gr5

200% would be more reasonable. You want to squish the first layer quite a bit. There are MANY ways to achieve this and I don't know which method is most common but I think it's by lying about the Z calibration. Let me explain. I'm going to explain with Ultimaker but it's the same idea for any printer. Normally if Z=0 the nozzle is just barely, perfectly touching the bed everywhere. The first layer in cura is typically 0.3mm (or 0.27mm often - close enough). Cura normally then moves the Z to 0.3mm and prints with enough filament to fit into a 0.3mm height. So far everything sounds normal/perfect/nominal. However that will result in prints not sticking well and popping off the bed during printing. So during the "leveling procedure" instead you use a piece of paper (the calibration card) that has a thickness of about 0.1mm. When you "level" at the 3 points with the calibration card you are basically telling the printer that this is 0.1mm above the bed. So you would think tha the printer would store internally that here is Z=0.1. But no it cheats at this point and (depending on firmware version and so on) I believe it actually uses 0.07 on the UM2 (not certain as I never really cared all that much). So when you print at 0.27mm above the bed you are actually printing 0.24mm above the bed and overextruding by .27/.24 or 12% (much more than your suggested 2%). However this is horrible - parts still pop off the bed. I recommend about 200% overextrusion on the bottom layer. I just level it at the start of a print and leave it there for weeks at a time without needing to adjust it. Here's my guide to the perfect squishing on the bottom layer. You can fix it on the fly when the print starts. Using "initial layer flow" is too much work as far as I know. Except maybe for the S5 which doesn't allow you to disable auto level: ========== Examples of how bottom layer should look. These pictures are from the internet mostly and I am just guessing that the bottom layer height for all of these is 0.3mm (cura defaults to 0.3 or 0.27 depending). The numbers below indicate my best guess as to how far the nozzle was from the bed (again assuming .3mm is nominal). The yellow one is a bit too squished but at least the part WILL NEVER COME LOOSE! lol. Danger is it might pull up glass when you try to get the print off. The blue one is about perfect if you want it to stick really well. This is my goal squishiness. The red one is okay for small prints but for large prints they may come loose. This is typical if you use the calibration card or auto level. The black one is leveled much too high above the bed. I aim for red or blue depending on my goals. If I want it to stick better I squish a little more. If I don't want the bottom layer to have a tiny brim I usually just set initial horizontal expansion to -0.2 or so but sometimes I instead go with what you see in red below. For example if I'm printing tiny gear teeth, initial horizontal expansion of -0.2 will ruin them so I might go with -0.1 I.H.E. and level like the red one. I usually adjust the 3 screws (BY EQUAL AMOUNTS) at the start of every print although once it's perfect I don't have to touch leveling for months.

The S5 has multi-point leveling. The S5 and UM3 use the same firmware. I'm hoping they release multi-point leveling for the UM3 some day and then it won't matter as much if your glass is warped a little. Although of course the bottom of your (very large) prints will be shaped like the glass so this isn't a perfect cure. But it will at least allow parts to stick well in the rear 2 corners.

You might need to get a um2go if you want dimensional accuracy better than this. I'm not sure which of the json files to edit either. You can do a grep or guess. I'd just try editing both. You should definitely buy this recovery cable from olimex. I sell some in my store but only for USA customers. The company is in europe. Since you know Linux I can explain that if a json file is corrupted (missing a comma, or semicolon, or quote) the linux box will boot but fail before ssh shell is started so recovery is impossible with ssh. But if you have the olimex cable it's easy to login and fix the json file and reboot. There are 3 ways to recover I believe - olimex cable gives you a serial console. You can connect a small HDMI cable to a computer monitor and watch it boot and I think you can probably connect a keyboard to the USB and you can then type. Finally if linux is completely crashed you can reinstall all the firmware with a micro sd card. These are explained here: https://ultimaker.com/en/community/51752-recovering-a-bricked-um3 As far as play. You want the long belts to be at about 100-120Hz. If they are too tight it might increase friction and increase backlash/play. You shouldn't have to mess with them as they have a spring inside the black box that the belts go through but you might have to loosen a pulley to equalize the tension if the belt jumps a tooth (happens to me when other people transport my printers and they grab it by the cable). ALSO make sure the short belts are tight. You can loosen the 4 screws on the stepper from the outside and then push down hard - maybe 2kg of force - on the stepper while you re-tighten them. You don't have to remove covers to do this - just use a stick to push down on the stepper. It's also possible to have play in the print head bearings - one of the bearings isn't all that tight in the print head on some UM3s. But I think that's been improved - not sure. Ultimaker keeps improving their printers without making a big deal about it. Printers from China of course are going to be better quality for the price than a european printer because labor is cheaper there so you won't always get better quality. Many China printers are really crap but some are probably quite excellent. Smaller printers tend to give you better tolerances. I love my um2gos as they print so beautifully. My UM3 is great because it prints PVA.

meduza instructions and files here: http://gr5.org/med/ iRoberti feeder is here: https://www.youmagine.com/designs/alternative-um2-feeder-version-two I also bought a cheap 20 euro heater and soldered in my own pt100 temp sensor. I prefer not to ship to europe but if you want my new custom um2go heater for $30 USD and my $19 USD meduza kit (just as the screws and pulleys and belt and so on - the hardware) I'd be willing to ship to europe. Shipping approx $20. I also installed an olsson block for my UM2gos so I can do different nozzles. 3dsolex sells those. I absolutely love my UM2gos. I have 3 of them and they get used the most.

The filament is coming out of the nozzle just fine. You don't need to do cold pulls. What I see in the video is exactly what you expect when the leveling is too high. Just run the leveling again. Or don't even do that - just turn the 3 screws CCW as seen from below about a half turn to raise the bed up a bit. You can play with these as it's printing. Or you can push up on the bed to see that I'm right for a few seconds. I demonstrate this at about 3:14 in this video:

You don't need to measure this 14mm distance. Watch this video and skip to the part where I level the UM2. UM3 is the same except there is a final extra step for the second core. Jump to 4:54 (4 minutes 54 seconds): For my UM3 I usually use the calibration card - I make sure I feel the same resistance at the rear, at the front two screws and for the second core. Also around 3:14 you can see how to level "on the fly". Once you get the leveling good it will stay there for weeks and months without needing updating. That's one reason I don't care about autolevel.

On my UM3 I actually bent the aluminum plate up in the rear corners. I just used my hands and pulled very hard. Later I realized that on the other side of that aluminum plate are circuit traces for the heater bed and I could have broken a trace and broken the heater for the heated bed. But anyway it worked and my glass is very flat now on my UM3.

In the last photo that looks like too much warping of your glass. This will be a problem if you try to print in either of the 2 rear corners but not a problem for small parts. Please remove the glass and test the glass by itself (the aluminum plate can pull the glass into this shape so see if it's the glass or the aluminum that has most of the warp). If the glass has this much warp then tell your reseller you want another one or you'll return the whole printer. Meanwhile keep using this glass until you get the replacement.

Autolevel can be nice but I recommend manual level. Use the calibration card. Then once print starts I often change it a little bit by turning the 3 screws an equal amount to move the glass up or down depending. You want it squished well or the part won't stick well. If you squish too well it's hard to get the part off (if it squishes until transparent then that's too much). === Examples of how bottom layer should look. These pictures are from the internet mostly and I am just guessing that the bottom layer height for all of these is 0.3mm (cura defaults to 0.3 or 0.27 depending). The numbers below indicate my best guess as to how far the nozzle was from the bed (again assuming .3mm is nominal). The yellow one is a bit too squished but at least the part WILL NEVER COME LOOSE! lol. Danger is it might pull up glass when you try to get the print off. The blue one is about perfect if you want it to stick really well. This is my goal squishiness. The red one is okay for small prints but for large prints they may come loose. This is typical if you use the calibration card or auto level. The black one is leveled much too high above the bed. I aim for red or blue depending on my goals. If I want it to stick better I squish a little more. If I don't want the bottom layer to have a tiny brim I usually just set initial horizontal expansion to -0.2 or so but sometimes I instead go with what you see in red below. For example if I'm printing tiny gear teeth, initial horizontal expansion of -0.2 will ruin them so I might go with -0.1 I.H.E. and level like the red one. I usually adjust the 3 screws (BY EQUAL AMOUNTS) at the start of every print although once it's perfect I don't have to touch leveling for months.

Also stay at least a half meter away from the printer. Just touching the glass will mess up leveling and being 2 cm away can mess it up. Other electronics within an inch of the printer can mess this up. However usually these things mess it up the other way - it never gets to the glass before it thinks it does. I've never seen this particular situation. Even with a broken wire in the head going to the leveling sensor.

This is a common problem because when UM printers are shipped I think they use elephants to move your printer from one truck to the next. Seriously though - it's a common problem and you just have to realign it - it's pretty easy to do. More serious is if the entire frame is warped so check that the frame is also perpendicular/square.

I don't think I've ever printed HT. I certainly don't know it very well so I can't help you. If it's splitting then it's indeed bad layer adhesion and you want to enclose the whole printer and lower the fan speed to maybe 30% but then the tops of your horizontal holes will sink down a bit.

That circled relay is K1. That died on my printer (it was intermittent - I actually have a video about it demonstrating the problem - I could wack it and it would work again). But I'm 90% sure that's unrelated. You hear it click because the meanwell brick is turning off for a few milliseconds. I have about 6 of these meanwells. Some are GST like yours and some are just GS. The GST ones aren't as reliable. Anyway you should get a new one. They will turn off for just a few milliseconds but long enough to crash the arduino (and make K1 make that click sound). If the problem was only K1 you would lose some steps (if you were printing) but that does not affect the screen or the arduino - they have power in either position of K1. Don't worry about K1. It's the power brick. You can buy these on the internet or if it's still under warranty get one from your reseller for free.

I don't know your printer but if it was an Ultimaker I'd say one of the belt pulleys are loose. Each axis has 6 (six! not 4) pulleys. Usually the ones that need tightening are the two on the short belts. In fact it's usually the one on the stepper which is the hardest to get to.

I suspect it's the top half of the hole. I suspect this is normal for FDM printing because you are printing overhangs. Usually I just design the hole to be a bit larger on the top or I drill it out after I print it. One solution is to use PVA support but that would require a UM3 or major upgrades to your Um2. Cooler temps help somewhat: more fan helps - make sure both side fans are at full power. If you can lower the room temperature that won't hurt (put the printer under the air conditioner). You can try lowering the nozzle temp a bit - maybe 10C. But then you might get underextrusion so you could experiment. You could lower the bed temp a bit but then parts are less likely to stick. Some filaments print better than others. Maybe try regular colorfab? Or if you need high temp then try nGen (but with nGen I wouldn't mess with the temperatures -- just maybe try faster fan - maybe).

Two things: 1) All gcode files created by Cura 15.* have all the settings stored at the bottom of the file. Cura has a wonderful feature called "load profile from gcode" in the file menu. So find an old working gcode file and load the profile from that. 2) I suspect this is not a cura issue. I could be wrong. It is more likely a problem with data transfer to the arduino. If you print through USB then I suspect the USB cable. If you print through SD card then I suspect the SD card or the SD card reader has dirt in it (that happened to me - a hair in the SD card reader). Is it possibly you changed something in machine settings (in cura)? If so then that won't get set back to the way it was if you load a profile. I suggest you print an old gcode file that you know works for sure and if I'm right (#2 above) it will also fail to print.

Also maybe set all the printing speeds to the same speed and set them all to 35mm/sec just to see what happens. If inner shell is higher speed than outer shell then you can get a bit of overextrusion when it switches to the lower speed because it takes a while for the pressure to equalize in the print head.

Try turning off "outer wall wipe distance". (set to zero). Let me know if that helps. I'm a bit skeptical about that feature but haven't messed with it. Let us know. Also you can try cutting the speed in half just to see if that helps. That often helps. You can do it in the TUNE menu. Not just the outer shell speed but all the speeds as if it's printing faster on inner shell then it tends to over extrude when it hits the outer shell.

So in the very last photo the right edge looks worse than all the other 3 corners visible. I'm thinking that is probably the "z seam" which is a bad name - basically where the layer's outer shell starts and ends - for those layers where that edge sticks out more. Could you confirm that in cura layer view? So make sure coasting is off and I'd turn off "outer wall wipe distance" to see if that helps. The other thing to try - I may have already mentioned this - is to cut speed in half and lower temperature by 10C just to see if that helps. Even if it's not a good final solution it may lead you to new ideas.

Okay good - that's normal. Did you try both sides of all 4 belts? (8 tests)? I suspect the belts are fine anyway. when you pluck a belt it's best to pluck in the center as that limits the over tones. If the primary frequency is 120Hz then there will be also frequencies at 2X 3X 4X 5X 6X 7X etc but those are fainter. So 240HZ or 360Hz would be what are called "overtones". The tuner app is supposed to ignore those but it has trouble hearing tones below 200Hz I think (possibly the microphone sensitivity falls off). Or maybe the rubber belt just isn't an efficient way of transmitting lower frequencies to air.

The temperature limit is in the eeprom on the core itself but I think there may be a second limit at 350C for the head. I've never tried going over 350C. Supposedly the main reason for limiting to 350C is not because UM cores can't handle it but because the head itself can't handle that much heat (obviously the head doesn't reach 350C but it reaches a temperature that is a problem for either the plastic or the electronics - not sure which). Inside the UM3 is a linux computer. You can connect to it through ssh and all the source code for that linux computer is visible python code. If you grep around you should be able to figure it out eventually. I can't help you more than that. There is also a second computer (arduino) on the larger board and that might also have limits but I doubt it. Or if it does then they are probably controlled by some new gcode from the linux computer. So concentrate on the linux computer and let us know what you find.

@Torgeir I'm not sure what your point is. The only difference I think between UM and UM+ firmware is the steps/mm for the extruder, right? If that's the only difference then you just need tinkerMarlin (as mentioned in the post above). In TinkerMarlin you can specify both steps/mm and also direction of rotation. So switching extruders is very easy if you have tinkerMarlin. I have tinkerMarlin on all 4 of my UM2 printers and it works great. You have 333 posts on this forum so I would think you know tinkerMarlin by now so maybe you are trying to make a different point?

Yes. gcodes have a checksum apparently and if a bad checksum gets to the arduino it requests the same gcode again but the checksum is only 8 bits and once every 256 times the checksum passes even though the gcode has some bad bits. The problem can be in the SD Card reader (mine had a hair and blowing that out fixed things a lot). Or it can be the SD Card itself (try a different one) or it can be in the cabling between card reader and main board. If you haven't moved that cable or taken the covers off around when this started happening I'd concentrate on the SD card reader and the SD Card.

310Hz when the head is in the corner? I... don't think so. Try making a 5 second video please - I want to hear it. Put the speaker so close the belt almost touches the speaker. I'm talking about the long belts - not the 2 short belts. Those should also be tight - 310Hz for the short belts may be fine actually. If you can show better pictures of the part in the area of the problem. Is it sticking out more? In more?

What "aluminum pieces"? I believe everything you need to buy is listed in the bom.xlsx file. Everything else is stl files that you can print. One of the items in the bom links to the feeders which have their own bom.