Weird horizontal banding

[miglo]

Hi guys,

Been off 3D printing for several months due to frustration with calibration of this thing. Recently got back into it and tried some calibration prints. This is the result (I don't remember seeing this when I was printing stuff awhile back)

[img]http://googledrive.com/host/0B_SQvOfEZ- ... JjSkVMWkpJ[/img]
[img]http://googledrive.com/host/0B_SQvOfEZ- ... 9jWVV2Vngw[/img]
[img]http://googledrive.com/host/0B_SQvOfEZ- ... BUZXhsTmhZ[/img]

These bands, or very very slight depressions, are exactly 3.6mm apart and about .8mm wide. I tried using different filament from different makes (all ABS) and it demonstrates the same thing. Looking at the edges it doesn't look so bad.

This is on a Rostock v2 with tricklaser arms, and flying struder. Hotend is a brand new E3D V6. Software used was Simplify3D. Board is a Smoothieboard with 626Pilot's latest firmware for the autocalibration.

HE Temp is 225-235. Bed Temp is 110 (has a thin PEI bonded to the glass).

Retraction Dist 4.0 mm
Retraction vert lift 0.5 mm
Retraction speed 2000 mm/min
Coasting 1.25 mm
Wipe 2.0 mm
Default print speed 5400 mm/min
Outline Underspeed 50%
Solid Infill Underspeed 80%
XYZ movement 10000 mm/min

Thanks in advance!
Mig

[mhackney]

Slow down and reprint. You are printing at 90 mm/s which is way too fast until you have things sorted out. Start with 25mm/s (1500 mm/min).

folks have reported odd banding issues like this with S3D. Have you tried another slicer?

[miglo]

I've printed it at several different speeds, and it seems like speed affects the interval of which these lines appear. Faster speed, smaller the interval in between. One of the things I read while researching this is that it *may* be related to my heater bed, which is setup as "BangBang" instead of PID. I had to use BangBang because PID shuts down the power supply after 15-30secs of having it on.

Going to try and put a load on the 5v line tomorrow and see if this helps it any while I switch back to the PID.

[IMBoring25]

I know I posted in the other thread...Not sure why it showed up here...

[mhackney]

The issue is not bang-bang control of the bed heater. I run all my machines that way and never see a problem like this. It is either mechanical (my 1st guess) or slicer (S3D specifically).

[miglo]

[img]http://googledrive.com/host/0B_SQvOfEZ- ... lRyeVlBNU0[/img]

This was printed using Kisslicer with your profiles mhackney. I don't even know where to start in terms of mechanical. I guess I'll check the belts again as well as install the new e3d metal mount when it comes in and try again.

[Jimustanguitar]

The issue is not bang-bang control of the bed heater. I run all my machines that way and never see a problem like this. It is either mechanical (my 1st guess) or slicer (S3D specifically).

Not entirely related, but kind of...

A friend had a Prusa with a quite literal bang-bang mechanical relay on his bed, and he found out the hard way that the switching coil causes a jolt when it turns off. He had very strange issues, banding included, that were a mess to track down. After banging his head into the desk, somebody else casually mentioned that he should put a diode across the relay... As soon as he did, his printing troubles were solved.

When a relay switches off, the magnet on the armature gets pulled back by a spring, and it sends a current through the switching coil... You need to put a diode on a relay for the same reason that you put one on a motor. Transducers work both ways, motors are generators, and so are solenoids...


I don't think that helps this exact problem, but it's related enough that I'm leaving this here for prosperity

[KAS]

Have you tried to print it hollow with no top to see if the banding is still present?


Here is a test print I use and the measurements. I mainly use it to calibrate the extrusion rate on a .4mm nozzle.

20mm Cube 04mm nozzle test.stl

(1.45 KiB) Downloaded 384 times

[mhackney]

@jim - very odd!

@miglo, so you ruled out the slicer, that's good. But that is a very odd pattern and I can't correlate it to anything mechanical. Very curious. You're positive it isn't in the STL file?

[Polygonhell]

If you want to do a single wall test print, your much better off using a solid cube model and configuring the single wall and lack of top in the slicer.
The problem with hollow models like that is slicers don't always handle thin walls like that appropriately.

A single wall print would be nice to see.
You can also rule out the bed if you print PLA on blue tape, you can do it with the bed turned off, although it requires some messing around, it would rule it out entirely.

With regularity like that it's probably mechanical, I'd start by moving the axis by hand, the spacing looks close enough to the belt tooth spacing that I'd start by looking at belt paths, paying special attention to the idler pulleys and the drive gears.

[mhackney]

I agree polygonhell - mechanical seems logical, I just can't imagine how trigonometric movement can result in such a regular pattern in the Cartesian space.

[Polygonhell]

I'm not sure what I'd expect, obviously all three towers move together in the Z move, so you'll see something close to horizontal, especially on small prints, but I'd expect to see some sort of striation on the linear moves, so who knows. You have to start looking somewhere and that's where I'd start.
Kiss is very good about always printing perimeters the same way, so that rules out the only other thing I could thing of, if you have a small amount of backlash it can generally be seen when an edge is drawn in the opposite direction, I've seen this with what look like offsets that line up with holes in the vertical face of the model, but I can't see how you would get the effect on a solid cube model.

[KAS]

Interesting, I've always just created a .4mm thick wall to match the nozzle and printed it as to verify the extrusion multiplier with calipers. That way you know exactly what thickness the wall "should" be and not leave it up to the slicer.

[nebbian]

Is your slicer set to do some solid layers every now and again when printing infill? If so then there's your problem, I can guarantee that the solid layers correspond to your bands.
Also related is using a bowden cable, and pushing more filament out per second when doing solid infill compared to doing hollow infill.

I had similar issues with my bowden setup.

To sort this out, adjust your travel speed and extrusion width so that the mm/s of filament feed is the same when doing solid infill, vs doing normal perimeters.

Alternatively install a flying extruder like I did, which cuts the length of bowden cable down to 20% of what it was previously.


I hope this helps.

[Polygonhell]

Interesting, I've always just created a .4mm thick wall to match the nozzle and printed it as to verify the extrusion multiplier with calipers. That way you know exactly what thickness the wall "should" be and not leave it up to the slicer.

Buy you should specify the extrusion width in the Slicer, and with one perimeter, you should get exactly that width, if you specify it in the model and it doesn't match the specified extrusion width, the width you get will be a crap shoot. Some Slicers try to reduce flow to match the model, others will just extrude a single wall at the extrusion width specified, or in some cases, not even extrude the perimeter.

[miglo]

[img]http://googledrive.com/host/0B_SQvOfEZ- ... lKbmxnNFRV[/img]
[img]http://googledrive.com/host/0B_SQvOfEZ- ... 5SbGhvRnZZ[/img]
[img]http://googledrive.com/host/0B_SQvOfEZ- ... RQcm1QYVp3[/img]

This was reprinted with KISSLICER using mhackney's profiles and KAS's hollow 20mm cube with .4mm walls. Looking at it straight on (or most angles) you can't see it at all. Only looking at very specific angles and lighting does it kinda come up, as you can see.

@nebbian - The bowden cable is really short, like around 5-6", since I'm using tricklaser's flystruder setup.

[nebbian]

Fair enough Miglo. Just thought I'd mention it as this has bitten me before and I spent an inordinate amount of time trying to figure it out.

Can you watch your print next time (the hollow one is fine) and see if it's swapping from clockwise to anti-clockwise where the bands start/stop?

[miglo]

[img]http://googledrive.com/host/0B_SQvOfEZ- ... RVdDM3MGE0[/img]
[img]http://googledrive.com/host/0B_SQvOfEZ- ... hPczZoNHdz[/img]

So I went back to S3D to try again, this time, I printed from the OUTSIDE-to-inside, instead of the standard in-to-out. And that horizontal bands disappear...wth?? (or maybe its much more subtle?) I get a bunch of other weird artifacts, which may or may not be related to wiping, coasting, and extrusion rates etc.

@nebbian - looks like it only goes in one direction, and doesn't switch.

[Polygonhell]

Printing inside to out is very sensitive to over extrusion, which can appear periodic on simple models, but what you're seeing doesn't look like what I'd normally expect.
If you measure the wall thickness of the single wall model does it match the extrusion width specified in the slicer?

[miglo]

@polygonhell, my last hallow cube single wall was exactly .4mm as expected.

[miglo]

The problem seems solved. It was indeed the bang-bang bed heater. It was more suspect as I was printing something large for my son, and notice that portions that required more printing time, the interval of these bands were narrower. I finally figured out how to have it run in PID mode without it shutting down the power supply. My first reprint of the cube using the same setup as above turned out great. I'll test it some more in the morning.

[img]http://googledrive.com/host/0B_SQvOfEZ- ... UwX2JlOVZN[/img]
[img]http://googledrive.com/host/0B_SQvOfEZ- ... I0Z3R3UG1V[/img]

[mhackney]

Very interesting and one for the knowledge bank. Are you running a single power supply and the bed directly connected to RAMBo?

[miglo]

Running off of a dual rail 750 watt ATX PSU. The board is a smoothieboard running 626pilot's latest firmware. Apparently the PWM default setting of 2000 was freaking out the PSU. I set it to 1000, and was able to run PID mode again.

[dsiegfried]

I had z-ribbing issues as well, also caused by the heated bed. But I was using PID. Solved by putting a dedicated PSU on the bed rail.

[JFettig]

I pointed out in my thread a while back that the bed physically moves when power is applied, I made a video with dial test indicators moving while the PID is settling in. I switched over to Repetier and dialed in PID properly and now it doesn't do that at all once its warmed up and settled in. Bang Bang is the same thing, 100% power, 0% power, 100% power, 0% power, etc. It'll flex the bed up and down, up and down.