Don’t mind the mess, I have been busy. 
7 Likes
Given that the number of unique parts is so low, I’m going to bet that those Z motor mounts are the exact same part as for the CoreXY motor mounts, which will make it a 16 tooth direct drive mount.
With 3 motors, and the motors that the other V1 projects use, I don’t think holding torque is going to be a problem.
One of the issues with belt driven Z axis is that we move in small but distinct increments. Most of the older stuff was all 20 tooth pulleys, and at 16X microstepping, it came to 80 steps/mm This was kind of an awkward spot to be in, and for some layer heights, the eye would pick up those very tiny variances in layer height because it would be a regular pattern. For example, I tend to print 3 layers/mm (0.3333333) which might still result in some visual artefacts at 100 steps/mm, but at 400 steps/mm is invisible. (I’ll see what it looks like at 200 steps/mm.)
I really love my Duet boards. The “gcode everywhere” firmware makes reconfiguration a breeze, and having direct control over stuff like the Z probe and homing sequences means that stuff like the fight that Ryan was having with homing/levelling gets way easier, and no recompiles needed for any of it.
That said, it’s a learning curve to get used to it, and there aren’t the same post-rocessors available for it as there are with Marlin. Estlcam’s output for my Primo needs some editing (Just to get rid of the “S” parameter when it tries to set spindle speed really) I haven’t looked at other programs for compatability yet.
Still the slicer support for 3D printing is a done deal, it all supports the Duet firmware native, so it’s all good there.
My only hesitation with this printer is that I have an SKR Pro board (Though it’s slated for my LowRider2) and I also have another Duet Wifi board, which I’d really like to use for it. The Wifi connectivity means that the printer can live wherever, so long as it gets wifi.
2 Likes
Thanks Ryan and Dan. My curiosity and concern have been sated.
I agree 100%. I LOVE IT! Once you get past the learning curve of a new system its so much faster to make adjustments.
The only problems I have run into like this are with Prusa Slicer. Setting K values for various Filament profiles or printer profiles. But those can be adjusted in Prusa Slicer so depending on what printer or profile you choose the correct gcode commands are used.
My issue with Duet boards is the price. But I have been looking at using SKR boards and just running RepRap firmware. Apparently. It works for several of their boards. Duet Firmware For SKR Boards - RRF On SKR 1.3 / 1.4 | 3D Print Beginner
The biggest unknown is if the resolution will be an issue, 0.01mm is the smallest step it can take in the Z and steppers do not like to typically take one (16th micro)step. I thought the leveling might have put that to bed, but a real print is a better test I suppose.
I have never tried to print a 0.1mm print in a very long time. I suppose I should see what happens on the MP3DP 2 vs the xi. I will slice something up and see it there is any resolution issues I can see. That leaves 10 steps per layer, if there was an issue it could possibly not move up every step and jump .2mm the next step if it was bad enough. I think.
I will give it a shot, the xi is running a longer print now but I will start it when it is done.
1 Like
So it sounds official then… corexi? Lol love Strider_Matrix’s meme for that.
1 Like
I’ve seen charts at 1/16 microstepping of commanded movement vs real movement, and it tends to look like a sine wave for error, in that the 1/16 microstep does work, but is decidedly not 1/16th of the movement of a full step.
It looks like a sine wave, but the amplitude is reasonably small.
As such, I remember that it used to be stated that for best results, Z layer heights should be as close to a multiple of a full step as possible. In this case, I guess that would be 0.16mm. (Hmmm. My layers are typically 0.3333333mm, which is pretty close to 2 full steps, and my “high res” slices are 1.6666666 – 6 layers/mm – which is pretty darn close to 1 full step.) 0.2mm layers should show error at thick-medium-thin-medium-thick given the stepping ratio. 0.1mm layers at 6 microsteps shy of a full step might also show some of this banding, but I’m pretty sure that the amplitude of the actual error isn’t going to let that seem like much.
Worst case would be 0.24mm layers for stepper motor error. 0.25 would be pretty close to that, so might also show some wonkiness.
Good luck on the 0.1mm! Getting down to those resolutions the bed leveling has to be pretty good. I have printed a lot at 0.1mm, but mostly with an 0.4mm nozzle so not the best. Bltouch with ubl definitely helps a ton. I think you need more like a 0.2mm nozzle or so for best results (so not great for production).
I would love to see that commanded vs real chart if you can find it. That is the sort of info I have not been able to find. So the ideal layer heights for me would be .16 and .32…I will try it out. What would you say is the worst case, I would love to test it and see if I can see anything. I can try a best vs worst.
The 0.1mm prints are doing okay on both printers, the MP3DP has noticeably more ringing and poor infill (cooling?)…I am printing at 50mm/s so that was expected, but the new printer has some ringing as well infill is fine. Pics soon.
Not yet. I really need to sit down and double check that count is right. I typically do not get counts right the first time around. I should have CAD open here pretty soon the double check.
1 Like
Well they both look crappy. 0.1mm is a bad test with a 0.5mm nozzle I think. Both at 50mm/s grey is from the new printer. Better all around and infill. Not happy about these prints.
Try again at 0.2? That is a non ideal layer height and these big nozzles have a better chance at a good print.
1 Like
IDK… The Z won’t start on a full step either. It will be wherever the endstop stops it. For each motor. The proof is in the pudding, but I am skeptical.
You can adjust the steps per mm for the tmcs to be 1/32. It is in configuration adv and then you need to change the steps per mm. I wonder if that would make a difference.
I think the only two places it will matter are:
- bed leveling. Which it sounds like you have close already.
- mesh level adjustments while moving in xy. If you have meat leveling on and you were drawing a line across the X, the z will move up and down. Will the width of the extrusion jerk up and down when the Z changes one microstep?
We could maybe trick it into thinking it was doing 2, by extruding a line like:
G1 X50 Z0.2 F600
G1 X100 Z0.25 E2 F600
0.2 * 0.5 x 50mm = 5mm^3
5mm^3 / (π(1.75mm/2)^2) = 2mm
Something like that…
If the Z resolution was very fine, I would expect a smooth width transition across the 50mm. If the Z was choppy, I would expect to see “steps”. I bet you it isn’t even measurable.
This is honestly a lot of guesswork. The voron datapoint is interesting. But it may very well be worth the extra cost (to them) just to silence the nay sayers, without any measurable difference.
So those previous pictures…disregard them I changed slicers, verified the settings and made sure that there was no layer slowing…flawless 0.1mm prints are happening right now. Feeling good now!!!
How I have seen this explained before is while it will not be on a full step since it is taking full steps the error would be the same everytime so it is taking full steps. Not sure how I feel about it. I saw a video where he measured the error with a laser and a mirror somewhere, I need to find that again to check for myself.
I’m running 2209’s so they can go to 64th…and are running 256 interpolation by default. It is easy enough to test…if we can find aproblem.
This one I full support. I am convinced there is a limitation here either drivetrain or BLtouch related. The terminal only goes down to 0.01mm resolution and it frequently hits that so I am sure we are pretty good here.
Now here we go, that sounds like a good test!!! Here I was ready to set up mesh leveling just to try it out and you came up with two lines of code that I can run as a test. I am very excited to test this.
1 Like
So provided this 0.1mm print comes out spectacular, the only issue we have is maybe possibly a first layer thing.
Any sort of taller print that might show a flaw with a coarse Z step other than first layer? A 0.1mm spiral vase?
1 Like
I personally would like to know more about the meat leveling. 
4 Likes
.1 layer, at 50mm/s. Ringing… slightly flash exaggerated but still not expected. I stopped it to do the other test.
I can’t find it. I think it may have been on the Duet forums, but the RepRap forums is another possibility, and you’ll never find anything more than a couple months old there…
I did find this diagram on linearity, obviously a vendor pushing their motor drivers as being more like the “good linearity” model.
Worst I would say is probably at the 1/4 step mark. 0.12mm layer is going to tickle any error in the motors to its maximum fairly regularly.
But if we’re looking for relative differences…
At the 0.12mm layer height (regardless of what microstepping is set at) it will be 0.75 full steps, so as far as the motors go it’s the difference between what it manages at full, 1/4, 1/2 and 3/4 steps that will show in the layers. I don’t think that this is a big issue, and the actual error bars are going to be manageably small, with them receding to relative zero every 0.8mm by that graph. At this 3/4 full step height, it doesn’t matter what point in the graph you’re starting, the “ideal” line would still be off by its maximum after 3/4 of a full step, making it the “worst case” for driver/motor error with microstepping.
In practical terms, I’d bet on each commanded pulse to be 0.01mm, it would in actuality be between 0.009 and 0.011. It might show to the eye because of being a regular pattern, but you’d have to be super picky to actually see it.
Increased microstepping won’t change this error, because it’s based on 3/4 of a full step on the motor. 0.9°/step motors would help though, as would geared reduction.
I am given to understand that this is not a function of the stepper motor, but of the drive system. Full steps on a stepper motor are when current is not modulated through the coils, aligning the magnetic fields of the coils with the permanent magnets in the stepper. Microstepping is modulating those fields so that the motor armature sits in between the permanent magnets. It is how well the driver chip modulates the two coils that determines how accurate the microstep is. The motor will respond with a change in position in any event. A 1/2 step (easiest case) is like a 50% PWM as a special case. (Gross oversimplification, but kind of how it works) which is easy for the stepper driver to do, and tends to be correct, hence the graph crossing the “ideal” line at about the 1/2 step mark.
Going by the pictures that have appeared since I started typing… I’m going to say that the 2209 drivers are probably plenty accurate to the 0.1mm layer resolution, and I don’t think you’d see anything at the 1.2 (theoretical worst case) resolution.
1 Like
Layer lines are always a pain. I fought weirdness with one of my printer’s designers for a month trying to get some wobble lines to go away.
I ran a giant square around the entire bed. No steps, just a clear indication the bed in not as flat as it seems. The line goes in and out in spots gradually. From what I can see it is always a smooth transition.
Super clear layer lines on the red, but the black is almost perfect. What was it?
The Z rider!
1 Like