Sunday, November 27, 2011

RapMan 3D Printer Modifications

I purchased a RapMan 3D printer from Bits From Bytes because it is, as BFB claims, a very cost effective 3D printer.  However, once I had it, I couldn't resist redesigning it. I began by replacing the plastic corner pieces with heavy duty aluminum ones, and didn't stop until I redesigned nearly everything.
The end result bears only a passing resemblance to the original RapMan.

I started by replacing the corner pieces, which consisted of 3 or 4 pieces of polycarbonate and about a dozen M2 screws, with single solid blocks of  1 1/2" x 1 1/2" aluminum.  Then I remembered I had several feet of aluminum C channel, 1/2" outside and 1/4"  inside.  This I decided would make great channels to hold side panels to replace the original X braces.  The result was this foundation for my new machine:

Wherever possible, I tightened the tolerances in my replacement parts.  The bearings now rest in bored holes which hold them in rigid alignment, while the round stock from the original RapMan is tightly pressed into holes in the corner blocks.  The C channel is also pressed into slots, and locked with screws.  The entire foundation is very stiff and square. A 1/4" aluminum plate serves as the motor mount.

Here is a side by side comparison of my aluminum corner assembly vs. the plastic RapMan corner.
The aluminum assembly is comprised of two pieces held together by a milled slot and a single 1/4-28 bolt, while the original consists of 5 pieces of polycarbonate, 12 screws with nuts, and a short piece of 8mm rod.

What must be remembered here is that my design is not economical to manufacture and sell, and that the RapMan design does work OK.  The point I'd like to make is that it is possible to take the affordable RapMan and turn it into the equivalent of a much more expensive machine.

Below: I replaced the idler pulleys made from a ball bearing-and-washer sandwich with a lathe turned pulley riding on an 8mm bearing.
The jack-screws are made from common hardware store threaded rods and nuts, with laser cut polycarbonate toothed pulleys sandwiched between washers. Everything is a sloppy fit, but I'm not complaining, for using common hardware enabled BFB to keep the cost low.  I found the holes in the toothed pulleys to be about .005" over-sized, or about the thickness of a Nine Lives steel cat food can.

This solved two problems: The sloppy fit, and the fact that friction by squeezing the pulley between two nuts was the only thing that locked the pulley to the shaft, for there is no key.  The cat food can made the pulley a tight fit on the shaft, enough to turn the shaft even if the nuts are loose.  At right is a picture of the jack-screw and pulleys, along with a strip from the can.

The original assembly is the one the left.  It has 2 small washers for spacing, then two large washers to form the sides of the pulley.  While the symmetrical appearance looks good, the resulting pulley is too wide.  I settled on the right hand version, which is asymmetrical in appearance, but the correct width for the belt.

I felt the motor pulley deserved special consideration, and made a pair of lathe turned discs to replace the washers. The assembly is held together with the original screws, and clamps to the shaft as before. 

The RapMan merely pinches the ends of the Y axis belts, which can shift over time.  I made clamps with slots that match the teeth on the belts for a more positive grip.

Then I decided that the nut assemblies which support the table were too flimsy for my requirements, so they were the next polycarbonate pieces to go.  Here is a picture of my aluminum replacements.
I retained the basic design, right down to the hex shaped hole for the nut. However, I changed from the two double guides like the original shown here, to four single guides, one at each corner. The double guides help prevent binding on the more flexible original design, but my rigid aluminum design does not require them. The slots that form the inverted "V" in the bottom of my guide are clearance slots for aluminum "C" channel which now frames the table, stiffening it considerably.

The_completed table and "Z" axis assembly is sturdy and powerful.  Here it is lifting 4 gallons of antifreeze, which is nearly 30 pounds.  Mechanically, it could handle more, but the stepper motor stalled at higher weights.
The completed frame is sturdy, and initially seemed rigid enough.  However, when I tried printing at higher speeds, the frame shook a bit.  Also, I had a misalignment of about .005 inches in the frame from one side to the other.  Adding diagonal braces solved both problems.

Also visible in the first picture is another improvement, a case for the electronics.  As another cost-shaving idea, the electronics come with only a top panel, while the sides and back are exposed. This does not present a hazard to the operator, but it does make the electronics vulnerable.  In addition, there is no strain relief for the wires.

Here is a rear view.  The wires go through the holes in the board, and that is it. 
I fabricated a case from the metal case of an old VCR.  VCR's are a great source of a variety of parts. I wished to keep the original RapMan electronics assembly as original as possible, but I did make 2 modifications. First I countersunk the screws in the top cover for a nicer appearance, and then I milled a relief around the perimeter so the cover would fit flush with the surface of my case. 

The electronics slide in from the side, and are retained by polycarbonate pieces which lock everything tightly in place.   The cable protectors are twisted into threaded holes in the polycarbonate and will not pull loose.

The picture below shows how the side pieces support the electronics cover and circuit board.

Below, a close up of the assembled controls. The side panels are .10" steel, and the whole thing feels very substantial.  There is plenty of room for air circulation underneath.
The screw terminal "D" connector will be replaced with a conventional one, further improving the appearance.
 With all of my experimenting, I have now fried two stepper motor driver chips. I repaired the board, but some of my repairs are a little unusual. For more on that, click here

More to come!  After going this far, I've decided to redesign the carriage the print head rides on.  There will be very little polycarbonate left, and the the machine will gain a few more pounds in weight.  It will likely tip the scales at around 60 pounds when complete.
The print head development took a while, for I developed an entirely new design.

The Rap Man printer head works OK, but is excessively large in size, and is complicated, with the body made from 10 pieces of laser cut Plexiglas. The head is over-sized because the motor , feed screw, gears and bearings are mounted at an angle while the raw material being fed to the printer is vertical.  BFB did this to move the drive gear and bearings out of the way of the path of the raw material.

In  the more expensive BFB3000, they made the head smaller by mounting the drive gear vertical and bending the raw material.  This also works OK, but is a nuisance to reload, as one has to snake the material through the curved path.

My idea was to have the best of both worlds by positioning everything vertical.  This would make the head small, and reloading simple.  But, of course that meant the bearings would be in the way.  The solution here was to move the bearings out of the way.  But how to do it?  Instead of having the drive screw go through the center of the bearings, as is normally done, I put the drive screw on the outside of the bearings. Now, instead of the center of the bearing rotating while the outside is stationary, the outside rotates, and the inside is stationary.

This required two bearings at each end.  The feed screw now rides in the valley between the two bearings, leaving the top of the feed screw unobstructed, creating a straight path for the raw material.  

Below is a picture of my feed mechanism. 
Very compact, and a much more attractive design than the original RapMan head.

Now that the material feed design was complete, it was time to develop a mounting system. Primary requirements were ease of removal, and a versatile design which would permit me to experiment with a variety of print head designs, including the ability to swap the 3D print heads with a mechanical or laser engraver head, or perhaps even a pottery extruding head.
I settled on a dovetail style mount, which eliminated the need for threaded holes in the carriage, and bolts to attach the heads.  A single screw on each print head expands the dovetails, clamping them in position quickly and securely.
The carriage is machined from a solid block of 1 inch thick aluminum.   This allowed me to make a very low profile carriage whose top surface is just 2 mm above the linear bearings.  This helps to reduce the overall height of the machine. Combined with the low profile print heads, my machine is about 4 inches lower than the RapMan.  My carriage is also smaller in both the X and Y dimensions, resulting in more x and Y travel than the RapMan carriage.
One design problem I struggled with was how to get power to the extruders in a way that looked tidy and also gave me the versatility to experiment with different styles of heads.  I rummaged through my collection of old Apple hardware and found a 50 conductor ribbon cable from a hard drive.  It was an ideal length, and I was able to cut the connectors off the hard drive and motherboard.

I cut one of the connectors up into several small ones, and made individual plugs for each motor and extruder.  This way I can easily remove them separately, or swap their positions for troubleshooting.  
Unlike  the RapMan, which required removing several screws and wires, I can remove each head in under a minute by loosening only a single screw.
Since ribbon cable has light gauge wires intended for carrying signals, and not large amounts of power, I used 3 wires in parallel for each extruder and stepper motor.  This arrangement works well, giving me the flexibility of a ribbon cable, and the current carrying capacity of a thicker, and stiffer, wire. Out of the 50 conductors available in the ribbon, I used 46.

Finally, I wanted a convenient means to hold several spools of material without taking up valuable table space.  I mounted the spools on top of the machine, where they are readily accessible.  A picture of the completed machine is below. The whole thing fits nicely on a small 2 foot x 2 foot  table.

  I ran the "Duck" test program provided by BFB, and made some very nice ducks in a variety of colors.  The surface finish is quite good, and I'm impressed with what can be done without using support structures.  These ducks are ABS, but I've also made them in PLA.
 I modified the duck file to make two color ducks as a test of the second extruder head:
After making a whole flock of ducks and other things, I realized that there is still room for improvement.  One thing I discovered is that PLA stays soft for a longer time than ABS.  This made it difficult to make small parts because the material was too rubbery and unstable when the extruder returned for a second pass.   I solved this by adding a second fan on the rear of my carriage.  Below are pictures of the brackets, and how they mount on the carriage.

 This solved one problem , but created another.  The RapMan extruders always struggled to reach maximum temperature, and now with the increased airflow, and the addition of a heat sink on the inlet side, they really struggled. It didn't help that BFB did not insulate them well, or used 3 stainless steel standoffs to mount them.  I turned the center portion of the standoffs down in my lathe to about 2.6mm to reduce the heat transfer.  As the cross sectional area of the standoff is now less than half the original, the heat transfer should also be reduced.
 I added additional high temperature RTV to the exposed back of the extruder, and covered the slimmed down standoffs. It did not make a huge improvement, but it was enough to allow the heads to maintain 260 degrees.

I ran these extruders for many hours, and they generally worked well, but would occasionally lose their grip on the filament.  Increasing the pressure on the filament against the screw helped, but then the motor would occasionally stall.  I realized that the blunt machine screw style threads took considerable pressure to bite into the filament, so I designed a much sharper tooth profile.  Below is a drawing representing the original profile and my design.

I made these screws from water hardening drill rod, which I hardened after machining.    The new screws are working well so far, and require much less pressure to maintain a grip on the filament.  It remains to be seen whether the hardened drill rod maintains a sharp edge.

Another improvement was made to the material feed arrangement.  Feeding from overhead quickly proved to be less than ideal, so I added a pair of turning pulleys to bring the material in from behind.  The top pulley removes quickly without tools.  With room for 6 materials overhead, material changing is far easier than most other machines I've seen.


With the machine now running well, I decided to tackle the issue of ABS shrinkage and build a heated bed.  Since my existing bed was made from 1/8 inch aluminum, I was off to a good start.  I experimented with a variety of heaters and decided that 300 watts was about right.  The heating elements were removed from food warming trays which had a nice length of Nichrome wire insulated with fiberglass.  I would not recommend going beyond 300 watts, for if your temperature controller sticks on, the bed would get extremely hot.
I had a old British made CAL 5000 temperature controller in my junk box, so I built an enclosure and used that.  It uses a type J thermocouple.  The heated bed was a huge improvement when printing in ABS.  It also helps when using PLA as a raft material, for it keeps the PLA slightly soft and sticky.  
     Next problem.  I was not satisfied with Kapton as a bed surface, and I heard of someone using a stone surface, so I decided to give it a try. I tried a marble tile and that has worked well so far.  I'm using the rough back side of the tile as my surface. As long as the tile is warm, the PLA raft sticks well. When it cools, the PLA pops off by itself.

Finally, here is my copy of Emmett's Heart Gears, a popular download from the Thingiverse site: