Absolutely wierd banding - Help please!

New mendel90 had finished. Bushings insteed of bearings. Arduino 2560+ RAMPS 1.4. Last original marlin from marlin repo. M6 threaded rods.
Problem - Z banding due to e-feed periodic fluctuation.
I tried different filaments and two hotends.
Yes it can be bad hobbed bolt BUT period is about 5 turns of hobbed bolt!
Upper parts of towers - I halved printing speed with M220 S50
What to try?

Short video with my printer on youtube

Edited 3 time(s). Last edit at 02/14/2014 03:45PM by karabas.

Could it be due to the lack of PTFE tubing running up to the reel and a mechanism to keep the torque constant? Is the reel exerting any resistance to the extruder as it feeds? On Nophead's design (which only works for the larger reels unless you modify it) the reel spins very freely on bearings. Just a thought...

I controlled reel by hand to be sure it's free feeding. And I tried another coil of filament without spool at all. Wades extruder is powerful enough to drive 5-10kg of force.

BTW I measured towers (10x10x50) with caliper and got 10.0mm between bands and 10.4mm max width. So changing in efeed is big enough. When printing raft I see it as well.

If you print a tower twice as wide then the bands will get closer together if it is a E problem but stay the same if it is Z.

---

[www.hydraraptor.blogspot.com]

It's definitely E problem - lower part is much bigger and bands are much closer (in fact every next layer).
Raft also shows E problem - periodiaclly changing width. But.
I cannot find the cause.
I use standart wade with selfmade hobbed bolt.

Edited 1 time(s). Last edit at 02/15/2014 10:08AM by karabas.

Maybe the bearing pushing filament against hobbed bolt could be the cause, but the ratio of rotation should be lower.
Another possibility is problem with gearing - there is long period before same teeth meet (11 turns of big gear), this could explain long period of the problem.

Are you sure it is five turns of the hobbed bolt? The only cause I can think of that would make so much error is that the hobbing is eccentric, but that would be once per turn, which is about 10mm of filament in.

---

[www.hydraraptor.blogspot.com]

If the bands are closer together at the bottom because the bottom layers are bigger, what causes them to get closer again further up the columns?

---

[www.hydraraptor.blogspot.com]

Another possibility is the bed is flexing up and down as the heater cycles. That would be a a fixed time period, so would be closer spaced on slower layers.

---

[www.hydraraptor.blogspot.com]

Quote
nophead
If the bands are closer together at the bottom because the bottom layers are bigger, what causes them to get closer again further up the columns?

because I slowed down printing speed with M220 S50 to test what happens.

I probably have located the problem - teeth on gears are not equally spaced but gradually shifted back and forth ( printed on bad printer) compared to new gears I printed.
I replaced them and tomorrow will print a test.

Edited 1 time(s). Last edit at 02/16/2014 03:25PM by karabas.

Doesn't slowing it down prove it is not the extruder? I.e. if it was related to the extruder gears, or to the Z axis they should slow down in synchrony so the bands would appear in the same places or at lease the same spacing.

If changing the speed affects it then it looks like the bands occur at fixed time intervals, so that points to my bed warping hypothesis. You could try turning the bed off and see if they stop. Eventually the print will probably come loose but it may be long enough to prove it. Or you could print on blue tape and a cold bed.

Edited 1 time(s). Last edit at 02/16/2014 06:11PM by nophead.

---

[www.hydraraptor.blogspot.com]

You are right. My bed is powered via relay(it clicks loudly) and I use 2mm antiglare glass.
After extruder reassembling efeed became better but after bed switching test it is now clear that you are right.
See pic

I let it go until about 1.5cm then off the bed manually - low black arrow.
After about 5mm high first colomn was becoming loose and I turned bed on balck arrow. No more relay switching until top black arc when I started to swich on /off manually trying to see height changing visually. No luck.

I wonder if I will power the bed with pulse mode - it will vibrate... Ok smoothing capacitor may help.
I think I will try thicker glass or alum plate between my glass and bed...

My bed is standart MK2 from reprap.me bolted to dibond plate. No springs on bolts.

Edited 3 time(s). Last edit at 02/17/2014 12:56PM by karabas.

When turning on, cold bed Z0 is getting higher approximately 0.1mm in about 10sec. When off - returns to initial height.
How to fix??
Idially I need intermediate rigid alu plate between heater and glass but does it help really?
My MK2 is not perfectly flat, it is lower in center so there is thin air gap between glass and heater.
I wonder what is exactly the cause of height increasing.

Edited 1 time(s). Last edit at 02/18/2014 03:56AM by karabas.

Is it mounted with the tracks on the bottom?

Do you have glass the same size as the BOM?

Do you have the glass and clips positioned and oriented as the model shows?

---

[www.hydraraptor.blogspot.com]

Quote
nophead
Is it mounted with the tracks on the bottom?

Sorry what tracks?

Quote
nophead
Do you have glass the same size as the BOM?

214x202 ( 202 - Y dimension)

Quote
nophead
Do you have the glass and clips positioned and oriented as the model shows?

clips as in PDF

see pic

I tried 3mm glass today(214x214 with diagonally cut corners). The same.
I played with loosing bolts and suddenly was able to change the direction of shifting.

Edited 1 time(s). Last edit at 02/18/2014 04:16PM by karabas.

By tracks I mean the zigzag PCB traces that form the heater.

You might need to switch from bang bang to PWM and PID to fix it. That needs a more recent version of Marlin than I distribute.

Edited 1 time(s). Last edit at 02/18/2014 06:35PM by nophead.

---

[www.hydraraptor.blogspot.com]

Quote
nophead
By tracks I mean the zigzag PCB traces that form the heater.

Yes heater traces are on bottom as in PDF.

Quote
nophead
You might need to switch from bang bang to PWM and PID to fix it. That needs a more recent version of Marlin than I distribute.

I think it's the only way.

I had a very similar problem to you. Here are things to try:

- Make sure there are no drafts in the vicinity of the printer
- Try putting the clips as close to the edges/corners as you possibly can
- Make sure you always have the corrugated cardboard under the bed

After doing all these things I got really good prints. Enabling PID for the bed improved them even more.

There has been a video posted on YouTube recently showing heated bed PCB flexing on a Mendel90. It looks like the poster was getting 0.05mm of flex at 70C with bang-bang heating. I have asked if he can re-test with PID and post his results.

I have just enabled PID for my heated bed and it looks encouraging for temperature control but I cannot yet say for sure what improvement it will bring in print quality. I can provide my Arduino tree if anyone is interested but there are a few deviations from classic Marlin e.g.:

• A Z_HEIGHT_MM #define is added to the start of Configuration.h to make setting the Z height a little easier.
• Post-homing Z position is 200.0mm (nophead's is Z height - 0.1mm which I find does not always clear the Z limit switch).
• Panelolu2 support may be left enabled without a Panelolu2 connected (currently undergoing testing).
• Panelolu2 status displays X and Y position with sign and one decimal place (you can move in 0.1mm steps, why not display them?) which requires loss of the X, Y and Z labels.
• A custom machine name of "Mendel90" is configured for display on the Panelolu2.

I have not enabled heated bed PID control by default or provided starting PID constants so a small amount of editing, and PID auto-tuning, is required to enable this feature.

Regards,
Neil Darlow

---

I try to write with consideration for all nationalities. Please let me know if something is unclear.
Printing with Mendel90 from fedora 25 using Cura, FreeCAD, MeshLab, OpenSCAD, Skeinforge and Slic3r tools.

Quote
neildarlow
There has been a video posted on YouTube recently showing heated bed PCB flexing on a Mendel90. It looks like the poster was getting 0.05mm of flex at 70C with bang-bang heating. I have asked if he can re-test with PID and post his results.

Yeah that was me

I enabled PID yesterday and tried another print. I had actually already got rid of a lot of my banding doing the 3 things I mentioned, but enabling PID made it a little bit better still. The temperature variance has gone from +/- 1 degree to +/- 0.1 degrees. I will hopefully be able to do the same deflection experiment again in the next couple of days - it would make sense that with PID there should theoretically be no deflection at all as there is no sudden heating/cooling.

Quote
probonic
I had a very similar problem to you. Here are things to try:

- Make sure there are no drafts in the vicinity of the printer
- Try putting the clips as close to the edges/corners as you possibly can
- Make sure you always have the corrugated cardboard under the bed

After doing all these things I got really good prints.

It's not enough.
See video bang bang
When on(click 0:07) - you see light in gap between caliper and X smooth rod
When off (click 0:23)- no gap.
As I understand - when on, bed becomes lower.
So my hope is only PID mode

Edited 2 time(s). Last edit at 02/19/2014 08:22AM by karabas.

Quote
karabas
It's not enough.
See video bang bang
When on(click 0:07) - you see light in gap between caliper and X smooth rod
When off (click 0:23)- no gap.
As I understand - when on, bed becomes lower.

When using bang-bang you will always get some movement - the only thing you can do is minimise the effect. The suggestions I mentioned for me greatly reduced the movement of the build plate. Everyone's mileage will vary.
The only real way to virtually eliminate the effect is to use PWM/PID.

it seems the problem solved. I rewired electronics and turned on PWM on bed. No more banding.
Double check your prints if you use bang bang mode.

Edited 1 time(s). Last edit at 02/20/2014 05:11PM by karabas.