UM3 Feeder limitations on print resolution

[erns_ict 0]

Does anyone have any experience regarding the minimum resolution I can print with? Any help or ideas shared are highly appreciated 🙂

 

According to the UM3 manual, the layer resolution is:

Quote

 

0.25 mm nozzle: 150 – 60 Mikron

0.40 mm nozzle: 200 – 20 Mikron

0.80 mm nozzle: 600 – 20 Mikron

 

 

Thus, I expected the minimum printing resolution to be x: 0.25mm; y: 0.25mm; z: 0.06mm However I'm not getting good results for my fine structures.

 

I then had a look at the Ultimaker 3 feeder:

• Moons 17HD6482_01N stepper motor (200 steps/round)
  • powering a 11-tooth gear
• Transmission to a 36-tooth gear
  • powering 10mm diameter feeder wheel

These characteristics lead to a minimum filament feed lenght of: 1/200*11/36*10mm*3.1416 = 0.048mm / step

For the 2.85mm diameter filament, I thus assume a minimum extrusion volume of 0.048mm*(2.85mm/2)^2*3.1416 = 0.306mm³ per feeder step.

For a line printed with the 250 Mikron nozzle and 60 Mikron layer height I get a single line minimun printing lenght of 0.306/0.250mm/0.06mm = 20.4mm per feeder step

This value seems quite high I think. Am I really limited to print such long lines when it comes to a small printing layer hight? Is there maybe an errror in my calculations and assumptions?

 

Does anyone have any experience what happens when the amount of required material for a structure is below the minimun volume to be extruded by one feeder step? (e.g. printing a single line, 5mm long, 60 Mikrons layer height, 0.25mm nozzle)

I have the impression that in this case, the feeder stays still and the print core moves along the printing path, but doesn't extrude/print significant amount of matrial. However, I couldn't confirm my suspicion so far. Probalby I need to check my printed structures under a better microscope to measure width and hight of my line/layer.

[gr5 2,193]

0.048 mm/step is what you came up with.  Usually this is shown the other way around:

20 steps/mm

However the actual value on the UM3 I think is 369 steps/mm (according to some json file in the firmware I just checked).  UM2 is 282 steps/mm and I think the UM2+ and the UM3 are similar and have more steps than the UM2.

 

The difference from what you came up with (20) and 369 is about 18X.  I think it's because there are 16 microsteps per step so the actual value is probably 369/16 or 23 whole-steps/mm which comes out to 0.071mm/step.  Your math seems pretty good but the effective diameter of the feeder sleeve is probably closer to 8.7mm which makes your formula work out.

 

Anyway, 16 microsteps is about right - that's the same value used for X,Y and E for UMO,UM2,UM2+ and the Z drive for some of those is 16 microsteps and for some it's 8 microsteps.

 

Anyway what's a microstep?  Well you can look it up in wikipedia servo articles.  Basically there are 2 sine waves sent to the stepper and you can change the phase slowly and get as many substeps as you want but the chips that Ultimaker uses in the UM3 I think only go to 16 microsteps.

[gr5 2,193]

Please note that you can also get 0.15mm and even 0.1mm cores for the UM3 at my store (yes, I'm biased) or at 3dsolex.com.

 

Printing at 0.25 and smaller nozzles requires lots of experimentation to get the settings right.  I've printed some amazingly small stuff with 0.15mm nozzles.  One problem is cooling - I had to print many objects so that each object had time to cool while printing the next object over.

 

I think these were 0.1mm layer and certainly 0.15mm nozzle.

 

[erns_ict 0]

22 hours ago, gr5 said:

However the actual value on the UM3 I think is 369 steps/mm (according to some json file in the firmware I just checked). 

 

 Your math seems pretty good but the effective diameter of the feeder sleeve is probably closer to 8.7mm which makes your formula work out.

 

Anyway, 16 microsteps is about right - that's the same value used for X,Y and E for UMO,UM2,UM2+ and the Z drive for some of those is 16 microsteps and for some it's 8 microsteps.

That's good to know, thanks for your help! Both the microsteps and s slightly smaller feeder sleeve diameter make sense. As the feeder sleeve has some sort of bumps on its surface (probably to increase grip) I might have measured the outer diameter (including the bumps) rather than the actual diameter of the feeder sleeve.

 

With 369 steps/mm, the 250 Mikron nozzle and a layer height of 60 Microns I would get a single line minimun printing lenght of ((2.85mm/2)^2*3.1416) * (1mm/369steps) / 0.250mm / 0.06mm = 1.15 mm per step

That seems a lot more realisitc than the 20mm I calculated in the first place and might fit the detailness of your model prints. I was hoping to create even smaller structures, but probably that's how it is 🤔 I will print some more test lines...

[gr5 2,193]

Well you could convert to 1.75mm filament.  In your specific case - if you are printing a lot of 0.25mm parts - I think it's worth doing.  3dsolex sells a conversion core.  It's a bit tricky since you have to change the bowden but really - after you've done it a few times you can swap cores in just 3 minutes.

 

3dsolex also has a 0.15mm nozzle and 0.1mm nozzle for that conversion core.  I sell this stuff too.

 

The existing UM3 feeder works fine for 1.75mm filament.

 

Assuming 1.15mm per step calculation is correct (it sounds right) then with 1.75mm filament you will get 1.15*(1.75^2)/(2.85^2) or 0.43mm per step at 0.06 height (again - I prefer 0.1mm layer height - I'm not sure why I need higher resolution in the Z than I'm getting in X and Y.  But maybe if I tried it I would have my answer 🙂

 

[johnse 31]

Doesn’t retraction also play into this? If you print a shorter line than one step’s worth of feed, it will retract at the end of the feature.

[gr5 2,193]

Yes.  Retraction helps.

[erns_ict 0]

On 6/14/2019 at 1:32 AM, johnse said:

Doesn’t retraction also play into this? If you print a shorter line than one step’s worth of feed, it will retract at the end of the feature.

 

Could you explain that in some more detail, it sounds interesting. If I understood your point, the printer will retract the filament by one step if the line printed requires less filament than one feeder's step to prevent excess material on the print plate/object.

 

Let's say I want to print two separate lines, each of them requiring an amount of material that is just 50% of the minimum feeder' step. For printing the first line the filament is moved by one step. Supposing I print at a speed that exceeds the rate at which the material flows out of the nozzle (what would that be aprox?), I can stop printing by retracting the filament by one step at the end of the line. This step's material is then only used by half, so for the next line, one step will just extrude the other half of the volume that is left in this step. Did I get the principle right?

 

It might work like this in theory, but doesn't the print cores have some internal volume of melted filament within them whicht might affect such extrusion of such tiny bits of material?

[erns_ict 0]

On 6/13/2019 at 5:46 PM, gr5 said:

3dsolex also has a 0.15mm nozzle and 0.1mm nozzle for that conversion core.  I sell this stuff too.

 

The existing UM3 feeder works fine for 1.75mm filament.

Smaller nozzles might indeed be very helpful to increase precision, but it doesn't solve the issue with my feeder (minimum extrusion volume). 🤔

Or would you recommend switching to 1.75mm filament when printing with smaller nozzles.

[johnse 31]

Welcome to the whacky world of 3D printing. Your analysis of the flow of material makes perfect sense, but assumes instantaneous translation of feeder motion to material flow.

 

My understanding (and I’m pretty new to this as well) is that the purpose of the feeder is to provide a consistent pressure on that pool of melted plastic in the nozzle. 

Think for a moment about your reasoning. If moving a step of the feeder instantly pushed that volume of plastic out the nozzle, it would come out as a blob. Then another blob a few mm away when the next step occurred. This might be the case if there were no viscosity, and no flex in the Bowden tube, and myriad other factors.

 

During the printing of a line, it is going to average that 20mm/step you computed, but if you could measure the pressure on the plastic, it would fluctuate a bit between steps.

 

Another factor to consider is how stepper motors work. Let’s take the simple case of full steps of a 4 pole stepper. The two windings create two electromagnets, winding 1 has poles up/down, and winding 2 has poles left/right. In this simplified version, we only have switches to turn voltage on/off and control the polarity to each winding. We start with winding 1 positive which attracts the North pole of the armature to the 12 o’clock position. If we turn 1 off and  2 positive, it would swing to the 3 o’clock position. But what if we leave 1 positive AND turn 2 positive? The North pole of the armature is drawn halfway between the poles, to the 1:30 position. The following table shows the sequence of steps in a full cycle:

 

1 2 direction

+ 0 12:00

+ + 1:30

0 + 3:00

- + 4:30

- 0 6:00

- - 7:30

0 - 9:00

+ - 10:30

+ 0 returns to 12:00

 

The actual steppers have more poles per winding and more armature magnets to make that 8-step sequence move only a fraction of a circle, but the principle is the same. The key here is that even though the inputs to the motor are digital (analog control adds the microsteps) the armature takes time to move between positions. And if something prevents it from moving all the way, the magnetic forces pull towards that desired position with a force controlled by the amount of current flowing through the windings.

 

So, in a perfect world with no flex anywhere in the system, that one step of the motor takes as long to move as the head takes to move 20mm. If you want it to draw 2 10mm lines, you would step forward to start pushing filament and the motor would get 1/2 way to the commanded step at the end of the line. In this perfect example, you then back up to the step where you started (retraction) which happens quickly because there’s no resistance going back, but now the end of the filament is 1/2 step away from the end of the nozzle. You move the head to the start of the next line. Again you command to go forward a step. There’s no resistance for the first 1/2 step, and it prints the 2nd line, finishing with the motor completing that step.

 

In the real world were there’s lots of flex, compression, viscosity, etc., things are a lot more squishy if you’ll forgive the pun.

 

Retraction is done by moving the filament several mm to remove pressure and try to suck molten plastic away from the nozzle end. Then it pushes several mm to re-prime the nozzle as it starts the next line.

 

It might help you to read about “hot end”, “heat break”, “direct extruders”, “Bowden extruders”. The physics of these machines is pretty amazing.

Edited June 19, 2019 by johnse
Typos

[gr5 2,193]

36 minutes ago, johnse said:

it is going to average that 20mm/step you computed,

He was of by 16X because he didn't know about microstepping so it's closer to 1mm/step.

[johnse 31]

2 minutes ago, gr5 said:

He was of by 16X because he didn't know about microstepping so it's closer to 1mm/step.

Yes, but I was using a non-microstepping example to illustrate.

[gr5 2,193]

As @johnse says retraction is a few mm.  Typically 4 to 7mm.  It should be *only* enough to relieve the pressure, not actuall retract and let air into the nozzle which would be bad.

 

But say as you calculated in your scenario you are printing and it's about 1 step per mm which is what you were talking about I think with the 0.25mm nozzle and the 0.06 layer height.  If you printed two half mm long lines it would step once and print the first half, then relieve the pressure, move to the new line and unextrude.  This unextrude puts the pressure back in the nozzle cavity and it starts extruding again.

 

feeder resolution isn't as big a problem as controlling pressure in the nozzle.  The bowden acts as a big spring.  Pressure in the nozzle versus resistance creates extrusion velocity.  Typically once you fill the area below the nozzle, the friction goes up so nozzles tend to extrude the right amount automatically.