Sunday, May 31, 2009


Tips, Tips and more Tips

I've recently ended up with a whole bunch of welding tips and some tattoo tips.
I thought it might be interesting to compare them.
From RepRap_bits
From RepRap_bits
Halfords are the smallest : an M5 thread, and a short body (18mm). B&Q have a larger M6 thread and are nearly 24mm long. Both welding tips are copper, and the tattoo tips are stainless steel. The tattoo tips have a long barrel, but a huge hole for a rubber band in the side. (white line is 3mm standard ABS filament)
From RepRap_bits
The nozzle holes are all similar sizes. the first two are 0.6mm Halfords. The next two are 0.8mm from B&Q. The tattoo tips come in various sizes. These are the two smallest round holes: I think they're designated R1 and R2, and look to be a similar size to 0.6/0.8mm. Some of the tattoo nozzles are much larger.
From RepRap_bits
The shapes are very different: the Halfords tip has the 'best' shape, a 45degree taper. The B&Q has a flat tip - it might cause the extrudate to stick, although it would be easy to reshape with a drill and file.
Both the welding tips need to be carefully drilled out - typically 3mm - to within a whisker of the exit hole - so that the thin section is as short as possible.

The tattoo tips seem to be made by drilling a 6.8 or 7mm hole, then a 2mm? hole right through - then a tube of the required dimension is (brazed?) attached to bring the diameter down to the exit hole, and the tip shaped(guessing). From visual inspection, it looks like the 2/2.5mm hole extends almost all the way to the tip.
From RepRap_bits

I think that if the excess tube is 'trimmed' it will form a good nozzle shape internally.
The tattoo tips need to be attached and/or the big hole in the side sealed. a few initial thoughts are:
I'll have a play with them (I've got 18, so plenty of spares) and see if anything works.

Of course, if 0.6 or 0.8mm is too big for you, you can always use the nichrome wire/high temp solder trick to get a smaller exit hole.
Since solder is a generally softer, has anyone tried blocking the hole and drilling the solder with a small drill bit to get a 0.4 hole?

Labels: , ,

Wednesday, May 27, 2009


Sourcing low-cost parts for RepRaps/RepStraps

The key to finding RepRap or RepStrap parts: keep your eyes open, and don't be afraid to ask: experiment, try things out, and tweak your design to what you can find. I've found a few useful bits over the last couple of months.

I wanted to add an encoder to my GM3 drive motor, but I've previously blown my three spare opto-switches.
Asking around at work, I found an old ball mouse with a broken USB plug, and got permission to take it home from sysadmin.
Logic dictated that running off USB (5v) it might contain some opto-switches that I could recycle. Taking it apart, a quick inspection revealed two components either side of a slotted wheel that looked right:
From RepRap_bits

I desoldered them, ripped out the broken IR sensor from the x-max endstop board, and just soldered them in.

From RepRap_bits

Hooking it up to the arduino, it works perfectly! (even the pins are the same)

Other extruder bits:

I found some thin-walled alumininum an brass tube in a Model shop in Westbury. It's imperial (5/32) but the internal diameter is exactly right for my 3.1mm filament. (£4.50 for a pack of 4)
From RepRap_bits

I think it might work well to make vik's no-lathe/ariel extruder, or to make a Nophead-style thin-walled extruder. I've got a large heatsink from a core2 CPU to go with it.

I've been looking for a block of metal to make nophead's resistor heater for a little while - it's not something you can buy in B&Q, and I don't happen to have aluminium bar stock in my garage. :-)

In the middle of Bath, down by Sainsburys, I walked past a little tool shop that was selling scrap metal outside. I went in and asked, and...
From RepRap_bits

Hey presto, the chap disappeared out the back and came back with a 10mmx30mm bar - and cut me off a couple of 30mm lengths from it. The best bit : cost 25p. So, a shout-out to the guys at Avery Knight and Bowlers, Bath - thanks guys. They've been consistently helpful and the shop is full of tools and other useful bits. I also grabbed some m6 coupling nuts (50p)- might be useful to attach nozzles - and a 7.5mm drill (£3.24), which is the right size for my resistors.

I've thrown a few quid away on some tattoo tips from a dodgy overseas company (like these ones). I wanted to have a look to see if they would be useful to turn into nozzles. When (if) they arrive, I'll post some measurements and photos.

Edit: half-hour after posting this, I noticed a big split in the mains cord insulation for my wife's hairdryer. 'It's not safe, we'll have to you get a new one' - of course, nothing to do with the fact it's full of nichrome wire....


Friday, May 22, 2009


Want to Slice Step files or bad STL Files?

I have produced a script that will slice STEP files. It will also handle really bad STL files, which is a nice bonus.

The script generates an svg file that is both human viewable and contains the path information for the slices. This file can be imported into skeinforge to generate the gcode.

Here's an example of an output file. Open with firefox or a javascript-enabled viewer, and you can scroll through the layers.

If you have been frustrated by the "create STL-->have skeinforge/rr host puke because there are holes/etc--> fix by black magic" dance, this is the tool for you.

Is it the solution for you? You can download the tool here:

Give it a try, and let me know your thoughts! Fair warning, installation requires a few dependencies; they are documented in the install.txt file in the zip file.

More Detail
As soon as I started playing with the RepRap host software and skeinforge, the first thing that I realized was that my modelling tool ( and old student version of solidworks ) creates horrible STL. Neither skeinforge nor the host would handle it my STL.

The next thing i realized is that STEP is a much better format for storing solid objects-- smaller, more accurate, etc.

I set out to make a slicing program that would handle STEP and bad STL files, and, ideally handle them very fast.

I did not want to re-invent the work Enrique has done-- what was needed was a file I could create that had only the hard part--the slicing-- and then let his tools do the rest.

Enrique and I collaborated on the slice file format. It seemed like a good idea to use a standard way to communicte 2D paths, and to provide a way for humans to understand the files contents.

Needless to say, reading STEP files and fixing bad STL is quite hard-- I needed to avoid writing this. So, I ended up getting involved in -- a python binding to the OpenCasecade open source modelling kernel. This is really the tool for this job-- an open source full-fledged modelling kernel!

For the cost of having to install OCC, i got lots of benefits-- including a really handy, really fast viewer for 3d objects that can handle STL, STEP, and IGES!

Going Forward
I hope to expand the tool to provide a more interactive environment for slicing and getting objects ready to build. i would like to integrate skeinforge into the tool.

Monday, May 18, 2009


Hi, I have been using a RepRap to make ephemeral forms out of light for an installation. Thank you to everyone who answered my questions on the forum, it really made the process of constructing the RepRap enjoyable. I am not sure what I can add at the moment in terms of RepRap development, but thought that you might be interested in this modification. I have some Processing code that converts 3D Turtle geometry into GCode which I think is interesting for generating recursive algorithmic forms but it might get a bit complex with solid forms as the form itself could get in the way. I need to clean it up a bit before I post it online, but if anyone is really interested and feeling brave give me a shout and I can send it to you.

Anyway thanks again for all your help, it would not have been possible for me to make this without such a vast resource of schematics, code and inspiration being shared.

There is a bit more information about the installation here.

cheers simon

Saturday, May 16, 2009


Joining the ranks of 3d printer owners

Hey all,

I have completed my makerbot cupcake and have now joined the ranks of the repstrap brigade (photos here).

The cupcake from makerbot is a great kit , I spent about 40 hours all up building it across a week (getting little sleep and RL, the GF and a full time job just kept getting in the way :) ).

The mechanics in the cupcake are easy to assemble and the instructions on their site are a great reference as you put it together.

The electronics were a far more serious task , as they are surface mount you need a magnifier and a steady hand.

All in all a good kit and would recommend it to reprappers who want a complete kit that they can just get started with.

Right- i'm off to print more stuff. :)

Tuesday, May 12, 2009


The circuit behind induction heating

Here the schematic of my induction heating circuit (power stage PCB):

It is a "quasi resonant converter" design, where the induction coil (connected to the screw terminal X1) is the inductance of the LC-tank and C1-C5 are forming the capacitance. The principle of induction heating is explained very good in this Fairchild-document:

Here the description of my circuit:
The electrolytic capacitors C6 and C7 together with the inductance L2 are smoothing the current seen by the used PC power supply. Therefore a high peak current can flow through the induction coil without burdening the power supply with this peak value. C6 and C7 have to be low-ESR (Equivalent Series Resistance) in order to reduce the losses and to prevent overheating caused by the high AC-current flowing through these capacitances. For the same reason two capacitors are switched parallel instead of using only one with the double capacitance. The induction coil is connected to the screw terminal X1. If a cable is used the wires should have a high cross section (minimum: 1.5mm²) and it should not be longer than one meter in order to reduce the losses. C1 to C5 are forming the resonant capacitance, again several capacitances with low ESR are used in parallel to reduce the losses. R5 is a current sensing resistor, the current through the induction coil can be measured with an oscilloscope connected to JP2. The Power-FET Q1 is driven by the MOSFET-driver IC1. Depending on the signal "SWITCH_ON" at JP1 (which comes from the microcontroller board) a inverting or non-inverting driver can be selected through the solder jumpers SJ1 and SJ2. The driver is necessary to increase the TTL-high-level of the switching signal at JP1 to +12V, because Q1 requires a minimum level of +10V to be switched on properly. Furthermore the driver has an output current up to 1.5A which loads the gate-capacitance of Q1 fast at switching on. Therefore a very short rise time of the gate voltage is achieved which helps to reduce the switching losses. The circuit consisting of D1, R1, D2, D3, R7 and C12 cuts the drain voltage of Q1 seen at JP3 to a maximum of about 4V (Z-diodes D2 and D3) and a minimum of 0.3V (Schottky-diode D1). Two parallel Z-diodes are used to make sure that the signal at JP3 does not exceed 4V even in the case that one Z-diode fails. The resulting signal "ZERO_VOLT_SENSE", which is connected to the microcontroller board, is used for zero voltage detection and makes zero voltage switching (ZVS) possible. ZVS means switching on the FET when the drain-source voltage becomes zero. This reduces the switching losses, because no power is generated in the FET at this time. As it can be seen in the priviously published photo of the PCB there is no heatsink for the FET necessary because of the low losses.

Saturday, May 09, 2009


Induction Heating

For my Final Year Project (I am studying Master of Engineering) I have designed an extruder with induction heating. This technique uses a varying magnetic field generated by the induction coil to heat the surrounded (ferromagnetic) metal. Below a photo of the extruder, the induction coil can be seen left at the end of the extruder pipe.

The thick blue and brown wires are connected to the induction coil, whereas the yellow ones are leading to a thermistor for temperature measurement. A cutaway view of the extruder pipe is shown in the next picture.

The induction coil heats the sleeve made from mild steel (ferromagnetic) whereas the stainless steel is not heated directly because it is not ferromagnetic. The main reason why I have used stainless steel for the pipe is that this metal has low thermal conductivity compared with e.g. mild steel (about 16W/mK versus 59W/mK). This, along with a wall thickness of only 1mm in the middle part of the pipe, increases the temperature gradient from the heated zone to the right part of the pipe. This helps to reduce the 'filament sticking' problem described in the HydraRaptor Blog (thanks to "nophead" for his work). I have measured the temperature inside the pipe at different distances from the nozzle, the result can be seen in the following diagram.

The higher the temperature the steeper the temperature gradient at distances > 2cm (end of heated zone). This can be explained by the higher difference to the ambient temperature at high nozzle temperatures.

At the next photo the used electronics for induction heating is shown. The top PCB is the power stage circuit, connected below (the green PCB) is a "44-Pin Demo Board" (with PIC16F887) from Microchip which was handy. It is used for driving the FET (see picture centre) of the power stage and to control the heating. It also controls the DC-Motor used for filament feeding. The two potentiometers on the lowest board are for setting the extruder temperature and filament speed.

Friday, May 01, 2009


Extruder design musings

I'm going to throw down some random extruder thoughts, untested theories, and probably re-iterate most of Nophead's proper research. I apologize for any complete rubbish or plagarism


ABS isn't really like chocolate, or ice, which is a simple solid or liquid.
When cold, it's solid and can withstand quite a lot of force in compression. This allows us to force it down a tube and use it a bit like a rod or piston to force the plastic into the nozzle.
When it warms up, you can bend it easily.
Warm it more, and it starts to soften and expands, getting wider, pressing against the sides of the tube causing friction. At some point, it becomes 'molten'.
'Molten' ABS isn't really liquid. In some ways, it's like plasticine or clay - it needs to be forced through a nozzle, it doesn't really 'flow'.
It's also a bit elasticy at times - when you stop pressing down, it can dribble on slowly for a bit, until it's evened out the pressure.
And finally it's a bit like melted mozzarella - pizza cheese. Pull the melted filament out of a heater, and you often get a really long, thin stringy bit - just like a Domino's advert.

Designing an extruder

Lots of my extruder designs have worked to a greater or lesser extent. It's not *too* hard to make an extruder work once, it's reliability and repeatability that cause my problems.

A larger heater chamber (4mm/5mm diameter) works OK, but the extra ABS takes longer to warm up and melt.

The shorter the 'melted', hot section the better, - preferably as near to the nozzle as you can get. Also, the less metal you have, the less thermal inertia you have, so it heats up quicker. The quicker it heats up, the quicker it cools down while extruding - so the heater power has to be controlled.

PTFE is a good insulator, and it's easy to machine - you can cut an 8mm rod with a sharp craft knife, and it's easy to drill. You can cut a thread easily with a bolt or nut. PEEK is much stronger and requires proper machining - use a tap/die to cut threads.

PTFE is soft, and when it's hot it gets softer, so it really needs extra support (external bolts and washers, etc) to hold the join to the heater barrel still, and extra pressure to avoid leaking plastic out the side. 8mm PEEK rod is strong enough to have a welding tip screw-in (tap an m5 thread) and easily holds the pressure.

The join between the barrel and the insulator is very important. This is where the ABS becomes softer and the filament will get wider and fill any gaps in the channel.

Quite often, my extruder experiments work the first time, but after cooling and re-heating are much stiffer and difficult to get restarted. After inspecting the cold filament, this is because the ABS flows into any small gaps and then solidifes when cold. The entire assembly will then be stuck until the gaps have melted.

If the extruder is hot but not flowing for a long time, the heat will flow up the ABS, softening and widening it further up into the insulator. At least PTFE is very slippery and it helps getting it restarted.

I think NopHead's design, with the thin aluminium tube insulator, excellently avoids the join problem. If you can cool the barrel further up, so that the melt occurs inside the barrel, this problem doesn't occur.

Nichrome / Resistors

Both Nichrome wire and resistors both work reliably and reach the temps needed. Nichrome wire needs more preparation and a layer of insulation - Fire cement, jb weld, or Kapton / kraken tape. If you connect your normal wire to the nichrome by knotting it, or using a crimp connector, it's much easier to wind on and you can hold the join safe inside the fire cement or tape. Nichrome is very flexible, you can wind it on the barrel, nozzle, or both.

Resistors are nice, self-contained heaters. They work well, but you need to transfer the heat from the resistor surface to the nozzle. This means either embedding it in a drilled block (like NopHead) or my low-tech solution of shoving them inside m8 coupling nuts (as I don't have any solid blocks lying around). Either way, they're larger and need more metal (therefore more thermal inertia, slower heater), and not quite as flexible as the assembly is big and chunky compared to the nichrome wire.

Both methods work well.

Ideally, the nozzle should be removable.
Ideally, to keep the heated section as short as possible, the heater should be close to or on the nozzle.
Both of these requirements can conflict :-)

Welding tips (0.6mm) from Halfords, £5 for five, work pretty well as nozzles or even combined barrel/nozzles. They need to be carefully drilled out to 3 or 3.2mm almost to the end, but be careful. I've wrecked four trying to drill to 3.5mm to get a better heater entry, the walls are just too thin. The central hole helps align the drill centrally, even drilling by hand works OK, but I'd use a 1mm then 2mm then 3mm bits first.

I've had reasonable success with a PEEK insulator, a welding tip, with nichrome wire wound directly round the welding tip.

Have a go and see what you think :-)

Labels: , , ,

This page is powered by Blogger. Isn't yours?

Subscribe to
Posts [Atom]