Diamond hot end potential print speeds and pressure advance testing



  • Herewith a link to my latest blog post https://somei3deas.wordpress.com/2017/06/22/exploration-of-print-speeds-with-a-diamond-hot-end/

    The results may surprise you (they did me).

    Warning. The associated YouTube video is 27 minutes long (but worth watching IMO).

    Ian

    EDIT. Title edited to include pressure advance testing. Further links are in a later post (page 2)



  • Interesting read I haven't watched the video yet but will try to I'd like to see it. I have a volcano clone currently sporting a 0.4mm nozzle but I have 0.6,0.8,1.0 and 1.2 I run that direct from a titan, I'll see what it can do. Can you post the object I can print 300x200 so might try it at 90% scale.

    What wattage is your heater cartridge? Or perhaps what resistance across it and voltage do you run it on, might be more accurate?



  • Hi Simon,

    The object was just a simple cuboid, 300mm in X, 30mm in Y and 30mm in Z. That's 300mm long, 30mm wide and 30mm tall. Nothing special at all and if you can print 300x200 then you'll have no difficulty. Sliced at 100% infill and various speeds.

    Heater is 40 Watt. About 14 Ohm at 24v (so 41 Watt). I didn't notice any drop in temperature, either when gradually increasing the speed or at the highest filament flow rate.

    Try and find time to watch the video and if you get chance, it'll be interesting to see what a Volcano can do as a comparison. Now if E3D did a 3 input mixing version of a Volcano….......

    Cheers

    Ian


  • administrators

    Hi Ian

    Very interesting, future experiments with pressure advance will be interesting to see, in theory you should be able to achieve even extrusion pressure through the acceleration parts of each move.

    Another think that will be interesting to see is if the layer adhesion is noticeably changed at the higher speeds.



  • Hi Tony,

    Yes I think I need to look into pressure advance next. I have a feeling that it may need to vary depending on speed. That is to say, the faster the print speed, the higher the pressure might be, so the more compensation might be needed. I don't know what the algorithm is (David - feel free to jump in here). If it reduces by a fixed amount or as a proportion of the length of the move, maybe it needs another term that will vary it according to extruder speed as well. I'll have to do some testing to find out. Also, I suspect that I might need different pressure advance settings depending on whether I use a single filament input or all three. In theory, if I set all three extruders to have the same pressure advance setting, then the mixing ratio should adjust each one proportionally but I don't know if that's how it works (David?) So lots of combinations to try…......

    Ref layer adhesion - hadn't thought of that. I'd have thought that providing you don't run into severe under extrusion, then the filament must be coming out at the right temperature so there ought not to be an issue. Maybe, at higher speeds what has already been laid down is warmer, so it ought to stick better? Dunno. Wait - I'll check........

    ...........There is certainly nothing wrong with the "sticks" that I produced. Even when I aborted the print once I had run into severe under extrusion, I cannot separate the uppermost layers. Just tried jabbing an awl in between the layers - all I can do is make dents and holes. The last one which went to almost the full 30mm height at 100% infill would make a handy cosh (albeit a rectangular one).



  • Yeah I find pressure advance useful on any extruder setup but minute amounts for direct extruders and large amounts for bowden, a touch seems to help with cable drives too it compensates for rotational elasticity in the drive cable.

    I print a thin circular object and observe whether you get a blob at the start of a perimeter or at the end. If it's at the start it's too much, if it's at the end too little.

    I used values from 0.2 with flex41 filament to 0.025 with pla.

    As for volcano to diamond maybe it's time to machine a mixer to bring 3 volcanos together.

    My heater is 38W will try a few runs now.



  • I did very briefly try pressure advance but it didn't at the time seem to make any difference with my mixing hot end. Just wondering if there is some sort of interaction thing going on (between the filament inputs). Won't know for sure until I test in earnest.

    Ref bringing 2 volcanos together - it'd have to be 3 for me and then all feeding into a single nozzle. I really couldn't be doing with all this loading and unloading filament malarkey. 🙂



  • Well time is short but I have acceleration set to 2000 so to save time and filament, I printed a 200x20x15 cuboid. To hit 300mm/s would require 20mm travel with this acceleration.
    I used a volcano hotend (clone) with 0.4mm nozzle, bed 65 deg C, initial temp 205 deg C generic white PLA

    Machine is a custom corexy (based on smartrapcore alu originally), duetwifi, with direct mounted titan on the x-carriage. I dont know my gantry mass but its linear rails for y and x and printed PETG functional parts, it prints 300mm in x and 200mm in y.

    Up to 80mm/s no issues.
    At 100mm/s there was slight thinning on the short infill lines but the long infill lines were fine.
    At 120 mm/s the same
    at 140mm/s I decided it would be better to up the temps to 215 (I feel that fast printing should use higher temps) and it stopped on a heater fault.

    One interesting observation though, I was getting the slight over extrusion at the ends of a long infill line, running quite minimal pressure advance of 0.0125 so I upped it to 0.025 and it noticeably slowed down the printing speed, not much effect on the overextruded areas. So I disabled it and it went much faster, not much effect on the overextruded areas. So pressure advance is tied in some way to speed and for this type of overextrusion seems to make little difference.

    I will if I get chance have another go, I rarely print this fast so it was quite good fun.



  • Yes, it's fun. For my investigation I was trying purely to ascertain what the maximum melt rate was for different nozzle sizes and filament input configurations (single input and all three together). So I felt it important to keep everything else the same and left the temperature alone. For sure, if you increase the temperature, it'll increase the melt rate, so for comparison purposes, can you run it and leave it at 195 degC - the same as I used. I managed 160 mm/sec before hitting serious under extrusion with a 0.4mm nozzle and single input. I'd be surprised if you can get anywhere close to the 260mm/sec that I got with 3 inputs (so 3 melt zones) and 3 Titans sharing the work load but let me know….....



  • It's interesting as I always presumed we are limited by the heater cartridge power output and/or the thermal conductivity of the hotend assembly. If that were the case then a 40w heater in a volcano or a diamond would perform the same. Even with its larger thermal mass the diamond would begin to cool at higher rates.

    Perhaps it's the ability of the extruders to push the filament though, be interesting to see if someone with a bondtech is willing to give this a try too.



  • One minor thing having skimmed through the video, the calculation of flow rate, I don't think you ought to apply 0.95 extrusion multiplier as that's arbitrary as it's 0.95 multiplied by your steps/mm which might be different on one machine to the next. I'd just drop that term since your working out the volume and the time without it.



  • Simon,

    Ref the factors that affect melt rate, it's not just heater power and conductivity of the metal. It's more about the surface area of the filament that is in contact with that metal (and the time that it is in contact with it). If you look again at the drawing of the Diamond hot end, taking just the short 0.4mm diameter sections there is a 3mm long part then a another 2mm long part. A quick "back of a fag packet" calculation reveals that the surface area of those "tubes" is about 6.2mm^2. Now if we open those holes up to 0.9mm, the surface area increases to 14.1 mm^2 which is why the 0.9mm version can melt more plastic than the 0.5mm version. Of course, there is also the long 21mm section but it has a bore of 2.0mm and the filament is only 1.75mm so it only touches the metal here and there - probably more at high speeds because the pressure will make it buckle more inside the tube. Without a camera inside that tube, it's impossible to know so it's difficult to know how much of a contribution to the overall melt rate that section makes.

    Ref extrusion multiplier - Ideally, I should have used an extrusion multiplier of unity but forgot to change it (0.95 is kind of my default). However, for comparing the differences between single input, 3 inputs, 0.5 nozzle 0.9 nozzle etc, the important thing is to use the same extrusion multiplier throughout. If you think it should have been left out of the calculation then it's easy enough to simply divide all the results by 0.95. I was extruding at 95% so personally I think I did the right thing by taking that into account. 95% extrusion should be the same on any machine I'd have thought. The calculation in any case is a bit arbitrary as it assumes there was no under extrusion. I reality, at these high speeds, there probably was some degree of it even though I picked a point below where it had obviously fallen off a cliff.

    In general, if we are going to compare different hot ends, then we need to use the same test methodology. It'll be difficult in any case because every machine will be different but using shorter moves (200mm instead of 300mm) and higher temperatures will further "muddy the waters".

    Cheers

    Ian



  • Those are fair explanations sir. I'm not sure what the surface area in the volcano nozzle is. I will try to have another go at the test later I'll just start layer 2 at 140mm/s and go from there have you verified the speeds you were moving based on how far your carriage travelled using the frames in the video? I only ask as rarely do printers move actually at the speed expected, acceleration, jerk, coordination of axes, and pressure advance all seem to affect actual head speed.

    I have seen the volcano run out of heating ability (if that's whats happening) with 0.8mm nozzle at about 45mm/s, and I was using flex3drive with 40:1 gearing so it wasn't lack of extruder drive torque.

    What motor current are you using on your titans? and what motors?



  • Hi Simon,

    Ref verification of speed - that's why I chose long moves. At 300mm/sec, and 1200 mm/sec^2 it takes 0.25 secs to get up to speed and 0.25 secs to slow down and travels 37.5 mm during each acceleration and deceleration phase. Meaning that it runs at 300mm/sec for 225mm which takes 0.75 secs so the total time to move 300mm @ 300mm/sec is 1.25 secs. At 50 frames per sec that should be 62.5 frames. Zooming in on the time line for ones of those moves, it's difficult to say precisely when it changes direction frame by frame but it looks like it's around 68 frames from start to stop so it ain't far out. Looking at it, I'd say any error would be in the acceleration and deceleration phases taking a little longer than theory would dictate but that the linear speed section is accurate.

    I'm not running any pressure advance. Why would that affect print speed? I thought it just slowed down the extruder - not the carriage (which if true, would have the opposite effect of reducing extruder pressure relative to axis travel). Also, I thought jerk was only used for short instantaneous speed changes such as segmented circles. What do you mean by coordination of axes? That's also why I chose 90 degree infill so all moves are either pure X or pure Y or are you thinking something else?

    Motors are these http://uk.omc-stepperonline.com/nema-17-bipolar-step-motor-29v-07a-18ncm255ozin-17hs100704s-p-260.html. Not the really short pancake ones that some people use but not the larger ones either. Rated at 0.7 Amps, driven at 600mA and 16 x micro stepping with interpolation.

    Cheers

    Ian



  • I'll take some video when I've got that heater fault sorted, to check my speed is accurate.

    I have no idea why pressure advance slows down the printing speed but it does, try it on a dry run (or with filament) the higher you go the more it slows down, more noticeable on shorter moves. I suspect it has something to do with the next paragraph, as its reducing the extruder feedrate.

    Coordination of axes - the highest speed a move will be run at is determined by the lowest maximum feed rate and acceleration of any axis taking part. So if you ask for a G1 X100 Y100 Z100 then the move has to move at the maximum z feedrate as this will usually be the slowest axis. In your experiment the limitation would only be the extruder maximum feedrate being coordinated with x and y. So if the extruder hits max feedrate the x and y axes of each coordinated move i.e. G1 X10 Y20 E0.05 would only move at the extruders max rate, not their maximum rate.

    I'm using the smaller 20mm pancake stepper http://uk.omc-stepperonline.com/nema-17-bipolar-step-motor-35v-1a-13ncm184ozin-17hs081004s-p-101.html but its actually rated 1A and I run it 550mA normally for PLA/ABS and 700mA for PETG, I have a nice light ally heatsink bonded to the back of it. I've been using 700mA for this test but I suppose I could go to 750 or 800 as a maximum reasonable current. Perhaps fairest test condition are to use 600mA then we are both inputting the same power to the extruder (presuming were comparing 1 extruder to 1 extruder rather than to 3). Be interested to see what happens if you set each of your 3 to 200mA and then run them all. Whether its the same result as with 1?


  • administrators

    Pressure advance demands extruder jerk every time the acceleration changes. So enabling it can reduce printing acceleration if the amount of pressure advance is high and the configured maximum extruder jerk is low. This is probably the reason why you observe that enabling pressure advance increases print time.



  • @DjDemonD:

    Coordination of axes - the highest speed a move will be run at is determined by the lowest maximum feed rate and acceleration of any axis taking part. So if you ask for a G1 X100 Y100 Z100 then the move has to move at the maximum z feedrate as this will usually be the slowest axis. In your experiment the limitation would only be the extruder maximum feedrate being coordinated with x and y. So if the extruder hits max feedrate the x and y axes of each coordinated move i.e. G1 X10 Y20 E0.05 would only move at the extruders max rate, not their maximum rate.

    I'm using the smaller 20mm pancake stepper http://uk.omc-stepperonline.com/nema-17-bipolar-step-motor-35v-1a-13ncm184ozin-17hs081004s-p-101.html but its actually rated 1A and I run it 550mA normally for PLA/ABS and 700mA for PETG, I have a nice light ally heatsink bonded to the back of it. I've been using 700mA for this test but I suppose I could go to 750 or 800 as a maximum reasonable current. Perhaps fairest test condition are to use 600mA then we are both inputting the same power to the extruder (presuming were comparing 1 extruder to 1 extruder rather than to 3). Be interested to see what happens if you set each of your 3 to 200mA and then run them all. Whether its the same result as with 1?

    Ahh yes. When you said coordination of axes, I didn't realise that you were considering the extruder as being an axis, but of course you are right in saying that if the maximum extruder speed is too low, it would limit the maximum print speed. However, I've just checked and there are no worries on that score. Max extruder speeds are set to 3600 mm/min = 60 mm/sec (for each one). So more quick "back of a fag packet" calcs (why do I say that - gave up smoking years ago).

    Anyway, the area of 1.75mm dia filament is about 2.4mm^2. So at 60mm/sec means 144 mm^3/sec and the best (guestimated) melt rate I saw was about 32. That was using all three extruders. I don't know if the maximum E speed is per extruder or for all 3 combined (David?) but assuming the worst case, maximum extruder speed was still about 4 times higher than the demand. For info, max X and Y speeds are set to a bit above what I could reasonably expect for the longest diagonal of my bed, at the accelerations I use. Namely, X is set to 50,000 mm/min = 833 mm/sec and Y is set to 35,000 mm/min = 583 mm/sec.

    Ref steppers. Personally I think the torque is more important than the current rating. The ones I settled on are rated at 18Ncm 0,7Amp whereas the small ones, although rated at 1 Amp only have 13 Ncm torque. so mine are about 38% more torque but of course they are a bit bigger and heavier. 13 Ncm is probably more than adequate but if you do get skipped steps where I didn't, that might be an explanation.

    Cheers

    Ian



  • @dc42:

    Pressure advance demands extruder jerk every time the acceleration changes. So enabling it can reduce printing acceleration if the amount of pressure advance is high and the configured maximum extruder jerk is low. This is probably the reason why you observe that enabling pressure advance increases print time.

    David.

    Can you explain (preferably in layman's terms) how pressure advance actually works. Even the name is a little confusing as we want to reduce pressure, so perhaps "pressure retardation" or some such might be a better term to use (if indeed that is what it does). My limited understanding is that the extruder (or rate of extrusion) will slow down towards the end of a move. But at what point does it start to slow? What does the "S" parameter actually do? Does it change the absolute amount that the extrusion rate will slow by, or does it alter the point within the move when it start to slow? or a combination of both or some other factor?

    Can you also confirm if and where jerk is used when accelerating an axis from rest, up to a given speed, decelerating, changing direction and accelerating again in the opposite direction.

    Thanks

    Ian



  • Okay so managed to get up to some higher speeds

    Settings as above but kept extruder temp at 205 which does seem about right for this specific PLA.

    140mm/s working fine long infill without any issues.
    160mm/s some holes appearing in the long infill
    180mm/s a few more holes, but it would still work as infill. Outer perimeters were still okay as they were being laid down at 90mm/s.

    Didn't manage to get to 200mm/s as the baby started crying, and I'd already reached the point where it had already "failed".

    As for extrusion rate I make it:
    0.13mm2 nozzle area x 0.3mm layer height x 160mm/s print speed = 6.24 mm3/s

    I did a couple of videos but they have screaming baby in the background and I haven't time to edit them.


  • administrators

    @deckingman:

    David.

    Can you explain (preferably in layman's terms) how pressure advance actually works. Even the name is a little confusing as we want to reduce pressure, so perhaps "pressure retardation" or some such might be a better term to use (if indeed that is what it does). My limited understanding is that the extruder (or rate of extrusion) will slow down towards the end of a move. But at what point does it start to slow? What does the "S" parameter actually do? Does it change the absolute amount that the extrusion rate will slow by, or does it alter the point within the move when it start to slow? or a combination of both or some other factor?

    Can you also confirm if and where jerk is used when accelerating an axis from rest, up to a given speed, decelerating, changing direction and accelerating again in the opposite direction.

    Thanks

    Ian

    Ian, have you read https://duet3d.com/wiki/Pressure_advance ?

    Jerk is not used when accelerating an axis from rest, or decelerating to rest. Its primary purpose is to allow the print head to make small changes in angle without s!owing down, such as when printing a sequence of short line segments to approximate a circle.


Locked
 

Looks like your connection to Duet3D was lost, please wait while we try to reconnect.