Avisos
Vaciar todo

First layer is consistently poor in the same area  

  RSS
JP Powers
(@jp-powers)
Eminent Member
First layer is consistently poor in the same area

Hello all. I usually lurk on these forums but I've been slamming my head against the same issue for months and it's time I ask for help from the community.

My first layer is consistently something of a mess, in my opinion at least. I've contact Prusa Research support and they said it looks nearly perfect, but I disagree. I'm attaching a couple photos. The first (IMG_20181226_134213.jpg) is a full plate print of this object: https://www.thingiverse.com/thing:2741800 , the second (MVIMG_20181226_134551.jpg) is what it does when I remove it from the plate.

first: https://imgur.com/3eF9idA

second: https://imgur.com/xlaVv2c

The right edge is pretty perfect, exactly what the first layer I want to see should look like. The left edge isn't as good but good enough for me considering what the center looks like. The center is a complete mess. I printed this at .2 layer height, the left is measuring at .19-.21 in my calipers, where as the center is ~.3. That's a 50% variance in layer height, creating a washboard effect. More pictures are available here, but those two are the key ones I think. The pictured print is at least the 40th time I've printed this test print, and every time it shows the same issue: the center of the bed is dipped.

The big problem I have with this isn't that the first layer is a rough surface in that area of the bed, it's what happens later. That first couple layers going down so poorly means that area of the bed is higher than other areas as the printer gets taller. I get to around 20-30mm object height and the nozzle is pushing down into the object, which tears it up and risks a jam. It's possible I've been chasing the bed level "issue" as the root cause of this larger problem incorrectly, but I cannot explain what else it could be.

I've tried the following:
1. using the firmware's bed level correction - has little to no effect. As it is adjusting the edges but the problem is the center, it's not making enough difference to alleviate the problem.
2. I changed the spaces out for M3 nylock nuts as suggested here https://github.com/PrusaOwners/prusaowners/wiki/Bed_Leveling_without_Wave_Springs . I was able to get my total bed variance from ~.5 down to below .1 after lots and lots of adjustments. The same exact problem continued to happen, though. The center is just dipped too severely and creates that wash board effect.
3. I've factory reset and updated my firmware a couple times since I started experiencing this problem, in hopes that if there was some software issue (a bug or data stored in my printer specifically that needed to be wiped) it would be corrected, but it's made no effect. I've also rerun XYZ and other calibrations repeatedly in an attempt to work around this, and it's consistently telling me everything is nearly if not perfect.
4. Other minor things: Yes, I've used both sides of the steel sheet, both have the same effect. Adhesion itself seemingly isn't the problem, the pictured test print peels off like scotch tape where it prints properly and is incredibly satisfying to remove. I mostly print via my OctoPi setup but I've also done this test print via the SD card to ensure it's not OctoPrint messing up and get the same problem spots. I've readjusted the PINDA probe a few times, and that helped a tiny bit initially. During XYZ calibration I used to have problems with the right side's paper test but after adjusting it's never a problem now after PINDA adjustment. I had a Raspberry Pi camera arm attached to the X axis motor on the Z, which also had some LED lights on the arm, so I removed that to confirm it wasn't causing sag or something, same problem. If anything I'd have assumed the left would be better than the right if were the case anyways? I've probably done a lot of other little things over the last few months that aren't coming to mind at the moment.

I'm reaching wits end. I got my printer in August of this year, and I started experiencing the problem in mid September and I haven't had a successful print since. I contacted support hoping for any assistance in at minimum getting pointed in the right troubleshooting direction if not possibly replacement parts, but was told (direct quote): "I was checking the pictures and it really looks almost perfect to me." That second picture, where you can easily see how each line is completely separate and not merged at all, is "almost perfect?" I've gone from frustrated with spending what little free time I have chasing down this problem to complete disheartened.

Am I wrong? Is this first layer "almost perfect?" Am I chasing the wrong problem? I'm incredibly open to suggestions as to what else could be causing the nozzle to drag against the print, or to help fix this first layer, or anything else, I just don't know where to go from here. I'm tempted to look for a replacement for the spring steel sheet, but I'm concerned that's only part of my problem and/or a band-aid fix when I want to fully correct this. To be frank, if it comes to replacing the heated bed itself, I'm not sure what to do... I don't want to purchase a bed for my printer if what shipped with it warped so quickly, I feel like that should be replaced by the manufacturer but it certainly feels like Prusa Research customer support won't be doing that for me.

Genuinely, thanks for any help anyone can provide... I've been chasing this for months and I just can't figure out where to go next trying to get this problem behind me. I just want to go back to happily printing again, not letting this relatively large expense collect dust.

edit: Thanks randolph.l for pointing out the images didn't attached/appear right. Uploaded to Imgur and linking to them.

jp-powers.com
Respondido : 27/12/2018 9:43 pm
Mustrum Ridcully
(@mustrum-ridcully-2)
Honorable Member
Re: First layer is consistently poor in the same area

your pictures did not post here is how to make that work for these forums.

https://shop.prusa3d.com/forum/viewtopic.php?f=64&t=26660&p=121855#p121855

Respondido : 27/12/2018 9:48 pm
JP Powers
(@jp-powers)
Eminent Member
Topic starter answered:
Re: First layer is consistently poor in the same area


your pictures did not post here is how to make that work for these forums.

https://shop.prusa3d.com/forum/viewtopic.php?f=64&t=26660&p=121855#p121855

Ah, thought it was not showing in the preview for some reason other than it just didn't work. Uploaded to Imgur and linking. Thanks!

jp-powers.com
Respondido : 27/12/2018 9:56 pm
Mustrum Ridcully
(@mustrum-ridcully-2)
Honorable Member
Re: First layer is consistently poor in the same area

your live Z is too high use the method here
https://shop.prusa3d.com/forum/assembly-and-first-prints-troubleshooting-f62/life-adjust-z-my-way-t2981.html

and of course even though your plate seems to be clean give it a clean with a dish washing up liquid like Dawn in north america or Fairy Liquid in the EU wet wad of paper towels and squirt the stuff on it and scrub then rinse under the running faucet and either dry with paper towels and then hold only by the edges or flap back and forth to dry... the skin oils from a finger print will spread over time... you can then wipe with Isopropyl Alcohol between prints for while but I rewash about every 4 or 5 prints I tend to take the plates to breakfast with me and was them as I do my breakfast dishes so they get a wash every day that I am printing.

Respondido : 27/12/2018 10:13 pm
JP Powers
(@jp-powers)
Eminent Member
Topic starter answered:
Re: First layer is consistently poor in the same area


your live Z is too high use the method here
https://shop.prusa3d.com/forum/assembly-and-first-prints-troubleshooting-f62/life-adjust-z-my-way-t2981.html

I saw that thread and I think I tried it but I don't recall the results. To my understanding, the Live Z effects the entire bed, right? If so, although I might be able to help the center of the bed, the nozzle would now be too close to the parts of the bed that are currently acceptable or great. Also, I don't have a lot of room to play with, mine is currently set to -0.820. Considering, although this might help my center area of the bed, it'd severely limit my printable area. Am I misunderstanding how Live Z works?


and of course even though your plate seems to be clean give it a clean with a dish washing up liquid like Dawn in north america or Fairy Liquid in the EU wet wad of paper towels and squirt the stuff on it and scrub then rinse under the running faucet and either dry with paper towels and then hold only by the edges or flap back and forth to dry... the skin oils from a finger print will spread over time... you can then wipe with Isopropyl Alcohol between prints for while but I rewash about every 4 or 5 prints I tend to take the plates to breakfast with me and was them as I do my breakfast dishes so they get a wash every day that I am printing.

I wipe the plate wipe acetone about once every 3-4 weeks and IPA 90+% wipe the plate after every print. I don't think going beyond that is necessary. Where the plastic sticks to the bed it sticks very, very well, and the entire plate is cleaned the same way. Mind if I ask what you expect this further cleaning to do? I'm just not sure what advantage it'd provide as adhesion is great when nozzle is aligned to the bed properly.

jp-powers.com
Respondido : 28/12/2018 1:00 am
Mustrum Ridcully
(@mustrum-ridcully-2)
Honorable Member
Re: First layer is consistently poor in the same area

dish washing liquid works best for me the iso just seems to spread the skin oil, from the edges where you touch to flex the sheet, about the plate. I have not yet found the need to use acetone. the 75 x 75 method if you follow the authors instructions will work a lot better than the standard prusa canned op in the firmware... if you have to use the supplemental ± .050 adjustments after that you can; if all of that does not work then you can do the manual adjustments with or with out springs where your bed spacers are. or just mike the spacers and put the short ones where the bed is high etc.... when I built it I lapped them all to the same length ±0.001mm and the used a torque driver to set them to just under 0.8 newton meter (10 inch pounds) lots of ways to skin the cat.

you can go as far as -2.000 with live zero I run in the -1.785 range now but support diagnosed a variable trip hight in my pinda and have sent a replacement it came today; so tomorrow I have the fun of changing it I picked up some 12mm split fabric for the whip and I will use it rather than that darn spiral stuff from my build but I still don't look forward to things... this was my least fun part of the kit build, since I will be doing it anyway I plan some mods to the hot end and the filament sensor as well with luck I will be printing again on the weekend.

Respondido : 28/12/2018 1:23 am
JoanTabb
(@joantabb)
Veteran Member Moderator
Re: First layer is consistently poor in the same area

live Z adjust sorts the centre of the build plate,

Bed Level correct, sorts the periphery of the build plate, afterwards

Joan

I try to make safe suggestions,You should understand the context and ensure you are happy that they are safe before attempting to apply my suggestions, what you do, is YOUR responsibility. Location Halifax UK

Respondido : 28/12/2018 2:14 am
JP Powers
(@jp-powers)
Eminent Member
Topic starter answered:
Re: First layer is consistently poor in the same area


live Z adjust sorts the centre of the build plate,

Bed Level correct, sorts the periphery of the build plate, afterwards

Joan

OK, that's good to know. I'll give it another shot by going more negative on my live Z then adjust the edges after.

jp-powers.com
Respondido : 28/12/2018 5:37 am
JP Powers
(@jp-powers)
Eminent Member
Topic starter answered:
Re: First layer is consistently poor in the same area

I printed 6 of the live adjust z test squares, making changes to both the live adjust and the bed correction values to attempt to get the top left of the square right while also not letting the bottom left get too low.

https://imgur.com/a/9UoOh4T

First I just took my live adjust Z from -0.820 to -.920. My thought was to account for the .1 difference I was seeing with my calipers. It's hard to see here because of the lighting, but the top left is the first and still has a good bit of line separation in the top left (you can see the gaps along the edge between infill and perimeter, and obviously one of the perimeter lines completely lifted apart). The bottom right is pretty good, though, so I felt like I was on track.

The second I left the live adjust Z but tried using the bed level correction to see if it would have a significant impact. I went to +25 on both the left and rear. I can see a difference but I needed to go higher.

The third I figured I'd try both going up (down? negatives are weird) to -0.970 on that half, after also adjusting the bed level correction up to +40. Again, progress but not enough.

The forth I went to -0.985. I knew the bottom right would be effected so I figured I'd attempt to "tilt" the bed with bed level correction. +25 on left and rear, -25 on right and front. The bottom right is plainly too low still but this is the best I've seen the top left and it's exactly where I'd like it to be. Time to try make adjustments so the bottom right can match. Caliper measurements were .2 on the top left and .13 on the bottom right.

I dropped back to -0.960 and increased the bed correction for the left and top to +50. The top left is a little worse than the third, and it's still too low on the bottom right.

For the sixth attempt I just went full on +50 or -50 where applicable and tried to split the difference on the live adjust Z. Top left is a bit better than the fifth but not as good as the forth, and bottom right is nearly as bad as the forth.

No point in going back to the test circles yet. Tomorrow I'm going to try the other side of the sheet just to see if there's a difference. Assuming there isn't, I'm going to investigate getting another sheet, maybe one of the BuildTak ones. I'm prepared to invest a bit of money just to prove to myself whether or not the sheet itself is the problem. I'm going to go back to the M3 nylock nut leveling method and attempt these corrections again, then worst case I'll look at getting a replacement sheet.

edit: thought about it, figured I should try going back to the M3 nylock nut leveling first before committing money into a new plate.

jp-powers.com
Respondido : 28/12/2018 7:39 am
Mustrum Ridcully
(@mustrum-ridcully-2)
Honorable Member
Re: First layer is consistently poor in the same area

https://shop.prusa3d.com/forum/zusammenbau-und-fehlersuche-bei-den-ersten-druckve-f273/wiki-howto-live-z-die-last-mit-dem-first-layer-fir-t26806.html

Take a look at this page in the Chrome browser and hit Chrome's translate button (unless you are fluent in German...I can manage "I would like another beer please" and "which way to the toilet" in german so the huge advances in machine translation are a minor miracle to me :mrgreen: )

Respondido : 28/12/2018 5:42 pm
JoanTabb
(@joantabb)
Veteran Member Moderator
Re: First layer is consistently poor in the same area

is this any help guys?

use 75mm square to get middle as good as possible first,
let the 200 mm square file run a few circuits and see what happens.
stop, remove evidence, adjust bed level correction settings, re run, rinse repeat!

when happy, check your 75mm test square again

Joan

I try to make safe suggestions,You should understand the context and ensure you are happy that they are safe before attempting to apply my suggestions, what you do, is YOUR responsibility. Location Halifax UK

Respondido : 28/12/2018 9:47 pm
JP Powers
(@jp-powers)
Eminent Member
Topic starter answered:
Re: First layer is consistently poor in the same area

I'll give the 200mm square a shot to confirm my changes are good after I finish another pass or two with the circles test I was doing at first.

https://imgur.com/a/8dwV551

After reapplying the M3 nylock nuts I was able to get my reported bed variance down to 0.025 at cold/room temp and ~0.08 at 60C on the first go. Printed another 3 test squares to get that sorted for the most part and tried another circle test to check the edges. The center is pretty great, FAR better than it had been in my book, but the bottom center mostly as well as bottom right of the bed is too close and the very top left is still a bit too far. I'm running some further corrections with bed level correction at +50 on the left and rear and -25 on the right and front, but I'm way better off than I had been. I'm hoping I can back off the corrections but for this go I wanted to go somewhat extreme to ensure I'm seeing a difference.

Thanks to both randolph.l and joan.t for the input. Frustratingly (frustration directed at myself here) I had tried your suggestions previously but not all together. I think the push to just do them again all in one go made a difference.

I'm still perturbed that with such a very flat surface according to the PINDA probe I'm still needing to apply corrections and such minor adjustments. It feels like the mesh bed leveling is working against me somehow. I'd think that getting my variance below .1 even at printing temp the mesh bed leveling should be able to do the heavy lifting and bed level correction of the edges shouldn't need to be so drastic.

edit: just stopped the circles test I was doing when I realized... I think I was setting bed level correction backwards? I was setting positive where I should have been setting negative and vis versa. The problem was getting worse in the corners I was correcting for. Trying again with flipped values and it already looks better. Can't help but wonder if that dumb mistake has been a major component of chasing this down the whole time...

jp-powers.com
Respondido : 29/12/2018 5:39 am
JoanTabb
(@joantabb)
Veteran Member Moderator
Re: First layer is consistently poor in the same area

Good morning JP,

edit: just stopped the circles test I was doing when I realized... I think I was setting bed level correction backwards? I was setting positive where I should have been setting negative and vis versa. The problem was getting worse in the corners I was correcting for. Trying again with flipped values and it already looks better. Can't help but wonder if that dumb mistake has been a major component of chasing this down the whole time..

Yup, that would make matters worse...

Good luck with your retry!

Joan

I try to make safe suggestions,You should understand the context and ensure you are happy that they are safe before attempting to apply my suggestions, what you do, is YOUR responsibility. Location Halifax UK

Respondido : 29/12/2018 12:53 pm
JP Powers
(@jp-powers)
Eminent Member
Topic starter answered:
Re: First layer is consistently poor in the same area

Another test circles print this morning was still a bit too close on the bottom of the build area, but as close to perfect as I can reasonably expect to get everywhere else, so I increased the level correction a tiny bit more on the front and I'm printing some real stuff now and it's already looking far, far better than it had.

Thanks again for the assistance and encouragement, I think I'm (finally) out of the woods with this problem. Now it's time to see if my assumption that the overly thick first layer in certain spots was was caused the problems later in prints or not with larger objects, but that'll have to wait for this first "real object" test to finish.

jp-powers.com
Respondido : 29/12/2018 10:34 pm
JoanTabb
(@joantabb)
Veteran Member Moderator
Re: First layer is consistently poor in the same area

Fingers crossed for success!

Joan

I try to make safe suggestions,You should understand the context and ensure you are happy that they are safe before attempting to apply my suggestions, what you do, is YOUR responsibility. Location Halifax UK

Respondido : 30/12/2018 1:10 am
JP Powers
(@jp-powers)
Eminent Member
Topic starter answered:
Re: First layer is consistently poor in the same area

For the record for anyone that stumbles on this via searches:

I'm pretty happy with the changes. My process to get good results is listed below, but to review what was going on: I have/had what appeared to be a dip in my bed. Just to the left and up from the center point. Mesh bed leveling wasn't account for it and it caused that area of the bed's first layer to not be squished down well.

  • Switching to the bed leveling without wave springs method of mounting the bed first helped dial in the bed to a very low variance as detected by the PINDA probe. Run the Calibrate XYZ, then I used Octoprint with both the Bed Visualizer and Prusa Mesh Leveling plugins. Prusa Mesh Leveling lets me see the variance, Bed Visualizer lets me see values of individual points more easily (confirm it's "flipped" properly versus Prusa Mesh Leveling, I had to Flip the Y Axis).

  • Note about the above: I had tried applying the firmware changes listed in the github wiki above the first time I made the change to this method. It worked decently but I prefer running a "stock" firmware. The second time I applied the changes I just did some basic math. If a point was showing as higher, I'd subtract the mid point's value from the high point's value and attempt the minor change necessary, or if it was lower I'd subtract it from the midpoint's value.

  • I set the Live Adjust Z value to .850 as that's where I had it previously and got decent results in areas. I started printing the live adjust Z squares from jeffjordan. The dip was still visible, so the PINDA probe is still not accounting for it. While it at first appears like the leveling method above doesn't work, it DOES make it easier to adjust for in the firmware. I first increased the live adjust z value to get as much of the square squished as I could, then started adjusting the bed level correction values to account for changes. I got the square really close and switched to a full plate test that prints circles. Two notes about that: 1. Printing circles is a bit tougher than squares, hence why I prefer it. 2. When slicing in Slic3rPE I uniformly scale down to 99% than scaled the Z axis of it back to .2mm. Otherwise the skirt prints off the plate. I then made further adjustments to the bed level correction. Here is where I started playing with the live adjust Z and bed correction values more, getting it as fine tuned as I could. In the end I burned through like... 100-200 grams of plastic, and for something that's a single layer it's a bit nuts how many squares and circle tests I ended up doing. It was worth it though, I'm getting beautiful first layers now.

  • I flipped the steel sheet over and changed to a different roll of plastic and tried printing something pretty difficult. The second layer and beyond was going down pretty poorly but I let it go anyways, and it caused failures later in the print hours later. I tried printing something easier and the second layer was worse, so I stopped the print right away. I flipped the sheet back over and changed back to the spool I was testing with, printed the same gcode (filament settings are the same for both spools), and it was great. While it was printing I was looking at the failure of the second layer. I figured the filament was printing too cold because it wasn't going down as smoothly and curling up in some spots. Turned up the heat in Slic3rPE for that filament by 10C. Tried another model and the second layer was much better, let the print finish and it's great.

  • I've printed a few more things and it's much, much better than it had been. I tried a sample spool I had from MakerBox and while super stringy and some of the supports failed it came out pretty great. That's actually why I'm not a fan of MakerBox and canceled my subscription: You don't get enough filament to actually get it right and get something out of it, it's one or the other with only 16m of plastic.
  • So, to conclude: Things seem better. I did a quick maintenance pass (wiped down everything, lubed the bearings, checked the electronics box for loose wires, etc.) Printing another sample spool from MakerBox and it's looking good. I wanted to review what I did here for anyone that stumbles on this trying to adjust for the same problems. Thanks again for the assistance and encouragement, guys.

    jp-powers.com
    Respondido : 02/01/2019 12:11 am
    Mustrum Ridcully
    (@mustrum-ridcully-2)
    Honorable Member
    Re: First layer is consistently poor in the same area

    And thank you JP for your report. So many folks come for help and we never hear from them until they have another problem. And yes the maker box sample are too small to develop a full set of notes on how to use them... I think they intend you to print something like a calicat as reference for what it looks like when printed but about 35 or 30 meters would be a much better sample size.

    Respondido : 02/01/2019 5:51 am
    Compartir: