gr5

I made this list for someone else this morning - you might want to compare all 3 designs - although I really do like Robert's quite a bit. Actually all 3 designs have different advantages. Ian has a 4th design - I think Robert stole some of Ian's ideas. Ian's latest design is secret until his talk is over at a big conference. Geek's design for UM2 feeder: https://www.youmagine.com/designs/extruder-um2-version-2 Robert's design is here: http://umforum.ultimaker.com/index.php?/topic/4393-ultimaker2-feeder-system-improvements-and-ideas/?p=45758 Post #402. Read Robert's posts below that one for more details on assembly and such. Ultimaker's design is in these two posts Post #279: http://umforum.ultimaker.com/index.php?/topic/4393-ultimaker2-feeder-system-improvements-and-ideas/?p=42330 post #268 http://umforum.ultimaker.com/index.php?/topic/4393-ultimaker2-feeder-system-improvements-and-ideas/?p=42162 I think this second post is more useful. Anyway, contact Bas for a free metal grommet.

I found that putting the spool on the floor helps enough that I don't notice any specks of black ABS in my prints (even though they are in there somewhere).

This is fine. You do it with steppers off, but power on. For example when you first turn power on the steppers are all off. After you stop a print or pause a print the steppers turn off within a minute or so. It's okay to turn the big extruder gear by hand also. Probably not - it was probably because of your lack of strain relief (hopefully we have diagnosed the problem correctly). Sounds good. I added some glue from a hot glue gun on top of the strain relief because that is a very flexible but firm material.

Hit power switch. You misunderstand. We are talking about 2 different things: 1) It is possible to break a part up into multiple sub parts using CAD software. Then print the different sub parts. You have to choose carefully where to break up your part. 2) If a print fails, you can continue where you left off with my explanation above.

So for example, in your gcode above, if you wanted to continue at layer 48 (z=5.1mm) you would delete everything up to layer 48 and start the gcode with this little snippet: ;FLAVOR:UltiGCode ;TIME:1360 ;MATERIAL:2282 ;MATERIAL2:0 ;Layer count: 78 ;LAYER:0 G0 F9000 X0 Y0 Z5.1 ; home head and move straight up so as not to hit part G11 ;unretract M106 S255 ; fan on full - not sure if this is smart G92 E1749.75789 ;very important step - this is the last extruder position on layer 47 ;From here on nothing is changed... ;LAYER:48 G0 F9000 X134.40 Y120.90 Z5.10 ;TYPE:WALL-INNER G1 F3000 X95.60 Y120.90 E1751.30989 G1 X95.60 Y119.10 E1751.38189 The important two steps are to move Z to where the next layer starts (5.1mm up) and to zero the extruder at the current extrusion amount (1749.75789mm of filament).

TO TEST SIDE FANS: 1) Select PRINT (even though you won't be printing anything). 2) Before anything heats up go to TUNE menu. Printing will not start until you exit TUNE menu. 3) In the tune menu set fan speed to 100%, they should start spinning immediately. 4) When done you can either hit the power switch, or exit tune menu and quickly abort the print before it starts.

It's starting to sound like you found a bug maybe. But be aware that the bottom most layer speed you have set to 20mm/sec and Cura doesn't just jump immediately to the nominal speed on the next layer but goes up a little more gradually. Maybe over 3 layers? Also be aware that for short movements - say less than 10mm, it is unlikely that the X and Y axes even get up to full speed due to acceleration and jerk settings. So Cura's time estimate might (correctly) not change much at all even if the nominal speed is much higher. I assume you are in ultigcode mode, but in reprap mode, the retraction speed is specified with the F (feedrate) command and you can't have different feedrates for different axes so in reprap mode Cua has to insert the nominal feedrate after every retraction. In ultigcode mode it doesn't have to but perhaps it does anyway. That might be why on some layers you seem to get many more F commands. Also on every Z axis move (at the end of every layer) you might also get F commands there.

1) It helps if you can break up a part into multiple pieces and print them separately like this one which is split into 4 pieces: http://www.thingiverse.com/thing:268757/#files 2) Pronterface is free and useful for experiments and tests and fine control. For example if you want to experiment with different accelerations on the Z axis: http://koti.kapsi.fi/~kliment/printrun/ 3) Don't worry so much. Just start printing. Understand that if it fails part way through then: A: Don't let the bed cool! B: Understand that you can continue it later. You can even wait a few days. More details here: http://umforum.ultimaker.com/index.php?/topic/4213-ideas-for-recovering-failed-prints/

The search ability for this forum kind of sucks - for example it ignores anything 3 characters or shorter even when it is likely to be the most important search term e.g. ABS. The simple solution is to use google. You can tell google to only give results from a particular website like this: site:umforum.ultimaker.com ABS

There's a tiny board on top of the UM Original test head that converts the tiny thermocouple voltage into a voltage from 0V to 5V where 0V is 0C and 5V is 500C (and 20C is 200mv). The wires that leave this board go along the bowden tube and are moved every time the test head moves. The problem is usually at the connector of that wire bundle. I think the wires are ground, power, and signal. The wires are supposed to be crimped into that connector but UM typically solders them which is bad because some solder wicks along the stranded wire and inside the insulated part and having a partial wick causes a stiff strand which can break when the head moves back and forth. There is a black "F" shaped strain release that is supposed to help this but you (like me) may not have used it. This commonly causes MINTEMP and MAXTEMP errors. You can usually duplicate the intermittent error by pushing the test head to all 4 corners a vew times or if that doesn't work, push on the cabling (at least 30 times) in every different direction right near the connector at the top of the print head. The quick and easy fix is to switch to the other cable (for the second extruder). You have to switch at both ends of course (both on the head and on the PCB).

Which STL exactly? There are 7 or so STLs there. Is it tenall? sixall? sixpieces? eightall? I figure likely one of those 4.

I have done prints > 24 hours. And I had a failure (nozzle clog) on one. I was able to continue the print - it was on the UM2 with heated bed so the most important advice is to keep the bed hot and don't panic. You can wait a few hours before continuing. If the bed had cooled the part would have popped off the bed. To continue the print I had to use pronterface to move the Z axis up and down until I found the exact layer that had just finished. Then I had to edit the gcode - deleting all the begining part and adding a G92 code for the Z axis. Let me know if you need more details.

Probably too many polygons. You'd be amazed by the thousands of details you can get with only a 100k polygons. So take your HUGE model and run it through this process and then try meshmixer again: (meshlab is free) http://www.shapeways.com/blog/archives/226-polygon-count-reduction-with-meshlab.html

In the USA, I like printbl.com But be warned - their filament spools won't fit on the back of the UM so you will have to make something to hold the spool. Not a big deal.

Well the UM1 and UM2 can definitely achieve .05mm but it's more difficult. Do you have an Original or UM2? I forgot, sorry. But my point is that I suspect Z axis issues. Maybe acceleration or jerk is too high. Maybe you need to add more grease. Maybe something is sticking. For the Original there are all kinds of issues with the Z nut and how it is held in place that can cause banding like this. If you want high resolution, it seems silly to use a .4mm diameter nozzle hole and then use .05mm layer height. To get really good resolution in Z seems a little bit of a waste if you aren't getting it in X and Y. For really amazing resolution it would be good to get a smaller nozzle - maybe a .25mm nozzle.

I think your z axis is acting up. Sometimes it is moving too far and sometimes not far enough.

I'd rather see a photo of what you are talking about before looking at gcode. Could it be that your Z axis is getting stuck occasionally and you get over extrusion? When you get a chance turn off power and push the Bed up and down and see if there are stiff spots. By the way an alternative to the second tower is to check the "cool head lift" feature. And 10 seconds per layer should be plenty of time to cool (even 5 seconds might be just barely enough).

Wow. I think you just created an open source makerbot! Maybe you could post this part on Youmagine with a free-to-copy license?? Ultimaker is working hard on solutions to underextrusion issues and I've seen a few things they are working on and it's very promising.

Cura thinks of each layer as a separate entity of closed shapes with holes in them or gaps between the shapes. So a hole in this case is an area of a given layer that is considered *not* infill.

The smaller the layer height, the more likely you can hit Z accuracy issues. It's why I prefer printing at .2mm layer height for most prints. There is usually a layer height where if you go any smaller it starts to get worse again. Anyway so my point is that I'm guessing that your Z layer movement isn't consistently .05mm and sometimes it is a little less and sometimes a little more. But this theory could be wrong - your layering in your photo is not quite what I expected. But at this thin a layer height maybe this is how Z accuracy error looks - because a layer that is thinner than .05 will overextrude and create a thin tiny bulge line and a layer where the Z axis moves more than .05mm will look just like underextrusion.

Well then, that's not the problem. I don't know what the problem is then. Are you sure it's related to the fan? Could it be a coincidence? The MAXTEMP error is very common on UM Originals after (very roughly) 100 prints are printed. But in that case it is related to the wiring on top of the print head and has nothing to do with the fan - it will happen when the fan is on or off. You can sometimes duplicate this by pushing on the wiring at the top of the print head. Does it work okay at 100% fan? at 1% through 99% the PWM circuit is turning the fan on and off many times per second and this can cause possible electrical interference but at 100% (setting 255 on ulticontroller) there is much less interference. Also consider just swapping to the backup temp cable at the top of the print head (and of course you will have to change the wiring at the other end also on the PCB).

That's how the robot looks if you use no fan. Did the side fans come on by the 5th layer? Important settings related to perfect overhangs (like the robot's ear and belly overhang). 1) Fan at 100% by 5mm for sure - this is probably default but test that the fans work. You can test them without printing by telling it to print and then before it starts printing go to TUNE menu and turn them on there. 2) Bed at 65C max. For this part I would use bed temp of 60C. 75C is only useful on very large parts (>100mm across) that have warping issues. 3) Keep nozzle cool also. 220C should be fine but if you still have issues, you could try 210C. 4) Layer time - did you have infill on? If you had 20% infill then that means each layer had plenty of time to cool which is good. But if you printed a hollow robot then you need to slow things down - maybe set minimum layer time to 10 seconds instead of the default of... 5? 7? 5) Print slower - quality always improves as you go slower. Usually you can get excellent quality at 30mm/sec and okay quality at 100mm/sec. But the slower you go the better the quality.

It looks strange. I think I would have to see it from another angle. My first thought was that one of your axes slipped and needs tightening and that it slipped only once on a particular layer except that The "tail" support didn't also slip. Before printing it again, check the "bad" layer carefully in Cura layer view. Although maybe you got some underextrusion? The stringing isn't too severe but you can probably eliminate it. You might want to experiment with retraction distances on a test piece first.

Right. Also twisted wires to the fan are better than non-twisted wires. I am talking only about the wiring inside the head - not above the print head.

For layers with hatch like infil, combing is no big deal and usually saves time because you skip a retraction step. But for bottom and top layers combing can make an ugly scar. So for example if you are printing a cell phone case where you want the bottom layer to be the best layer, then you probably won't combing off. For 90% of my prints I have combing on.