High speed, high volume flow rate printing.

  • I've been doing some more work on high speed, high volume flow rate, printing.

    I've managed quite decent largish prints at 300mm/sec with a 0.5mm nozzle and 0.3 mm layer height. That equates to a volume flow rate of around 36mm^3/sec (if my maths serve me correctly. I've also attained a speed of around 100mm/sec with a 0.9mm nozzle and 0.6mm layer height which equates to a volume flow rate of around 54 mm^3/sec.

    I've covered many aspects, including use of pressure advance etc. Be warned, it's a long post. Link here https://somei3deas.wordpress.com/2018/10/14/real-3d-printing-at-high-speeds-and-even-higher-melt-rates-with-a-large-nozzle/

    There is also a video. It's about 25 minutes long but the first 12 minutes are probably the most interesting. https://www.youtube.com/watch?v=rUV5IZxfAxU

  • @deckingman
    I would like to draw your attention to a very special PLA filament, whose thermal properties have been modified with diamond nanoparticles, Carbodeon's uDiamond. This filament doubles, at least, the extrusion rate for a given hot-end/nozzle.

    I experienced this during mechanical speed tests (and not extrusion flow test), to check the behaviour of my Zatsit hinged delta at high speed, with regard to ringing, and the speed/print quality ratio.

    The hot-end I use usually delivers, with a standard 0.4mm nozzle, about 8mm3/s. With the uDiamond PLA, the throughput exceeded 15mm3/s, which allowed the printer to be tested at up to 560mm/s, with standard Hot-End and nozzle.

    This shows the huge influence of the filament itself, for questions of extrusion throughput.

  • @zatsit Interesting. I'd be a bit concerned about the abrasive nature of having diamond nanoparticles passing through my brass nozzle though. Did you measure the nozzle diameter after using it? Was the reason why it increased the volume flow rate because your 0.4mm nozzle is now a 0.5mm nozzle? ☺

  • administrators

    Ian - thanks a very good detailed analysis. Makes the case for a variable temperature based on extrusion volume feature in (probably) the slicer. Tune the hotend extrusion rates to the filament and then modulate the hotend temperature with suitable look ahead to achieve the maximum flow rate, without under extrusion, or cooking.

  • @t3p3tony said in High speed, high volume flow rate printing.:

    Ian - thanks a very good detailed analysis. Makes the case for a variable temperature based on extrusion volume feature in (probably) the slicer. Tune the hotend extrusion rates to the filament and then modulate the hotend temperature with suitable look ahead to achieve the maximum flow rate, without under extrusion, or cooking.

    Hi Tony,

    Yes, probably - possibly. BUT........

    There are diminishing returns as you push the speed higher. For the reasons that I set out in that post, you can't push the acceleration too high. So as you increase the maximum speed, the constant speed phase becomes less significant as it takes progressively longer to accelerate and decelerate to and from that speed. So using "normal" temperatures at 120mm/sec max speed, the overall print time was 4 hrs 19 minutes (at say 60 m/sec it would have been a lot longer). Doubling the speed to 240 mm/sec but keeping the temperature the same, reduced the overall print time to 3 hrs 02 minutes. So saving over an hour or about 30% in time. But then raising the temperature in order to achieve an even higher volume flow rate and allow printing at 300mm/sec maximum, reduced the print time down to 2 hrs 52 minutes. So saving another 10 minutes or less than 3% of the overall time.

    What I'm trying to say is that raising the temperature to achieve that bit higher volume flow rate, has minimal impact on the overall print time compared to the print time that can be achieved with normal temperatures. So I would question whether it is worth the time and effort to develop some sort of look ahead algorithm to vary the temperature in relation to volume flow rate, which is likely to only reduce print time by about 3% or 10 minutes for a 3 hour print.

  • @deckingman Thank you for the detailed blog post and video Ian. Always appreciate your work.

    I think the huge melt chamber certainly helps you sustain the high speeds at normal temperatures. My regular V6 hotend is not nearly as forgiving. I definitely have to boost the temp beyond normal to really push volume. Failing to do so leads to the extrusion being kind of cloudy and it doesn't adhere well. Raising temp solves that, but introduces its own problems. ie. increased stringing, poor overhangs and bridging, etc.

    Whether or not an adaptable heater temp could help, I'm not sure. Would depend on the responsiveness of the heater. I wonder if the copper block and nozzle from E3d would help in this for those without a large melt zone.

    Like you, I am using low acceleration, and high top speeds. While it's true you don't always hit the top speed, this is actually desirable I think, because it allows for adapting to small short sections. Since the top speed isn't reached in those sections the printer doesn't shake itself to pieces, but on wider areas it can stretch its legs.

    One question that arose for me while reading, was what is your extrusion width usually set to? Is it always equal to the nozzle width? Recently I've been experimenting with much wider extrusion widths. I used to always use the nozzle width, and had fine results, but Slic3r has defaults of around 1.125x the nozzle width, which for my .6 nozzle is .675. Print times are reduced and quality seems about the same. One issue is getting the larger circumference into the corners for top fill. I've managed to extrude a 0.5 layer height 1mm extrusion width vase mode that looked quite good and solid. I'm wondering if you've done any experiments with regards to that.

    Thanks again for the excellent write up.

  • Overtemp lookahead is physically impossible for complex print models, eg a print that alternates between a voronoi islands region and a boxy region. The delay between heater action and temp sensing (or heat reaching filament) is quite large, on the order of 3-10 seconds maybe, while extrusion flow rates change meaningfully on the order of tenths of seconds. There’s no way to vary hot end temp as fast as you want to vary flow rate. Sure, you can crank the heater power up on long extrusion runs to compensate for the heat leaving the hot end through the filament, which will help maintain constant temp, but you can’t really raise the hot end temp for extrusion then drop it back down for every travel move. Not very effectively, anyway.

  • @phaedrux Thanks for the feedback - always nice to know that the time and effort I put in to this stuff is appreciated.

    Ref melt zone. It's not just the fact that the combined surface area of 3 melt chambers is comparatively huge, it's also the fact that there are 3 of them. So each filament spends 3 times as long in the melt zone compared to a single melt chamber. So at 54mm^3/sec volume flow rate the feed rate for a single 1.75mm diameter filament would be around 20mm/s but using 3 melt chambers, that feed rate drops to around 7 mm/sec. So as well as the bigger surface area, we also have a longer melt time. That helps because filament itself is a poor conductor, so it takes time for the heat to find its way from the outer surface to the core. Also, with a single melt chamber, the extruder itself might start to struggle at those sort of feed rates.
    Ref accelerations, it's good to come across a like minded person. Apart from anything else, it's a hell of a good way to limit the speed of those short zigzag infill moves as you move at 45 degrees into the corner of a rectangle.
    Ref extrusion width. Up to a few months ago, I always used to set the width to the same as the nozzle diameter, for no better reason than that's what RepRapPro did when I built my first printer from one of their kits. Recently I've been using the default value and letting Slic3R do it's own thing. I need to do more testing but at the moment, the jury is still out as to whether it makes much difference either way.

  • @deckingman
    This filament containing nanodiamonds does not wear out the nozzles, because firstly these particles are practically round, and secondly their size is homogeneous around 10 nanometers (supplier's data).

  • @zatsit Yes, sorry. I have since looked at their web site and seen their claims about nozzle wear.

    I do note that they recommend a print temperature of quote "Nozzle temperature 220°C - 250°C (depending on speed)". That seems a little contradictory to their claims of enhanced thermal conductivity and improved lubrication. - i.e. one still has to increase the temperature in order to achieve higher print speeds, which ought not to be the case if those claims are true.

    I try to keep an open mind but age has taught me to be cautious and sceptical. I apologise if this comes across as being negative. I might give this filament a try some time (when I'm feeling wealthy).

  • @deckingman I think there are a few benefits to increased extrusion width.

    Firstly, it reduces the total number of print moves by covering more area per pass. This reduces the number of direction changes, and therefore slow downs. So you're in the peak print speed zone more of the time.

    Secondly, and I could be wrong, but it would seem that the increased pressure out of the nozzle and the smoothing effect of the nozzle tip on the hot plastic could increase layer bondage. Now this increased nozzle contact may be undesirable for some sticky plastics that tend to gum up on the nozzle, like PETG, or filament containing fiber fill.

    With such a high melt throughput I would be interested to see what kind of max layer width you could get with the 0.9 nozzle size. Even if speed isn't 300mm/s, the print will complete faster if your sustained melt rate is still just as high. Average melt rate would actually make a decent way to estimate print times in general. I think you do just that in your post.

    At any rate, I look forward to your next write up.

  • @phaedrux I think you must have misunderstood me somewhere. Apologies if I didn't make myself clear. What I meant was that I don't see much difference between setting the extrusion width to the nozzle diameter vs letting the slicer decide on the extrusion width, where the difference might be between say 0.5mm and 0.52mm.

    I am however a huge fan of using a 0.5mm nozzle over a 0.4mm nozzle. (By default, the Diamonds come in 0.4mm but RepRap.me will make whatever size you ask for).
    0.1mm in width isn't much but it's a 50% increase in area which is huge. You don't loose all that much in terms of detail but you gain a hell of lot in terms of flow rate, reduced pressure, increased speed, better inter layer adhesion and all the other stuff you mentioned.

    But, you rightly say that with the 0.9mm nozzle, I could probably push that to say 1.1mm which might be a worthwhile exercise at some time.

    I'll have to wait until my next pension payment though - do you know how quickly you get get through filament with a 0.9mm nozzle at 100mm/sec? ☺ Watching the reels of filament go round is like watching the wheels on a slot machine - and your money disappears just as fast.☺

  • I have yet to see any really convincing argument on extrusion width vs nozzle width. You can see a difference under a microscope in strand shape / squash if you under-extrude, but it’s not clear that there’s much advantage to one or the other. The EngineerDog guy did some strength tests and found that wider strands from the same nozzle made weaker prints (all else being equal) but his data was noisy enough that I’m not sure I trust his methodology to produce representative results.

    Slic3r’s volume calibration does assume your extrusion width is at least [nozzle size + layer height] so the single-wall strand develops a full squashed oval shape. It’ll be off a bit if your extrusion width is smaller than that. Still shouldn’t make much difference in practice though; most people can’t tell the difference between “correct” volume and -10% volume anyway.

  • @rcarlyle I've been using the Prusa slic3r profiles for the MK2 using 0.6 nozzles from their GitHub that varies the extrusion width between 0.6 and 0.68 or so for different moves. This is different from just leaving it 0 and letting it decide. Print times are a little less. Infill is a bit more solid. Print quality has been good.

    I would think that if part strength is less perhaps it's due to trying to push too much plastic maybe? Purely anecdotal on my part as I haven't done any destructive testing but I'd say the parts feel stronger. The infill especially even looks stronger.

  • As an aside, I haven't yet thrown out those test parts because I want to devise some method of testing their relative strengths. I'd have thought the layer boundaries would be the weak points. I also suspect that higher speeds might lead to stronger parts. My reasoning is that at higher speeds the volume flow rate is higher, and layer times are lower. Therefore, the previous layer will not have cooled so much so the next layer should fuse to it better (maybe). Anyway, I'm open to suggestions as to how best to test that - given the resources I have available (like a big hammers and vices)☺ I could crush them in a vice and measure the distance between the vice jaws at the point where they break. Not very scientific but it might provide some sort of comparative data.

  • @deckingman simplest quantitative test protocol is to print some kind of beam shape with a notch in it, clamp it to a table, and hang weight off the beam until it yields or parts.

Log in to reply