Sunday, 10 January 2010

Golden wonder

My first attempt at extruding ABS onto hot Kapton had "all the stops pulled out" to make it stick, i.e. 120°C bed, nozzle height 0.1mm too low, very slow outline and infill on the first layer (4mm/s). The adhesion was very good so I decided to back off a bit. It is not a good idea to change more than one thing at a time but I did anyway. I got rid of the -0.1mm Z offset and sped up the first layer infill to 32mm/s, leaving the outline at 4mm/s. I also dropped the bed temperature to 80°C. That was too low, the corners lifted about 1mm during the build, but I think the part will still be usable.

The base is still glossy but you can see and feel some valleys between the extrusion "lanes". The next test was a binary chop with the bed at 100°C.

This is perfectly flat, even when off the bed for a day, but the extrusion lanes are still noticeable. The next test was at 110°C.

The extrusion lanes are gone in most places but a few are just visible. The first one that I did at 120°C has no extrusion lanes on it all, just some very slight graining from the Kapton tape that you can also see on the picture above. The tape lines and grain go from bottom right to top left. The extrusion infill slopes bottom left to top right and is only visible on the right hand side of the object. I think perhaps Z has to be a bit lower to get rid of them completely, but it is only important if you want to make something aesthetic, like an instrument panel, for example.

Of course there are the tape join marks. I used unbranded polyimide as it seems to be about half the price of branded Kapton. I got it from here, which is very cheap and free shipping if you don't mind waiting a while. You can get polyimide tape up to 250mm wide, but it is always on a 33m roll, so it gets very expensive. I have ordered a 150mm roll to cover the working area of HydraRaptor's build table. It was £53.71 from here, so very expensive, but a small price to pay for perfection! I don't know when it will arrive as post is a nightmare at the moment. I am still waiting for things from the 17th of December. Parcels are not being delivered because of the snow, so you have to go and collect them, but several letters and packets seem to have disappeared.

Here are all the tests side by side, notice the colour change with temperature, it is a bit exaggerated on the photo : -

I now have a full set of Mendel vertexes including two that I made in PLA that warped slightly (on a cold bed). I moved onto something more ambitious on the warping front: the Mendel x-carriage-lower_1off part. I don't think this is printable in ABS without a heated platform, or air stream, unless you use the apron method developed by Forrest Higgs. For this test I started the bed at 120°C and dropped it to 100°C after the first layer. The logic being that 100°C seems to be enough to prevent warping, but 120°C is needed to get a perfectly smooth finish. It takes a few layers before the temperature has dropped to 100°C as I don't wait for the plate to cool down.

Unfortunately the ancient version of Skeinforge that I use gets one layer wrong on this part. The layer has the central hole missing. The filament didn't span the void very well as it is a very big void, I have no fan running and there is a lot of heat rising from the bed. That caused some filament to stick up and collide with the head. It spun round 90° unscrewing it 1/4 of a turn. Amazingly it did not leak but the nozzle hole must be slightly off centre with respect to the barrel thread, so I got an offset in X and Y above the layer that went wrong. Still, the objective was to test warping and it came out totally flat.

The corners have a dimple that looks like an air bubble, but must be something to do with them trying to lift I think. Apart from these the base is as flat as glass and had it not been for the Skeinforge bug it would have been usable straight off the bed. I cut the membrane out with a knife and drilled through the blinded holes before taking these pictures.

I tried the x-carriage-upper_1off starting the bed at 120°C for the first layer and dropping to 90°C. Again Skeinforge got it wrong, not surprising as the topology is very similar. This time I also dropped the filament temperature to 220°C, so it spanned better and the head did not get spun. A longer snout on the nozzle might be a good idea to avoid collisions with build defects.

Again here it is with the membrane removed.

The corners lifted very slightly but the rest of the base is completely flat. It doesn't rock on a flat surface like an object made on a cold bed would. In fact, the raised corners made it easier to remove from the bed.

So it looks like 100°C bed temperature is the minimum to prevent warping when using Kapton. 120°C for the first layer gives a better aesthetic finish, perhaps with a small negative z-offset. Having the object kept warm seems to allow a lower filament temperature without losing strength. I used to build at 240°C and use 0.5mm for stronger objects. I can now use 220°C and 0.4mm with no sign of de-lamination so far. The lower temperature is good because the ABS out-gasses less and so smells less.

I can't recommend Kapton on heated aluminium highly enough. It has transformed my experience building with ABS completely. I no longer need a raft, which saves a lot of plastic, time and labour to remove it. My objects can be completely flat, smooth and glossy. Together with using a geared stepper extruder drive to completely eliminate ooze it means I just print an object, remove it from the bed and it is ready to use. There is a slight meniscus of plastic around the base, which you might want to remove with a file or a knife.

It has several advantages over acrylic: -
  • Acrylic is a good insulator, so even 3mm reduces the surface temperature by about 15°C, making it take longer to warm up and harder to control.
  • It tends to warp as it has a similar glass transition temperature to ABS.
  • It can be hard to remove the object as it can be permanently welded if you deposit the ABS hot enough.
The way ABS sticks to hot Kapton is different. The Kapton does not melt at all so you don't get a weld no matter how hot the ABS is. I don't know what sort of bond it makes, but it is always peel-able.

While I have been writing this article a friend came up with a brilliant suggestion. Why not use non-adhesive Kapton film, clamp it on the table, possibly with a vacuum? When the build is finished just release it so it can be peeled off the object with ease. I realised that would enable a conveyor belt table to be made. People have suggested this would allow a machine to churn out parts unattended. E.g., stretch a band of Kapton over a heated plate and rotate it when the object is finished and has cooled. The object will then drop off the end.

I still have a couple of problems to solve with the heated bed. The heat spreads downwards and warms my X-Y table. It is not much, I haven't measured it but I would guess to mid 40's C. That is enough to expand the aluminium that the table is made from and open up a gap in the ways so that it has some play and starts rattling. I removed the foam-board to leave an air gap (the logic being that the movement of the table would generate some cooling airflow) and covered the top of the bed with aluminium foil to reflect the heat back. That helped, but not enough. I think I will need to blow cold air over the top of the table with a sheet of something like PTFE to cover the bottom of the heated bed.

The other issue is that having heat around the object rather than cold air blowing on it means that void spanning and overhangs don't work as well as they did. I think I need a jet of warmed air directed at the end of the nozzle to cool filament to freeze it quickly.


  1. wow great progress, as usual (:

  2. I think your going to have to get a better camera now.... ;)

    Nice work.

    I have been trying to see if I can find out how to thermal model a table but I didn't study thermodynamics. I had wonderd if I could learn about the temperature distribution on the table with some form of model.

    Would be cool to do it for a plastic part as well on the table.

    Still I have learnt a bit about thermodynamics now, thanks.

  3. wow. A ton of great info in this post. I can't wait to build a machine using what I've learned from reading this blog.

    I also have a background in CNC control, so I'm planing on using a different controller that will give me a lot of power for the logic you've described.

  4. i think you are on to somthing here, so far i think you have the best results.

  5. Hey, great!! I'm using acrylic on top of my heated bed, it allows you to use lower temp, which is great since my RepMan is made from acrylic (even 75°C heats the contruction a lot) But I absolutely will try the Kapton tape on top of Alu, its great :-) You can check my progress on and watch our printer live on ustream here

    Good work!!

    Can I ask you for your settings of "suck-back" feature at this speed? How many steps backwards your extruder goes? I'm using 5000 back and 500 forward at 32mm/s and 62RPM, but i still have blobs of fillament :-/

  6. Hi Josef,
    Yes Acrylic does have the advantage that it grips much stronger so you can use a lower temperature and it will still be able to resist the warping. Kapton has a very weak grip so the ABS has to be kept at a temperature where it is too soft to exert any warping force. That then makes it easier to remove though.

    The suck-back is currently set to 0.75mm (of feedstock)and the speed is equivalent to extruding 0.5mm filament at 128mm/s. I.e. much faster than the extrusion speed of 0.4mm filament at 32mm/s. I fast forward by exactly the same distance and then drop to the extrusion speed. There is a little backlash in my gearing so filament will not withdraw the full 0.75mm.

    Details of how I tuned it are here:


  7. > I also dropped the bed temperature to 80°C.
    > That was too low, the corners lifted about 1mm
    > during the build, but I think the part will
    > still be usable.

    Every distance on the frame setted using threaded rod. So there are almost no significance how much the object warped. I have some really warped frame-vertex and seems to go all axes nicely.

    > You can get polyimide tape up to 250mm wide,
    > but it is always on a 33m roll, so it gets
    > very expensive. I have ordered a 150mm roll
    > to cover the working area of HydraRaptor's
    > build table. It was £53.71 from here, so very
    > expensive, but a small price to pay for
    > perfection!

    Ehh, you was a bit impatient... I would definietly buy the 250mm wide one (80GBP). Im thinking to sticking to household folie:

    That way it could be chopped up to 150mx150mm pieces.

    Best regards,

  8. Chris, how thick is your Aluminium bed?
    Looks like 5mm thick or even more.

    Im thinking to this bed setup:
    3mm thick Aluminium bed or even 2. Soldering small rods to its back a cutting threads on it. So the power resistors could be mounted.

    Putting an another *thick* aluminium or mdf or acrylic sheet under it, and to maintain the equal distance (ie. avoid bending) between the two sheets I would use a grid of screws.

    The thick sheet would then connected on three points with the base sheet which is screwed to Mendel's four sliding parts. So the levelling could be made.

    A more advanced idea is reducing the sliding parts to three and make in a way that an M6 screw would point out from it. So no base sheeld would than required.

    Maybe modify it to accept linear bearings instead of 624 bearings, to better quality.

    So with my modification I would lessen the thermal capacity, that way I think the radiation too (or the radiation is only depend from the surface area?). Anyway it would help to your soldering project too.

    Also having an another sheet under the heated sheet would prevent any heat to go below.

    I think its a bit hard to manufacture, as soldering aluminium requires special equipment.

    Maybe some picture would help to better imagine it.


  9. Impressive results, as always! :-)

    We have found 1/4" thick cork sheets to be an excellent thermal barrier in an application similar to what you describe with the carriage under the heated bed, you might want to give that a try. It is resilient (for a porous material), cheap and easy to work, but the adhesive bonding cork particles might degrade in time if the temperature is too high. Didn't bother us because the cork was supported, but you either have to plan for this or be ready to tolerate some cork grains falling off.

  10. Hi Lazlo,
    I think cheapskate rather than impatient. 150mm is big enough for HydraRaptor for producing aesthetically perfect parts. On Darwin and Mendel I will use two strips so I will still be able to make parts that are 150mm wide and as long as the bed with no seams. That is a very large part indeed for FFF fabrication.

    Have you tried cling film, do it work?

    The bed is 6mm thick. Radiation depends on temperature, surface area and the emissivity of the material. Black things radiate better than shiny things. I don't think thickness has any effect apart from increasing the warm up time.

    Thanks for the tip about cork.

  11. Impressive results, do you ever need to replace the layer of Kapton tape on the bed or will it last for several prints?

  12. Chris - Excellent progress, the best yet!

    I'm sure if you wanted too people would buy these from you ?? ;-)

  13. Robert,
    Small objects can be plucked off by hand. Tall objects can be knocked off with a gentle tap from a hammer. There is no damage to the tape in these cases.

    Large low flat objects need prising with a knife under one corner. I have nicked the tape when doing that a couple of times, but it can be done with care.

    Large objects could have a notch in the base near a corner to make them easy to remove.

    Using non-adhesive film held down by a vacuum seems like the way to go for easy removal.

  14. > Have you tried cling film, do it work?

    No, not yet. I have ordered like two days ago the Kapton tape. But I will definietly try it.

    Best regards,

  15. > Using non-adhesive film held down by a vacuum seems like the way to go for easy removal.

    Vacuum tablet at such high temperature?


  16. Just some small holes through the AL plate, some channels milled in the back to join them up and a sheet of something over the channels, Kapton tape should be ideal here.

    The Kapton will not need much holding down and it will be completely sealed, so not much vacuum needed. Maybe something like a big syringe and a valve, or an aquarium pump like I used on my jukebox.

  17. Really very nice yet again.

    Just looking at the number of objects printed here are you now printing these much much faster than you were before?

    Is this due to your extrusion speed?

    Or is it the in fill % you are using?

    Rough times/timings on the various prints would be interesting.

  18. Not a lot faster.

    I printed Darwin with 0.5mm filament @16 mm/s, 25% fill.

    I am printing Mendel parts with 0.4mm @16mm/s with double outlines and 25% fill @ 32 mm/s. The reversing slows the build down quite a lot. I need to add an optimisation so does not do it when the move is very short.

    The vertex brackets take about 2 hours each.

    There is also nearly 10 minutes warm up time and similar cool down with fan at the end.

  19. In terms of labour though it is much less because I used to spend hours scraping off bits of raft and strings.

  20. There are no rafts now, which saves a lot of time as I had three raft layers the size of the bounding rectangle plus a few mm overlap. They could take up to an hour on large parts.

    Also I can build multiple parts one layer at a time now without getting strings between them. That means I will be able to fit far more on the bed and avoid multiple warm up and cool down times.

  21. I'd build a hot air jet but never mounted it to the machine. Going to try that again, your suggestions give me some more confidence that it will be worth the trouble!

  22. Yes the idea is to blow air at a similar temperature as the bed so that the plastic freezes quickly but is not cooled further and the bed is not cooled. Hopefully it can be made directed enough to not waste too much energy and doesn't cool or heat the extruder too much.

  23. Hi Chris

    Definitely looking more and more like commercial quality!

    Since you try a lot of things have you considered testing the approach that tofletcher used?,24119

  24. Hi Freds,
    No I didn't try it because I didn't think it would work for large objects, or a bed full of objects. It is easier and cheaper to make though and obviously works well up to medium sized items. Looks like you still need a raft though.

  25. Do I have it right that you are proposing using Kapton film to print on? I checked the price here and for 45 square feet (1 lb) is $244. Seems a bit dear.

  26. It does seem a bit pricey for film but still that works out at less than $5 for a Mendel bed. Film is easy to share, unlike tape. But we could also fold tape in half length wise to make a non-adhesive film and that could be shared.

    Possibly another high temperature film would work, perhaps polycarbonate.

  27. I guess I am a little bit puzzled as to why you don't just slap a piece of cheap acrylic sheet on top of your aluminum and print on that? Lots of other people seem to be doing that successfully. :-?

  28. I tried 3mm acrylic first but is lost 15C and inserted a big time lag so temperature control would be a bit more complicated. The temperature difference between top and bottom made it curl up so it needs to be held down with a frame, but I also had problems with the middle buckling upwards.

    It is much harder to remove the object. A fine line between it sticking enough to prevent warping and permanently welding. The surface of the acrylic gets damaged.

    With Kapton I can use a higher temperature such that there is virtually no warping force. That is fortunate because there is very little force holding down. The up side is that it is easy to remove and I get a perfectly flat, smooth and glossy base.

    It also makes a conveyor belt bed feasible. Because the base of the object is kept above the glass transition temp it doesn't need much force to hold it down, so we can deposit onto a belt of Kapton stretched over a hot plate with a vacuum to hold it down. When the object is cooled the belt can revolve and the object will peel off at the end. That means we can have a machine churning out parts 24/7 with only one item on the bed at a time, so no big risk of wasting parts if it breaks down.

    I think this is a major advance. I haven't heard of commercial FDM machines that can make smooth finish surfaces without post processing. I also haven't seen one that can remove objects from its build table and make more. In fact I haven't see one build without a raft.

  29. The 3mm acrylic supplied with the Rapman is useless. My experience is that you need at least 6mm. So you get a big deltaT across the thickness of the plastic then?

    As to alternative films, I checked and it appears that the nylon film that they use in oven roaster bags would be adequate to your needs and much cheaper than Kapton. I haven't found a supplier for that film in rolls yet, though.

  30. Yes I haven't tried it, but I assumed thicker would increase the temperature drop in proportion. The problem is then that the bottom becomes too hot for the acrylic. There is only about 20C or less between the two glass transitions.

    Yes other high temperature films may work. I don't know exactly what makes ABS in its rubbery state cling to polyimide. I intend to try PLA, PCL and HDPE to see if they will stick. I would like to try other films but holding them down might be tricky until my friend Paul (who made the Al plate for me) and I develop a heated vacuum table.

    The advantage of polyimide is that we can go up in temperature for higher melt point plastics. I expect some like PET will only work with a hotter bed for the same reasons as PMMA.

  31. This comment has been removed by the author.

  32. Hi Erik,
    The bed takes about 1kW/m^2, i.e. 40W. You need at least twice that for a reasonable warm up time and control.

  33. I see, thanks!

    I was thinking, can't you hairspray the bare aluminium to increase its stickiness? I did that with my MDF base in the early days. It's cheap and ubiquitous...

  34. NopHead, I just follow your steps and I got the same results :-)

    See more here:

    Thanks for sharing ;-)

  35. I stumbled onto this blog while searching for a photo of a reel of smt components. What incredible and inspiring work! Unbelievable!

  36. Commercial machines use an entire heated build envelope and they build on a Polycarb sheet sucked to the inch thick aluminum platform. The sheet gets soft and seals onto the platform and makes a semi melted surface to build onto with a raft and then a bed of support material. I think the heated platform is doing what the heated envelope does and the kapton sheet is a really good idea on a heated platform.

  37. Hi Brad,
    I wonder if anybody from Stratasys reads this blog. Perhaps their next machine will use Kapton and not need rafts. Or extrude onto a Kapton conveyor belt and dispense the part when it has finished. ;-)

  38. Soldering aluminum is actually not too big a deal you do not need to tig it. Instead take your AL plate and put it across your stove top. Place your parts you want to solder on the aluminum and let areas come up to 900°F. You can get a special casting aluminum at any welding supply store that liquefies at 900°F and is in a 3/16" rod. That is not hard to do, I solder pieces of thick aluminum foil all the time whit that method. My problem is burning through the foil. A thicker piece will be harder to manage but you don't have to worry about being too hot. This solder rod is cool, it acts just like solder but when cooled, bonds very well and is very hard. You know you are up to temp when the rod starts to liquefy. Don't go any higher it actually changes the chemistry of the rod and all you get a white and black flakes.

    Hope that helps.