steps/mm dilemma: is more than 0.1mm/microstep necessary?

Hi guys, I have a dilemma regarding the number of microsteps per mm of Z movement.

If the minimum resolution that a 3d printer can commonly handle today is 0.1mm and it already looks good enough for the time consumerd doing the print at such high resolution, why would anybody need more than 0.1mm per microstep? So a printer that can do 0.1mm resolution for each axis will have 1/0.1=10steps/mm.

But I see the leadscrews have huge numbers for this metric, north of 1000 or 2000 even. Is this just a consequence of the helicoidal trajectory or is it really needed for the actual resolution? As far as I can tell it only helps with torque.

I see then printers that have addopted timingbelts for the Z ( smartrap core for example) and they achieve far less steps/mm values yet I don't think their prints look worse because the nozzle is still 0.1mm in diam.

I kind of intuitively feel that having more steps per mm can help somehow, maybe the accumulated errors are going to be less with gentler multi-revolution movement then with jerks that take a bed 0.1mm up in one go.

I'd like to hear your opinion on this because I'm tempted to go the timebelt-way and I don't want to involve any gearbox for Z nor too heavy microstepping.

Edited 1 time(s). Last edit at 11/06/2015 03:19PM by realthor.

---

RepRap Lander concept on Concept Forge
RepRap Lander concept on RepRap Forums
My Things, mostly experimental stuff

Following your thinking...

TR8 Leadscrew
2mm pitch
4 starts

steps per mm = (motor_steps_per_rev * driver_microstep) / thread_pitch*starts

1.8 degree stepper motor ran in half steps with a TR8 Leadscrew would be 50 steps per mm with each step being 0.02mm.
5 steps would give you 0.1mm

Two reasons why 0.1mm/microstep is nothing like enough:

1. Not all microsteps are the same size. One reason for this is that the torque due to motor current has to compete with the detent torque.

2. The incremental torque per microstep is very low.You may have to command the motor to move more than one microstep in order to ovecome friction.

A Cartesian printer that had 10 microsteps/mm would suffer from severe Z banding.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Quote
dc42
Two reasons why 0.1mm/microstep is nothing like enough:

1. Not all microsteps are the same size. One reason for this is that the torque due to motor current has to compete with the detent torque.

2. The incremental torque per microstep is very low.You may have to command the motor to move more than one microstep in order to ovecome friction.

A Cartesian printer that had 10 microsteps/mm would suffer from severe Z banding.

ok, two questions:
1)How much is enough? Is it the more the better? Is there a formula/process by which we can assess the lower limit that we can achieve to have a good performance out of our printers? Smartrap Core achieves about 80steps/mm at 16x microstepping. Is this a minimum?
2)Is there any workaround for a printer that has a low steps/mm count? I've seen counterweights used for both deltas and cartesians but those affect speed.

@MadMike: is your statement the same with dc42's that not all steps are made equal? otherwise i seem to fail to follow

---

RepRap Lander concept on Concept Forge
RepRap Lander concept on RepRap Forums
My Things, mostly experimental stuff

I just was just trying to show a leadscrew example instead of belt. Half step would be higher torque than microstepping. You could use Full Steps at 0.04mm per step, but you would have to use .08mm(2 steps) or .12mm(3 steps) layer heights. It would be easy enough to try to see the results.

Edited 1 time(s). Last edit at 11/06/2015 07:41PM by madmike8.

I've done a few prints at .05mm that looked quite nice. I expect .1mm would be fine for Yoda heads.

ok Mike, i get it that you lose torque when microstepping. I am trying to set the corect dimensions for a new build in order to achieve a minimum resolution of 0.1 with as many steps as possible without too much microstepping and to avoid a gearbox. My specific example will have to have x4 or maybe even x8 microstepping to achieve 40-80 steps/mm and I wouldn't want to go further than that with microstepping.

What I am after is to understand what parameters are affected when having less steps per mm rather than high step count per mm and so far I have:
1) less steps/mm means difficulty in overcoming friction upon starting the move(dc42);
2) less steps/mm means that the Z increments will not be very smooth, the bed will maybe jerk during the steps as the motor tries to cover the higher distance in less moves and it is possible to even overshoot because of the higher torque needed for these moves (this can introduce errors in the print if overshooting is happening). QUESTION: can this be adjusted from software with accelleration or smth?
- [MORE TO WRITE HERE ...]

- on the other hand less steps per mm means faster speeds possible, although I am not quite sure it has any implications for Z, which moves once per every finished layer, but for some situations like leveling compensation it might be needed.

I have read a couple of articles on fine machinery moving stages one of which is this one: http://m.machinedesign.com/archive/resolution-resolved, but if you can point me to other explanations on the subject i'd be more than happy.

---

RepRap Lander concept on Concept Forge
RepRap Lander concept on RepRap Forums
My Things, mostly experimental stuff

I was wondering same things...
I have leadscrew as above, 16th microsteps (but wondered about half 1/8th steps)
I thought I had things setup ok, I thought I would be happy with a machine with 0.1mm accuracy for a first build,
but also thought about the possibilities of greater accuracy on the layer heights as this thing seems so simple, up a bit, up a bit more..
for the amount of hardware incorporated into my Z, the Z was the least of my worries.
when testing I needed the thing to move Z up 10mm when jogging in host software and that distance was out at first,
so I got it about right, ie 10mm, ish, can't remember now what value I had, so pulled up what I think was my most current FW,
and the values dont look like how I remember them, but I think I altered the settings in Matter control.
I thought for the X&Y the values should be 80 (belt) and the Z maybe 40 ish
(but it was over a month ago I was setting this, so the numbers arent to hand)
by the looks of the settings below i've set z feed rate at 3 but steps at 200?
Looks like I could also reduce maxfeed rate on X&Y,

also wondering if a fixed coupler would perform better than flexi?

---

#define DEFAULT_AXIS_STEPS_PER_UNIT {78.7402,78.7402,200.0*8/3,760*1.1} // default steps per unit for Ultimaker
#define DEFAULT_MAX_FEEDRATE {200, 200, 3, 25} // (mm/sec)
#define DEFAULT_MAX_ACCELERATION {3000,3000,100,3000} // X, Y, Z, E

#define DEFAULT_ACCELERATION 3000 // X, Y, Z and E max acceleration in mm/s^2 for printing moves
#define DEFAULT_RETRACT_ACCELERATION 3000 // X, Y, Z and E max acceleration in mm/s^2 for retracts

// The speed change that does not require acceleration (i.e. the software might assume it can be done instantaneously)
#define DEFAULT_XYJERK 15.0 // (mm/sec)
#define DEFAULT_ZJERK 0.4 // (mm/sec)
#define DEFAULT_EJERK 5.0 // (mm/sec)

---

Dont all laugh at once

Edited 2 time(s). Last edit at 11/07/2015 08:51AM by MechaBits.

A few other things...

Unless you want to not just level the bed but also set Z=0 using manual adjustment screws, you need a far finer adjustment than your layer height to optimally set the height for your first layer.

Fine resolution in Z would be needed if you are using autolevel or printing spiral vases.

What are you hoping to gain with coarser resolution? Since Z only moves a layer height at a time during a conventional print and always has an accompanying retraction, it will never be fast.

Quote
IMBoring25
A few other things...

Unless you want to not just level the bed but also set Z=0 using manual adjustment screws, you need a far finer adjustment than your layer height to optimally set the height for your first layer.

Fine resolution in Z would be needed if you are using autolevel or printing spiral vases.

What are you hoping to gain with coarser resolution? Since Z only moves a layer height at a time during a conventional print and always has an accompanying retraction, it will never be fast.

I don't quite follow. Nor do I have much experience with 3D printers so please bear with me and if you would like to explain more about this manual adjusting the bed level being affected by the coarser Z increments i'd be thankful. You level the bed independently of the Z movement and, once level, you still have the same increments.

---

RepRap Lander concept on Concept Forge
RepRap Lander concept on RepRap Forums
My Things, mostly experimental stuff

For successful prints, you need to get the first layer "just right". The gap between the nozzle and the print bed typically closes as you heat the machine, so you need to adjust the Z0 height to make the gap precise if you change your print settings (eg printing with ABS vs PLA). Even changes in ambient temperature can affect the gap, as the frame expands and contracts. It's much easier to adjust the Z0 if you have microsteps.

What I don't understand is what is the perceived problem with micro-steps for the Z axis? What does it matter whether it takes 10 or 100 or 1000 steps per layer change? There are definitely lots of advantages of being able to control your Z height in very fine increments - how would you print a 0.25 mm layer if your Z axis only moves in 0.1 mm increments?

Edited 2 time(s). Last edit at 11/07/2015 08:55PM by julianh72.

---

Follow my Mendel Prusa build here: [julianh72.blogspot.com]

Quote
julianh72
What I don't understand is what is the perceived problem with micro-steps for the Z axis?

The main issue with microstepping is loss of torque, so less microstepping is desirable right?

Quote
julianh72
What does it matter whether it takes 10 or 100 or 1000 steps per layer change?

This is my exact question although I am sure you ask it from another prespective. If your question is "whay not 1000" then my answer is that it's not possible with my current plan/design. 10 steps per mm without microstepping is all I got. I can x16 it and go to 160 steps/mm with loss of torque.

Quote
julianh72
There are definitely lots of advantages of being able to control your Z height in very fine increments - how would you print a 0.25 mm layer if your Z axis only moves in 0.1 mm increments?

I never thought it that way. I don't know why would you print 0.25 and not 0.3 or 0.2. It's like you would ask: "how would you print a 0.33 layer if tou can only do 0.05/microstep".

---

RepRap Lander concept on Concept Forge
RepRap Lander concept on RepRap Forums
My Things, mostly experimental stuff

I think the first question to answer is how much torque do you need? Even if you have less torque with more microsteps, is there still enough to do the job?

Quote
realthor
I never thought it that way. I don't know why would you print 0.25 and not 0.3 or 0.2. It's like you would ask: "how would you print a 0.33 layer if tou can only do 0.05/microstep".

Once you start printing and calibrating, you quickly get an idea of what works well for your machine. I get results that I'm happy with using a first layer of 0.35 mm, and second and subsequent layers of 0.25 mm. If I could only position in increments of 0.1 mm, I guess I would try 0.3 mm / 0.2 mm, but that would be slower than 0.25 mm layers, and I don't know how successful it would be.
Whatever works is great, but I'm glad that my machine doesn't impose a relatively coarse Z increment that I would need to work around.

---

Follow my Mendel Prusa build here: [julianh72.blogspot.com]

all I know is it goes up, I picked higher torque motors, didnt check the pots on stepstick, I dont know if Im driving 2A motors at .4A, I have no idea, I didnt need to tweek or fiddle, I was waiting for the ceramic screwdriver to arrive and just went ahead without it...it works, move 10mm seems 10mm could be 11, I wont know until its fully setup. slowly slowly print a monkey
I got a long way to go yet though with much tweeking to learn.

Edited 2 time(s). Last edit at 11/08/2015 03:38AM by MechaBits.

I am yet to start creating a Z stage still in the design stages but I will not use screw-type Z and will be quite limited in the steps/mm area. I have 2.5A motors that can do quite a torque but in theory I can print a 600mmx300mm area which would make a pretty heavy bed if it is machined tooling aluminum plate.

I would really try the corrugated aluminum the felix printer uses but haven't seen anywhere that so I can order to size (https://www.youtube.com/watch?v=3-Saisv_Z78).

---

RepRap Lander concept on Concept Forge
RepRap Lander concept on RepRap Forums
My Things, mostly experimental stuff

Here's a source for honeycomb aluminum panels: [www.aliexpress.com]

p.s. I think your initial premise about the resolution of 3d printers being .1mm is wrong by an order of magnitude.

Quote
etfrench
p.s. I think your initial premise about the resolution of 3d printers being .1mm is wrong by an order of magnitude.

I might be having difficulty understanding Z resolution and layer thickness but in terms of output I have this understanding of resolution:



Is it that the mechanical Z resolution refers to the positioning precision and the layer resolution is layer thickness? A Z positioning precision of 100 micron will still print at 0.1 but the positioning accuracy of 0.002 would ensure a more correct/consistent/repetable result?

---

RepRap Lander concept on Concept Forge
RepRap Lander concept on RepRap Forums
My Things, mostly experimental stuff

From that pic 0.1 looks cool,0.5 looks a little rougher, 0.3 looks good enough, you might want a more diffuse specular highlight, and faster print, i'll be quite happy with either of them.

Quote
etfrench
Here's a source for honeycomb aluminum panels: [www.aliexpress.com]

p.s. I think your initial premise about the resolution of 3d printers being .1mm is wrong by an order of magnitude.

I am about this kind of stuff, which is way trickier to secure: [goo.gl]

---

RepRap Lander concept on Concept Forge
RepRap Lander concept on RepRap Forums
My Things, mostly experimental stuff

If a large number of steps/mm is indeed a good thing after say 100steps/mm, then a gearbox is my other solution. My question would be: as long as almost all gearboxes introduce backlash, but few that are very expensive (harmonic drive for example), is there any such gearbox design that can be mostly 3d printed and offers a high gearing ratio so that the effort on the plastic per each step does not deform it?

I have looked around and besides the classic herringbone gears used for the extruders and for some one or two other printers for their axis (Sculpto for example), there aren't many that would probably advice for using them for the Z axis.

There are also 3d printed cycloidal gears that have some backlash and the plastic solid pieces are quite easy to print and look sturdy enough. I think they might be suitable for a very slow axis like Z. Maybe have them lasercut from some hard plastic?

And there is the Harmonic Hyperdrive by OskarPuzzle , a concept that gets rid of the Harmonic Drive's flexible cup and only uses solid pieces, again suitable for 3D printing. Unfortunately I can't find anywhere the design in STL format (it can be ordered at Shapeways I guess for $75 or so) and I would have a pretty hard time reverse engineering it to a CAD design. It provides 300:1 gearing ratio which would take my 10steps/mm with no microstepping to a whopping 3000steps/mm, overrunning the best leadscrew.

If I were to buy a stepper gear, the cheapest are planetary gears. The worm gears are very expensive on stepper motors, unless I buy the worm and the gear as two pieces and make a case myself.

The printer I am going to emulate, Slid3R by Rich Rap, uses for the Z motor a 5.18:1 planetary gearbox (not 3d printed) so I am looking for reprappable alternatives before going for buying a gearbox.

Any useful advice? Can we use 3D printed gearing for Z?

Edited 4 time(s). Last edit at 11/10/2015 05:11AM by realthor.

---

RepRap Lander concept on Concept Forge
RepRap Lander concept on RepRap Forums
My Things, mostly experimental stuff

As I understand it, you're trying to reduce the cost of a printer mechanism by using a belt to drive the Z axis instead of a screw, and now you're looking for a solution to the lack of fine steps that comes from using the belt drive. The thing you have to understand about the Z axis is that it moves in both directions, and its position at the end of either an up or a down movement has to be accurate and precise. When the printer is running if you turn on retraction/z lift (you'll need it to be on most of the time if you care about print quality- spiral vase mode being an obvious exception) the Z axis moves up and down multiple times as it prints each layer. Imagine the effect it will have on print quality if the nozzle doesn't come back down to the same Z level after retraction/z lift on each layer of the print.

If there is any backlash in the mechanism- either from the bed/X axis binding on the Z axis bearings or from the drive mechanism (belt or screw), the layer heights won't be consistent and print quality will suffer. Using a geared motor drives the cost up, unless you have one laying around. Low cost gear mechanisms usually have backlash.

I think it would be possible to use belts to make an almost backlash-free drive mechanism, but it would be expensive, or you'd need a very well calibrated printer to print pulleys for it. You don't have to use GT2 belts that are difficult to print pulleys for- use 5mm or even 3/8" pitch belt. You can easily(?) print pulleys for that and make your own belt based reduction mechanism. If you keep the belt tension up, and use ball bearings on the pulley shafts it might even be pretty low loss and have minimal backlash. The problem with printing pulleys (or gears) is getting them to come out round. If your X and Y axes aren't perfectly orthogonal and accurately calibrated, the pulleys won't be perfectly round. Using inaccurately printed pulleys would be a disaster. Belt tension will vary depensing on the rotational position. Ugh!

You might be able you make a compact, low cost, and backlash free reduction mechanism by wrapping low-stretch fishing line around smooth pulleys...

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

If you at looking at a belt driven Z axis, you might want to take a look at the CoreXZ mechanism. It uses additional pulleys to provide 3:1 gearing for the Z axis belts.

Deltas printers use belt drive for all the towers, so in effect for the Z axis. The minimum resolution used is 80 microsteps/mm (i.e. 20 tooth pulley and GT2 belt), but 100 steps/mm (16 tooth pulley) and 160 or 200 steps/mm (0.9deg/step motors instead of 1.8deg/step) are also used.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

There has to be an element of trial and error here. I went from having m5 threaded rods moving my z axis on my i3 rework to acme leadscrews and my steps per mm went from 4000 to 400. I was concerned I would have insufficient torque to turn the screws and overcome friction so tried in vain to change from 1/16 microstepping down to 1/8 and 1/4 etc... but I could not get it to work. So I went back to 1/16 and calibrated it and as the leadscrews are well made and (reasonably) low friction when properly lubricated (gt85) it works very well and I have no complaints. After autolevelling and during z-lift there are z axis movements both up and down with good reproducibility and good printing results, albeit the amount of actual movement is minimal, whereas with m5 threaded rods the z-motors were making full rotations back and forwards just compensating for an un-level bed. Surely enough torque is enough and you only find out by building a prototype and running it.

Isn't 400 too many? Cant remeber my exact settings at the moment, will interrogate the printer later

I asked a question or 2 about this earlier in the thread but no clues yet.

Edited 1 time(s). Last edit at 11/10/2015 07:30AM by MechaBits.

Quote
the_digital_dentist
When the printer is running if you turn on retraction/z lift (you'll need it to be on most of the time if you care about print quality- spiral vase mode being an obvious exception) the Z axis moves up and down multiple times as it prints each layer.

Is this Z retraction really used? Isn't it more sensible to use filament retraction instead? I have seen prints where the Z-axis retraction (Lift) was never used and they don't have any quality issue. All my printer's parts were made without retraction on the Z axis on a solidoodle. Stringing can also happen when the nozzle lifts up the print, why bother?

Even in Slic3r's manual they say:
"
Lift Z - Raises the entire extruder on the Z axis by that many millimeters during each travel. This can be useful to ensure the nozzle will not catch on any already laid filament, however it is usually not necessary and will slow the print speed.
"

Quote
the_digital_dentist
The problem with printing pulleys (or gears) is getting them to come out round. If your X and Y axes aren't perfectly orthogonal and accurately calibrated, the pulleys won't be perfectly round. Using inaccurately printed pulleys would be a disaster. Belt tension will vary depensing on the rotational position. Ugh!

This is always a problem that is why I am thinking on a few solution to solve this: one would be to measure toe "ovaliness" of the resulted pulley and then print suficient perimeters to be able to file it down round with a fine metal file. Second is to print the hub of a pulley/wheel/etc 5-8mm less than the actual size and, having a proper size pipe section aligned, pour some resin in between to obtain the perfect circle. Sand that one if necessary and voila you have perfect round. This will improve the precision of the first 3d printer and the second one will be able to be 3d printed at better tolerances.

Quote
the_digital_dentist
You might be able you make a compact, low cost, and backlash free reduction mechanism by wrapping low-stretch fishing line around smooth pulleys...

The only problem with spectra is that it needs thick pulleys/drums because you have to account for string walking. If the string walks only on one side, then the tension is impacted and also the geometry of the transmission. I solved this for the SCARA part by having calculated heights of both the bolt and the drum so that the string walking on the bolt is mirrored on the larger diameter drum. The gearing down is done in this case with no backlash and no geometrical error.

For Z I was thinking to use again Spectra like the Slid3r 3d printer from Rich Rap (it doesn't use belt but fishing line) and i have a solution in mind but will have to be prototyped and tested.

My actual question is that, considering that I have room at the base panel of the printer to have a very large size round (25cm diam), if I use let's say a herringbone gears like they use on many extruders, any backlash would be rather quickly dissipated due to the large size of the gear, am i correct? I was even thinking about friction gears like they use on many hobby telescopes for best positional accuracy. The gearing can be big enough (300mm -maybe even 400mm- gear/10mm pinion => at least 30:1) so that the friction doesn't need to be extreme to engage.

Edited 6 time(s). Last edit at 11/10/2015 08:57AM by realthor.

---

RepRap Lander concept on Concept Forge
RepRap Lander concept on RepRap Forums
My Things, mostly experimental stuff

Quote
realthor
Is this Z retraction really used? Isn't it more sensible to use filament retraction instead?

Z lift is commonly used on delta printers, because it can be used without sacrificing speed. Another reason may be that when using slow 8-bit electronics, the segmentation algorithm used by most delta firmwares causes the head to dip a little on fast travel moves. Z lift is not a substitute for filament retraction.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Quote
dc42

Quote
realthor
Is this Z retraction really used? Isn't it more sensible to use filament retraction instead?

Z lift is commonly used on delta printers, because it can be used without sacrificing speed. Another reason may be that when using slow 8-bit electronics, the segmentation algorithm used by most delta firmwares causes the head to dip a little on fast travel moves. Z lift is not a substitute for filament retraction.

Does this apply to cartesian or scara where Z is dissociated from the XY moving? (Wow look i've found similarity between cartesian and SCARA ).

---

RepRap Lander concept on Concept Forge
RepRap Lander concept on RepRap Forums
My Things, mostly experimental stuff

I use both retraction and z lift and get excellent results with both set to 0.5 mm.

You can calibrate your printer and get the axes to be orthogonal and use it to print pulleys- I have done it for a chocolate printer project that is slowly making progress.
Here's one - 60 teeth, 3/8" pitch (I found the belt on the hack rack at the makerspace). IRIC, that pulley is about 8" in diameter with the flange.



[www.youmagine.com]

Here's a 15 tooth pulley to drive the big one: [www.youmagine.com]

The final pulleys are mounted on the machine and working fine, but the application isn't nearly as backlash-critical as yours.

You can find closed loop type L belts on ebay for very low prices.

You could measure the "out of roundness" and then file/sand to correct the error, but it would be better to align the printer's axes properly anyway. Aligning the axes in your printer involves printing a square/rectangle and measuring the diagonals, then making appropriate corrections to the positions of the guide rails. And, of course, if you're going to print a pulley that's 8" in diameter, you need a flat print bed with no print adhesion problems.

I wrote a spreadsheet that calculates the corrections to the guide rail positions based on diagonal measurements of a rectangular print. You can DL it here: [www.youmagine.com]

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

If you mean the head dipping on long fast moves, it doesn't apply to SCARA. Nor does it apply to deltas that use segmentation-free firmware. Even with delta firmwares that do segmentation, the dip is probably insignificant when using 32-bit electronics, or a low travel speed and 8-bit electronics.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].