The idea for the MegaMendel came in the early 2010. The project was spurred by an idea of printing parts for a small wind turbine.
It is a scaled up Mendel with a build envelope of 766mm x 453mm x 497mm. The X and Y dimensions are the freedom on the respective axes. The 497mm on the Z-axis is actually the height of the tallest object printed on it so far, please refer to the image below (the ruler is 600mm).
It is possible to tweak it to print at least 550mm in height, but that will have to wait as I don't see the point in doing that right now :-)
The machine's dimensions are 1160mm x 880mm x 830mm and my guess is that it has a weight of some 30 to 35 kg's.
The whole project made me rethink parts of the construction. I had two reasons for that:
1 - I made the "plastic" parts in MDF - So there was a need for a design that would compensate for the weaknesses in MDF.
2 - I don't agree with some of the solutions in the standard Mendel, so I changed lots of details to what I consider simpler and better solutions.
So the MegaMendel is more of a RepStrap.
This page is meant to document the build process, and most important design decisions for others to adopt or comment.
What I really like about the Mendel concept is the frame and the idea of extruding material instead of removing material.
The design of the MegaMendel has been with emphasis on keeping things simple as well as ensure easy access to vital parts for repair or maintenance.
The first practical decision was the size and dimensions of the steel rods. My first thought was to simply double the size of the frame, but then I learned that threaded rod come in 1 meter lengths. That decided the length of the X-axis. The triangles are sized so the whole thing fits on a 80cm wide table. The rods themselves are 12mm instead of 8mm, that seems to be a good decision as the machine is fairly stable without too much weight.
The frame vertices were a bit of a challenge as I was working with a 9mm MDF sheet - I made a lot of small pieces and glued together to a larger piece that I cut into shape so I got the thing in the image below.
After some more cutting and drilling I had a frame that's just a big copy of the standard Mendel frame as seen in the image below.
Please note the beer can and the 60cm ruler.
After the frame construction the turn came to the by far most complicated part - the Z-axis.
I made a few changes to the design here: The 60 degree angle between invites to tension, thus causing more friction than necessary, so I changed the angle to 90 degrees.
I also decided to mount the angled bearings on the Z-carriage itself, this is due to the fact that the angled bearings are the most important constraint, so I don't like having them on a part that has been weakened for adjustment purposes (I applied a similar philosophy to the other axes - more about that later).
It took two iterations before I was happy with the Z bearing arrangement:
The one on the left was my first attempt, then I decided to go for a lower profile, giving about 10mm of extra length on the X-axis (I don't know what I was thinking, but I went for those stupid 10mm). This design will be changed a bit as I think a single retainer bearing is more than enough - That change will be made when the machine is taken down for repair/maintenance anyway :-)
I decided to keep the lead screws in the 8mm dimension, but I added bearings to the top.
The belts, I got, are 10mm wide so I had trouble finding space for them at the bottom of the machine. The solution was to put the whole Z-drive on top of the machine, so the lead screws are actually hanging in the top bearing rather than "standing" on the bottom bearings.
This proved to be a good solution as I get a lot smoother run this way, and it runs very nice at 100mm/min.
At this time the MegaMendel found its way into my study, where my cat took it into possession:
Please don't mind the date in the image - My camera was a bit off that day :-)
The main reason for the move was that more delicate parts were about to be added, so the dusty environment in my workshop was not the best place any more.
The big challenge with the Y-axis was to prevent sagging, so I started out with a design applying six!! bars to the Y-axis. Then I came to my senses and designed it with just four bars. With three support points on each bar I would have just about the same distance between supports than the standard Mendel. The image below hasn't got all the bearings installed yet, but it pretty much illustrates the concept:
Apart from the four bars there are two major differences from the standard Mendel here: The Y-chassis is running on top of the bars - This brings the chassis close to the print bed, thus allowing the support screws to be mounted in holes in the chassis rather than spending plastic and time on printing the holes.
The next difference is the second bar from the left - That's where the angled bearings are located. This means that gravity helps us out here. In this has worked quite nice so far with no retaining from beneath the bars! (with the middle support bearings installed, that is).
Again the angled bearings are in 90 degrees as opposed to the 60 degrees - I even printed the height record with this set up!
The image above also have the X-axis installed - Again the angled bearings runs on top of the bar. Please note this design has got only seven (7) bearings on the X-carriage.
The X-carriage is not yet finished in the image above as I was awaiting deliveries from MakerBot Industries. One guy there has my special gratitude - Isaac Dietz has been helpful beyond any reasonable expectation - Thanks a lot for that!!!
Having a bachelor's degree in mechatronics I could easily have built the electronic parts myself. But I saw lots of challenges elsewhere, so I went ahead and ordered the electronics and the steppers from MakerBot Industries in Brooklyn.
The electronics should be known by anyone in the community, MakerBot were the only ones to have them in stock at the time. Furthermore they were able to supply the mighty NEMA 23's for the big machine.
My deliveries was somewhat delayed by a volcano that decided to burp just as I was creating big things here - That's most annoying, they should put a cork in that thing!
The drives - Big machines need lots of torque. Or maybe not - I've simplified the drive mechanisms for two reasons: To get simpler mechanics and also to get less friction.
Common to all drives is that I bought the pulleys in standard sizes, and I had to drill them myself - One came out catastrophic, so I had to get new one. The others came out lousy, not too good and perfect. I decided to put the lousy one on the Z-axis as it has got the the longest belt and a bit of wobbling is not really that bad due to the high gearing ratio. The X-axis carries the least weight, so it got the not too good one. The perfect one ended up on the Y-axis as this is the one that potentially carries the most weight.
All the pulleys got glued to the stepper axles with some two-component epoxy stuff - And they have stayed there for quite a few hours of building time. I did scratch the axles as well as the holes in the pulleys before gluing them - That might be why they just work flawless, or maybe they would've done just fine without the scratches :-)
The choice of the NEMA 23's may have been relevant for the Y-axis only - With a huge build volume that is. Furthermore the stepper drivers go into thermal shutdown when running the NEMA 23's at the full current. That could lead to losing steps during a build. I remedied this by turning down the current limiters a bit.
The Z-drive was the first to be created, and it is kept quite simple:
Please note that the are two 624 bearings in the guide and tensioner parts, this due to the 10mm belt. The guide bearing could be omitted with a little redesign. And yes, it takes a bit of the build envelope, but this is the far end of the X-axis and my guess is that I will never need that. It would be fairly easy to redesign this so the stepper goes on the side of the frame so the build envelope would be complete.
I've heard arguments that this makes the construction more top-heavy, but that stepper has a weight of only a fraction of all the steel up there already.
In fact I'm really happy with this design as it also keeps vital parts in full visibility and it is easy to access for repair or maintenance - I may add that this design has worked without a single glitch for many hours of building time already.
X and Y
The X and Y drives do not need much explanation:
I don't think this can be done simpler than that.
The X-axis - This is where the NEMA 23 stepper appears somewhat ridiculous as my X-carriage has a weight just about the same as the standard Mendel's X-carriage - Then again, my Z-axis has a NEMA 23 to handle it :-)
Then there is something I just don't get about the standard Mendel, it has six bearings in the X-drive - MegaMendel does that quite a bit simpler. The Y-drive is also somewhat simpler.
X and Y tensioners
You may not believe it:
Please note this after you're done laughing: They work like a charm - Never had a problem with them!!
A word on bearings
The MegaMendel will have a total of 49 bearings when the Y-retainers are in place - The same ideas applied to the standard Mendel would make it possible to build a standard Mendel with a total of only 30 bearings. I know some people don't care as they get the bearings very cheap. I see reasons to reduce the number of bearings:
- Less friction - I have seen standard Mendels with even the Y-axis running a lot tighter than the MegaMendel's Y-axis!
- Simpler/easier assembly
- Possibly less plastics is needed as well
More about my design changes later.
The USB2TTL cable
I bought the FTDI converter cable together with the other stuff from MakerBot. That one turned out to self destruct before it even worked. I know the MakerBot guys have seen this kind of problem before - They might be interested to know if anyone have identified a pattern here.
I was aware that the extruder is by far one of the biggest sources of problems, so I thought I played it safe by ordering MakerBot's Plastruder. I got a bit wiser. The incident occurred after some ten minutes of test extruding - The idler wheel broke. Later on the thermal barrier deformed and leaked plastic (I think that was my own fault as I tried running it at 240 degrees to avoid filament jams). That may in turn have lead to the heater retainer breaking.
I made a few hacks to the Plastruder, and had it running without any problems ever since.
The Plastruder hacks
The idler wheel
The Plastruder comes with a spare idler wheel. I decided for some reinforcement instead of just supergluing. I took the two-component epoxy that I glued the pulleys with, then I smeared that up along the sides of the wheel:
This photo shows the broken wheel to the left and my hacked version to the right - I never had a problem with one to the right :-)
I wasn't happy with the original pulley as it has to "bite" quite deep into the plastic in order to get a firm grip. I also thought that the teething was too coarse and I thought that it would be good if the pulley would "embrace" the filament to some extend.
I achieved that by giving the pulley a treatment with a 3 mm treading tool and got the result below.
It might not look like much of a difference, but it works - Actually at the same time Makerbot introduced a similar philosophy with their MK5 pulley.
The heater barrel
I had a feeling that some jams were caused by the filament colliding with the edge of the heater barrel, so I gave the entry of the heater barrel a slight funnel shape with this tool.
The heater barrel retainer
The image below shows my biggest disaster so far as the plastic started to leak between the heater barrel and the thermal barrier.
The plastic retainer is broken, I believe that happened due to my running the heater at 240 degrees. This caused the PFTE to deform so the plastic could make its way into the gap and cause more stress.
The X-carriage itself became the replacement part:
This construction should be strong enough. It is made in combination with a change to the thermal barrier.
The thermal barrier
The image above shows that there still is a bit of leaking between the thermal barrier and the heater barrel, but I consider it at an acceptable level.
What I have done to the thermal barrier is quite simple - I simply cut the threaded part off and gave the remaining part a gentle treatment with this tool:
This resulted in a 45 degree angled "funnel" that serves a guide for the heater barrel. So now it's the tension of the mounting screws that ensures the tightness of the assembly. I believe that the leaking is due to the fact that the tool doesn't leave a perfect surface, that will be corrected when the extruder comes apart for some other reason.
The build bed is currently made of masonite and has worked OK so far, but I have not yet tried to use the full build area. This is due to severe warping issues with my ABS plastic. I will try with PLA at some time to see how that works.
Some people say "heated bed" when the talk comes to warping, but I have decided not to go for that solution, the main reasons being: A build area of that size would be very likely to warp itself due to the heat. It wouldn't work for tall objects as they start cracking due to the warping stress.
So what I really need is a way to heat the top of the object without getting the drive mechanism heated as that would just cause a lot of jams.
The build surface turned out to be a bit of a challenge, which kind of natural as it has to serve opposite functions:
1 - Making the object stick to it during the build.
2 - Still allowing the finished object to be easily removed.
I have seen masking tape recommended, but I couldn't get the plastic to stick to it. My next attempt was the kind of double sided tape, used to make carpets stick to the floor - That was virtually impossible to get of the objects. A good compromise turned out to be the glue side of packing tape, but it has glue on one side only. So my build surface is currently a sandwich of masking tape because it is easily removed, then the carpet tape and the packing tape on top of it all with the glue side upwards.
Some double sided packing tape would reduce this to one layer only.
The first prints
This is my first almost successful object:
I had some problems with the raft being too small and some warping even with this small 20 mm X 20 mm object. The top of the object shows that the extruder decided to jam in the last seconds of the build. But I'm still quite impressed with quality of the object - Not bad for a first.
Then I experimented with various shapes and settings, but soon I decided to design something useful like these:
A nice solution to a daily problem.
Printing new parts for itself
The first prints were made without retainer bearings on the Y-axis, I had an idea that gravity would help me out for the first prints. It actually worked and saved the print head during a head crash as the Y-chassis just jumped of the bars, thus leaving space for the print head.
But the plan has always been to have retainers on the Y-axis - Printed ones, that is and printed bearing assemblies as well. I made my own design and here is the complete set for the Y-chassis:
A standard Mendel would need only two of each, but this set is for the MegaMendel.
Please note that the retainers are open to one side. This design allows for the whole chassis to be removed after only detaching the belt.
This photo shows the chassis with the bearings installed (It's upside down in order to show the bearing assemblies).
Some may ask how this design gets adjusted. The answer is that it doesn't need adjustment because the design leaves only 11 mm for the 12 mm bars, so the chassis serves as a spring that keeps the bearings in contact with the bars. The chassis the photo is made of 9 mm MDF, but it has later been changed to 3 mm masonite in order to save a bit of weight.
A project like this is bound to lead into unexplored territory where new ideas are needed. Some ideas are good and others are bad.
The bad ideas
Trying to remedy extruder jams by raising the temperature to 240 degrees Celcius - Don't even think of it, especially not if your thermal barrier is made of PFTE.
My electronics are mounted at the bottom of the machine - Seemed like a good idea at the moment, but it's less accessible and I have had screws coming lose on the Y-chassis - electronics and metal is not a good combination.
The stepper drivers are not really a bad idea, but they are not capable of driving the NEMA 23's to their full potential and that might be needed in the future.
The size of the machine looks a bit like a bad idea for now as warping issues currently prevents me from using full build area, but it is my intention to turn that into a good one.
The good ideas
First thing is that I decided base my work upon the Mendel concept - The triangular frame ends makes it a basically stable structure, the separation of the X and Y axes is brilliant and finally the combination of a screw driven Z axis together with direct drive X and Y axes is very good. I'm thankful towards the Mendel designers for showing the way, and I hope they can forgive me for stepping off the path a few times.
The choice of MDF to create my "plastic" parts was good as it is easy to work with.
The changes in the angled bearing set's positions works very well.
Having the Z drive located on top of the machine leaves the whole drive mechanism easily accessible and it tends to run smoother when the screws are hanging.
With the warping issue solved it has the capacity of printing two complete sets of standard Mendel parts in one go.
Ideas for the future
I've got a few ideas for the future in order to improve MegaMendel.
The warping issue
The warping issue prevents the MegaMendel from working to it's full potential. The heated bed solution will not work for tall objects as it heats only the bottom of the object, but the heat is needed in the build zone, which is the top of the object. I have had tall objects crack all the way up and I don't think that a heated bed will have much effect 40 cm above it.
I have experimented with a heat gun, but it seem that the extruder's drive mechanism gets too hot and jams. A solution to this might be some kind of heat shield and possibly a cooling fan on the drive mechanism.
Another possibility is some sort of upside down "toaster" that would radiate heat towards the object. That would be mounted so it's always a few mm above the build zone.
I should also try to get PLA in order to see how that works, but my initial idea was to print blades for a wind turbine, and with PLA said to be biodegradable - well that spells: "Not for outdoor use" - at least in my book.
It seems like weight is about to be added to the X-carriage, thus justifying the NEMA 23 stepper.
Designing new parts
My redesign of the Y-axis lead to small parts, that prints without warping. I would like to simplify the Z-axis construction by designing plastic parts for it and I would also like to have a plastic X-carriage, but those parts are bound to be of a size that invites to warping.
I am considering a design assemblies of several narrow parts that can be printed vertically.
I did actually print the plastic parts for a Wade extruder. It's warped, but I have it assembled and it appears to work. But then I learned that the extruder controller's stepper driver can't be used with it.
So I looked into the firmware and discovered some issues that needs some looking into, but that will have to wait until I find the time to completely grasp the firmware.
The build bed and Y-chassis
The build bed is levelled with 12 adjustment screws, so it's a complicated process. The reason being that I wanted to have support points at the same interval as the standard Mendel.
I have come to think of that my 12 mm bars might be stiff enough to carry the Y-chassis and the build bed. Then I could reinforce some sheet for the build bed so it absolutely straight even with only four adjustment points. I have my doubts about it, but it could be worth a try.