Top Layers not touching - UM2

[Labern 774]

Haha

Good spotting

[Nicolinux 288]

They shouldn't have used your printer for the demo

(sorry, I have to make fun of this issue or else it would drive me mad).

[ultiarjan 1,223]

To me this just shows you need an x number of layers to get a good surface.... no issue here I think...the top layer looks good.

[Nicolinux 288]

I think we have a very different definition of a good top surface...

The test object hast six layers for the bottom/top surface. It should be enough.

[Labern 774]

This even happens on bottom layers and doesn't matter how thick you make the top/bottom thickness. Also the side walls have the same issue.

There is a clear difference between a printer with this problem and one without as i have had both and this gets rather annoying.

[SandervG 1,521]

Weird thing is that we even haven't been able to diagnose it as hardware or software..

[ultiarjan 1,223]

I think we have a very different definition of a good top surface...

The test object hast six layers for the bottom/top surface. It should be enough.

 

I was referring to the still of the Ultimaker video

I totally agree there's room for improvement in the red cube I printed.... I'll try to find time for some maintenance on this machine in the weekend and see again...

[Nicolinux 288]

 

@SandervG: I think it is partially a software and hardware issue (the board mostly).

[Nicolinux 288]

Meanwhile I tried something else. I tightened the spring on Robert's feeder by 4 turns. Looks slightly better:

Then another 4 turns. It looks _way_ better:

And another additional 4 turns - looks worse:

Well I'd say:

Good thing this is kinda like a palindrome and works both ways :

Edited October 14, 2015 by Guest

[ultiarjan 1,223]

How many mm was the spring compressed when it was at its best? And you use the original UM2 spring I assume?

[Nicolinux 288]

It is the original spring but there have been some revisions over time. I remember some original feeders required the user to have the tension set a few notches up, where as with my spring, I had to set it to the most relaxed tension.

The compressed spring is 9mm long. Since I have added a few shims, it is a total of 12mm between the feeder body and the screw cap.

[shurik 87]

Some updates that might be helpful.

After having lots of trouble,I removed the small black ring that was between the brass hotend and the white teflon.

The things did improve but not quite sufficiently. There were yet some strange under-feedins of the material. So, I took the tension yoke off my Robert's feeder, and it looked very interesting:

Once the yoke being replaced, the extrusion test went from this

to this:

It started to click on 9mm3 and stopped with "Temp. sensor error" in the mid-10mm3.

And here is the cube of @Nicolinux:

Bottom layers:

The whole cube:

Colorfabb White, 210C.

Sorry for the quality of the phone camera.

While not 100% perfect, it looks OK in the reality.

So, could it be that your feeder motor or the yoke from the other side of the filament are not aligned properly or something else obstructs the feeder flow?

Not quite sure what to do with that temp. sensor error, though. It appeared once or twice before, during the nozzle cleaning heatups, so I disregarded it up until now.

Edited October 14, 2015 by Guest

[Nicolinux 288]

Nice @shurik, so there is some improvement.

And good thing you mentioned the feeder. I looked at mine and found something odd too. In its nominal state it looks like this:

I noticed sometimes the filament is not centered on the bearing and moves to the front. Usually this would make the filament snap off the bearing but I have printed a modified yoke that has a small guard to prevent it. But this guard increases the friction and may lead to under extrusion too.

Besides the yoke is quite beat up. I think I should replace it.

Printed the extrusion test. The first one failed at 10mm^3/s which is ok. But for the second one, I remember it failed later on.

[Labern 774]

My printer had the issue with original feeder and new Roberts feeder. It also would print the extrusion with no problems.

[Nicolinux 288]

Ok, I have tested the extruder accuracy. Took the bowden from the head and used PrintRun (also known as Pronterface) to extrude manually. Don't forget M302 to allow for cold extrude. And there is a bug on the Mac version where you have to enter values with commas "10,0" for 10 mm _and_ press enter twice after entering the value...

I set it to extrude 50mm and measured 50.7mm and 50.4mm. There is a little slack through the bowden tube so it is not entierly possible to measure very accurately. But still, this value tells me that there is at least enough material extruded. If it were to be less than 50mm then I'd think about this as a cause for under extrusion.

[IRobertI 521]

Try doing the same thing but while extruding material. I think you'll find that the distance will be reduced quite a bit. If you look at the teeth marks on the filament you'll see that the spacing between them gets smaller and smaller the faster you print. At least that's the case for me. Easiest way to do it is to simply make a mark right at the entrance of the feeder and then another mark after you've extruded Xmm of filament, open the feeder up, yank the filament and measure.

Also, I'm not surprised that people aren't seeing any improvement with my feeder. Honestly the intent was more to make it more convenient to use than anything else

I've also noticed that the filament moves a bit more across the bearing when using the grommet version of the body. The bowden-snippet version doesn't seem to have as much of a problem with that. I assume that's because the bowden diameter is much smaller than the grommet so it doesn't allow for as much play. I'm thinking of switching the feeders on my XT and Go to the oldest version I can find on my computer because honestly it worked better before I started trying to please everyone

[Nicolinux 288]

Alright, I understand the problem. Will redo the test.

Speaking of the feeder - I do sense a slight hint of bad vibes around the pleasing everyone issue. You need to keep in mind that you still made the best feeder for the UM2. Success produces headaches at times...

[IRobertI 521]

Yeah, it's just that after you've gotten the 20th PM asking "What screws do I need?", "could you change this to fit printer X?", "This piece is too long, can you make it shorter?", "This is too short, can you make it longer?", "This didn't come out right on my printer, can you change this part?" it gets old...

My favourite is when people ask how to assemble it though. Even after I've posted several pictures and even a video showing how it goes together.

But, I'm de-railing the thread so I'll shut up now

[Nicolinux 288]

Ok, I got a bit desperate here. No matter what I tried, there was constant under extrusion. This gap between the PTFE coupler and the nozzle is the culprit (I think). (highlighted in red)

When I do an atomic pull, the filament tip looks weird too. Could it coincide with the gap?

I have swapped the Olsson Block with the 1.75 version that I have here. The same .gcode file printed with the very same settings looks perfect... WTF? (I had to play with the colors a bit in order to make the "no gaps" visible...)

[Nicolinux 288]

Good news everyone,

The problem might be solved. @neotko had the killer tip to add a ferrite core to the feeder motor cable. I did it and the result is marvelous:

Before adding the ferrite core I installed the 2.85mm Olsson Block and did a test print. The surface was awful and there was under extrusion on the outside walls too. After adding the ferrite core the next print looks like the image above.

Having ordered too many of those thingies I added them everywhere

Printed again. This time it doesn't look as good as before. I don't fully understand how the ferrite cores work but can too many of them be bad too?

[neotko 1,417]

Sorry but seeing that many ferrites block hanging just made my day XD!!

So happy it prints better!!!

[Dim3nsioneer 557]

 

@Nicolinux: I wonder if you would have the same problem with a different infill pattern? The new Cura e.g. allows for concentric infill. That might give some new hints about the origin.

[neotko 1,417]

On my setup I added just one 3-4cm away from the extruder and also warped it with a cable plastic stuff (like the one that goes around the cables on umo but thinner). But I keep the cable away from the stream just to forget about this error. I don't know if many ferrites are good since they are little magnets warped in plastic. I don't think that near the board should be good. But again congrats on printing close to normal

[SandervG 1,521]

Could you explain why this helps?

My expertise doesn't reach that far

[neotko 1,417]

Could you explain why this helps?

My expertise doesn't reach that far

 

My expertise it's probably the same XD (designer, accountant and 3d printer for a year with zero electrical knowledge, except what I have Learn since I got this printer. (i clap to myself by having desolder and solder an smd transistor chip)

My theory it's explained on the previous page:

https://ultimaker.com/en/community/view/6375-top-layers-not-touching-um2?page=18#reply-120951

Edited October 20, 2015 by Guest