If you're looking at this blog, I'm guessing I've either told you about it in person, you have a RepRap, or you are looking at building one.
I'll try to keep away from going on and on about how I built my RepRap through the whole blog: there are hundreds of other blogs that do that and are all basically the same but do each have a few new tips, if you are starting out or thinking about it, I would recommend you look at some. Since I have already built my first machine, I'll be able to compact all that I've had to do and learnt into one post. In future, I am aiming to have more projects than just RepRaps (although they probably will remain central to things), most of my projects will probably include some electronics, mechanical and some coding stuff. However, I wouldn't be expecting anything too impressive: I'm pretty much learning all this stuff as I go and don't have much previous experience with any of this.
Before I embarked on the journey of acquiring a RepRap, I had been following and researching the RepRap project for about three years, at the time, I believed it would be much simpler to just buy a Makerbot Cupcake CNC, so I went off and investigated that instead for a year and a half. When I finally did take a proper look at the RepRap project again, it had improved a lot and things were improving even more rapidly, it also now had many advantages over the Makerbot. Confident that RepRaps were now the leading edge and much easier to use, I decided to finally bite the bullet and buy a Makergear.com Prusa Mendel kit. However, a few weeks before I placed the order, nophead, a highly regarded contributor to the RepRap project, posted about a new machine he was developing and pointed out the problems with machines that use threaded rod as the structure of the machine. That really steered me clear of spending a thousand dollars on a machine which had inherent problems which I was also probably going to upgrade from. Nophead didn't release the files and the amazing script which could generate any configuration of machine you wanted for a while after the post which was fair enough: he wanted to get everything perfect and iron out all the problems. I wasn't going to use his exact design anyway because I didn't have access to a 3D printer to make the parts, so that was no big loss or setback anyway, I decided to design and make a RepStrap version of the machine.
I built, or more like bodged the RepStrap together to last long enough to build a more reliable machine. I wouldn't recommend people make one like it if you want a long-lasting, reliable machine: it often requires running repairs. I should also add, that at the time that I'm writing this post, I'm already half-way through making my next RepRap, a Mendel90 made using Nophead's designs, which should be more reliable and precise.
 |
| My RepStrap machine |
The RepStrap machine itself is made from MDF; uses the weaker, lower current NEMA 17 stepper motors; has a Budaschnozzle extruder; has a PCB heated build plate with glass; uses Sanguinololu electronics; uses M10 zinc plated rod for the Z axis, not the best but it works. The design also uses a lot of hot-glue to fix things together, not an "acceptable" engineering practice, but it makes valuable parts retrievable and it works for my needs, only just. On the axes, I used 8mm "precision ground, chromium plated rod" from VXB bearings which in total, costed about $60AUD. For the linear bearings, I used LM8UU bearings.
 |
| The MDF panels I cut for the machine with a pretty inaccurate jigsaw, luckily the original sheet had enough straight edges and 90 degree corners on it to be used where accuracy mattered |
 |
| The whole frame assembled with screws and wood-glue |
The machine is usable enough to create reasonably accurate plastic parts for my next Mendel90, but all printed parts have mainly cosmetic "ribbing" on the surface which could be caused by a number of things. Though I think it is mainly caused by the massive wobble on my Z axis and possibly because the pitch on my threaded rods isn't constant enough, however, I never expected miracles from hardware store zinc plated rod. I got some stainless 8mm threaded rod for my next printer from a local fastener store, which, to my luck and amazement, had every single fastener on the Mendel90 BOM, except for the springs, hex pillars and tiny number 2 size screws for the endstops.
 |
| Some printed parts, you can clearly see the "ribbing" on them. I used the "brim" setting to create a skirt around the base of the object to eliminate warping. |
In the process of making the machine, there were quite a lot of set backs, the first thing was that the linear bearings have an outer diameter of 15mm, unfortunately, it appears that 15mm spade bits aren't actually made, so I couldn't make the right sized hole for them in the bearing holders. Instead I had to make do with a 16mm bit which left a lot of slop which I had to fill up with hot glue.
Also, as soon as I plugged in the electronics together for the first time, after checking everything for shorts, a motor driver blew up, or more accurately, an SMD capacitor on the driver board.
Here's a slow motion shot of the capacitor blowing up that I managed to film:
After buying a new StepStick, triple checking for shorts and connecting it up, it worked and things were good to go. After doing some calibration, which requires having the motors on for a while, my motors melted the hot glue I was using to attach them to the machine, I ended up fixing them in using cable ties which worked fairly well. I never bothered trying to fix them in using their mounting holes because that would require me to drill 4 holes very accurately with no room for adjustments.
The last problem I faced was that my heated bed couldn't get up to temperature for three reasons, the voltage on the power supply dipped by 1 volt whenever I powered the bed, there wasn't enough insulation under it and the resistance was slightly too high, causing the power to be too low.
The insulation problem was easily fixed by putting some wool carpet underneath. To fix the voltage problem, I tried to put resistors on my ATX power supply's 5v rail, but I was only gaining 0.2v per 5 watts of load on the rail. The resistors were also getting very hot and it was hard to find places to put them safely; it was an ugly solution.
To fix this issue, I bought a so called "Ebay power supply" which are normally used to power LED light systems and security cameras to replace the ATX, it worked great, had good voltage regulation, was adjustable and was meant to power things like this, unlike the ATX computer power supply. However, a couple weeks ago, it suddenly stopped working, there was nothing I could reset, and upon opening it, there wasn't anything that was obviously blown up at all, the fuse was also fine. The only odd thing, which I had ignored when I received it, was that the warranty sticker was cut and part of the metal case was dented. When the power supply stopped working, I went back to the seller and asked for a refund or replacement: according to the Ebay page, it was meant to have a one year warranty, however, the seller would only give me a 10% discount if I wanted to buy another one. A new one is currently in the post from a different supplier, I definitely wouldn't want to support the person who sold it to me by buying another off him. Hopefully the old one was just a one-off dud and the new one will work for a long time: everyone else seems to have no problems with them. As for the problem with the resistance, that turned out to be a fault with my multimeter.
With all these problems fixed (at the time), I managed to calibrate the machine, print a few small improvements for it's accuracy and start printing out my next printer. However, as I mentioned before, things are currently at a halt and only half the parts are printed as I write this.
I sliced the objects using Slic3r, which required almost no fine-tuning to get good prints out-of-the-box and I suspect was much less of a pain than the alternative, Skeinforge. As host software to send g-code instructions to the printer from the computer I used Pronterface. Everything was printed out of ABS plastic, famous for its use in Lego bricks, at a temperature of 235 deg C. I also found a good local supplier of this plastic, their postage was quick and reasonably priced (for a person in Melbourne), the plastic was in good tolerances (less than 0.1mm), the plastic was of good quality and the plastic was also good value ($50AUD for 3mm diameter x 300m or about 2.2kg).
I used 3mm plain old window glass heated to 110 degrees C by the heated build platform as a print surface, all I had to do was clean it and coat it with sugar water before each print which of course evaporated leaving the sugar. One coat of the solution at 2 parts to 50 applied with a brush while the plate was at 70 deg C worked fine for me. However, I did have to spread the coating of sugar (which is impossible to see but possible to feel), by slowly moving a damp, non sugared, piece of paper towel over the surface when it was at about 80 deg C because when the sugar-water is first brushed on, the surface tension of the water causes the solution to not cover some areas.
This sugar-water solution works, as far as I can tell, about as well as the very widely used, relatively expensive kapton tape: like kapton, it has the odd edge that lifts, but since using the "brim" I haven't had any trouble. I think sugar-water is a much better alternative to kapton as it is so cheap and widely available, unfortunately, the plate does have to be cleaned with some glass cleaner and re-coated on each print to avoid spots with too much sugar and to avoid baking it on. Parts also pop right off with some crackling sounds when the print is finished and the bed is cooled, this happens between 40 - 50 deg C.
If you attempt to use this method, one thing to note is that you can apply too much sugar, in which case nothing will stick but instead slide around. The surface, once coated, should feel "dry" and almost feel even less sticky than clean glass, rather than sticky/slippery. I have found that I need to squash my bottom layer down quite a bit to get filaments to stick on the first layer, however, I think this may be down to the fact that I currently cannot level the bed accurately enough on the RepStrap.
Here is my very, very first print:
It took many attempts to get it to stick which is how I discovered that you can coat on too much sugar: each time I tried to put down the first layer, nothing would stick but rather a big ball would form and drag over the surface, taking away some excess sugar till, on the next try, it finally stuck.
If I had to make a RepStrap again, I would definitely do it differently. I would probably scale up a version of the Mantis CNC as they are a lot more suited to being made with simple tools in a home shop and, by the looks of things, would be a lot more accurate and reliable (as a RepStrap). The Mantis would probably only need a few simple modifications, such as belts instead of threaded rods and larger axes.
All up, I spent about $600 AUD on the machine including replacements for defective parts, etc. I now know that, in hindsight, you can get parts much cheaper than what I did: I just didn't look around enough. You can get a full Sanguinololu electronics kit off Ebay, including stepper drivers, endstops and heated bed for a good price and you can also get 5 stepper motors for about $60 including postage, half of what I paid for my motors!
To fix this issue, I bought a so called "Ebay power supply" which are normally used to power LED light systems and security cameras to replace the ATX, it worked great, had good voltage regulation, was adjustable and was meant to power things like this, unlike the ATX computer power supply. However, a couple weeks ago, it suddenly stopped working, there was nothing I could reset, and upon opening it, there wasn't anything that was obviously blown up at all, the fuse was also fine. The only odd thing, which I had ignored when I received it, was that the warranty sticker was cut and part of the metal case was dented. When the power supply stopped working, I went back to the seller and asked for a refund or replacement: according to the Ebay page, it was meant to have a one year warranty, however, the seller would only give me a 10% discount if I wanted to buy another one. A new one is currently in the post from a different supplier, I definitely wouldn't want to support the person who sold it to me by buying another off him. Hopefully the old one was just a one-off dud and the new one will work for a long time: everyone else seems to have no problems with them. As for the problem with the resistance, that turned out to be a fault with my multimeter.
With all these problems fixed (at the time), I managed to calibrate the machine, print a few small improvements for it's accuracy and start printing out my next printer. However, as I mentioned before, things are currently at a halt and only half the parts are printed as I write this.
I sliced the objects using Slic3r, which required almost no fine-tuning to get good prints out-of-the-box and I suspect was much less of a pain than the alternative, Skeinforge. As host software to send g-code instructions to the printer from the computer I used Pronterface. Everything was printed out of ABS plastic, famous for its use in Lego bricks, at a temperature of 235 deg C. I also found a good local supplier of this plastic, their postage was quick and reasonably priced (for a person in Melbourne), the plastic was in good tolerances (less than 0.1mm), the plastic was of good quality and the plastic was also good value ($50AUD for 3mm diameter x 300m or about 2.2kg).
I used 3mm plain old window glass heated to 110 degrees C by the heated build platform as a print surface, all I had to do was clean it and coat it with sugar water before each print which of course evaporated leaving the sugar. One coat of the solution at 2 parts to 50 applied with a brush while the plate was at 70 deg C worked fine for me. However, I did have to spread the coating of sugar (which is impossible to see but possible to feel), by slowly moving a damp, non sugared, piece of paper towel over the surface when it was at about 80 deg C because when the sugar-water is first brushed on, the surface tension of the water causes the solution to not cover some areas.
This sugar-water solution works, as far as I can tell, about as well as the very widely used, relatively expensive kapton tape: like kapton, it has the odd edge that lifts, but since using the "brim" I haven't had any trouble. I think sugar-water is a much better alternative to kapton as it is so cheap and widely available, unfortunately, the plate does have to be cleaned with some glass cleaner and re-coated on each print to avoid spots with too much sugar and to avoid baking it on. Parts also pop right off with some crackling sounds when the print is finished and the bed is cooled, this happens between 40 - 50 deg C.
If you attempt to use this method, one thing to note is that you can apply too much sugar, in which case nothing will stick but instead slide around. The surface, once coated, should feel "dry" and almost feel even less sticky than clean glass, rather than sticky/slippery. I have found that I need to squash my bottom layer down quite a bit to get filaments to stick on the first layer, however, I think this may be down to the fact that I currently cannot level the bed accurately enough on the RepStrap.
Here is my very, very first print:
It took many attempts to get it to stick which is how I discovered that you can coat on too much sugar: each time I tried to put down the first layer, nothing would stick but rather a big ball would form and drag over the surface, taking away some excess sugar till, on the next try, it finally stuck.
If I had to make a RepStrap again, I would definitely do it differently. I would probably scale up a version of the Mantis CNC as they are a lot more suited to being made with simple tools in a home shop and, by the looks of things, would be a lot more accurate and reliable (as a RepStrap). The Mantis would probably only need a few simple modifications, such as belts instead of threaded rods and larger axes.
All up, I spent about $600 AUD on the machine including replacements for defective parts, etc. I now know that, in hindsight, you can get parts much cheaper than what I did: I just didn't look around enough. You can get a full Sanguinololu electronics kit off Ebay, including stepper drivers, endstops and heated bed for a good price and you can also get 5 stepper motors for about $60 including postage, half of what I paid for my motors!
No comments:
Post a Comment