Some minor issues with The Duet WiFi

  • HI,

    I have built a generic corexy based on 2020 extrusion, linear rails with original E3D Titans and an E3D Chimera, 220x220x350, 24v Heatbed, as well as the Duet Wifi, and the IR detector, all running on 24v.

    Most bugs have been worked out with the excellent wiki and trolling the forums. But, I have 2 minor bugs .

    1/ Using the standard X Y Z cube, I cant quite get perfect edges. I have tried retraction, both hardware and software, extrusion percentage, speed, and also pressure advance.

    Settings that seem to work are 0.9mm coasting in Simplify 3D and a 5mm wipe whilst moving, layer height is 0.2mm, as well as these settings in config file :

    M350 X16 Y16 Z16 E16:16 I1 ; Configure microstepping without interpolation
    M92 X80 Y80 Z400 E400:400 ; Set steps per mm
    M566 X1000 Y1000 Z500 E1000:1000 ; Set maximum instantaneous speed changes (mm/min)
    M203 X50000 Y35000 Z1000 E5000:5000 ; Set maximum speeds (mm/min)
    M201 X5000 Y5000 Z500 E1000:1000 ; Set accelerations (mm/s^2)
    M906 X1000 Y1000 Z1000 E700:700 I30 ; Set motor currents (mA) and motor idle factor in per cent
    M84 S30 ; Set idle timeout
    M572 D0 S0.4 ; pressure advance
    M207 S2 R-0.3 F5000 T5000 Z0.2 ; firmware retraction
    M204 P2500 T5000 ; Acceleration during printing, and air moves

    Print Speed is 10000mm per minute default…. Slowing doesn't appear to change things.... Increasing speed gives ringing artefacts.

    2/ Mesh bed Levelling ...It works perfectly, except .... Some points are moved to and scanned at a different speed to the rest ...

    M558 setup as : M558 P1 X0 Y0 Z1 H5 F1000 T15000

    Which for 80% of points works fine .... some near center of the bed seem to work at approx half speed.

    Ideas please ?
    (I'll add pics as soon as I can figure out how to do it LOL )


  • 10000mm/m is 166mm/s witch is really really fast for printing speeds, is that correct? Most of my printers are in the 50-60mm/s prints speeds and 100-150mm/s for travel nonprint moves.

    It also looks like you have a good deal of under extrusion going on, may want to tune that before moving to other issues, this could be caused by really fast print speeds if the extruder cant melt the plastic as fast as requested.

  • administrators

    If using coast-to-end helps, then pressure advance should work even better. Don't use both at once.

    If you are using the IR sensor to probe the bed, when it detects that it is close to the trigger height it slows down the probing speed to 1/3 of normal. The exact point at which this happens may vary.

  • Thanks Kezat and DC42,

    To answer some of the points made :

    1/ Printing speed was set at this speed to ensure that I had a wide window of operation. I deliberately made the printer extremely rigid, metal parts, linear slides etc. to cope with this. The gantry isn't the lightest ( aluminium extrusion ) but, there are no ringing artifacts at fast speeds. This suggests that the acceleration and travel / print speeds are ok in terms of mechanical stability.

    2/ The under extrusion is a function of what I need to do to get an almost perfect print. Setting E-Steps to 418 per mm as per Titan manual gave 110mm instead of 100mm of extrusion…Adjusted that, then Simplify 3D suggests 90% extrusion for PLA, and I ended up with 87% extrusion to get the results you see there.

    Coasting V Pressure Advance. As an engineer I'm aware that experimentally you can only change one variable at a time to find a root cause. I must have printed about 30 or 40 cal cubes to date ( each with only one change in parameters ). What I am seeing is that the blob is being created at the end of a move, rather than the beginning, even with large and fast amounts of retraction. So, I set the retraction at 3mm ( approx 0.75% of bowden length ), then incrementally increased pressure advance ( now 0.4) . I was still seeing blobs and then spotted the coasting option. For some reason the combo of coasting and retraction during a wipe move, together with Pressure Advance is working.

    I will reset the extrusion percentage to 100%, and remove coasting etc, leave retraction at 3 - 4mm, and try pressure advance on it's own, as well as reducing print speed.

    3/ IR Sensor, looking more closely, you are right DC42, it is a slow down when the Z axis moves towards the sensor. The bed is black, non reflective and clean, and hence I wasn't expecting variation between probe points given the evenness of the surface. ( Yes, IR spectrum sees things the naked eye can't ).

    I have to say, I love the Duet and the IR sensor - I have 2 of each!! ..... Much better than anything else on the market. But. like anything else there is a learning curve to get best use from it.

    One last thing, do the settings I have chosen in config.g seem appropriate ? jerk, acceleration etc ?


  • I notice that you are using a Z hop in your firmware retraction. It's possible that this might cause a slight pause while the bed moves. Try disabling the Z hop to see if that improves the blobbing.

  • administrators

    Does your extruder use a Bowden tube; and if so, how long is it?

  • Bowden tube is 400mm long … tried using regular and capricorn tubing. No difference.

    I have done some tests this morning.... I reset the extrusion percentage to 100%, and removed coasting etc, left retraction at 3mm, and tried pressure advance on it's own, as well as reducing print speed.

    Currently running 95% extrusion rate ... with these settings
    M207 S2.5 R-0.1 F5000 T5000 Z0.0
    M572 D0 S0.9

    No Z hop, but speeds still at 10000mm per min ..... dropping the speed has ZERO effect... in fact it makes blobbing slightly worse at the 90 degree corners.

    I had to push up the pressure advance to what I think is an insane level (0.9) to get rid of the problems.

    My thoughts is that Pressure Advance value being so high is a symptom of accel and Jerk settings ? Am i missing something ?


  • You should dial in pressure advance without retracts. Then once you have good results start adding retracts to find the right setting.

  • administrators

    With 400mm Bowden tube you will probably need about 5mm retraction if you don't use pressure advance, and 3 or 4mm if you do. Less if you use Capricorn tubing instead of regular 2mm id PTFE. As Tim says, it's safest to tune pressure advance with little or no retraction to get corners etc. looking good first; then add retraction to get rid of the tiny blobs you get just before a travel move.

  • The best I have now achieved is with

    M207 S3 R0.4 F5000 T5000 Z0.0
    M572 D0 S1.7

    The pressure advance seems insanely high …. Is this indicative of needing to change acceleration and Jerk settings ?

    If I add more retraction (S) I need to compensate with more (R) .... So, i left it at this manageable value.

    Seems to be working well at 10000 mm per min.

    Thanks to you all for your help .... I'll post when I have some pictures to upload.


  • I'd be interested in your resolution of this, as I have exactly the same edge issues after my rebuild!

  • administrators

    Yes that is a lot of pressure advance. It implies that there is a lot of slop in the filament path between the extruder drive and the hot end.

  • David, I will try the capricorn again this evening, and also reposition the extruder closer to printhead to see if that helps.

    Biscuitlad, I have a theory that DC42 is right …. but, also another thing popped into my head. I was scanning the extruder info on the wiki, and I realized that Im not using a pancake motor, Im using an 84oz motor .... so, torque is high, and acceleration low because i have lowered the motor current. ( pancakes should arrive tomorrow ).

    I'll keep you both posted.

  • Well, maybe I'm missing something but the quality looks pretty good on the vid?

    I have to disengage from the debate unfortunately, until I can be sure there's no z wobble on my set up. Noticed a regular pattern on my test cubes that can only mean one thing. It's an unpleasant re-occurrence of something I thought I'd cured several years ago… 😞

Log in to reply