Saturday, May 18, 2013

My first customized part: A new Filament Holder

One of the most annoying things once I had everything aligned was controlling the filament feeding. I was doing it by hand but I had to constantly be checking on a print to make sure that it was feeding properly and when some prints can take 3 hours or more, it quickly gets pretty annoying to deal with. I came across a design by Garfield on thingiverse for a roll holder for the mendel max 2 that bolts to the top of the machine and accepts a 10mm tube.


Since I am in the US, metric sizes can be hard to come by but I did find a 5/16" solid rod that would fill the job quite nicely. I redesigned the ends and it worked perfectly after it was printed.

You may notice that the print quality of the left holder is horrendous and I'll talk about that more in a future post that details print errors I ran into and their solutions.  Stay tuned!


The Design:
I designed these in google sketchup and started with the .stl files of the original spool holders.
1) Get google sketchup if you don't already have it! It is free!
2) Download the stl importer and put it into your plugins folder. (Mac users click on finder and type apple+shift+g and then type: /Library/Application Support/Google SketchUp 8/SketchUp/plugins)
3) Import the stl file. You will probably notice that the part is way larger than it should be. I am not sure why this is because I have mm set as my import stl unit, but I often have to scale the part down by ~0.03948 or so depending on the part. I suspect that this is the 1cm/1ft conversion factor but I am not positive. To calculate this conversion yourself just measure a distance in sketchup and then measure the same part in an stl viewer like netfabb basic.
In netfabb you click on Extras->New Measuring and then select the point-point option on the right pallette. 
In sketchup you click on the measuring tape and measure the same distance. Note for this model I have already scaled it which is why both distances are the same. 

4) To scale the objects, select them all using the cursor and press 's' and then type in the scaling factor.
5) Now you have your model imported and the correct size. The trouble is that you probably have way too many extraneous lines going all across the model. You need a method for stripping away all the excess lines and a different plugin called cleanup will do this. Dont forget to install the ttlib plugin that they recommend as well. I apologize if you need to make an account to download those files, but it feels a little wrong if I hosted them separately.
6) Once you have simplified the model you can start adjusting it! For the left spool holder model I show above, I was able to minimally adjust it by simply decreasing the radius of the hole. I first created an offset of the semicircle with the 'offset' button and then used the 'push/pull' feature to fill in the whole path.
7) After making all of your adjustments, you now need to export it as an stl file. I used this plugin but I am sure there are others.
8) Double check your stl file by opening it up with netfabb. I am a huge fan of repairing all stl files using cloud.netfabb.com. There is really nothing worse than printing out something with errors in it so it is really worth it to triple check that your stl files are exactly what you think they are. 
9) It is also worth checking in slic3r or cura that each layer is what you expect too. In the right spoolholder version, it took several iterations for sketchup to properly export the cone that the rod rests on. At first it just eliminated that section of the design which would have been annoying to catch earlier. 

For those interested here are the left and right google sketchup models.




Setting up Rumba electronics and first prints


Once I had the Mendel Max together the next step was to load the marlin firmware. First download the arduino software so that you can edit the marlin.pde file.

I then pulled the marlin rumba configuration settings from Maker Tool Works. I just went through and made the necessary changes to the configuration.h file in the marlin to make it match up with the configuration.h settings from the Markers Tool Works file. I initially tried to just copy the whole file over but that does not work because apparently they used an older version of marlin.


I then just compiled and uploaded the firmware onto the rumba electronics. This is done by plugging in the rumba via USB, pressing the reset button on the rumba board, and then pressing the upload button in the arduino software.  

Once uploaded, I opened up pronterface or Cura and checked to see how the steps/mm match reality.
Cura jogging interface.


Pronterface movement interface. 

I told the X and Y axis to travel 100 mm and then measured what they actually traveled. It was almost spot on so then I tried the Z axis which was also correct. To make small corrections to the calibration I did the following:

Find this line in the configuration.h:
#define DEFAULT_AXIS_STEPS_PER_UNIT {80,80,1600,470} // default steps per unit for ultimaker
The numbers correspond to X, Y, Z, Extruder respectively. 

To calibrate X and Y: 80/(distance it went) * (distance it should have gone)

Replace 80 with that new number you calculate to 2 decimal places and then save the file and re-upload it to the rumba. Repeat the process several times until the movements reliably go the distance you are telling it to go +/- 0.05 mm. For me, I ended up with a calibration of:
#define DEFAULT_AXIS_STEPS_PER_UNIT {80.5,79.83,1600,470} 
so it is quite close to the original values that I used. It took about 5x before I was satisfied with the movement. 

Next, I wanted to check the extruder steps/mm setting. I measured 30 mm above the extruder and told the extruder to move 30 mm. It was spot on so I didn't bother changing that value. 

I then decided to try my luck by printing off a Z end stop so that I didn't have to worry about crashing my extruder anymore. I made sure to turn off any axis homing to prevent crashes. You can find these codes in the Slic3r or Cura software. Make sure these are not in the start gcode!

G28 X0 Y0  ;move X/Y to min endstops
G28 Z0     ;move Z to min endstops

I did not previously have experience operating a printer without endstops and it definitely felt like I was living on the wild side because any false movements led to the extruder crashing into the bed or the X or Y motors going to their extents. It is important to have your hand ready to unplug it if necessary!

Due to the custom size of the Rumba endstops, there were not any available so I had to adapt Ohmeye's design for RAMPS to work with the Rumba electronics. Here are some X, Y, and Z endstops I designed that work perfectly for the Mendel Max 2.

To start I decided to use PLA as my first print material because it can be extruded at a lower temperature and does not require a heated bed. I had good luck using blue painter's tape that was just stuck on the glass top. In setting the Z height, it is important that the gap between your nozzle and the bed is ~2 pieces of paper thick. This can be a huge pain to set up initially but is easy once you have the Z endstop mounted.

Bed Leveling is another critical aspect to printing well. You can start by using a spirit level but that requires first leveling the printer itself. For precise leveling first make sure that the nozzle is at the front left and there is ~2 pieces of paper thickness between it and the bed. Next start marching across the bed in the +X axis and adjusting the two leveling screws on the front as appropriate. When looking from above, counter clockwise raises the bed and clockwise lowers it. Once you have the front edge leveled, you can start going in the +Y direction and using the back leveling screw. Do this a couple times until you feel the same amount of resistance on the paper as you drag it under the nozzle.
Start by getting the height right on the front left. The paper should have a little bit of friction in between the nozzle and the bed.

Next, go to the right front of the bed and level using the screw underneath or by twisting the Z lead screws. It is a good idea to independently level the X axis with the Z leadscrews and then leave that and just adjust the bed after.


Your first print:
Before printing out the endstop, it is better to first make sure a calibration print comes out correctly. I like this calibration set by coasterman.
1) Print out the box at 0.3 infill and check the dimensions. Is it 20x20x10mm3? If not, apply correction factors to the marlin firmware. i.e. if it is 20.1 mm in the X: 20.1/20*steps/mm=new steps/mm.
2) At this point you are probably ready to print out the Z endstop. There is a good chance that this print won't be ideal but at this point you are mostly concerned with getting a piece of plastic that will hold an endstop long enough for you to really hone your settings.
3) After you have a Z endstop in place feel free to play around with the other things in the calibration set. I particularly like the precision block because I find that models often tend to undershoot holes so this provides a way to try and fix that. The Bridge test is a more advanced one but setting it right is critical for more advanced objects. Once you have those two set up, you can try the hollow test cube and see how flat the top comes out. 

Friday, May 17, 2013

Picking a 3D Printer: Mendel Max 2


Welcome to my rep rap blog! I hope to document my adventures building my second 3d printer. I have previously built a self sourced a prusa mendel 1.0. Although I eventually was able to get it printing smoothly, it took a ~6 months to iron out the bugs. Here is it printing some parts on a good day:





Here are some of the chief complaints that I was hoping to avoid the second time around:

1) The triangular design of the Prusa Mendel enables it to be very strong but unfortunately the direction that the X axis moves is exactly the direction that the printer is weakest. This prohibits faster print speeds because the frame itself has a tendency to flex. 
2) I would like the possibility of printing with multiple extruders which further necessitates having a very stiff frame in the X axis as well as having enough room to fit both extruders.
3) My previous extruder was a maker gear hot end and a wades geared extruder. Although it was great when it worked, there were many times that the hobbled bolt would lose its grip or the hot end would be otherwise unreliable. The ceramic heat core is also very fragile so I am looking toward a more robust resistor or heater cartridge solution. I am looking for an extruder solution that is bulletproof in its reliability.
4) The Prusa frame is not very open and can be finicky to get everything squared up correctly. The threaded design has a tendency to come undone from vibrations loosening the nuts holding everything together.
5) I am not a huge fan of hanging the X axis from the Z lead screws because they have a tendency to  pull out from the printed Z couplers. This leads to the whole X axis falling down.
6) Leveling the print bed was always a huge pain on the Prusa version 1. The design called for a plate with the lm8uu bearings and then a second plate that you actually print on that is attached to the first plate through 4 screws and springs. Although the springs are a nice buffer in case you crash your hot end, they also prohibit precise positioning of the height of the print bed so this inevitably leads to more crashes. Newer designs tend to have a fixed print bed that you level one and forget about so that is a definite must for my new printer. Also, leveling with 4 screws makes it over constrained- 3 is much better.
7) I am not a fan of 8mm rods because they flex and after using lm8uu linear bearings with the rods for about a year, the bearing actually cut trenches in the rods. Going forward I want to stick with 10mm rods and possibly migrate to a motion system where i won't have to worry about damage to these expensive rods over time. Stainless steel should be a good choice here but buying precision ground rods from Mitsumi goes for upwards of $100 for all the axis.
8) I have a preference for 24 V over 12 V so that the bed will heat up faster and the motors will have more torque. My previous heated bed was made from 120 V silicone heaters attached to a 1mm aluminum plate. This heated up quite fast and the aluminum plate helped distribute the heat evenly.

With those properties in mind I had several printers that I was considering:
1) Prusa i3
     Pros: The aluminum plate makes the x axis super rigid
     Cons: Aluminum plate must be custom made unless you buy it from Europe
2) Lulzbot TK-0 (now TAZ)
     Pros: Huge 300x300mm2 print bed, Aluminum extrusions
3) Ord Bot Hadron
     Pros- Kit for $500 or so, linear rails make for a smooth ride
     Cons- Very little support for it in the reprap community. Some people are building in buildlog.net
4) Mendel Max 1.5
     Pros- Very rigid
     Cons- Same triangular frame as prusa mendel
5) Mendel Max 2.0
     Pros- Orthogonal frame design, All aluminum
     Cons- Expensive

Coincidentally the Midwestern Reprap Festival was going to happen around the time that I was making this decision so I was able to see these printers in action. Here are my thoughts on the different types:

1) Prusa i3- Josef Prusa was there demoing his machine and although his prints were silky smooth, the printer sure vibrated like crazy while it was working! There were some people with SGraber's laser cut version of this printer that have additional stabilization pieces which seemed to help with the vibrations but I was really looking for something that was more professional in appearance. In addition, the fixed size of the x axis meant that it would be more challenging to fit a second extruder without drafting a wider frame. Finding a water jet cutter to make an aluminum version of this frame is not cheap and would run over $100 for sure.

2) Lulzbot TK-0 is a very nice printer but since it has a print bed of 300x300 mm2, the thing was just enormous and would easily take up a whole desk. They were however printing out pretty large objects with no trouble at all. Their layer heights were small and the parts looked amazing.

3) The ord bot is made out of makerslide but the connections between the rails is not great so the printer was not very sturdy. Also, the main guy selling the printer was more interested in selling bulk quantities of his motors so he gave off the feeling that the printer was just a side deal and not his real concern. Even though this kit was $400 for the mechanical stage, there wasn't really a community behind it so it seems like you would be on your own.

I was pleasantly surprised by the performance of the Roostock style printers but they are quite tall! When on a table it is about eye level and I am 6'3"! Given that I live in an apartment this probably is a little too big for my purposes but as Richrap has shown, you can print some pretty impressively large vases with it:

At the end of the day I was most struck by the Mendel Max 2.0. Maxbot's, the designer, was extremely knowledgable about all aspects of the design and it was obvious that he really did his homework in selecting each piece of the printer. The 90 degree frame made it very open and easy to get in there and the aluminum extrusions made the whole printer rock solid and it didn't flex at all. The printer uses a combination of 10 mm linear rails, and these troughs with plastic cars that slide in them. 

It took about a month to arrive but here is the hardware kit unassembled:
I was going to document each stage of the build, but since it went together pretty smoothly and Maker Tool Works already made a great instruction manual with plenty of pictures it seemed a bit excessive. I had previously purchased Rumba electronics from the Indiegogo campaign so I was sort of stuck getting the Mendel Max 2 hardware kit and then I purchased the motors from ultimachine, and a 24 V power supply from amazon.com. Here it is mostly assembled:
Despite it going together well here are a few stumbling blocks I had:
  • For the Y axis, they provide additional sliding inserts that have a tighter tolerance than the ones they provide. You attach the inserts by pushing them in and they secure with a little knob on the back of the insert that goes into a hole in the pillow block. The problem is that this knob is off centered so when you trim the inserts to size, you have to trim them from the correct side or else they won't fit into the pillow block! My recommendation is to test fit the uncut insert first and mark which side needs to be cut.
  • It took a bit of patience getting the Z copper sliders to work nicely. The instruction manual has a nice part about how you need to work the copper bearings on the rod to increase their diameter. This is critical because although just 1 slider will go fine, it gets harder as you couple more together. For example, it has you tie two of them together on a side and I had to really play with the bearings to get them to slide nicely. It was the same thing once the whole X axis was assembled and you were trying to coordinate 4 bearings moving. Once set up however, they slid like a dream with almost no friction so just be patient! 
  • Setting up the X troughs that the cars slide in to move the axis is a little confusing. It appears that they clarified the manual since I assembled it, but it is important not to modify the cars with the bowed plastic because they are indeed supposed to be bowed. I thought it was a defect and I bent them to be less bowed and consequentially offer less resistance, whoops! 

Using custom RUMBA electronics had several other issues:
  • The electronics mount they provide is only for RAMPS so Rumba folks are out of luck. The universal electronics mount they provide with their STL files does not fit the RUMBA either so I am in the works of making a new one. I plan to take the enclosure by rznag and add some tabs so it can be bolted to the frame. 
  • Keep in mind that the max cable length for the LCD screen is 10cm. I placed the electronics in the back of the printer, which is what is recommended but the LCD cable length will not reach to the front where I'd like to put the LCD. 
Here is a video of it moving!