More likely back to the mixtape mendel project. I have had, in my head, the ideas I want to implement into version 0.2. Many of these things are fundamentally sound. I also wanted to test out some of the hardware and software I would be using on testing mixtape mendel. Partially to eliminate variables. Most of this project is focused on the mechanical design of a highly optimized printer for PLA bushings and low hardware count. One of my first steps involved some minor maintenance on Isaac. I replaced the X carriage with Buback’s version, here. It works well. I also went ahead and replaced the dying 30mm fan on the original carriage with a new 400mm one, which is positioned mush nicer. the original has a corner cut off and was positioned to blow air from one side of the carriage to the other, which ended up going downward. I also added heatsinks to my Pololu drivers. I have been running them since August and been breadboarding with them since last May. I also added a 90mm fan, which is awesome. Together, i can really crank up the extruder current, which is nice.
I have also been busy helping my friend, Brett, get isaac’s child working. Last month I printed him a set of Prusa parts. He did a great job on the mechanical construction, and i have been helping him get the electronics running. I am really surprised the PLA bushings work so well. I dunno if it is the PLA bushings, or the general acoustics of the parts, but it seems to transfer a lot of vibration and noise through the frame. He is using brass bushings on the Z axis and PLA bushings on the X and Y axes. I am really impressed at how well they work and constrain the axes, so I feel much more confident in using these on Mixtape. I have shown Brett the improvements I made, and his favorite is the glueless bushings on the X ends for the Z axis.
Here is what I have learned from version 0.1.
– Tapping plastic works well.
– I need to make my parts a tad bit thicker.
– In OpenSCAD, I should never use scaling factors to describe walls, but rather hard millimeter values.
– Having a configuration file would be awesome.
– I should flip the smooth and threaded rods on the z axis.
– The cross bar isn’t necessary.
What I am going to do in version 0.2
– See above.
– Hollow the vertexes (maybe)
– Try vertical x bars like in the eMakerShop Huxley.
– Try longer, 3 contact point, bushings.
– Try to make bowden extrusion work well.
– Try these out.
– Actually test it.
I figure I will just post up some pics. It turns out I need to rework the top piece and bottom piece, because the motor hits the side rods on the triangle. This effectively makes my x travel around 2cm. 6cm if I pretend the print bed is non existent. I am going to flip the smooth and threaded rod like Christopher Olah did so I can eliminate the cross bar. I have known this since last week. Why I am posting this now, I don’t know. I think my Version 0.2 notes are about done as well. I will be posting those soon. Like 15 minutes soon.
Oh and my Arduino Mega got here. I ordered 10lbs of filament and other goodies from Ultimachine, which should be here soon. My smooth rod ship date got pushed back from the 1st to the 15th, which sucks, but I figure I can get version 0.2 done without it. I am still debating on releasing version 0.1. As a whole it does not work, but, parts do. If anything I might release parts on request.
I have been preparing for my Science Olympiad competition the past week.
I already have a road map for next week, and have started thinking about version 0.2 for mixtape Mendel. All I can say is expect alot next week. My smooth rod will be here possibly, and I will be printing and designing version 0.2. That means I might have a functioning second printer pretty soon, and then I can start playing with print quality.
Most importantly, I have a few fails I have been documenting in version 0.1 that I will be posting as soon as I can get pictures up. Expect that stuff Sunday. 🙂
Yes, lots to blog about about.
First, the top piece. I got most of my ideas from hacklab.to for this part. Most of it isn’t interesting, except for an accident in the design I made. I had originally intended to shift the motors outward to allow for room for the nuts. As I was finalizing the design for printing I realized I had neglected to do so. I decided to quickly add another variable called “beforetap”. What I then did was insert this into the radius parameter. So the script would behave as if the hole were 8mm, but it would print a 7mm hole i would tap for M8 during post processing. So the mistake turned experiment was tapping the holes. This could potentially cut the nut and washer count in half. By tapping the hole we are not loosing structural integrity, nor adjustably. As long as the opposing side is adjustable, by using nuts, we are fine. Also considering the “overlap” nature of the design in many of these part, the rods are in contact with each other, which helps aid in structure.
And it is starting to look like a printer :). So far I have designed all the part necessary without building or using advanced CAD software to visualize. I used the other openscad scripts to keep track of parameters. At this point the only thing left to design is the X carriage and maybe extruder and MixTape Mendel version 0.1 will be complete.
Below you can see on the right top piece, there are no nuts holding it together. It is entirely tapped.
Some pictures comparing Mixtape to Isaac.
I think this one has an optical illusion because the Y travel seems identical in person but isn’t so in this picture. It is probably because Mixtape is much more compact.
These pieces were very simple to design. I had three main ideas I wanted to implement. The first was glue free bushing holders. My idea was to wrap the plastic around the bushings, so they get pressure fit within the cavity. You can still pop the smooth rod off, but now they align better and put more pressure on the bar.
The second idea I had was to flip the belt over. Doing so increases the z travel one could attain. The only issue is that the smallest possible nozzle from the top of the carriage will have to be 37 mm long. This is fine because it will still easily work with the makergear nozzles, which are generally considered the best in the United States. For the nozzle, I am doing something else though, which will be documented later on. It will be very close on architecture to the simple PTFE and brass nozzles done a couple of years ago.
My third improvement was bringing the bushing and nut closer together. This is part of the “optimization” I am making. bringing these two closer is a more efficient use of space and material.
Some other features I added were adjustable idler positioning, hybrid lead nuts and pressure fit x smooth rod. The adjustable idler positioning means you can get the belt perfectly flat, which is nice. My hybrid lead nut is as follows; a regular m8 nut takes most of the load, but for half of the bit underneath, it was printed as a 7mm hole which i tap to M8. The plastic puts enough pressure on the lead rod that we no longer need anti backlash systems. I think this is really going to be needed to get tested before I decide if it is better than a full anti backlash system like the Prusa Mendel. Last thing I decided to do was use pressure fit x rods. The Prusa Mendel uses an interesting system to ensure parallelization between the bars. What I am going to be doing is drilling these holes out to 5/16′ and pressure fitting the X smooth bars in. Again, this is something that needs to be tested.
I am calling these version 1. I already have a second potential motor mounting system which might be better. I might also use prusajr’s x rod clamp system if I decide pressure fit is not parallel enough. I already have a few ideas for version 2, so these are definitely not final yet.
Ok, so sometime last week I was messing around and realized that I could totally eliminate a bearing from the printer. The realization was that the pulley inner circumference was a tad bit over 8mm, so a M8 rod could go right between the belt if everything lined up just right. Well it took up until last night to fully get this idea to work.
There were three main issues in designing this. The first was that I had made a minor mistake on the hole spacing, which was easily corrected. the second mistake or issue was creating the supports. I still honestly do not know If I totally like the amount of support on this mount, but it is exactly how I imagined it. I imagined it being simple and using little material. It flexes a bit, but I made the “beefyness” parametric, so I could easily increase it later.
I am going to start a new paragraph on the third issue, because it was BIG. It took me a few nights to figure out why my mount was not fully working. It ultimately came down to my OpenSCAD code being right, I just did not fully understand how the vertexes were parameterized. To imagine, on the vertex, there are 4 rods entering it. when you set the angle to 60 degrees, only 3 of the rods can be contacting. the fourth needs to be angled slightly, so it is only in contact with the horizontal rod. The horizontal rods and the two rods going in the Z direction create a 45 degree angle. If you do not have the correct vertex for each corner, this will be confusing, because the two X direction bars make a very slight angle.
Long story short, each vertex has a specific corner. And I am just about done with my Y axis. Some pictures for the end…
This is going to be short and sweet. I have decided to use Whosawhatasis’ bar clamps in addition to his vertexes. They are here. I have done this for two reasons. The major one is the bars perpendicular have zero gap between them, as illustrated here.
There is a gap between the two bars on the Prusa Bar clamp, which creates a flex point which makes them break. I think this is for better spacing on the belts. Secondly, segueing into the next point, it is a very efficient use of material when they are fully touching. Overall I like this bar clamp better, and I believe Josef Prusa used it on his original Mendel remix with Igus bushings.
Oh, did I mention I am going to be over engineering and thinking every part of this printer? 😛
I have to say what really got me started on this project was this. The design is nice, parametric and simple. I had been messing with hollowed out vertexes and trying to make them lighter. Whosawhatasis’ vertex uses as little material as possible and is well executed. I posted up a updated scad file here. There are some comparison photos on thingiverse, but to try and interest my readers I will repost them below (more words after the pictures).
These vertexes are not compatible with the Y motor mounts used on the Mendel. I have given it 3 tries but my math was off. I am going to be reevaluating my OpenSCAD script tonight and will hopefully figure out how to mount it correctly. After I verify it works it will be up on thingiverse. I am going to go back a relook at my hacked Prusa Zmotor mount that took ideas from hacklab.to’s remix. I got really excited over the weekend and was hacking the files so I could get my printer done quickly. After Jmil kindly directed me to the Google Science Fair, I have decided I will try and flesh out the development and do something useful for the community, rather than hack together my second RepRap.
The Mendel I am going to be building is aiming to be simpler, faster and more precise. To achieve the faster and more precise part I am going to be pushing the limit of my firmware. I have forked off the Tonokip-Firmware I have been maintaining the past few months and am going to be optimizing it for use with gcdump and step gcode. Gcdump has been my sender of choice lately, because it is simple and works well. Coincidentally it doesn’t resend, do checksumming or line numbering. This means that the sender and arduino do a lot fewer communications and their communications are smaller in size.
Some of the first changes I made to the racetrack firmware was to remove all the checksum, resend, echo and failsafe stuff. Next I began to play with skeinforge gcode step generation and figure out exactly what it is. It turns out the step code generated is best described as the steps from origin. So if one line of code was “G1 X6000” and the next was “G1 X3000”, we would be going 3000 steps in the negative X direction. This makes the math really easy. Racetrack is up on github here. I am going to be working on it slowly. I just wanted to get a head start on the software side as I work on construction and design. This way I can think about it and improve it when I get bored with OpenSCAD. 😛