gr5

I strongly recommend you do these experiments with just a single letter. And make it very thin - say 1mm. No reason to print an entire part for these tests. If you can get 10X as many tests in the same amount of time then you can learn 10X faster and get 10x more done before potentially getting discouraged.

Okay well you should look at this document here @jonatanrullman and look at how one person did what you describe. https://drive.google.com/file/d/0B7A1MEUY39HCaE44NnExeDJqRlE/view Which I found in a link on this web page: http://shop.bondtech.se/ec/upgrade-kits/ultimaker/ddg-extruder_ultimaker2_en.html

The power supplies for the UM3 have been known to be flakey. I'd get a new one shipped to you asap. I had a bad one myself with my UM3. Meanwhile to prove this and to get started you can put blue tape on the glass (should have been a roll of blue tape with the printer), clean the blue tape with isopropyl alcohol (to remove the wax that keeps the tape from sticking to itself and prevents parts from sticking well) relevel and print with the bed turned off (set bed temp to zero in cura). This will reduce your power usage by about 160 watts and you should be fine. The printer is often near the limits of power requirements for the power brick that comes with the printer. Don't think that having the bed at 40C instead of 60C will fix the problem. The bed, even at 40C, will occasionally come on full power for a few seconds and it only takes a few seconds of over-power to trip that sophisticated, computer controlled, power supply. You need that bed off 100% until you get a better power supply. Your reseller should be able to take care of getting you a new power supply for free. Alternatively there are power budget features on the UM3 but adjusting those is complicated and involves advanced stuff like ssh'ing into your UM3 and if you aren't comfortable with that then definitely simply get a new power brick.

That's a nice feeder but it won't work on plus style feeders that they have on UM2+ and UM3. The plus feeders have a gear at the end of the motor shaft instead of the knurled or splined sleeve.

In the near future, print this thing out: https://www.youmagine.com/designs/wedgebot-for-ultimaker2 That should be your very next print. Well I assume you held the lever up and pulled and couldn't get the filament out. Did you try while the core was hot? You need to do this with the core at 150C (or hotter but 150 is safer). If that still doesn't work you have to remove the bowden at either end. Best to do it at the head. Push the head towards the front of the machine. Remove the small white clip around the collet. Push down and hold down the collet firmly (use pliers if you have to) and at the same time lift up on the Bowden tube. While doing all this someone needs to hold that lever on the feeder so you need a second person or you need some kind of wedge like the one above (a screw driver might work - make sure the filament slides up and down a few mm before deciding your temporary wedge is working). If it's really really stuck in the core even when the core is at 150C for a few minutes then get a hot flame and an exacto knife - (first practice) heat the blade up until water is boiling and keep going another 5 seconds then use it to slice the pla just above the metal tube at the top of the core. Now you can slide it out and disassemble. Don't disassemble without watching my video! You can damage the core with almost zero force when taking it apart so watch my vid first.

If you get a UM3 and have tried it out with 2.85mm filament and are ready to convert, contact me. You might be able to get a free conversion from 3dsolex but contact me first and let me talk to Carl at 3dsolex. I think Carl would be happy to sell these "conversion cores" but he needs some people to experiment with it first before he can make it an official product. I actually have a customer in USA who wants to convert to 1.75 so I should know much more in a few weeks. For example how good the "plus" feeder (found on UM2+ and UM3) will work with 1.75mm. It should be fine but I haven't tried it myself.

After you replace the teflon part with a TFM, if you still have issues let us know here and I'll post my big list of possible underextrusion causes.

It's not a firmware issue. Does it work at 20% if you start slowly at 0% and move towards 20%? where EXACTLY does it turn on. You want to pick a value above that because what happens is the first layer will be at 0% and then gradually increase as per cura settings until it gets to full power which may indeed be 20%. BUT ALSO it's multiplying these values with the Material Fan setting in your material menu on the printer itself. You want these 2 values to multiply out to a value well above where the fans come on if they are increased very slowly. Note there is a feature in the firmware if you go from 0% to 10% it gives the fan 100% for some portion of a second to get them kick started. But if the final value is 10% and the fans then stop, and then later the fans are increased to 15% it doesn't do that kick start feature the second time. It only does the feature when starting from 0%. So Cura makes this feature useless because cura increases very slowly each layer for the first few layers. Is all this clear? Please experiment more. I want to know where your fans come exactly on WHEN YOU INCREASE SLOWLY - NOT WHEN YOU DECREASE AND NOT WHEN YOU INCREASE QUICKLY FROM 0%. If the answer is 19% then you need a higher value than 20% to be safe (maybe 25%) but in cura you probably should set it to 100% and in your material settings you can set it to 25% for ABS. But when in the TUNE menu it doesn't do these multiplications. It just goes directly to what you request. Also when you enter TUNE menu it shows the % fan it was told to do (cura gcode multiplied by material fan %).

There's a few possibilities. But first ALWAYS view your part in layer view (as shown in screen shot) before printing it so you don't waste time as about half the time I slice I catch something (like this) in slice view. The first obvious possibity is that the part is less than 0.3mm thick. I know you have Cura set to layer height 0.1 but the first layer (initial layer) defaults to 0.3mm. So try setting that to 0.1 and see what happens in slice view. Or thinner. how thin is that part? Another possibility is that the part is not one solid object (part isn't manifold) and that the part that isn't printing either has holes in its walls or is missing a bottom or has extra internal walls. View your part in xray view to see if this is true. Any red areas in xray view means the part is not a solid. It's not manifold. You can fix it with the netfabb free repair service here: https://service.netfabb.com/login.php Another possibility is you set Z to a negative value. Click on the part in normal view and some icons appear at the bottom. Click on the move icon and it should show the Z position to be zero. If not set it to zero.

It's definitely not a bug. If you want the initial layer height to be 0.1 then it sets Z to 0.1 so the nozzle is 0.1 off the glass so there is room for a 0.1mm layer. If you set the first layer height to 0.3 then it sets z to 0.3 so the nozzle is 0.3 off the glass so there is room for a 0.3mm layer. So be aware that not only will you have the first layer starting closer to the bed, you will ALSO have 1/3 the extrusion because 0.1mm is 1/3 0.3mm and so you are getting 1/3 as much filament coming out of the nozzle. So I'm not sure you are done with your adjustments. I still think you need to level better such that Z=0 has the nozzle touching the bed.

Google "acetone smoothing abs". See if the pictures that come up interest you.

I don't think so, no. Clean all the rods and add one drop of oil to each one. I was going to say "clean the glass in the sink completely" but you probably don't have the heated bed kit. Make sure you clean blue tape with rubbing alcohol before you print on blue tape. I'd get the ptfe part from fbrc.com where Erin works. I sell these also but I'm out of stock.

Oh and you can get Um3 covers from printedSolid.com. I don't know if they ship to Germany. If they don't and you want one, let me know and we can arrange shipping (I run a store in USA but don't normally ship to Germany).

I never monitor this part of the forum so if Sander hadn't flagged me I wouldn't have seen this post. You may not get any replies from anyone else because of this issue. But here we are. Wow - lots of questions. So there is no 1.75mm mod yet but... I've been looking for someone to try a mod in USA but you are in Germany. But if you contact carl at 3dsolex.com you can probably talk him into sending you a free conversion kit if you promise to test it. His solution and my solution are different. His solution puts an inner bowden all the way to the feeder. I don't like that - if he sends that to you then you might consider cutting off just enough bowden to fit inside the core. Anyway - I don't recommend you try this until you've used the UM3 for a while and are an expert. You will also have to fudge the "flow" parameter to 265% (ratio of the two filament diameters squared). The feeder can handle 1.75mm filament - it's a pretty amazing feeder. Or you can tell Cura that it's 1.75mm filament and keep flow at 100%. Anyway - other questions - I don't know much about support. It depends which reseller you use. Consider using dim3nsioneer who has amazing support but is in Switzerland. Or igo3d whose support may be worse but they are very professional and do a lot of business selling UM printers. You can get great advice here on the forums. As far as printing polyflex - well this is about as flexible as ninjaflex (slightly more flexible) - so this is a very difficult material to print on a bowden printer. I don't think you can print 1.75mm filament (it's like pushing on a string - the feeder has to push the filament through a 1 meter bowden tube) of this type but 2.85mm you can print because the larger diameter makes it a little stiffer. You will need to add oil to the filament to get it to slip through the bowden. It's a tricky filament but you can do it. I've done it with ninjaflex. However there are easier materials like Ninjatek Cheetah or Ultimaker's TPU95 which are quite flexible but no where near as flexible as polyflex. More material info here (I will make sure to add polyflex to these tables but it appears identical to ninjaflex)

So Cura assumes that Z=0 has the head touching the glass and because your first layer is 0.3mm thick (that's the default bottom layer thickness even though the remaining layer heights are all 0.1mm). So is there a way to level your head closer to your bed before starting the print? If not you might be able to do: G0 Z0 G92 Z0.3 ;tell marlin that we are now at position z=0.3 even though we know we are at position z=0.

Maybe I should instead ask: why ABS? Does this need to withstand high temperatures? how high? You might want to try PLA. I printed the number for my house in PLA and it has done very well out doors for the last 2 years.

Good points. Also keep in mind this is ABS. I don't know if ironing works on ABS @neotko

How old is this printer? You might need a new teflon part if you've been using this for a while (hundreds of hours). Anyway the bottom layer thickness defaults to 0.3mm so assuming the future layers are .2mm or .1mm obviously you are printing significantly more plastic on the bottom layer. That is part of the issue. The other issue is how much you are squishing the filament into the bottom layer - leveling. I know you said you re-levelled but there are many choices as to where you position things for the bottom layer. I personally like to level such that the bed is extra close and I get extra squish. But that also means I have the danger of happening what I see here. Anyway I have a um2go with the original black feeder and I always do 0.3mm on the bottom layer and I squish the bottom layer badly so I think something is wrong with your printer most likely. How old is it? You can check hours printed in the menu system.

Perfect! This is called a cold pull (but with a second material) and it's also easiest to remove the bowden tube from the print head before doing this. but not mandatory.

You'll have to learn how to get parts to stick very well if you will be leaving your printer unattended: You can get all that plastic out using a hot air gun but it's going to take you a long time. A hair dryer might work also - not sure. You need to get it up to around 60C.

Did you play around with the fans in TUNE menu during the print? Maybe you didn't experiment enough? How do they do at 100%? What is the minimum speed if you start at 100% and go downwards? What is the minimum speed if you start at 0% and go upwards?

1) I still think it's the set screws. Use a sharpie to mark the shafts and pulleys on the short belts (4 pulleys) to prove it. 2) It's possible you simply have too much friction. with steppers not engaged (most of the time) push the head around and see if X and Y resistance is equal. You should be able to use your smallest finger on each hand pushing on the two end blocks and move the head easily in X or Y. Add one drop of oil (any light petroleum oil such as 3 in 1 or sewing machine oil) to the 4 outer rods. The two rods through the head shouldn't need any oil but maybe a half drop of oil might help. That it always happens in the same spot is not surprising. The planner inside the printer is complicated and quirky. There are ways to send the right gcodes such that you have higher acceleration than normal (because of these "bugs" or features in the planner). In fact the firmware has two accelerations: one should never be exceeded and is around 5 meters per second squared. The other is to truly NEVER exceed (lol!) and is around 9 meters per second squared. Just knowing that there are two different max accelerations tells you something right there. I've looked at the code carefully and understand why they did this but the point is this: the right pattern of movement can exceed the lower of the two accelerations. But this is rare and might only happen a few times within a print. Lowering the acceleration and jerk by half will indeed help but then the quality of your print will go down. Better to fix the hardware so you get good quality prints than to compensate your hardware issues by messing with the firmware settings.

It probably doesn't work for ABS. The technique was created by @neotko and Ultimaker included it into Cura without consulting him (he was annoyed I think because they know him personally but didn't get clarification from him on how to do it right). So you could google "neosanding" and get more info maybe. Anyway I'm guessing it only works well on PLA because it has to reheat the surface and PLA only needs to get up to about 60C for this to work but ABS has to get up to about 120C for this to work and so you basically would have to move the nozzle at maybe half the speed on these ironing passes. I'm guessing. When @neotko does it himself he uses S3D (simplify 3D) and he changes a bunch of things including the speed and he also extrudes a tiny bit (something like 5% or one twentieth normal extrusion - or maybe less) although it works without any extruding also. Instead since you are printing with ABS you could try sanding with sandpaper followed by acetone vapor deposition. Get a paint can or sealable container large enough for your printed part. Put 1 cm of acetone in the bottom and then put some kind of spacer to keep your part from touching the acetone - metal spacers are good like aluminum. Then put the part on top of that and seal it up. Google how long to leave your part in there - I think an hour at room temp maybe? I think just a few minutes if you heat it - not sure.

You could upload it to dropbox or one of hundreds of other free services and then post a link to the video.

It's the set screws. Probably on one end or the other of the short belt. Tighten the hell out of those. Enough that you are scared you might break something. Enough that if you used an L shaped allen wrench your fingers would hurt afterwards. Hurt quite a bit.