gr5

GREAT PHOTO! Each layer is supposed to move the bed down by 0.15mm. If the bed instead moves down 0.1mm only then you will get 33% over extrusion. If the bed catches up on the next layer and moves 0.2mm then you will get 25% underextrusion. This is why I keep expecting it to be a Z issue. The Z axis on the S5 appears to be 400 steps/mm. Or 0.0025mm per step (actually .0025 per substep). Due to "zebra stripe" issue I could imagine maybe an error of up to 3 steps or 0.0075 but that's about it. That's still only 5% error maximum. You are seeing more like 20% error. So I don't think it's a rounding issue (where it skips a step) or a weak servo driver (where it can have a step or two of play). Because the pattern is the same all around I don't think it's backlash/play. I don't think it's the gcode as I've looked at that through multiple viewers. And flow color would be different on every other layer. So that still leaves temperature or Z backlash. Z backlash makes total sense - it approaches every other layer from the opposite direction. What I don't understand is why it would be better with a larger cube or why we can't see it when I manualy some hops. This is just such a good fit to the observations! But doesn't fit all the data. Temperature - well this makes a lot of sense - you get better/more flow when the nozzle is hotter. What's happening is at colder temperatures there is more pressure in the nozzle so the feeder works harder and the stepper doesn't lose steps but the metal wheel that grips the filament slips a bit more so it extrudes less. But it still has the problem even when you printed a larger cube that has more time to adjust. FAN So the fan goes to 50% power on the PVA layers. While it's printing PVA it is cooling both nearby towers. One of the towers might be directly below the fan hole. Maybe the "tower" shrinks more when the fan is on higher power? I don't know this doesn't make much sense but when the bed is close to the nozzle the air bounces off the bed and back to the nozzles and can cool the nozzles a bit. Try setting the fan to be always either 100% or always 50%. There are two tabs (1 and 2) for the left and right cores. Make the fan speeds the same. My printer may be unusual but there is no difference in fan speed at 50% versus 100%. You have to get down to around 30% where you start to hear a difference in fan speed. I usually print with 100% or 10%. But my printer is probably unusual regarding the fan as I got an early "beta" S5.

I hate mailing these printers around as the postal services can easily warp your printer enough that it looks fine but now prints everything crooked. So much easier to just replace the Olimex board. I think it's 4 screws to uncover it on the bottom? And then maybe 4 more screws to disconnect the board. There are a few cables - just take a photo before disconnecting anything - maybe 4 cables? It's probably easier to do this than to package it up properly for shipping back to the reseller.

I'm guessing the print bed temp is well over 60C for these prints? The bed has to cool to I think 50C before heating up again which is for safety reasons but really crazy in this example which happens to me often. I was even putting a towel over the print bed to slow down the cooling because I didn't realize it was waiting to cool the print bed (message was not clear - didn't mention that it was waiting to cool) so I even made the problem worse until I figured it out. If this is the issue you are seeing then you can save a LOT of time by disabling this cooldown feature using this very simple tool here:

Note that it's only vertical holes. Horizontal holes are fine.

Okay so It's interesting that the speed thing made a difference. So you didn't notice any difference in play between the rotations around the two axes through the print head? Even on my printer there is a tiny bit difference. It's noticable. But I don't think it's a problem. If you look at your prints that have the problem - is it equal on all 4 sides? Do the sides that stick "out" continue or does an "out" side on one face turn into an "in" side when it goes around the corner?

I can't duplicate this every-other-layer-flow-change. I tried messing with nozzle size and speed and using your original project and I can't duplicate this. 1) What version of Cura are you using? 2) Please post a project file (menu "save" "project file as") that shows this issue.

This is expected. It's not underextruding - it's just printing slower. Because of the setting minimum layer time. What??????? That looks like the problem right there. Let me try to duplicate. By the way in cura you can show speed as well. How come I don't see that in your earlier "red" photo? Okay let me experiment...

So I don't usa a lot of PVA. I didn't know about the 55C issue. But I dry Nylon at 70C maybe 10 times per year and overnight at 70C doesn't penetrate very far into the spool but if I unspool it I can get it nice and dry in just 4 hours. I suspect drying PVA at 50C overnight while tightly spooled will not penetrate more than the outer layer. I also suspect that even if you unspool it's going to take 4 to 15 hours but I don't have any data for PVA. Just Nylon. So I recommend you do 50C, unspooled (in Cura you can see how many meters are needed and just unspool that much, overnight, on the bed under towels or in an accurate (non gas as that is humid) oven as a test to see if this solves your printing issue. Note that PVA and PLA do not stick together all that great. PLA usually prints over PVA reasonably well but the other way around not so much. It usually works but it is a bit tentative. This is why the PVA profiles in cura all have "support horizontal expansion" set to 3mm - to get the PVA to stick out and reach down to the print bed (for most models but not all).

Usually (always) the built in flash drive gets some bad blocks in a position preventing it from booting (these same bad blocks in a different location probably caused the strange errors you saw earlier). You should try repeating the unbricking procedure several times. Never letting it sit doing nothing for more than 5 minutes. Try as many as 10 power cycles. It's amazing how many people say it works now after 3 power cycles. But if it isn't getting to the progress bar and only showing you linux text and the robot then repeated power cycles won't help and something else is wrong. I describe a few issues (link below) in my guideline. If this is a company printer and you have more money than time then just order another main computer board. The "olimex" board. Aka the "linux computer" board. If you have the time and want to save some money you can read my guideline. Or at least buy a serial-cable-f to see what is going on. Or you can try my alternate unbricking suggestions. Or try the latest firmware. Which version of the firmware did you try to install? The newer, the better as the newer versions re partition the internal flash drive but the older firmware versions don't usually do that. Try to get version 7.0 or newer. So here is a LOT more detail on getting your UM3 running without spending more than 20 euros: http://gr5.org/unbricking/

What? That's crazy. 1) What version of firmware do you have on the printer (you can see it in the maintenance menu on the front of the printer). 2) Could you please post both the ufp file and the project file here? 3) Do you print using USB flash or through digital factory? I'm wondering if you have something wrong with your printer profile. For #2 project file (which will include your model - I hope it's okay to publish) do "file" "save project as". For the #3 ufp file hopefully it's still on your flash drive so you can send it to us - a ufp that definitely has the issue. By the way at the end of the print it should RETRACT. Not purge. It cools down both cores to a specific temperature. When it reaches that temperature it retracts. This keeps the nozzle cleaner and the shape of the filament cleaner and it makes it easier to remove a core or to remove the filament later without having to heat things up. It should not move the head around while doing any of this.

What issue exactly? You aren't happy with the quality? Maybe you should be more specific. For example in the arcing area - underneath the arc. That looks typical. It's difficult to print overhangs like that. It's normal for the quality to be worse under overhangs like that. the arc shows how steep you can print up to (looks like you can print right to the end! Very good.) and the other shows "bridging" which looks reasonable there. You can probably do better by tweaking some things in cura. And printing farther from the print bed might help as the heat might be part of the problem (not sure).

I reread your original post. You are right - it is strange that it can read the USB flash drive somewhat but not completely. So you can go into the maintenance mode and you can tell it to write all the log files from the UM3 onto the USB flash drive. And then post all those log files. Particularly the ones that end ".log" or ".log0" (or is it .0.log? something with a zero near the end of the filename) which are the newest. I'm wondering if the USB drive is fine and Linux is messed up. I would try reinstalling the firmware using the "unbricking" method. I'm wondering if your "hard drive" on your UM3 (it's actually eMMC memory) has some bad blocks. Fortunately it will reroute the blocks automatically as long as you reinstall the operating system (aka firmware). Instructions for unbricking: https://community.ultimaker.com/topic/20024-recovering-a-bricked-um3s3s5/ There are newer images but here is an older one. UM3 V5.2.11 image: https://software.ultimaker.com/releases/firmware/stable/5.2.11.20190503/recovery-5.2.11.20190503.img Image for um3 extended is the same as for um3. I think the log file might distinguish if the problem is with the firmware or with the USB wiring. It could be loose and intermittent.

Really good points! yes. It should probably be removed from Cura as I think it's one of those causes for: when people say "Cura is really slow". Sometimes I think it's trying to talk to their weather station or their mouse or something. Removed or at least have USB searching disabled by default. Oh my god. What a drama queen. The arrogance was only in your own mind. I hear your anger but it's just that no one here listening deserves the focus of your ire. Yes let us hear your frustration and motivate us/someone to fix it or at least think about fixes. But don't get mad at *us*.

Looking at my part again - the left side is the worst, then the front is slightly better, then the right side is definitely better and the back is definitely the best. By far the rear side is the best. Why? It prints back first, then left, then front, then right. What??? If this is a Z issue it should have equal issues on all 4 sides. If this is a "play" issue it should be worse on two opposite sides. If it's an underextrusion or overextrusion issue related to the hot and cold ends only then it should get better or worse as you go in order (not best at the first line segment and worst at the second). It could also be a speed change issue (partly) as it is printing the infill at a different speed than the outside. SPEED ISSUES Try setting all the speeds the same. Search for "speed" and change the first 10 results to 35mm/sec. Once you get to the "travel speed" leave the rest the same. Other flow rate issues - line width Also search for "line width" in the settings search box and change them all to 0.4mm and change the wall width (last setting I think if you search for line width) to 1.2mm so that it is a multiple of the line width. Anyway do all that and look at the part using "flow rate" colors in PREVIEW mode and make sure it's all the same color and try printing again to see if your lines go away. It will definitely change things as when it starts the outer shell it won't have a speed change which requires a change in nozzle pressure which takes a little while to equalize. I have to say that Ultimaker gets these "genius" (not) ideas - they changed the line width on the S5 to be 0.35mm for a 0.4 nozzle because they found that on average nozzles tended to be a bit too small versus too large and also they tend to get buildup of caramelized filament inside the nozzle tip. But in my experience, 0.4 works better than 0.35. And they seem to agree now because they came out with the "engineering" profiles which use 0.4mm and parts are more dimensionally accurate. But once you create a profile with lines at 0.35 you can never change it! 5 years later they will all tell you to use 0.4mm but they can't change the profiles because if you change a profile it will mess 10% of the people up and make their part different which will piss them off ("after 5 years you change the profile on us! Now we have to redesign the part! You just cost us millions of dollars in downtime!"). So they can't improve the old profiles and rarely (very rarely) come out with better profiles like the engineering profiles. I'm glad it's not my job to create the profiles that can rarely ever be changed. And can't incorporate new features that improve prints. Before changing line width and speed: After:

Have you tried using a different USB flash drive? Just because it works fine with your PC doesn't mean it's not a faulty USB flash drive. The UM3 has a linux computer in it. If you are good with linux you can ssh into the computer and look at log files. I could help you with that but only if you are good with linux. However I suspect it is going to just report errors with the USB flash drive. I'd check the contacts (blow on them? Check cabling maybe? I don't remember if the usb connector is part of the PCB (soldered right on the PCB) or if there is a cable to the USB jack). Two possible paths to go down (of many): 1) Get a new linux board for your UM3. They aren't cheap - contact your reseller for a price. 2) Put the printer on the network and going forward always use digital factory to print. There are more involved debugging paths that aren't worth it if this is a business.

I'm not convinced it is head play as it should only show up on front/rear face of the part. But it's worth testing. It still looks like a Z issue. It makes total sense that it is a Z issue. If Z doesn't move enough you over extrude a little as there is less space to extrude into. If Z then catches up it underextrudes a little because now there is more than the nominal 0.15mm of space to extrude into. Z issues make sense! Z play. Z backlash. In the PVA region the Z axis is approaching the new layer from opposite directions (it moves the bed down away from nozzle while changing cores so moves the bed up to the layer 2mm a moment before printing again. Then when printing the next layer it just moves the bed down by 0.15mm to print the next layer. This repeats approaching every other layer from above and below and I'd be surprised if it didn't have every other layer sticking out like that. Yet you tried to add weights and it didn't help. And I added 2 or 3 zhops to a tower gcode file and it didn't have the layers sticking out. So strange.

Okay so it's not temperature. It's not Z axis. I think it's head play. I think it's the bearing between the cores. I have a test for you. So I don't understand this but the problem happens when you do the core change for the right core. There is more stress placed on the print head when changing cores. There is a known problem with UM3, S3, S5 print heads with the bearing between the cores. It is quite short. There are 2 rods that pass through the print head. The one that passes left/right goes through a bearing that is quite long - as long as the print head is wide. However the other rod goes through a short bearing. This bearing is held in place by the plastic of the print head. The plastic can get old and crack and get a little loose. The most common symptom is head backlash where the head rotates around the rod that travels left/right through the print head. This shows up in complicated patterns where some layers jut out just like in your photo of the larger part. If you print a layer clockwise versus counterclockwise or if you move the head inwards to print some detail, the head may be in a different angle/position and you get some layers that sitck inwards or outwards. This was my very first theory but it didn't make sense for your part so I didn't mention it as I was looking at the pattern of printing (which corner was printed first, did it approach from the left or the right) and didn't think about the lift switch applying more force. Also the pattern usually only appears on the sides facing the front and rear of your printer. So I could be wrong about head play. Grab the print head and twist it gently back and forth using a "roll" motion like the print head is an airplane and it wants to bank left and right. This is quite stiff. Now do a "pitch" motion where the airplane wants to pitch nose up and nose down. This always moves slightly more but not enough to cause even tiny errors on your print. Unless your bearing came loose. I don't know how to describe how much is okay and how much is bad. The fix is to open the head and put in a shim using paper or something. Some brand new print heads already have the problem but most acquire the problem after months of printing. I've been printing for years and my S5 and my UM3 are both fine. A friend had the problem from the first day and he fixed it within the first week with a shim. So here's my test. Print just a single tower. A 10mm wide tower (maybe 30mm tall). After it has been printing for at least 2 mm, push the "lift switch" on the right side of the head so that the right core moves downwards. As long as your tower is only 10mm wide and you are at least 2mm above the glass, the right core shouldn't hit anything. Let it print 2 layers and then push the "lift switch" back again to raise the right core. Let it print at least 2 layers. Repeat this a few times. See if you get the same issues. However keep in mind that my S5 head play is pretty good and I get a similar pattern as you do and you think mine is not as bad but I am not convinced.

I wouldn't recommend preheating PLA. I print old PLA all the time. Some brands will get brittle if you leave them unspooled for a day or two but only the portion of the filament that was unspooled. It has to do with the mechanical stress of straightening the filament. I think it probably gets microcracks which simulates brittleness. I'm not sure if your parts are typicall of what is expected or not. But I think your printer is probably working fine. Not certain but it looks okay. Did you order a new sensor or heater? More often it is the sensor the breaks but it's possible for the heater to break as well.

Everyone will recommend what they already own. By now you probably already bought something. For around $700 USD you can probably get a used Ultimaker2go. It's a small printer but I think maybe big enough for your needs? It's very durable and easy to use. Very good quality prints. But if your budget is even smaller I recommend a Creality Ender3. This company is chinese and among the "crappy, cheap, chinese printers" it is the best choice. If your business doesn't do well you won't have lost much money. If your business does very well then consider getting the UM2go.

Not parasites, no. You are all very welcome to use Cura and Ultimaker tries to support what it can. And Ultimaker appreciates the feedback to help them make Cura better. Some of the contributors to Cura (for example ahoeben above) are not employees of Ultimaker. Sometimes companies that sell printers (like Lulzbot) contribute to Cura. But it's rare. Most companies can't even be bothered to contribute a profile (often those come from users of the printers). But most of Cura is done by Ultimaker employees and their bosses are Ultimaker employees and their bosses have a HUGE list of desperately need features. Features needed now or it could impact sales. Printing over USB is definitely not on that list. Sometimes a UM employee might fix a feature on their own time. During non work hours. Sometimes someone who wants to add a feature will do it and the Cura team will pull that feature in. But printing over USB is trending downwards. TCP/IP (ethernet and wifi) is this more popular, newer technology. I hope all the printers standardize on that and I hope Cura interfaces with it.

This sounds like a very recent hardware problem. Try printing the exact same file through something else like pronterface and see if it's a problem that just started today and unrelated to cura.

I thought of another potential test! Set these all to 200C. Leave the 60's alone but set the 190,185, and 100 to all be 200C that way it shouldn't cool down when printing the PVA. The downside is you get some leakage. Hmm. If you get leakage that CAN ALSO cause the next layer to be underextruded. No that's a bad test. A better test is to print something with more infill - like 10 seconds worth - so the nozzle gets to equilibrium before printing the outer shells. OH ONE MORE THING - there are 2 parameters that you didn't touch: outer before inner walls infill before walls both are unchecked so it does inner wall, outer wall, then infill. So amount of infill doesn't matter much. So the advantage of the 30mm cube is you have more inner walls to print to get up to the right temperature and pressure. This is advisable because if you print infill first you can (just barely) sometimes see it on the outer walls. But still you could try printing infill first and do more infill. As an experiment. To prove it has nothing to do with the Z axis. I would just try printing a 30mm cube. 20% infill is plenty (to support the top skin).

The authors of this feature (cool one head while the other prints) did it for good reason and they assumed the part was a bit bigger and it would stabilize on temperature while printing the infill. It's okay if the infill is slightly under or over extruded so you print that first. They didn't count on your infill taking about 1 second! Lol. Just not enough time for things to stabilize. Cura does all manner of speedups related to the nozzle temperature. When it is getting to the end of a layer and about to go print the pva it starts the cooling something like 6 seconds early so that the nozzle won't leak as much. When it is finishing up the pva it starts heating the pla nozzle something like 15 seconds before it is needed. Then at the actual moment of switching there is a wider temperature tolerance - it can be something like 3 degrees to cold. Something in all that is not working - they weren't expecting 3 seconds to print one tower's layer! So it's doing the cooldown later than it's supposed to (as the PVA layer doesn't take 15 seconds - it takes 3 plus maybe another 3 to switch nozzles) and heating up too late as well (although it should pause the printer if the nozzle is significantly too cool). Anyway this all doesn't work so well when the layers are less than 15 seconds to print. So there is a PID controller that controls the power to the nozzle heater based on requested temperature. There are 3 values: one for each P, I, D (proportional, Integrated, Differential). You can actually play with these values for an individual core. The values can be stored in the core. Ultimaker does not do this - they set it the same for all AA 0.4 cores. but if you get a heater that is a few watts higher or lower then the PID could be slightly out (versus mine which exhibited similar issues). Anyway the point is not necessarily every core is the same but most should be very close. For example the 3dsolex cores have different PID values than Ultimaker cores because they have much larger heaters (I think they are 35 Watts? Versus something like 25 Watts? I forget). So the PID values are stored right on the eeprom chip on the cores for 3dsolex cores.

So at first I was like - "wow that's smaller than I expected" and also "Nope - looks fine - my machine doesn't do this". Also by the way I rarely use PVA lately (I like breakaway) and it's been about a year since I used it and my PVA got "wet" despite being in a sealed bag with lots of dessicant. It needs to be dried but that doesn't matter because the PVA isn't the issue. By the way you set "support horizontal expansion" to zero. It defaults to 3 for a good reason - this brings the support down to the print bed and makes it MUCH less likely to fail. but your print didn't fail so whatever. As you can see it recovered. So then I took a more careful picture (and looked with some magnifying glasses) and I could see - my printer does indeed do it. Also I noticed that on the upper layers it was 3 seconds per layer per tower. 3 seconds!!!! This is important. Anyway so my printer does do it as you can see: I'm not sure if it is as bad as yours or about the same?? Anyway - it could be partly Z axis but I think it's temperature. At 3 seconds - that's not really enough for the print head to come all the way back to the eariler temperature and as it does the second layer (it does 2 in a row in the PVA section) it's now warmer - it has had 3 more seconds to heat up. So that's what I think it is. To prove it do a 30mm cube (instead of 10mm) and make sure there is at least 30% infill to let the nozzle heat up to proper temperature before printing the outer walls. ALSO - I didn't realize how subtle the issue is! I can't even see it without reading glasses. Plus I have to twist it in the light just so (which everyone naturally does when inspecting something but if no one is really looking for defects they are unlikely to notice this).

I just printed it. Wow. Yeah mine does it also. Okay so...