Z Axis Worm Drive

Overvierw of new Worm Drive System
Bottom view of the holder for the worm drive

After some thought, it was determined that a previously infeasible Z axis drive solution, is now not only practical, but essential.

Originally, this method would require several expensive mechanical pieces to get working; driving up the cost and complexity of the solution. It was then determined that a simple electromagnetic brake would suffice; since these can be sourced from omc-stepperonline along with the necessary motors. The electromagnetic brake, will prevent the shaft of the motor from turning so long as it is not energized. Therefore, an electronic sensor would have been needed to detect when the stepper motor was no longer energized, so that the braking system could be engaged. A complex solution to say the least.

DC Electromagnetic Brake; available from
omc-stepper online

Now, with the new frame design, an old idea has become practical. Initially due to the size of the frame, a D-Shaft of approximately 27 inches in length would have been needed. This length cannot be obtained off the shelf. As a result, a custom length would have to be ordered, or a longer length cut down to size. Ordering custom length is cost prohibitive. Cutting the stainless steel rod down to size is no easy feat. Not only are they stainless steel but they are also hardened. Further, the cut piece would go to waste, as there would be no part of the machine that could be serviced by it.

D-Profile Rotary Shafts, available on Mc Master-Carr

Now that the frame’s size has gone down to a more standard 24-inch length, an off the shelf 24-inch D-Shaft rod can be used. To that end, a worm drive from ServoCity will be used to turn the driveshaft, that turns the timing pullies, that lifts the CoreXY subframe.

27:1 Worm Gear Set, available on ServoCity

This approach eliminates a stepper motor and it’s associated driver, replacing it with a 24 inch D Shaft, and a Worm Gear set and six 1/4 inch flanged bearings. Incidentally the total cost is only marginally higher, even negligible. This will take PolyNC from having a 4 stepper drive system to a 3 stepper one. in total this is a marked reduction from the original 7 stepper motor design.

Update: The parts do not fit together

Sadly, the parts do not fit together. The hub and worm gear from servo city do not fit the d-shaft from Mc-MasterCarr, nor any of the stepper motors in my collection. Here is why:

The flat width of the d-profile is 1/8 of an inch.

The flat width of the profile from McMasterCar (and most likely the stepper motors) is 1/8 of an inch. This means that the distance between the flat part of the shaft and opposite site of the wall is almost exactly 6mm.

ServoCity doesn’t specify the flat width

The parts from ServoCity, however, do not specify the flat width. As it turns out the distance from the flat to the opposite side of the hole is only 5.73mm; too small for the various d-shafts.

All in all its back to the drawing board.

Update: A new worm gear has been sourced, and the solution works perfectly.

Ebay listing of a suitable worm gear set.

https://www.facebook.com/permalink.php?story_fbid=1112878895728295&id=764430967239758
Comment on facebook!

Hotend Ejection Mechanism

PolyNC hotend ejector mechanism, with red ejector handle

To facilitate the rapid changing of nozzles, I designed a method by which the entire hot end (the nozzle, heater ,thermal barrier and cooling shroud) can be removed in a tool-less manner. The system consists of a keyhole shaped stand into which the hotend snaps into place. The red handle is used to push the hot end forward just enough so that it snaps out of the keyhole.

Hotend dislocation after red handle is depressed.

Although somewhat costly, this method would vastly decrease the amount of time normally taken to switch from one nozzle to another.

Fan located behind hotend

The hotend cooling fan is connected to the back of the mount so that it does not need to travel with the rest of the hotend. There is a hole that allows air to flow through to the heat sink.

The scalability issues of various types of FDM 3D printers

ThreeD Printer Home
MacroMAKEiT: Affordable Large Scale 3D Printer launched on kickstarter September 18, 2018

One of the major concerns with owing and operating a large format 3D printer, is that you would prefer to get the largest possible build volume for the smallest possible foot print; where footprint is the volume of space that has to be allocated to the printer and can be expressed as an imaginary cubic bounding box around the outside of the printer. It is best visualized if you had to build an enclosure around the 3D printer to keep out drafts, dust or “curious fingers”. Since in some facilities/homes space is at a premium, this build ratio would have to be as large as possible.

The build ratio is defined as the build envelope / the space required to enclose the printer.


There are several variations on the two main co-ordinate systems; Cartesian and Delta.

The RepRap style

(or as I call them: traveling beds, because the bed travels back and forth)

RepRapPro Mendel, the progenitor of modern desktop 3D printers, uses a traveling Y axis bed within a gantry style cartesian system.

Core XY, Floating Bed

HyperCube Evolution by SCOTT_3D. An example of a CoreXY Floating Bed 3D Printer

Core XY, Static Bed

The voron2.1 An example of a CoreXY with a static bed

Delta Printer

Monoprice MP Delta Pro

All delta printers have a static bed. They also tend to lend themselves to being scaled in the Z axis very easily. They also have the worst build ratio of any style of printer.


The problems of trying to print big

The main problem with scaling up a 3D printer is having to increase the size of the build plate. Additional heating requirements can be overcome by switching to an AC mains heated bed, however having to move that bed either up and down or front to back exposes a problem that becomes all to clear when the full build volume of the machine is being used: The build plate becomes heavy.

Very heavy. It is expected that a large format 3D print will require many kilograms of filament to build the entire model. In the case of the world’s largest 3D printed object, as produced by the US Department of Energy’s Oak Ridge National Laboratory in 2016, the final part weighed in at around 750 Kg. That printer, incidentally was a gantry style cartesian with a static bed. In addition to which, the bed itself must be robust enough to carry that mass without itself warping. This would mean even more mas in the form of the build plate just to to keep from warping under the strain.

The Keenovo Integrated Silicone Heater; available on Amazon

Most 3D printer filaments require a heated bed. One notable exception is PLA. Aside from PLA, everything requires a heated build platform. At larger scales, it can be expected that just heating the platform will require a thousand watts of power (depending on size and target heat). Drafts and other sources of heat loss, from a non enclosed printer will cause an increase in that power requirement. Keeping the bed insulated from all heat losses is paramount. With a non static bed insulating it is non-trivial.

Since the prints are expected to be large, it is understandable that the user would not want to wait weeks for a single print, even if it is of high value. Print times on a large format should be comparable to desktop printers. The biggest limiting factor to print times is the rate at which the filament can be heated and extruded. A powerful hot end will be required so that enough thermal energy can get into the filament before it is extruded. If it is extruded at the wrong temp, then the print could fail.

A Maxiwatt, hot end. Its main feature is that the heating element surrounds the filament as it passes through

Multiple heating elements are needed to bring the filament to temp at prints speeds above 60 mm/s (desktop 3D printers generally operate at 30 mm/s). Additionally, large format 3D printers use a large diameter nozzle; around 1mm or more. Since time is of the essence there is no need to print a large object at with a .4mm nozzle unless the design calls for it.

Finally, there is the cost concern. Large format 3D printers typically cost two orders of magnitude more than desktop printers. The Anet A8 set the bar at the absolute lowest cost 3D printer that a person could buy. To achieve this, they had to cut back on some creature comforts that should not normally be left out (e.g. a power switch). At $150 USD it sets the benchmark as the lowest cost point that a machine could cost and still be considered a compete 3D printer (and it didn’t come with filament).

The 3D printer shown at the start of the article, the MacroMakeiT, kickstarted at $7,000 USD for a print area of 12 times that of the Anet A8 at 46 times the price; and this is one of the lowest cost large format printers you could obtain. A reasonable desktop printer, with the creature comforts such as the Ender 3, costs around $220 USD. At that price point the MacroMakeiT cost around 31 times for a 12 times build area. I are comparing the area as opposed to volume because most of the printer styles tend to scale well in the Z axis anyways.

What is needed is a print mechanism that starts at a reasonable point scales linearly with the build area. Doubling the print area should not cost 10 times the price.