Nophead's Extruder Tweaks
Release status: Experimental
The extruder design has moved on, getting ever more complicated, but I have stuck with the original version with a lot of minor tweaks to make it reliable. It gives good results with ABS, PCL, PLA and HDPE and has higher throughput than any of the geared designs.
It only requires this handful of RP parts.
Using a shaft encoder gives the following advantages: -
- Ensures a constant filament feed rate, independent of friction and temperature.
- Allows the speed to be reduced without stalling, but also does not reduce the top speed as gears do.
- Requires no trial and error calibration of feed rate as it can be calculated from screw pitch, etc.
I have designed an RP version described here. It works as an encoder but I have yet to determine whether it has sufficient resolution.
DC motors give off a lot of RF interference which can corrupt comms, crash nearby micros and affect radio and TV reception. I designed a small suppressor circuit, which mounts on the motor terminals and solves the problem.
- This must be the 12V version. The 5V motor has very thin brushes that quickly wear away. Eventually one bends back and causes a short circuit. If you use a Darlington driver, rather than a protected MOSFET, it can be destroyed.
The 12V version has far more substantial brushes and a bronze bearing rather than a plastic one.
- The clutch may need locking for tough plastics like PLA and HDPE. See Zach's instructions.
- I have found that the output shaft sometimes starts slipping (even without the clutch being locked). The plastic shaft slips on the splined metal shaft. This can be fixed by gluing it with a two part cynoacrylate superglue that states that it is suitable for polyethylene and polypropylene.
- Running in the motor by connecting to a low voltage supply for a few hours will allow the brushes to wear in before having to handle high currents. This reduces arcing, producing less RFI and prolongs the life of the brushes.
- Adrian Bowyer recommends lubricating the gearbox with silicone grease.
The flexible drive has a limited life due to metal fatigue and is not required for any plastics tried so far. Direct drive is simpler and more mechanically efficient.
The direct drive motor holder for the older design extruder is hard to dig out of SVN so I have posted it here: Media:Old-motor-holder.stl
I have designed a new shaft coupler that couples the GM3 to an M5 nut soldered on to the drive shaft.
The shaft and nut must not be stainless steel or they will not solder. Put some plumber's flux on the thread before soldering to help wick the solder through. Solder the top to form a meniscus and solder the bottom to form a fillet. An electrical soldering iron is plenty hot enough.
- The poly holder channel needs to be low friction, particularly for PLA. HDPE is ideal for this as it can be RepRapped and it is almost as slippery as PTFE.
- The best way to align the polymer guide is to use the shank of the drill that was used to drill the PTFE, rather than some filament.
- Lubricating the filament by passing it through an oiled felt washer can help with PLA and PCL. I have not needed it for other plastics and it tends to discolour the object being made.
Even when not oiled it seems useful in that it catches fluff, which would otherwise presumably end up in the extruder.
- The pump halves are best spaced apart at the top by two or three M3 washers so that the drive screw clears the filament. This gives a gradual lead-in, reducing the amount of friction needed to cut the thread. Only a relatively small section of thread needs to be engaged to move the plastic, any more just increases the friction.
- Two very strong springs are required at the bottom for hard plastics like PLA and ABS. Mine are 1.5mm x 9mm x 25mm and came from an A3 flat bed scanner's lid hinge. Note spring force is proportional to the fourth power of the wire diameter. I made end caps for the springs as the inner diameter is too big for M3. Ideally the studding would be bigger and the holes further apart.
If the springs are too tight the motor will stall, or the clutch will slip if it has not been locked. If they are too loose the drive screw grinds the filament rather than cutting a thread into it. I start with them too loose and each time the filament slips I tighten them a turn and push the filament with pliers for a while to restart it. When I find it is reliable I measure the amount of thread sticking out and make a note of it.
- Wingnuts make it easier to assemble and disassemble. The M3 studding wears out eventually. Keeping it lubricated with a little grease helps.
- I use tapped steel plates with lock nuts on the other end of the M3 studding to make assembly easier, i.e. you don't need a spanner.
- The screw drive works better if the thread is sharpened. This can be done by running a half round needle file along it while spinning it. Adrian Bowyer reports it can also be done with a split die on its smallest setting.
The friction clamp does not hold the PTFE barrel well enough to extrude HDPE. There are two solutions to this. Adrian Bowyer's is two 3mm pins inserted through the clamp and the barrel. I used smaller diameter pins out of pop head rivets for my first extruder. Vik Olliver puts a single self tapping screw through the clamp into the PTFE. Obviously it must not go too far in or it will cause a leak.
Again, I am using an older simpler version of the clamp, which is easy to replicate, tweaked to accommodate Vik's self tapper: -
I have made it in both HDPE and ABS.
The three smaller screw holes are HydraRaptor specific. The two larger attach to the Darwin x-carriage with M5 bolts.
At the temperatures required for HPDE and ABS, ~240°C, the PTFE barrel softens, which can allow the heater barrel to slip out.
- Keep the temperature below 250°C, note that the heater barrel may be hotter than the nozzle, see here
- Be sure to seal the joint with PTFE plumber's tape.
- A pipe clip can be used to hold the heater barrel in. It will require re-tightening from time to time as PTFE creeps.
- If you have a split die you can make a heater barrel with a slightly oversized thread.
- The PTFE expands by about 0.5mm when it warms up. That means the z-axis calibration must be done when it has been hot for some time. When I start my machine the warm-up routine waits for 1 minute after it reaches the target temperature. That allows the PTFE to reach its working length and also gives time for the filament at the joint between the PTFE and the brass barrel to melt. This is important as solidified plastic at that point can jam the extruder and cause the drive screw to strip the thread it has cut in the plastic.
Despite being rated to 600°F on the packet and 500°F on their website, JB Weld degrades at temperatures needed for ABS and HDPE, eventually turning to dust.
Various alternatives like BBQ paint and fire cement have been used but none satisfy all the requirements: -
- Adhesive to metal
- Able to stand high temperatures
- Good thermal conductor
- Good electrical insulator
I used Cerastil-H115 which is designed for the job and works perfectly, see.
The heater shows no sign of degrading after months of use. Unfortunately, although it is not expensive to make one heater, the MOQ is 1kg costing around £100. Forrest Higgs has found a cheaper alternative here.
I find bare nichrome easier to work with than the insulated stuff, as well as being more readily obtainable. The only downside is that you need a layer of insulation under it as well as over it, making it a two day job.
Two strands of 0.1mm nichrome twisted together gives a convenient length of 110mm for 8Ω. That gives 18W when driven from 12V, which seems about the right amount of power. It achieves 240°C in about 90 seconds, without any external insulation (only required if you want to be able to run the fan while extruding).
When using the RepRap electronics 2V is dropped across the TIP120 Darlington, so to get the same power, a 6Ω heater is required. Better to use a protected MOSFET, something like this, which is a drop in replacement and doesn't need a heatsink.
I connect tinned copper leads to the nichrome by twisting them together and soldering the joint with high temperature solder. This is tin/lead solder with the 60:40 ratio the other way round. It has a melting point of 300°C. I bury the joints in the Cerastil. If you do the same with ordinary solder, or unleaded, it is molten at extrusion temperature and eventually oxidises away leaving a dry joint.
The advantage of using copper flying leads is that they are less fragile than Nichrome and they don't get as hot. I insulate the heater and thermistor leads with PTFE sleeving and terminate them at a four way 0.1" pin strip with heat-shrink sleeving over the soldered joints. That conveniently mates with a floppy disk power connector.
It is vital that the heater leads are insulated from the thermistor leads to avoid destroying your electronics.
- The thermistor should be rated for 300°C.
- I use a 10K Epcos part with my 3.3V extruder controller that has an ADC input. 5V Arduino electronics should use 100K to avoid self heating and the PIC electronics should also use 100K. With a 10K thermistor the PIC runs out of drive current to measure it at about 170°C.
- I leave the thermistor wires at their full length. They are special wires with poor thermal conductivity to avoid cooling the thermistor.
- Mounting it on the heater barrel between the heater and the nozzle seems to be the best site.
- Cerastil works well for attaching it.
- The maths for measuring temperature with a thermistor and an ADC are here.
- I calibrate each thermistor with a thermocouple. The accuracy required for repeatable results is certainly less than 5°C and I think 1°C is desirable.
- Acorn nuts are the easiest way to make a nozzle, but the exit hole is very short, leading to incontinence. My preferred design is one with about 2mm exit path.
The end of the nozzle should be flat to give a smooth top surface on the object but it needs a chamfer at the edge to prevent it rucking up previous layers.
- The thread must be sealed with PTFE plumber's tape.
- The end of the heater barrel can be shaped to minimise the void inside the nozzle, see
The nozzle aperture only determines the filament diameter when extruded into mid air. The resulting filament will be the diameter of the aperture plus the die swell, which depends on the plastic being extruded and the speed of extrusion.
When the filament is being laid down it is simply defined by the plastic flow rate and the speed of the head. The resulting filament can be bigger or smaller than the nozzle aperture. I normally run with it the same (currently 0.5mm). Having it bigger is useful for the first layer of rafts. I make this 1mm and set the head height low so it is squashed onto the bed. This allows inaccuracies in the z-calibration and blemishes on the bed's surface to be levelled out.
There is evidence that extruding filament smaller than the orifice leads to loss of positional fidelity.