Tuesday, August 31, 2010

Mk4 extruder fail.

The makerbot MK4 extruder has been working fine as I reported earlier.... until today. While trying to print a small object, the extrusion stopped. The heater was working fine, the extrusion motor was working fine, but the filament was not moving anymore. Closer inspection showed that the ball bearing which is around the axle of the motor had shifted and almost fell out.

This was a reason to take the whole extruder apart to check what was going on inside. When taking the top plate of the extruder and removing the motor, the ball bearing came off... with part of the pulley.

The pulley should look like this (photo from makerbot flickr pool).

The thin flange of the pulley came off and after removing the small shreds of aluminium the flange is running loose around the rest of the pulley. I did put the extruder back together with the loose flange, but the bearing was pushed away again after extruding less than 5 cm and the filament stopped moving again.

I have seen many broken extruders in the blogs, but usually they where related to the insulator leaking, the retainer plate breaking, or the lack of grip on the filament because of the idler wheel slipping away. I did find some references to the same failure (e.g. here), but the best indication that this may not be such an uncommon fault is the fact that Makerbot is selling the pulleys also specifically "if you have broken your pulley"...

It is actually not so surprising that this part breaks. The idler wheel is pushing hardest in the middle of the pulley, so the filament will try to take either inside corner of the pulley to get away from it. This will create a serious sideways force. The flange of the aluminium pulley is only press-fit on the pulley and it looks like it is not strong enough. I guess the solution is to get the new MK5 drive gear upgrade kit which is made of one single piece of stainless steel and it looks much stronger.  Possibly I make something similar myself.

I had already started work on a new "Wade" type extruder, but this is not ready yet. I will try to finnish that asap to get a functioning printer again, but I do want to fix my MK4 too, since having a spare extruder will come in handy someday.

Some parts printed when the pulley was still ok: the Hemi-Demi sphere from thingiverse. These parts:

will make this sphere:

Tuesday, August 24, 2010

New extruder

The makerbot extruder is very good, but I really like Wade's extruder design which is using a stepper motor to push filament into the hot end. Having a stepper motor will give much more control over the extrusion which should help in combating the ugly plastic strings which are drawn between parts of an objects caused by the extruder leaking some plastic after the motor has stopped; a stepper motor can be reversed very quickly and pull the molten plastic back in.

It sounds simple enough to download Wade's design from thingiverse (here) and print it, but currently I am not able to print the largest part. The reason for this is warping. Any design which is larger than about 4 cm is warping so badly that the chance of it coming loose from the raft is about 80%. For larger objects the chance increases and it is double painful there, because when an object is large it usually takes a long time to print and therefore it may come loose after for instance two hours of printing...  Since my repstrap is different from the standard Mendel design (the horizontal rods from the x-axis are further apart), I am not able to use the design without modifying it a bit anyway. I decided to use the most complicated part of Wade's design and print that in ABS, use some 3mm polystyrene plastic sheet for the motor mount and use aluminium L-shaped profiles to attach it to the x-axis carriage.

The  extruder has several parts:
  • The extruder block  - Here I used a modified Wade's extruder block; I will upload it as a derivative to thingiverse later.

  • The extruder idler block  - For this I used Nophead's beefed up version from thingiverse.

  • A motor mount plate - Made of polystyrene having the same layout as Wade's extruder block for the motor but made a bit larger so that it can be fixed with bolts to the modified extruder block
  • A large gear - From Wade's design.
  • A small gear - For this I used Nophead's gear with set screw.

  • Standard 608 ball bearings
  • Aluminium profile and nuts and bolts to keep things together.
The bolt looks like this when inserted thought the extruder block:

The extruder is not ready yet, more information to follow later. I did test how well the hobbed extruder bolt grips the PLA filament (which is much harder than ABS) and it looks very good so far.

Saturday, August 14, 2010


After descriptions of the X and Y axis, the Z axis wants to be talked about too..

The Z-axis consists of:
  • 2 freestanding vertical metal rods
  • 2 separate stepper motors connected to threaded rods with a trapped nut in the x-axis ends
  • 2 different sliding mechanisms connected to the x-axis ends

The repstrap does not have the Mendel threaded rod frame, but is built on a wooden basis. The vertical rods for the Z-axis are free standing, they are not supported at the top. This is another part which was thought to be temporary, but which turned out to be good enough and may be a pemanent solution. A 2 cm thick piece of  hard wood is fixed to the basis and an 8 mm hole is drilled though the wood and halfway though the basis to give the rod a proper chance to stay put.

The 2 rods are connected to eachother with a thin strip of wood to make sure the distance between the rods is the same at the bottom and the top.

Two stepper motors were used for the z-axis when one would have been enough. It was easier to get an extra stepper motor than to get a set of pulleys and a timing belt to be able to use a single motor. So 2 motors were used (temporarily?); electrically the 2 motors are wired in series to the same stepper motor driver. This arrangement does not allow you to get the most power out of your motor (there is plenty of power available) but it makes sure the same current is running though the 2 motors and the the stepper motor driver will not get overloaded. The motors are connected to the threaded rods with a plastic tube which allows some movement to compensate for minor misallignment of the motor shaft and the threaded rod. Since steppermotors should not be loaded too much in the direction of the axle, a ball bearing is fixed in the wooden frame around the motors like this:

The motor is not resting on the basis, it is just very close, to avoid resonance.

The threaded rod is going though trapped nuts in the ends of the x-axis; the top of the threaded rod is not supported. The nuts are hot glued in some pieces of scrap plastic which are fixed with screws which have space to move horizontally. The hot glue is not very stong; this is on purpose. Already in several situations this has saved the printer form self destructing: when the z-axis is lowered too much the nuts will be separated from the plastic and will start to turn, thus stopping further damage.

The sliding mechanism around the vertical rods is close to Mendels, but the amount of ball bearings is less. One single ball bearing on the extreme left of the machine is the only bearing which is keeping the whole x assembly from moving to the right. The next 2 pictures show the arrangement on left (ball bearings in 135+90+135 configuration, allowing only up-down movement) and the right (180+180, allowing up-down and left-right movement).

The horizontal aluminium profile above  is fixed with 2 vertical M3 bolts but it can move in holes which are about 5 mm wide so that the M4 bolt a the top can be used to adjust exactly how tight the ball bearings are gripping the metal rod.

The metal rods of the x-axis are going through the white plastic which is used on both sides to make a very rigid construction and to prevent the x-axis to be twisted.

As usual one printed object at the end of a post. The object is for testing only: a 10 cm long single layer wall thickness cylinder with a diameter of 4 cm. The object shows no bad effects of the free standing z-axis. There are some artefacts visible in the object which should not be there: everytime the printer is starting a new layer, too much plastic is deposited causing some blobs. The comb module in skeinforge will move the position of this point causing it to look like some kind of staircase. I hope something can be done in software to prevent this in the future:

Thursday, August 12, 2010


The repstrap printer does not have a frame of M8 threaded rods like the Mendel. Instead, it is built with a wooden plank (80x40x1.8 cm) as a basis. Two pieces of wood (about 30x8 cm) are fixed with screws to the front and the back edge of the basis. These pieces have two 8 mm holes for two metal rods 18 cm apart which form the Y-axis. The Y-axis has a carriage with the platform on which the printed models are built and enables the printer to move the object forwards and backwards.

All metal rods in this printer are 50 cm long. This is the only length I could find in the local hardware store. Mendel is using rods about 30 cm long and cutting the rod would leave me with useless left over pieces of 20 cm. So I decided to use the full 50 cm for X, Y and Z axis.

The carriage with the build platform is made of aluminium profiles and is using skate ball bearings. This was actually the first piece of the printer I made, and since my order of smaller, Mendel sized ball bearings was delayed, I started with ball bearing for inline skates with an external diameter of 22mm and an internal diameter of 8 mm. The ball bearings are not fixed with M8 bolts as this would increase the size and weight of the construction considerably; only M4 bolts and M4 threaded rod was used with M4 nuts inside the ball bearings. The external diameter of the M4 nuts happened to be pretty close to 8mm. This may not be the strongest construction, but it is working very well in this particular setup. In general it would be better to use small ball bearings.
The ball bearings are fitted in the same configuration as on the Z and X-axis ( 90+135+135 degrees), but only one ball bearing was put under each metal rod to pull down the carriage with a spring (actually a rubber band, this should be changed still, as rubber bands have a limited life expectancy). The bearing on the underside are necessary as tall printed objects seemed to have the power to lift the Y-carriage a bit from the metal rods.

The top of the carriage is a 6mm acrylic sheet of 20x15 cm. The sheet is resting on springs around the M4 threaded rods with M4 bolts on top to allow for precise leveling of the platform (in the picture the springs on the left are missing).

The Y-axis is moving with a stepper motor, gears and a timing belt which were salvaged from an old scanner. Just like the scanner used for the X-axis (see previous post) there was nothing wrong with the scanner, it just got too old to be interfaced to a modern PC; this scanner still had a SCSI interface on it!
I have opened several modern scanners, but it seems modern scanners do not always contain steppers but may use DC motors with some kind of optical measuring system to provide acurate positioning. It may be possible to use those too, but that will require more special hardware and software.

The complete construction of stepper motor and gears (the black plastic) was sawed off from a larger piece of plastic in the scanner and bolted under the Y-axis carriage with 3 M3 bolts. The bottom of the carriage looks like this:

The belt is fixed to the wooden support pieces on the front and back of the printer and the motor is pulling itself (with the carriage) along the belt. This has the disadvantage that the system in moving the motor which is some extra weight, but the belt is short and easy to attach to the support pieces.

As with the X-axis, play between the gears between the timing belt and the stepper motor is minimal and does not seem to cause big quality issues.

The 6mm acrylic base plate is NOT used to print on directly as it is too difficult to remove. A separate 3mm acrylic plate is clamped on the base and printing is done on that.

Many prints can be made on the top plate, but it can be seen already that tiny cracks are appearing on the surface caused by large local temperature differences during prints.
Because of the arrangement with the acrylic the actual usable build surface is only 15x15 cm. I would be quite easy to rearrange the carriage and have a build surface of 25x20 cm. However, without a heated build surface, objects that large can not be printed yet anyway.

The largest object printed so far was the A-Mazing box. The box has a maze on the side and opening can only be done when you find the correct route! Infomation about this model at thingiverse.

Comments and questions are welcome, the next post will be about the Z-axis completing the overall hardware setup for the 3D repstrap printer.


Tuesday, August 10, 2010


The x-axis is the mechanical assembly which allows the printer to move the extruder from left to right and right to left. It is constructed using a recycled stepper motor, gears and timing belt from an old scanner and some materials from the local hardware store.

Two 8 mm metal rods spaced 6 cm apart are the basis. The Mendel printer has the rods closer together, so a Mendel carriage is not compatible with this repstrap. The change was done to make it easier to fix the makerbot extruder between the rods. The rods are fixed to wooden end plates using 2 mm aluminium strips.

Half round slots were made in the wood to make sure he rods stayed in the correct place.The half round slots were created by clamping two pieces of wood together and drilling an 8 mm hole though the middle.

A stepper motor complete with gears and a timing belt were salvaged from an old scanner. The scanner was working perfectly, but the latest drivers for it were for windows 98 and it did not work anymore with newer versions of windows!
Timing belts and the matching wheels for a particular diameter of a motor axle are hard to find, so using the complete set from the scanner was the best option. 

The motor is fixed to the side of the X-axis.

The off-center position is not a problem as the Z-axis ball bearings which prevent the x-axis from twisting are quite far apart and the construction is very rigid.
Because the stepper motor is on the side, there is no need to have lots of additional idler wheels as in the Mendel design. The Mendel design is more symmetric, maybe prettier, but it looks like is got a bit overcomplicated at this point.
I was afraid the gears (15 and 75 teeth) would introduce play and the printing results would suffer, but the play is small and the gear ratio gives a 5 times higher resolution and a 5 times higher torque.

The timing belt is going to the other side of the x-axis to a idler wheel made of a small ball bearing with large washers on both ends.

The belt is connected to the carriage for the extruder. The carriage is made of aluminium profiles (10x10 mm square and 10x20 mm L shaped).

The carriage is only running on top of the horizontal rods in true Mendel fashion with ball bearing allowing for some errors in how parallel the rods are (check out the video here). I did not fit any ball bearings under the metal rods. Initially this was because it was not ready and I was so eager to try to get my printer to do something that I just postponed it, but later I started to think that this has some advantages as I mentioned the post about the extruder.
  • when accidentally moving the Z-axis too low the extruder is not crashing into the build base, but lifts up automatically
  • when the filament gets stuck the extruder is lifted and does not (directly) damage the z-axis or the extruder.
The extruder carriage (and also the y-axis carriage, see future post) has 4 ball bearings running over a rod on one side. The bearings are placed at 90 degree angles. It would be ideal to have the angle 120 degrees to have optimal support in both sideways and vertical directions, but 90 degrees is easier to make. I studied the clever construction by Giles Bathgate (here), but came up with a slightly different implementation because of availability of aluminium profiles.  Holes are drilled at 45 degrees angles (from edge to edge) though the 10x10 mm square profile using a wooden jig. The wheels are not exactly opposing each other because the axles (m4 bolts) should not cross.

The x-axis is working fine with the carriage described. It may need to add extra ball bearing under the rods some day, but there are many other parts of the printer which should be upgraded first.

Another example of a printed object: the world famous whistle (thingiverse) which really works. The whistle is a digital design which was originally created last year by Zaggo and which within hours after posting on the website had physical implementations on different sides of the world. By now it has become a sort of standard print for new 3D printers.

Monday, August 9, 2010


A 3D printer consists of 2 main parts: an extruder which produces a very small flow of molten plastic, and a platform which is accurately moved in 3 directions: left-right (x), forward-backward (y) and up-down (z). In future blog entries I will try to address some specifics of each of these parts of the 3D printer, to start with the extruder in this post.

The extruder is the makerbot extruder MK4. This extruder comes as a kit with a motor pulling the filament in and pushing it into metal tube. Around the tube is a heating element which has a temperature sensor. The kit does not include the electronics. The included assembly is made of laser cut acrylic.
The extruder in general is excellent, it has been running for dozens of hours and it has never gotten stuck or given any serious problems. There are 2 things which are worth noticing though:
  • The motor is a dc motor with a fixed gear box. This works well when running at maximum speed. Since the motor is loaded very heavily it does not work very well at lower speeds. Sometimes there are situations in which I would like to slow down the flow of plastic, but I can not slow down the motor enough without risking it does not start properly. To prevent molten plastic to continue to flow though the extruder also after the motor has stopped, the software actually reverses the direction from the motor for a while. This works, but with a gearbox with a big ratio there will always be considerable play and reversing does not work very well. Newer generation of extruders in the reprap project are all designed to have a stepper motor to drive the filament, giving finer control of the flow of plastic which results in significantly better quality prints. I do plan to upgrade the extruder in the future with a stepper motor.

  • Filament is pulled in between a gear on the motor and an acrylic idler wheel. The idler wheel has to push really hard to make sure that the filament does not slip resulting in failure of a printing job. The idler wheel in the makerbot extruder has an M8 bolt and nut to prevent it from moving, but this is a weak part of the design. The bolt needs to be tightened really hard to make sure it does not slip, but since it is clamping acrylic sheets there is a serious risk that it is clamped too hard and the whole extruder is ruined. Tightening it too little causes the filament to slip so some solution had to be found. Since the 3D repstrap printer is built partly of aluminium profiles (10x10 mm square mostly), it was easy to add 2 bolts to the vertical profiles which prevent the M8 bolt from moving sideways. See picture, click to enlarge. An M3 bolt is going through a tapped hole in the vertical aluminium profile on the right; a second M3 bolt is present at the back of the extruder. Since fitting these additional bolts there has been no slippage and the extruder has worked fine.

I have read several reports of people ruining their extruders because of a software or control issues which caused the Z-axis to go too low. A strong motor in the Z-axis can ruin the extruder when it crashes into the base. In this printer the X-carriage is held down only by gravity. This has the advantage that in case the Z-axis is lowered too much, the extruder will be lifted of the rails and has a better chance of surviving. More in a future post about the x-axis.

Another example of an object printed recently on this printer is the dodecahedron as found on thingiverse:

Building a 3D printer.

First blog entry, ever...

After following the open source 3D printer project (reprap.org) for over 2 years, I finally decided to start building my own machine. The reprap project aims at building a printer that can print its own components. Since there are very few printers at this moment, a different printer is needed to make the initial parts of the reprap printer. This type of printer is called a repstrap printer (reprap+bootstrap). So without a complete plan, I ordered some parts in February 2010 to get started, and design my repstrap while building it.

From following the blogs I had learned that one part of the printer, the extruder, is particularly tricky to make without a very good set of tools ( proper drill press, lathe etc.) so for this I decided to order a kit from makerbot, the makerbot MK4 extruder. For the rest of the printer I wanted to use of the principles of the latest generation of reprap3D printers, the reprap Mendel printer, and use material which I already had or which I could buy locally. The Mendel printer, is designed with the limitations of home production in mind: you will never be able to drill to holes in exactly the right spot, 2 metal bars which should be parallel, will never be exactly parallel etc. But a clever design allows you to create a machine which is very precise, even though some of the parts have not been created as accurate as you would have liked to have them, more on this later.

After working on and off on the repstrap, I finally got the printer to work and the first test object came out of the printer in July 2010.
There was a plan to blog about the progress of the machine during the whole project, but there was too little time. I do plan to document now some parts of the design in this blog in the coming days showing some of my design decisions for later reference, but especially hoping it will be useful for others building a repstrap of their own.

So to show it really works the first picture of objects printed by this machine and the machine itself.