Bottoming

As well as repairing my extruder I aimed to bring it closer to the latest RepRap design. The machined heater barrel has been replaced by a drilled out brass bolt. Partly out of laziness, and partly out of a lack of confidence in my skill with a lathe, I bought a ready made barrel from BitsFromBytes.



I was surprised to find it came with a modification that I had blogged when I made my first nozzle. That was to turn down the end that fits into the PTFE holder to get round the fact that I didn't have a bottoming tap to make the thread go right to the end of the hole in the PTFE. Ironically, I had bought a set of taps which included a bottoming one in the meantime. Well at least I thought I had :-



The tap on the left is the one from my cheap set of taps which only has one for each thread. The boxed set are the new ones that I bought with a taper, second and bottom tap. The single tap seems to correspond to the second tap, which makes sense for a compromise. The thing I don't understand is why the bottom tap still has a point and some taper. I was expecting it to be straight with a flat end, so I made it thus with a grinder :-



To try it out, I made a test thread in a scrap of PTFE and cross sectioned it :-



As you can see the flat end of the heater barrel butts up nicely to the end of the thread in the PTFE. I think it is important not to have a void here as it will fill with molten plastic which will freeze when the heater is switched off. It might then be difficult to melt it again as it is insulated from the heater.

The end of the barrel that is turned down is then a pretty good fit for the acorn nut, although a simpler solution is probably to bottom the thread in the nut and the PTFE and then go back to a plain heater barrel.

Nuts

Since my first attempt at making the RepRap extruder the design has moved on to use a brass acorn nut as the nozzle. It has the advantages of making the extruder easier to fabricate, allows the aperture size to be changed by swapping nuts and allows blockages to be cleared. I have to say that I never experienced a blockage with my single piece nozzle, but I can see how it could easily happen if a bit of dirt bigger than the aperture gets into the barrel.

Unfortunately brass acorn nuts, otherwise known as dome nuts and cap nuts, are expensive and hard to get hold of. I got a couple of un-drilled ones from BitsFromBytes.



My plan was to start with the smallest hole I could drill and expand upwards to see what effect it had and then drill the other to the size I found to be the best. Stupidly, I overestimated how thick the dome was and put a centre drill right through the first one. So now I have one with a 0.3mm hole and the other is about 1.1mm.

This is the 0.3mm bit I used :-



If you use a drill or a drill press it is easy to snap drill bits this small but it is actually very easy to drill 0.3mm holes with a lathe. I spin the nut in the chuck and hold the drill in my fingers. I drill from the inside of the dome. The drill finds its own centre and then I apply light pressure. I expect the same could be done by spinning the nut in a drill chuck.

The RepRap design for the heater barrel is just a flat ended threaded brass tube made from an M6 bolt. This is easy to make but not the ideal shape. Brass acorn nuts seem to be machined from a solid piece of brass. The internal thread is made by drilling and tapping. Because it is a blind hole that means that the thread does not go all the way to the end. If you screw a flat ended barrel into it then it stops short of the end, leaving a void that will fill with molten plastic as can be seen here. Molten thermoplastics compress under pressure, so ideally the amount of molten plastic in the extruder should be kept as small as possible to make the start stop response as fast as possible.

I decided to sacrifice my over drilled nut to find out the inside profile by cross sectioning it :-



Not surprisingly, the inside profile matches a 5mm drill as that is the correct size for tapping an M5x1 thread.



This is how far a flat ended barrel can enter :-



This is my attempt to match the profile :-



And this is the improved fit :-



The chamfer at the end is not quite right. My DeWalt bits have a 110° angle but the standard appears to be 118°.

I decided to take a look at steel acorn nuts :-



These are a completely different animal. Rather than being machined out of one piece they consist of a nut with a dome pressed into it.



They have some advantages and disadvantages : -

• They are cheaper and more commonly available.
• They are smaller so less thermal mass.
• The dome is much thinner, about 0.4mm rather than 1mm, so it is easier to get a short hole.
• Steel has a much lower thermal conductivity than brass so plastic may cool down in the nozzle.
• Steel has a different thermal expansion rate than brass. Fortunately it is less so it should get tighter as the extruder warms up.
• The steel dome might spring out under the pressure of extrusion.
• The flat ended heater barrel goes in further but leaves voids at the side.

These are just the nuts I have managed to buy. I have no idea how much they vary from one supplier to another.

The heat insulator in the picture above is an experimental one turned from a bar of soapstone.

Extruder postmortem

So my extruder died at the beginning of the month. I have been busy with other things so I have only just got round to thinking about rebuilding it. Here is a list of things that failed :-

I got my extruder working in mid August. At the end of September it ejected its PTFE barrel from the clamp, breaking the heater wires. This is a common problem and stems from the fact that PTFE has the lowest coefficient of friction of any known solid material. I tightened it further and it didn't slip out again. However, when I came to dismantling it I noticed that the end that was in the clamp has been compressed by about 0.3mm. The 3.2mm drill that I made the hole with is now quite a tight fit so the hole may have shrunk slightly. I will drill it out to 3.3mm because some of my 3mm HDPE filament is slightly more than 3.2mm where it is a bit oval.

One curious thing is that it now looks to have a thread in the entry hole.



I am struggling to explain this. The only way I can think it may have happened is as follows :-

The filament has a thread cut into one side by the drive screw. Before I added my feed spool the filament used to rotate as it went through the extruder and had a thread cut all the way round. Perhaps it transfered its thread to the PTFE, which is quite soft.

The final failure mode was the nozzle jumping a couple of threads and leaking molten HDPE. It rammed the nozzle through the object being made and damaged the bed underneath.



The bottom end of the PTFE barrel has swollen by about 0.3mm. PTFE is known to creep, i.e. when subjected to a prolonged force it very slowly flows. It has no memory so it does not spring back when the pressure is removed. I think this is why the top of the barrel shrank and the bottom expanded. There is a lot of pressure in the barrel and it is close to PTFE's maximum operating temperature. PTFE melts at 327°C but it starts to degrade at 260°C. I have been running at 240°C which is a bit too close for comfort.

My barrel, at 12mm, is smaller than the current recommendation which is 16mm. This may have contributed to the failure.

The recommended solution is to fit a pipe clip, but I didn't have one small enough, so I pressed a short section of 15mm OD pipe over the end, which is a tight fit. It had the desired effect. When I first made the barrel the thread was a snug fit. When I dismantled it I noticed it had become quite sloppy. With the ring of metal in place it is now a tight fit. I will use some plumber's PTFE tape to seal it as well.



The other thing that was on the point of failing was the flexible drive shaft. Strands started breaking and the more that broke the more it flexed, so it was a kind of avalanche effect. I estimate that the shaft had rotated about 100,000 times so the flexing backwards and forwards must have caused metal fatigue. I probably have the most mileage on this part of anybody so far so it could be a sign of a design flaw. There are a couple of problems with my implementation which certainly won't have helped.

The first is that I used some 2.5mm cable I had to hand rather than the 3mm recommended. It doesn't seem like a big difference but I think the rotary strength is at least a cube law which would make my cable roughly half as strong. It was left over from a garage door installation but I don't think I have any worries there as 100,000 flexes is about 75 years use!

The other contributory factor is that the bearing lands on my drive shaft are a little bit eccentric. This stems from the fact that my watchmaker's lathe is not really big enough for this work. Fundamentally the hole through the headstock is not big enough to take the 5mm threaded rod.

Each time the shaft rotates it opens and closes the pump halves a little. This makes a big torque variation over a revolution because of the strong springs holding it closed. That caused the cable to wind up and unwind a little each revolution. It actually modulated the filament width and gave the objects a basket weave appearance. Here is a good example :-



I bought some new parts from www.bitsfrombytes.com, which wasn't an option when I first made the extruder.



That will allow me to give the flexible drive a fair test. If it proves to be unreliable then I will switch to direct drive. I found out from the core team that it is not required for any of the polymers currently used, only things like Field's metal.

Another thing that was starting to fail was the J-B Weld holding the heater wire to the nozzle. It is supposed to be rated to 315°C but it had started to crumble with my extruder running at 240C. The other problem I had with my J-B Weld is that it does not cure in the specified time at room temperature. I have to bake it to make it strong. I emailed J-B Weld but got no response apart from an automated reply.

David commented on my last post suggesting Thermosteel ,which is good for 1300ºC, so I will try that next. A couple of the core team use BBQ paint which handles 600°C so I will try that as well.

If it's not one thing it's another!

The third post in a row about my extruder breaking, not a good start to the year!

Now that I am making solid test shapes rather than hollow ones the extruder is working a lot harder and all it seems to do is break down. The drive cable started to disintegrate this afternoon, I could hear the strands breaking :-



It was still limping on however when this happened :-



The brass nozzle started to come out of the PTFE heat barrier. This was ironic because it was only yesterday that I said I had not had this problem in a forum discussion. Others have had it happen and the collective wisdom is to use a pipe clip round the end of the PTFE to secure it.

And the JB Weld which insulates the heater wire has turned to dust :-




So some rebuilding to do!

Running repairs

No sooner than I had fixed my heater, the extruder motor failed!

I bodged the heater connection by putting some more solder on it. It's not a permanent solution because the solder is molten while the heater is on so it slowly oxidizes away. The last time bodged it that way it lasted six months though. It really needs a crimped connection.

The GM3 motor failed by running slowly, getting very hot and drawing lots of current. It eventually caused the protected MOSFET that is driving it to shut down. Opening it up soon revealed how it had failed :-



It has two pairs of copper brushes. Three of them have holes worn right through and the fourth has broken off. Its stub was touching the wrong side of the commutator, causing a short.

More expensive motors have carbon blocks on the end of arms which can wear down a lot further before they fail. Bigger motors have spring loaded carbon rods. The gearbox shows no sign of wear so it is let down by the cheap motor.

This motor is not really up to the job of driving the extruder. It is being severely abused by running it from 12V PWM when it is only rated at 6V. I anticipated it would not last long and ordered a spare when I bought it. I fitted that and HydraRaptor is up and running again. Curiously the second motor seems a lot quieter than the first.

At some point I think I will upgrade to a stepper motor. They are more expensive but, as long as you don't load the bearings, they last virtually forever. In the long run they probably work out cheaper and I can also dispense with the shaft encoder and the interference suppressor.

Wear and tear

My extruder's heater went open circuit so I removed the heat shield to have a look at it. I have actually run it for many hours now and have extruded quite a lot of HDPE. I have about 200g of extruded test objects and scrap which represents about 13 hours operation. I only recently started saving my scrap so I must have extruded a lot more. The 2.5Kg reel of HDPE is noticeably smaller.

The heater has also run for a lot longer than the extruder has been extruding. I got fed up of waiting for it to warm up at the start of each run so my host software leaves it on. I keep meaning to put a timeout in the firmware to turn it off when there hasn't been any Ethernet messages for a while as I have left it on for long periods a few times.

The extruder is starting to show some signs of aging. The plastic shield which keeps the fan draft away from the nozzle looked like this when I made it :-



But now it looks like this :-



The nozzle itself now looks like this :-



The JBWeld that surrounds the heater wire has gone very dark and has several cracks in it. One of the heater connections broke off in a previous accident so I dug it out and joined a piece of copper wire by squeezing it tight and soldering it. There is now no sign of the solder which is why it has gone open circuit.

The black stuff which looks like bitumen must be slow cooked HDPE. I am surprised that long term heating to 240°C causes it to decompose. I don't know if the white surface on the shield is just due to its surface melting a bit or whether something boiled off the nozzle and condensed onto it or reacted with it.

Even the high temp insulation over the thermistor wires is starting to look a bit sad!

I also noticed that the steel wire that forms the flexible drive coupling is starting to break up. A couple of strands have snapped and there is a pile of black dust on top of the pump shell.



The heater connection should be easy to fix. I have a few planned improvements to make to the extruder but I will wait till parts wear out before replacing them with better ones to get the most use of it.

To raft or not to raft?

When extruding HDPE onto foam board a raft needs to be laid down first to increase the anchorage at the corners to reduce curling. It becomes part of the object and has to be trimmed back to its outline with scissors or a knife. Now that I am extruding onto polypropylene cutting board I wondered if it was still necessary.

The temperature at which I lay down HDPE onto the cutting board is important. At 180°C it does not stick. At 200°C it sticks well but can be peeled off with the help of a penknife. Higher temperatures make it harder to remove and do more damage to the board.

Here are a couple of 15mm test cubes made directly onto the PP board without a raft :-



The one on the left had the first layer extruded at 200°C and subsequent layers at 240°C. As you can see it curled badly, particularly at one corner. The one on the right had its first layer extruded at 220°C. It looked promising but when I tried my standard warp test block the result was not good!



So it looks like the raft is here to stay. Here is an example :-



I lay down the raft at 4mm/s with a notional filament diameter of 1.1mm with the extruder head 1.3mm above the board. This is to get the filament as round as possible so that it doesn't form a solid weld. In actual fact, gravity causes it to slump to about 0.9mm high and spread to 1.3mm wide. The oval area calculation would give 1.34mm and a pitch of 1.3mm is sufficient to get adjacent filaments to stick together. My rationale for making the raft as thick as possible in one layer was to make it strong without taking too much time. It probably does not need to be as strong now that it binds to the PP.

I put the raft down at 200°C, then I do the first layer of the object at 240°C with the fan off to ensure it welds to the raft and then subsequent layers at 240°C with the fan on.

I calculate the amount the raft overlaps the object with this completely arbitrary function :-

def overlap(x):
  return x + 10 + 10.0 * (x - 20) / 80

I halved the overlap when I went from foam board to polypropylene.

Cutting corners

When making solid blocks with 0.5mm HDPE filament I noticed that the corners are not very accurate. The right hand edge of the 20mm cube below shows this effect at its worst :-



The problem is that, although the machine makes a perfect right angle, the filament appears to have a minimum bend radius and so cuts the corner. The amount it cuts the corner seems to vary from layer to layer giving rise to the rough edge.

I think the variation is due to the fact that my extruder spindle is a bit off centre. This causes the torque to go up and down as it rotates, which causes the flexible drive cable to wind up and run down again. This causes speed variations despite the fact that the motor speed is well regulated. At some point I will get rid of the flexible drive.

I expect the fact that I am stretching the filament doesn't help with the corner cutting. I improved it a lot by slowing down the drawing of the outline to 4mm/s and leaving the infill at 16mm/s. Here is the result :-



Still not perfect, another thing to try would be to recognise that there is a minimum corner radius and make the nozzle follow an arc of that radius around the corner. At least that way it might be more uniform.

Here is a close up of the top face taken with a scanner:-



As it goes round the corner the filament has an external radius of about 1.5mm and an internal radius of 0.9mm. As it is 0.6mm wide that is probably not bad. You can also see that the zigzag infill sometimes ends a bit short of the edge, probably also due to corner cutting.

To get sharper corners I expect I need to use a nozzle with a smaller hole, so that the filament can be fine without having to be stretched, but that has the disadvantage of slowing down the extrusion rate for a given pressure.

Warped

Having got an idea of the HDPE warping for thin walled open boxes, I decided to start investigating solid shapes. I made a solid block 40 x 10 x 20mm to compare with the open boxes of the same dimensions.



Obviously there are many ways to fill the interior so I started with the simplest, just alternate layers of horizontal and vertical zigzags. HydraRaptor seems quite happy extruding 0.5mm diameter filament at 16 mm / second. If extruded into free air it would actually be 1mm at 4mm/s, but that is too course, so I move the head at 16mm/s which stretches it.

From trial and error I have found that a good layer height to use is 0.8 times the notional filament diameter. If it is more, then as the lower layers shrink, the nozzle rises faster than the object and a gap develops. Once that happens the filament squirms about and does not follow the path of the nozzle accurately.

So the extruded filament is constrained to 0.4mm high. Measurements show the width to be about 0.6mm. Incidentally, if it squashed to a perfect ellipse with a height of 0.4mm then it would be 0.625mm wide to have the same area as a 0.5mm circle. I extrude the zigzag with a pitch of 0.6mm so that adjacent filaments touch, but it means the object is not actually completely solid. The space occupied by each filament is a rectangular channel 0.4 x 0.6 = 0.24mm² but the cross sectional area of the plastic is π x 0.25² = 0.20mm², so about 18% is air. I confirmed this by weighing the block. It weighs 6.5g but if it was solid HDPE then 8ml would weight about 8g. It takes about 45 minutes to make the object including laying a raft.

Before I tried it, I always imagined the amount of plastic deposited would have to exactly match the volume of the extruded object otherwise it would sag or bulge. I could never understand how FDM worked reliably. Now I know that the volume can be a bit less and the difference is made up by air. That means the amount of plastic deposited is actually not that critical, which is why RepRap can get away with an open loop extruder.

I measured the warping with the three nail jig that I showed in the last post. The thin walled box is warped 0.44mm and the solid box has warped 0.87mm so that answers the question whether solid objects warp more or less. Note that the thin walled box is made with 1mm filament because 0.5mm filament is too thin to be self supporting.

I expect I can make a less warped block by extruding a thick base and then a less dense infill above that. Something else to try.

It is amazing how strong 10mm thick HDPE is. You don't often get to see plastics in that form. Most end products have optimised strength against cost by having thin walls and ribs etc.

Chopping up chopping boards

Up until now I have been extruding HDPE onto foam board because it was the only thing that it sticks to well enough. However, it has a couple of failings: It is not strong enough to completely resist the warping caused by the HDPE and it is not reusable because the surface gets ripped off.

I have tried many other surfaces including various woods and metals (with and without primer), melamine and several other types of foam board but nothing worked. Obviously HDPE sticks to HDPE so I decided to investigate that further.

My first idea was to use a thin sheet of HDPE cut from a milk bottle. This makes a nice surface to extrude onto but the problem is holding it down. I first stuck it down with double sided tape but the heat melts the glue. Sticking it to a sheet of aluminium to take the heat away improved matters and I was able to get slightly less warping than with foam board.



To compare the warping on different base materials I made a test shape that is a 40mm x 10mm x 20mm open box with 1mm walls and measured how much the corners lift using a simple jig.



With foam board I was getting 0.83mm lift between corners and the middle. With HDPE stuck to aluminium I got 0.76mm. Not much better because the glue of the sticky tape stretches under the curling force.

I needed a thick HDPE base and I had heard that plastic kitchen chopping boards are made from HDPE. I bought a new one from ASDA which looks like this :-



It is 5mm thick, opaque and quite rigid. I realised it was very different from the other chopping boards we have which I think came from IKEA.



These are 10mm thick and made from a softer, more translucent plastic. To find out which was HDPE I used the flow chart on this website www.texloc.com/ztextonly/clplasticid.htm. I concluded the thin hard one from ASDA is HDPE and the thicker softer one from IKEA is PP. HDPE seems to stick equally well to both of them but the HDPE one warped a bit when it was only held down with masking tape, so I decided to go with the PP one. I cut it up and bolted it down to my XY-table. It was a bit curved due to years of dishwasher use but bolting it down pulled it flat.



Surprisingly, if I lay down a raft at 200°C it sticks well but can be easily prized off again with a penknife. The board is marked slightly but it can be reused over and over again.



I extrude the object at 240°C so that it welds to the raft and itself, and I turn the fan on after the first layer so that the object cools to room temp as fast as possible.

The board is strong enough to hold the object completely flat while it is attached but when it is removed it does still curl a bit. I measured 0.44mm on my jig so that is about half the curling I was getting with foam board. Other than extruding onto a convex surface, I think that is the best that can be achieved for that shape with HDPE at room temperature. Here are the three tests side by side :-



Next I will look at different solid shapes to see if they warp more or less.

Extruder dimensions

I have been asked for dimensioned drawings of the extruder. I made these by manually inspecting the 3D models in ArtOfIllusion. It is not the easiest application for extracting dimensions so I made 2D drawings in Visio which does have good dimensioning tools. I then made Python scrips to do the milling. My dimensions may differ in places but I did make the extruder from these drawings and it does work. I tightened up some of the hole clearances because my milling machine holds much tighter tolerance than FDM.

This is the motor shaft coupler. I adjusted the slot to suit my GM3 motor. I think there are now two versions of this part. The official design is tapered but this is not necessary with the offset motor mount so I simplified it to a cylinder.



Here is the finished article milled with a 2.22 mm bit. The step on the outside and in the shaft slot are there because my milling tool's shaft is wider than the bit, so to go deeper than 9mm I need to have some clearance. The material is some sort of metal loaded resin.



Here is the clamp drawing I used. It has now been superseded by a larger design. Note that I adjusted the hole for the PTFE to suit my 12mm rod. I think the official design was 10mm but is now 16mm. I also widened the slot to allow the 2.2mm milling tool to get in and added some extra mounting holes to suit my machine.


I milled it from 9mm Delrin.



Here is the pump drawing :-



The poly channel on the official version slopes outwards at the entry but that is only needed for the version without the offset motor.

And here is the milled version :-



The material I used is not as slippery as CAPA so, to reduce friction in the channel, I smoothed it with emery paper, polished it with metal polish and sprayed it with PTFE dry film spray.

I split the motor mount into three pieces for milling from a sheet of 5mm perspex. I fixed the pieces together with M2.5 screws, tapped into the perspex.





If anybody wants the Visio source file it is here:- forums.reprap.org

Stretching a point

In his article: x-idler-bracket-continued Vik Olliver alluded to the fact that you can extrude filament with a smaller diameter than the hole in the nozzle. I did some experiments to see how fine I could go. In fact the final filament diameter is simply determined by the feed rate of the extruder and the travel rate of the nozzle, or in my case the bed. The filament stretches to the length that matches the rate of travel while it is still liquid. You can then calculate the mean diameter from the volume of material extruded. The nozzle height has to be a bit less than that mean diameter and then the width becomes a bit wider.

Here are three 20 x 20 x 20 open cubes with different wall thicknesses :-



The first was 1mm diameter filament extruded at 4mm per second with a height of 0.8mm giving a wall thickness of about 1.2mm.

The second was the same feed rate but with the extruder traveling over the bed at 16mm per second to give 0.5mm filament, the same as the nozzle hole diameter. The height was set to 0.4mm giving a wall thickness of about 0.6mm. As you can see it warps more but I expect it would behave if it was building a solid object. The bottom layer which was stuck to the table has better corner definition.

The third attempt was 0.35mm filament extruded at 16mm per second with a hight of 0.28mm and a width of about 0.5mm. As you can see holes started appearing but I think that was just because the sides buckled so badly. Interestingly the holes can be bridged by filament above that needs no support. Again, I think this would be OK making solid objects, or at least objects with thicker walls.

This is really good news as it means I can get down to the sort of resolution commercial machines get (0.25mm) without having to have a very small nozzle aperture, which would limit the flow rate. It remains to be seen what effect stretching has on the polymer but as it is still liquid at that point I think it wont increase the contraction much, if at all. It does mean I need very fast head movement to keep up the deposition rate, about 64mm per second. I think my machine will do that if I reconfigure the steppers for speed rather than torque, a simple one wire change.