Horizontal banding on UM2

[gr5 2,172]

btw: How stable is the temperature on your UM2?

 

If I set the nozzle temp to 220 it overshoots to maybe 230 but by the time I start printing it sits between 219 and 221 for the entire print. An exception might be when the fans come on but within 30 seconds or so it's back to 220.

 

[gr5 2,172]

Did you recheck the image? I uploaded it in full resolution. I think it's quite visible a the sides that it's not just a different color but different radial positions of the layers.

 

Which one? You have 5 pictures. Here is a crop of the most recent picture. It's a good picture but not good enough:

My eyes hurt from staring at this picture. The darker bands seem to have a somewhat regular pattern for example the 5 green lines - you have 3 bands equally spaced, then a gap, then 2 more. This kind of pattern seems to repeat all over the part.

Then looking at the lower 2 red arrows the part seems skinnier aligned with *lighter* regions but the upper arrow shows a skinnier area lined up with a *darker* region. I'm not sure what's going on:

Is it the photo?

Do you care about the dark/light?

Do you care about the thicker/skinnier issue?

Or is it the same issue but the photo isn't good enough for me to tell?

I have a $1000 camera but my cell phone tends to take better macro photography. If you aren't using a cell phone maybe you should try it.

Anyway, I tend to believe you that the problem is z screw related. Especially because the banding seems to be somewhat regular. What is your layer thickness again? I wonder if thicker layers (.2mm) have less banding because screw errors might be less obvious. But if the problem is on a scale of 1mm then .2mm layers won't be any better than .1mm layers and the green pattern appears to be much thicker than .2mm!

 

[Dim3nsioneer 557]

Please stop staring at the (last) picture before you turn blind... :eek:

I'll try to do some sharper ones later.

The print was made with 0.1mm layer height. I also did prints with 0.15mm and 0.2mm; they look very similar. I can also upload an image of them.

I think the color difference comes from the photo. I mainly care about the thicker/thinner issue because it's this effect which ruins prints with vertical walls.

Until now I considered the banding to be completely irregular but as you see some repeating pattern I have to check the prints again directly.

 

[Dim3nsioneer 557]

This is heading into an interesting direction...

This morning, I upgraded to FW 14.09.1. Of course the new firmware somehow has overwritten my PID settings for the hotend. The temperature was oscillating very badly (+/-13°C !). And the print just consisted of heavy rings... there was a clear correlation between the temperature and the banding. But you know, sometimes it has to get worse before it will get better.

Next thing was of course the PID autotune procedure. Having set the new coefficients temperature fluctuations were down to +/-1°C, i.e. the present temperature was switching between 209°C and 210°C for a target temperature of 210°C.

And this is the result for a 195°C print at 31mm/s:

(this time the picture was taken with my cell phone; I hope it's better visible now, I'm not a photographer... :wacko: )

 

As you can see, banding is still there, but it's less pronounced. I'm really thinking about the origin of the effect. If it is caused by temperature fluctuations then this filament is extreeeeeeemly sensitive to them...).

 

[Dim3nsioneer 557]

This is now getting interesting...

I reduced the current of the z motor to 900mA. Apart from the z axis being much more quiet the banding pattern changed!

It became much more regular than before at a higher banding frequency:

Then I remembered the suggestion by pm_dude to print it at different locations on the print bed and I decided to do a print (still with 900mA for the z axis) as far at the back as possible (centered in the x-direction). And this is the astonishing result:

The banding is almost gone... :blink:

As I cannot (and don't want to!) print everything at the back of the print bed I'm currently thinking about the origin:

- Vibrations? Why do they occur so regularly on a print at the center of the print bed? Do I hit a resonance?

- Can I exclude an issue with the z motor and the lead screw? (because it would show up the same way independent of the position on the print bed)

- Some excentricity of the x/y-rods or the pulley bores? But shouldn't the orientation of all those be the same for every layer?

Other thoughts are highly welcome... B)

 

[pm_dude 27]

Ho s***. Wow! What a difference.

Maybe not vibration but bounciness from the build plate could be greatly responsible for that. After all, It seems pretty normal if you have support of the buildplate only at the far back on a single axis. Maybe if the rods and the z screw where spaced out on the y axis, to create a triangle instead of a line, it would reduce that.

The faster the z speed the more follow-through will be induced at the front of the buildplate. If the plate is too flexible or springy (and it is) it will bounce at the front when the z change stop.

You could try to reduce the z speed to minimum.

Very nice result there

 

[Dim3nsioneer 557]

Might be worth to play around with the z jerk for a start...

 

[Dim3nsioneer 557]

OK, the case seems to be almost solved now. Not by me but by umdeviljur (see http://umforum.ultimaker.com/index.php?/topic/6613-firmware-surface-kwality-bug-1407-vs-1403/, especially post no. 37).

On the left you can see the result with the standard heating settings for the UM2 bed (bang-bang mode) as it is in firmware 14.09.1. On the left you can see the result with PID heating mode enabled for the bed (as it was in previous firmware versions before April 2014).

There are a few banding lines left but about 90% of the banding is gone now. I hope to get rid of the last few lines with the anti-backlash nut I will receive tomorrow.

And this is how I explain it to myself (and to anyone else who might be interested to read it... ):

When the bed is heating, thermal convection on the partially hollow inside is increased (warm air rising). This leads to additional heat transfer from the freshly deposited material to the air (as the transfer rate depends on the air speed; cf. pillowing). Stronger cooling on the inside leads to an increase 'rubber band effect'. Et voilà: banding if the heating happens in bang-bang mode. It's there on all the colors but light blue PLA seems to be the most sensitive to even minor differences in heating/cooling.

But for the moment this feels just a bit like early Christmas... :lol: Yeah!

 

[pm_dude 27]

Oh yeah I remember that post now. I didn't follow it back then but should have... Its quite impressive the difference. Hopefully that will get fixed soon in upcoming firmware versions

 

[thorsten-harzmeier 0]

Hello,

very nice, I see the same effect with the newer firmware.

could you post your new firmware hex file?

Thanks.

Thorsten

 

[Dim3nsioneer 557]

I think a new version of Cura is imminent. I expect that Ultimaker (Daid) will deliver a corrected firmware with this new Cura version if there is nothing serious which speaks against re-activating the PID mode for the bed. Posting my hex file (which also includes some other - experimental - changes) is therefore not the best thing I could do...

 

[thorsten-harzmeier 0]

OK,

I will wait again. Every new Cura Version a new problem... the newest 14.10rc5 the printer starts without heating Bed an noozle if I enable the tweakatz plugin.

Thorsten

 

[Dim3nsioneer 557]

the newest 14.10rc5 the printer starts without heating Bed an noozle if I enable the tweakatz plugin.

 

I'll have a look into this... can you please give me some details about your configuration (what machine do you have? It would be nice if you could put this info in your profile)?

 

[thorsten-harzmeier 0]

I change the info in my profile. I use a UM2 with 0.15mmlayer high. 2.85mm Innofill filiament. 220°C nozzle and 65°C Bed. Speed 30mm/s

Thorsten

 

[Dim3nsioneer 557]

...and UltiGcode or did you switch to RepRap flavor in the machine settings (maybe accidentally)?

 

[thorsten-harzmeier 0]

I use only UltiGcode

Thorsten

 

[Dim3nsioneer 557]

Can you please send me the gcode to Dim3nsioneer@gmx.ch? It will not going anywhere else and will be deleted after the issue is solved.

 

[thorsten-harzmeier 0]

Sorry, I deinstall the new cura versions and install Cura 14.09 and overwrite the gcode with new parameters. In 14.09 the plugin works without problems.

 

[gr5 2,172]

I expect that Ultimaker (Daid) will deliver a corrected firmware with this new Cura version if there is nothing serious which speaks against re-activating the PID mode for the bed.

 

Unlikely. Sorry. Because PID mode creates too much electromagnetic interference and violates CE certification. But Daid *might*. I mean most people don't care that much about a small amount of EM noise.

Meanwhile someone at UM will hopefully figure out if this is a rare problem (certain power supplies) or if this is a common problem to every UM2. Personally I have not updated my firmware in several months and am now reluctant to do so.

:wink:

 

[gr5 2,172]

By the way in other tests I found bang-bang mode still kept the bed (not a UM heated bed) at a constant temp within 1C. So I think the problem has more to do with the 24V supply. Or maybe it's a firmware bug related to interrupt service routines?

 

[Dim3nsioneer 557]

There are at least three UM2 which show(ed) the problem as http://umforum.ultimaker.com/index.php?/topic/6613-firmware-surface-kwality-bug-1407-vs-1403/ and this one show.

And it seems more likely to be a problem in the firmware.

Daid mentioned the EMF issue in the other thread. It's a pity. After knowing the reason I do not expect anyone affiliated with Ultimaker to post a hex file with the PID mode for the bed activated. And neither will I do as I consider this reason to be serious enough to make anybody think first before it's changed from the compliant state to the non-compliant.

I actually also found that the temperature in bang-bang mode is surprisingly stable. If you read one of my earlier posts in this thread you will find my thoughts on the temperature sensitivity of light blue PLA filament.

The case is somehow solved for me with the fact that I got rid of most of the banding with the switch to PID mode. I'll leave it to the software engineers to find the bug in the system called Ultimaker2Marlin. I hope to get rid of the last few remaining horizontal lines with the anti-backlash nut from Robotdigg I received today. Unfortunately one part of it is a bit too big for the available space behind the Ultimaker2-printbed cover but I figured out that with a combination of the new nut and the standard z nut including a new, already designed and printed part it should work as well and fit there.

 

[Dim3nsioneer 557]

Case solved.

I mounted the anti-backlash nut and the print quality got even better. The lines are now determined by the diameter accuracy of the filament I think. As it should be.

 

[pm_dude 27]

Impressive! Good work

[nightwish 0]

Can anybody give me a short info about these fancy mode names?

@Dim3nsioneer: Where did you mount that anti-backlash-nut?

 

[Dim3nsioneer 557]

Can anybody give me a short info about these fancy mode names?

@Dim3nsioneer: Where did you mount that anti-backlash-nut?

 

bang-bang mode means the bed is switched on until it exceeds the target temperature by I-don't-know-how-much, is switched off until it's more than I-don't-know-neither-how-much below the target temperature and then it's switched on again.

PID mode is a closed loop control with a (p)roportional, an (i)ntegral and a (d)ifferential coefficient...

The anti-backlash nut is mounted at the very same place the standard nut is mounted. Actually I had to combine the two...