I opt for 8mm threaded rod with 1.25mm lead which does land on full steps at 0.1 layer increments. (A Tr8*8-2, with only 2 starts instead of 4, would land on full 0.1 layers, or a Tr8*8-4 using a 1/32 stepper would too.)

I've never tried an acme screw for X or Y axis, though I see some designs that do. How fast they are, I don't know. But consider, a motor revolution on your Tr8*8-4 moves 8mm travel, while a revolution using a 20 tooth GT2 pulley moves 40mm travel. So, your motor will be running 5 times faster on a 100mm/sec print than if you use GT2 belts/20 T pulley. NEMA 17 can run that fast, but the driver is probably going to run pretty hot. Like I say, I've never tried it, but the math kept me from trying it.

 

You should be able to land on full steps with the 8mm lead acme screw that's offered by the parts store. With 1/16 stepping and 1.8 degree steppers, 8mm travel per screw revolution, you should be getting 0.0025 mm travel/step. This is based on the following math:

(1 rev/360 deg)*(1.8 deg/1 full step)*(1 full step/16 steps)*(8mm travel/1 rev) = 0.0025 mm/step or 400 steps/mm.

If you want 0.1 mm layers, then 0.1 mm/0.0025 mm/step = 40 steps. So every 0.1mm layer should be 40 steps. I would soon have the acme screw for less friction, straightness, and precision. Standard screw threads always bound up on me. My math may be wrong, but that's what I come up with.

 

+runningfarmer I use Prusa's method for a 1/6 step motor and Tr8*8-4 .
One revolution moves 8mm.
One revolution has 200 steps. (1.8 x 200 = 360)
8/200 = 0.04 mm per step.
2 steps = 0.08mm, 3 steps = 0.12mm, 4 steps = 0.16mm, 5 steps = 0.2mm. Consequently, 0.1mm would require 2.5 steps.

+Josh B It's hard to compare screws and belts on a milling machine vs printer. On a printer the X & Y belts are moving only the mass of the extruder or bed. On a milling operation a belt moves similar mass but also is applying the cutting force. On a miller, belts that are not engaging a stationary track should stretch under that load. That is why acme screws are ideal for X & Y on a mill. But, you're not milling at 100+ mm/sec, which is ideal for a belt on a printer. Per your link, that kind of lead gets up to the belt speeds, but, I'm thinking that a screw that moves an inch per revolution may require a NEMA 23 motor. That distance should require about 3 times more torque than a Tr8*8-4 which is moving about 1/3 the distance per step.

 

Yes, for a full step it's (8mm travel/1 rev)/ (200 full steps/rev)= 0.04 mm/full step. I said 1/16 stepping not 1/6th stepping, so you'd have (0.04mm/full step)*(1 full step/16 steps) which = 0.0025 mm/step. Most people run their 3D printers at 1/16th stepping, so you will not have half steps with the openbuilds leadscrew. So again, for one 0.1 mm layer, you get 0.1mm / 0.0025 mm/step = 40 steps.

 

Confusion typically comes with talking about stepping. Most people are using 1.8 degree/step motors which is where the 200 steps comes from. Dividing 360 degrees by 1.8 gives you 200 full steps/rev as Keith pointed out. You usually don't use full stepping because of poor resolution. As you increase your stepping, resolution increases. When I say 1/16th stepping, this means you've multiplied one full step by 16. This means you now have 200*16 which now equals 3200 steps per 1 motor revolution. This is much larger than the 200 steps per 1 motor revolution with full stepping. As long as you setup your equation so that units cancel where they should you'll be fine. I've calculated all my machines this way and never had a problem.

 

+Josh B Threaded rod always has only one start, therefore pitch and lead are the same. Acme threads can have multiple starts and lead (the distance traveled) = starts * pitch. In these calculations pitch is irrelevant, lead is what we are measuring.


+ runningfarmer You are correct - IF you rely on microstepping for Z layers. But let me quote Prusa at his calculator for "Optimal layer height for your Z axis" (http://prusaprinters.org/calculator/)

Helps you to select layer height in a way, that Z axis moves only by full step increments. Z axis isn't usually enabled during inactivity. If the axis is disabled during micro-step, axis jumps to the closest full step and intorduce error. This effect is occuring to some extent even while leaving the Z axis motors enabled. This is most usefull to machines with imperial leadscrews but also for unusual layer heights with metric leadscrews.

At first I doubted his assertion that "axis jumps to the closest full step". So I set up a printer using 8mm threraded rod with the X carriage belt removed and carriage parked to one side with a digital depth gauge mounted to the X axis. I loaded the SKP for OpenBuilds Universal Build Plate in Sketchup and raised it to 100mm. I then sliced it with .2 layers and set it to lift .2 on each non-print move. With 72 holes to have to Z lift around per layer, that totals over 36,000 Z lifts for the print, or 3600 lifts per 10mm of travel. I then measured the height of the X axis at bottom and ran the print with no filament. Pausing every 10mm I measured the height. Using .2 lifts, I consistently raised 10mm, as expected. But, if I set the lifts at .22 (forcing microstep parking) the difference could be as much as a layer off from 10mm. As Prusa had found, setting the motors to remain engaged during rest decreased the problem, but did not remove it. I then changed from 8mm threaded rod to 8 pitch acme screws and set the lifts to .3 (forcing them to stop on microsteps). The increase was so dramatic that I didn't even finish the entire print (about 24 hrs) before pulling the acme screws out of the machine.

+Kyo I am running at 60mm/s with infill at 80mm/s right now, with out problems. I suppose that would be a feature if you were pushing a 10 lb. extruder or build bed, but I'm running 300mm/sec for perimeters and infill using GT2 belts on my OneZ printers (PLA @ 215C w/Universal E3D) with motors and drivers "warm to the touch". Where is the advantage when the acme screw is not providing cutting force or moving far more mass than an extruder or build bed?

 

+ Josh B There are three thresholds for speed.

The first one is the frame. Old RepRap A frames, Prusa i3 Birch or Acrylics just can't deliver the rigidity. V-Slot extrusion frames certainly can because they are designed to produce rigidity for CNC milling. I have a BldrBot (http://www.openbuilds.com/builds/bldrbot.632/) built on V-Slot with X & Y belts that just logged it's 9,000th hour printing ABS and PLA at 100mm/sec and I've never touched a belt once I had them setup. If the frame is rigid and belts are tight, resolution is excellent.

The second threshold is the extruder's ability to heat up that much filament that fast. The BldrBot is limited to 100mm/sec because it uses a QU-BD 9 with a PTFE liner. That limits the "melt zone" to only 5mm. On the printer that is running 300mm/sec I'm using an E3D with a metal barrel which has a 13mm "melt zone" and another 20mm pre-heat "transition zone" above the barrel's heat break. Then above that I tap out the aluminum fin body to 4mm and insert a PTFE tube that runs from the top of the barrel to the drive gear. That prevents heat creeping up PLA and provides a channel all the way from the gear/idler to the nozzle for flex filament. Because the PTFE is above the "melt zone" & "transition zone" I can also print hi temp Nylon and Polycarbonate filaments up to 300C.

But, speed depends on the filament. PLA transitions from solid to liquid and I can actually print it at 350mm/sec. But ABS transitions from solid to plastic to liquid. That extra phase change requires more heat and I'm limited to 250mm/sec with ABS. On the down side, printing NinjaFlex is still at a pitiful 35mm/sec.

One you have a "melt zone" long enough to melt the filament the next threshold is temperature. It takes more heat so a higher temp is required. That's why I limit PLA to 300mm/sec at 220C because 350mm/sec at 230C is just too hot for PLA (perimeters around small holes will crystallize).

But, beware, speed isn't all it's cracked up to be. Example, I have a 7" x 3.5" plate I print for the RAMPS box. The difference between 100mm/sec and 300mm/sec is only 1/3 less time not 3 times faster. Using 45 degree infill the max stroke is 120mm, but only about 80mm of that is running at 300mm/sec. The deceleration and acceleration at each end eat up the remaining 40mm. Add some holes to have to work around with more deceleration and acceleration and the speed advantage diminishes quickly. With smaller prints you may never actually accelerate to full speed at all. If you can print at 100mm/sec you are doing fine pal. The only reason for engineering for faster speed is so you can say it does it, which is a biggy if you are designing a printer for sales . On the other hand, if the frame is rigid, belts tight enough for the "twang test", motors and drivers running cool enough, and an adequate extruder, that 7" x 3.5" plate looks and mics out as an exact duplicate whether it is printed at 50mm/sec or 300mm/sec.

 

Lots of good info! We are pulling together a bit of a giant printer with Monster. But pretty much the same overall goals as you are here. 3D printer that can also handle some possible CNC and laser work. I was at the Portland Mini-MakerFaire last weekend and one of the folks there had a small printer with a CNC spindle as part of the X-axis assembly. They were running Repetier with scripts to be able to machine the part on the bed either right after the print or during it. I didn't see which. (I'll add the link to their site when I get home - I can't remember it at the moment) Parts off the machine were very impressive.

 

I've tried them on a CNC Router. I discovered before I even finished the build that they were just not going to work out. By the time I got them adjusted loose enough that they traveled freely enough to not jam up, there was too much slop in the assembly to be of any use. For what you are spending on this build, "saving" $20 on these would be false economy.