Saturday, 4 August 2012

Lessons in Rapid Prototyping and Design

Never before did I realise what a huge difference software and firmware can make to 3D-print quality.

Not long after setting my reprap up again in its new home, I found it having problems moving in the Z-axis, where the motors would refuse to move half the time, which after describing it on IRC people helped me figure out that the controller was telling the motors to accelerate too fast.
The acceleration that had been implemented in June 2011 Sprinter firmware still caused a significant amount of 'jerkiness' in axis movements. By updating to the latest version for June 2012, not only did more appropriate acceleration, adjusted by a predictive 'look-ahead' buffer, improve this current problem, but it also cut out a vibration problem that broke my Y-axis last year, by pausing at the end of each move for a few ms, when making very short successive movements in order to draw a narrow zig-zag fill line. This helps because in most stepper-motor use there is no feedback control mechanism - the idea being to move a set number of steps then lock in place and hope for the best, which in reality can result in nasty vibrations. By stopping for a moment, the transient vibrations resulting from overshoot can settle, so there is less chance of them feeding into some resonant frequency and shaking the printer to bits.
Illustration of overshoot, via Wikimedia Commons.

In the new firmware a minimum temperature is now set by default, which might have prevented another previous incident that I had with over-heating when a thermistor connection broke.
I also upgraded to the latest version of the Printrun host software along with the latest version of the bundled Slic3r gcode-generating software; aside from pronterface's new easy-to-grasp user interface and slicer's big reduction in complexity over skeinforge, there are a couple of things that I like most about this: a function in slicer that sets a minimum distance the extruder must travel before filament will be retracted (useful where lots of repeated retractions on complex narrow infill used to make my extruder jam, such as with this part), and another one that defines a distance to move up in the Z-axis at the start of every retraction and drop back down at the end (which for a little extra print time can prevent the nozzle from ever dragging through lines already set down, which is important on the first layer while a bond to the print surface is still forming).
In summary, if you ever meet an open-source programmer, give them a hug.

Meanwhile, I recently discovered something that wasn't so well coded in a proprietary program, SolidWorks. When I had tried to open STL files in SolidWorks before, all it did was import 'STL graphics', which was essentially a picture of the object in question that you could rotate as eye-candy but had no solid surfaces that you could sketch on to cut extra holes or extrude extra structures. With lots of other people complaining about the number of things posted on Thingiverse with only STL files supplied, since they couldn't edit them in SCAD or some other programs, I pretty much wrote off the possibility of editing those designs for the last year, figuring that it wasn't an easy thing to do in any CAD software yet (though something at the back of my mind said it should be).
A few weeks ago while looking for SolidWorks-compatible files in order to make some edits that I'd been wanting to make to Greg's Hinged Extruder for months, I stumbled upon some options that allow SW to import STLs as solid bodies, and hence enable them to be carved up any way you want.
Solidworks STL Import - Default Options
Solidworks STL Import Options - how they should be by default.

I was kicking myself for not spotting this before, as this pretty much opens up all STL models on Thingiverse for fixing/editing. Since I'd already started, I continued to create a modified M4 version of that hinged extruder using Tom's SW file, but was now also enabled to edit Greg's 'guidler' part for a better fit.

Meanwhile I found that the old v5 extruder hot-end that I got with my kit from Mendel-Parts was increasingly creating problems for me. I was already annoyed that they stopped supplying replacement parts for it around the same time that the PTFE barrel on mine became warped at its threaded end, and I had been reading that more current hot-end designs using a thin PTFE liner that extended far down the barrel, minimising the contact&melt-zone and hence friction against metal parts, were practically never jamming by comparison, which was making me want to upgrade. For this reason I got hold of a J-Head nozzle from local supplier Reprap3D:
The J-Head Mk.IV-B looks damn sexy by comparison, with the heatsink-notched PEEK body and all-in-one nozzle-heat-block that improves heating efficiency. Shown here after I had applied some Kapton insulating tape.
Not only was the large force required to push plastic through the v5 nozzle a problem, along with the weakness of the PTFE used for its main body and the fit problems and extruder-body damage that came with the PEEK-block mounting system, but the original connections that I used to the heater and thermistor were giving out. The small bootlace-ferrule crimps were giving way one-by-one from months of vibrations, and as I soldered each connection back together, I was having other problems with short-circuits, broken circuits, and eventually, total failure of the thermistor resulting in overheating during printing.
There was a fluke of the kind favoured by cheesy hollywood writers, whereby one of the last things I was printing as the old nozzle failed, was a new extruder block to fit the new hot-end. Due to overheating it was badly warped and I had to cut some holes into better shape with both drill-bits and pen-knife, but here are the old and new extruders for comparison:
Left: Old'n'busted Mendel-Parts v5 in my M4 extruder body. Right: New-hotness J-Head Mk.IV in a prototype M4 body.
The left extruder body was already a temporary replacement for one that was breaking after previous overheating problems. Had the original design incorporated the zip-tie slot shown in use on the right, the ferrule connections might have lasted a bit longer. The new connections themselves I fixed in a different way. Instead of crimping wires twisted up along a straight line with the resistor pins and thermistor wires, I put them end-to-end, twisted the connections together, crimped a ferrule over them, bent that ferrule back on itself (which tears it slightly), melted lead-free solder over the ferrules for reinforcement, then tightened heatshrink around each tough little block. Hopefully this should stand up a lot longer, but I'm keeping the old extruder in a box as an emergency replacement (though I'll be needing a new thermistor for it anyway).
Theoretically it's not advisable to use lead-free solder (mine says it is 96.5% tin, 3% silver, 0.5% copper) so close to a heating block that routinely exceeds 230C when printing with ABS (or could be even higher with PC), since the melting temperature of tin is 232C, but in practice the little distance from the heating block and passing air keeps the temperature slightly lower, so all that heat is likely to do to them is keep the joints from becoming brittle.
Since starting to use the J-Head hot-end I've noticed a big increase in print quality, which may be mostly due to the slightly smaller nozzle diameter - the new one has a 0.4mm outlet while the old one was 0.5mm.

In other news, I've just released a few files with a glimpse of my cowled wind turbine design-in-progress that I've been mulling over for a couple of years now and sketching up ideas for in short bouts of inspiration. My hope is that with something more solid to look at with some very recent CAD work that I did for it, some other designers may be able to help me to improve it.
My 'Storm Turbine' design 0.2, about half the design work done now.

Once I get a moment I'll probably write up a wiki page for this project on Open Source Ecology.
I've also started doing my own tests with printing PLA onto paper, so will report back on that shortly, brb need to eat now.

No comments:

Post a Comment