Vertical Banding Revisited

[KAS]

That is super awesome. I hope the Rambo doesn't hiccup with the added steps on larger prints. (Are you still running a rambo?)

I read that it might be an issue on the Rambo, but I wanted to test it before installing the smoothieboard. That print was at 40mm/s, I'll try a larger print at 40 then start bumping up the speed.

Awesome. That looks great!!! I agree it will be interesting to see a larger print.

So it really does appear to be the steppers..... such that each step results in an amplified yet slight non-linear movement at the end of the arms. This non-linear movement, or vertical banding, is greatly impacted by calibration and print speed clearly as noted in this thread.

KAS, how much of the improvement do you think was due the steppers vs. the 16 tooth pulley size? With the standard pulley, how close can you get to 200 steps/mm without getting into trouble by adjusting the microsteps or etc? Just personally, not sure about those adjustments.

Just changing the pulley to a 16t would be 100mm/s. The .9 degree stepper on 20t is 160mm/s. Granted that's all on the Rambo so it's capped at 1/16 microstepping without going to external drivers. I plan on running this setup on the smoothie at 1/32 microstepping which is 400mm/s.

Honestly, if I can run 80mm/s towards the perimeter without issues, I might just stick with Rambo. Although curiosity is getting to me.

[bot]

I've been hearing a lot of nasty stories of Smoothieboard troubles. I'd tend to agree with you on sticking with the Rambo for now. Smoothie looks promising, but the hardware and micro-logistics (of setting it all up) are far from refined.

[KAS]

Ran a simple circle and I'm pretty stoked about the results. This was only a 200mm circle. ( I need to figure out how to remove the polygonal crap in DesignSpark). Anyways, started at 90mm/s then jumped to 180mm/s. Tried to increase the temp but it didn't want to go above 227c. The steppers stayed about 37c after 15mins of printing. I have the current somewhat low at 145.

The camera's auto focus was going nuts, sorry about that.

[youtube]http://www.youtube.com/watch?v=0fZw_zeHfGU[/youtube]

200_circle test.stl

(35.24 KiB) Downloaded 297 times

[Jimustanguitar]

I need to figure out how to remove the polygonal crap in DesignSpark

It's the nature of an STL file. There's no such thing as a curve because it's all triangles. You're seeing what's called faceting.

You can make DSM export finer triangles if you want. When you export to STL, click on options (right above the save button) and you can change the resolution and angle that new triangles are generated at.

[KAS]

I need to figure out how to remove the polygonal crap in DesignSpark

It's the nature of an STL file. There's no such thing as a curve because it's all triangles. You're seeing what's called faceting.

You can make DSM export finer triangles if you want. When you export to STL, click on options (right above the save button) and you can change the resolution and angle that new triangles are generated at.

I see, in sketchup you can set the number of segments and make it look smooth when designing. I wasn't even looking for it during the STL export in DesignSpark. Which makes sense now because the circle was smooth in DSM before exporting.

[KAS]

I finally setup the smoothieboard and this whole time I thought it was capable of 1/32 micro stepping, but I guess that's not the case without external drivers. This is my first print on the new controller and I must say that my calibration isn't quite matching what I was capable of doing with the Rambo and a dial indicator. I'll have to work on it a bit more with the auto calibration or try my hand at manually calibrating it.

Also need to figure out the layer lines that are visible on the bottom section and how to control the nozzle lifting that's creating the seam on the lower right side. No changes have been made to Simplify3D, but the nozzle purposly pauses and lifts at that location before coming back down and starting a new layer. Maybe it's normal, just never noticed it on the Rambo.

[DaveBot]

Hi Guys -
Great work by everyone, particular kudos to mhackney for the BandAid model!

Very interesting results with the 0.9 steppers!

A question: have any of you tested printing with 1 perimeter and zero infill? My Rostock was banding about average vs what's been shown here, but then I happened to print a 1 perimeter calibration brick w/o infill (I was checking over/under extrusion vs layer height), and lo and behold it was as smooth as the proverbial baby's bottom, with all other settings the same as before!

Doesn't quite make sense, as I don't think the banding has the same pitch as the infill grid, but I was shocked & surprised when I saw how smooth the hollow box was. Could it have to do either with accel/decel associated with infill printing, or maybe the infill deposition reheating the perimeters & making them bulge? The difference was so striking, it seems like a major clue to me.

Unfortunately, I've had my printer in pieces, putting Trick Laser arms on it, as well as a MakerHive quick-change hot end board. I've been traveling and work-stressed, so only get a few hours on a Sunday afternoon to work on it every couple of weeks. Right now, hotend adapter is miswired, so the machine is dead, hoping to be back running tomorrow, though. Not sure I'll have time to also get it recalibrate do too, but will try some test prints when I do.

Meanwhile, have any of you guys tried a 1-perimeter print with no infill? If so, what did it look like?

[KAS]

Hi Guys -

Meanwhile, have any of you guys tried a 1-perimeter print with no infill? If so, what did it look like?

Few pages back I printed a rectangle with a thin wall and no infill. It doesn't have the same effect as something with infill. You can see the 3 layer base still has the banding and it matched the object underneath.

http://forum.seemecnc.com/viewtopic.php ... =75#p62147

[Jimustanguitar]

A question: have any of you tested printing with 1 perimeter and zero infill? My Rostock was banding about average vs what's been shown here, but then I happened to print a 1 perimeter calibration brick w/o infill (I was checking over/under extrusion vs layer height), and lo and behold it was as smooth as the proverbial baby's bottom, with all other settings the same as before!

Doesn't quite make sense, as I don't think the banding has the same pitch as the infill grid, but I was shocked & surprised when I saw how smooth the hollow box was. Could it have to do either with accel/decel associated with infill printing, or maybe the infill deposition reheating the perimeters & making them bulge? The difference was so striking, it seems like a major clue to me.

An interesting thought... Perhaps the banding is the result of the slicer/firmware trying to achieve a specific wall thickness between multiple passes. Definitely worth testing.

[DaveBot]

After seeing KAS's print, I'm thinking we may really be onto something here. At first I thought his (her?) print disproved my hypothesis about interaction with infill being the problem, since the few layers of the top object would have been solid. But then I realized "duh, they just have solid *infill*!"

Perimeters are drawn smoothly, just tracing the outlines of the part, and we get nice smooth results if that's all we print.

Could the cure for vertical banding be as simple as just specifying a lot of perimeters? Would a print with, say, 5 perimeters provide enough separation between the outer surface and the infill that the laying down of the infill wouldn't affect the outer surface any more?

My machine's still down (had a bad thermistor connection after hot end mount change, now the PEEK fan's not running, & I still need to recalibrate after the arm upgrade :-/), so I can't test this myself yet, but if this isn't the answer, KAS's and my results sure look like clues. Wouldn't it be cool if we solved a problem the big boys haven't yet?

Can't wait for someone to try this, or for next weekend to get here & hopefully have my machine running again so I can test it myself!

[KAS]

his

More perimeters only seem to mask the banding effect. The speed also plays a large role in the visual banding.
To date the best "fix" is to increase the overall resolution. By either lowering the pulley size or increase number of steps per revolution ( or both in the case of the below picture)

Infill from 10% to 100% produce the same sized external banding regardless of the infill percentage. I've tested 10% infill with no top, even the internal infill has the same vertical lines in each pass. Zero infill with multiple perimeters produces it as-well

[DaveBot]

Thanks for the additional info, and wow, that's a gorgeous print!

So, it seems we know these things:
- Essentially no discernible banding with perimeters only
- Any amount of infill produces it
- Extra perimeters mask but don't vanquish it
- 0.9 degree steppers and small pulleys pretty much eliminate it

I'm having a hard time coming up with a mental model for what could be happening, sure would like to figure out a good-enough solution that doesn't involve taking apart my printer coz it takes me a month of Sundays (literally) to find time to do anything.

It doesn't seem like a fundamental issue of insufficient spatial resolution, because it doesn't show up on perimeters.

The only way I can think of the infill causing it is if it's smooshing out the already deposited plastic in the perimeters.

It can't be just a heat-related thing, else the 0.9 steppers wouldn't make a difference.

So it seems it has to be an overshoot issue, where the hotend nozzle is traveling some distance into the perimeter when changing directions as it hits the perimeter area.

Can't be play in the arms, universal joints, cheapskates or belt stretch or 0.9 motors & smaller pulleys wouldn't make a difference. (UNLESS the print speed changed in the process - but it sounds like you're slowing down, just not radically.)

That leaves the steppers themselves, and I'll put another hypothesis out there: It's that whenever we have a situation where the head needs to change course abruptly, if we're also somewhere between full steps, the weaker holding power of the micro step positions isn't enough to stop the head right where we want it, so it overshoots. Places where we *are* right on a full step position, there's enough holding torque to avoid the overshoot. So we get ripples, depending on where the infill bumps into the perimeter, and where the full steps of the motors fall.

Changing to a smaller pulley not only reduces the amount of head travel per motor revolution (which should produce ripples with a shorter period), but it increases mechanical advantage, so the transient load felt ny the motor will be less, and the wimpy microstep holding torque may be enough to stop the head where we want it. I'd imagine that half as many microsteps with 2x the stator poles would give the same res but better holding torque.

What can we do short of a motor/pulley upgrade?

It'll slow printing a good bit, but I'd think cutting jerk way down would help. Can we control decel separate than accel? (Noob here) If so, set decel low but let accel (when the head will be moving get away from the perimeter after a direction change) where it was.

Sorry for all the typos, waaay tired, gotta put don the iPad and get to sleep.

[bot]

I'm gonna be building a delta with the following specs, as soon as the parts get here:

0.9 degree steppers on xyz

16 teeth pulleys

smoothieboard (1/32 microstepping)

as well as a bunch of other fancy untested stuff that may or may not turn out to be good. But I'm really taking your lead on this high resolution stuff, KAS. Thanks for the work.

[teoman]

I wonder what printing one of these and adding them to the motors would do for resolution

https://www.youtube.com/watch?v=kYmUJVE6Vo0



[youtube]http://www.youtube.com/watch?v=kYmUJVE6Vo0[/youtube]

[bot]

This problem is 100% caused by segmentation of moves in repetier, marlin, and smoothieware. When using a duet with dc42's fork of reprapfirmware, the only small artifacts left seem to be from the extruder pulses -- the banding we see here is completely alleviated. This causes ringing to be more noticeable, but straight lines are actually straight save for small bits of extruder moire. See this photo:

From this thread: http://forums.reprap.org/read.php?178,509378

[Glacian22]

WHOA, that is a dramatic improvement!

[bot]

It certainly is, and it is a direct result of the way the delta kinematics and step pulses are generated.

[DSteck]

Damn. Looks like I need to think about switching to a Duet.

[Right to Make]

I get this same effect. kind of annoying really but I am running rostock max, stock, no modified electronics but with E3D V6 hot end. But if you get back to basics, we are running stepper motors which step in degrees. They just do it so smoothly we don't notice. And by using a high quality hot end we effectively increase our resolution we see on our prints as stepper motor/ vertical banding. The hot end is so precise we see the steps. With increased data processing, or upgraded / modded electronics they see better results, which make sense because it effectively lets the stepper motors buffer their steps faster so it will give a more fluent step rather than just a step, step, step. Its all so simple now. I paid extra to be able to print like poo. but at least i don't get any jamming.
All silliness aside, i really like the E3D nozzle but i did get a smoother outer layer print with the stock hot end, and this is why i think that is. or i just felt like talking randomly on the theory of printing plastics by electro-mechanical process by which we call, taking it to the Max, Rostock that is.

[3D-Print]

Damn. Looks like I need to think about switching to a Duet.

Also consider geared stepper motors for your extruder(s) (http://forum.seemecnc.com/viewtopic.php ... =75#p75760)

As I have time, I am planning to:

1. Change my extruder to geared steppers (in the boxs, waiting to assemble).
2. Change my motherboard to Duet (on my Christmas wish list)
3. X, Y and Z steppers to 400 step/rev with 16 teeth pulleys (http://forum.seemecnc.com/viewtopic.php ... 150#p68587)

[derzaubererer]

wow that a printer controller has soooo much quality improvements:O

[KAS]

This problem is 100% caused by segmentation of moves in repetier, marlin, and smoothieware. When using a duet with dc42's fork of reprapfirmware, the only small artifacts left seem to be from the extruder pulses -- the banding we see here is completely alleviated. This causes ringing to be more noticeable, but straight lines are actually straight save for small bits of extruder moire. See this photo:

IMG_6790_resize.JPG

From this thread: http://forums.reprap.org/read.php?178,509378

Damn you Bot! Ordering a Duet today....

[Right to Make]

Also think about rigidity. The stock hot end mount is extremely beefy, which makes it more rigid. Quite a bit more beefy than the mounting hardware you get for an e3d.

[DSteck]

Damn. Looks like I need to think about switching to a Duet.

Also consider geared stepper motors for your extruder(s) (http://forum.seemecnc.com/viewtopic.php ... =75#p75760)

As I have time, I am planning to:

1. Change my extruder to geared steppers (in the boxs, waiting to assemble).
2. Change my motherboard to Duet (on my Christmas wish list)
3. X, Y and Z steppers to 400 step/rev with 16 teeth pulleys (http://forum.seemecnc.com/viewtopic.php ... 150#p68587)

I actually did order a geared extruder last week to help with prints when I use my 0.25mm nozzle.

[bot]

This problem is 100% caused by segmentation of moves in repetier, marlin, and smoothieware. When using a duet with dc42's fork of reprapfirmware, the only small artifacts left seem to be from the extruder pulses -- the banding we see here is completely alleviated. This causes ringing to be more noticeable, but straight lines are actually straight save for small bits of extruder moire. See this photo:

IMG_6790_resize.JPG

From this thread: http://forums.reprap.org/read.php?178,509378

Damn you Bot! Ordering a Duet today....

Don't shoot the messenger! I had to beg my girlfriend to buy me a Duet as soon as I saw that thread -- gonna use the smoothie to make a little mill like Matthew's (travelphotog).