RE: Having a lot of trouble
I understood what you meant about retraction, I was simply unaware such an approach could negate the need for a wipe tower for proper priming on unretracting.
Also, as noted above, I haven't been having as many issues with one tool prints, as those are the only ones that finish reasonably well if the dreaded USB issue doesn't rear its ugly head. On the front of the USB issue, I have reseated my LCD cable at the suggestion of Prusa chat support, and in going in there to do that I noticed that the ground wire (the yellow one, not the ground line in the ribbon cable) had come loose from its connection tab. Mine was apparently very loose as putting it back on offered almost no resistance and it easily slid off again. I squished the metal of the connector a big with pliers and then reattached it and now it's properly and solidly connected. I also bought a brand new 16gb SanDisk Ultra and tried that. None of the things above have made any difference to this issue. This is an issue either with the Prusa firmware (issue seen in 4.7.2 and 5.1.0 at this point) in how they use their USB library, or the USB library itself in how it uses its filesystem library, or just the USB library itself. It seems like the sort of thing that you should be able to detect in code and do something that will allow it to recover without user interaction. Since pulling out the drive and putting it back in clearly resets it, but the same 5 electrical connections are made both before and after that process, it should be reasonable to expect that whatever that is resetting could be reset in software without the need to yoink the thumb drive out.
My biggest concern at the moment is that the bed levelling isn't working properly at the moment. The full sheet (single color) prints reveal that the mesh bed levelling isn't working as designed, even with a spotlessly clean nozzle and bed.
RE:
This must be a machine by machine issue. I haven't had any issues and have been printing some parts that take up a big portion of the bed. And others haven't had any issues as well That being said when I started reading about some of the issues I never updated from 4.7.1 or 2.6.2.
RE: Having a lot of trouble
This must be a machine by machine issue.
I accept that definitely might be the case. What I'm desperately scrambling around trying to figure out is how to make my machine NOT be one of the ones with this issue. Thus all the questions and long stories about things I've tried. Previously I'd have said maybe a poorly assembled tool head (by Prusa) or a bad load cell that's reading a little weird but not broken. Now I've done full bed sheet prints with every nozzle and they all show the same places that are too high. That means that if it's a load cell, it's all 5 of them, and I have strong doubts about that. Since I didn't assemble the bed, and don't even really know how to do so because the manual for that is internal only, I am left with simple things like using an actual bubble level on the bed in bunches of places and facing bunches of directions. Unfortunately, that also shows everything to be plumb bob level. Which means the differences I'm seeing are derived from the mesh levelling not doing something right I think. Whether that's on the capture end (a full sheet print did 12x12 probing for me) or on the usage of that date while printing (since it was a full sheet it should have used all of it) I really don't know.
RE: Having a lot of trouble
Could you take the G29 out of the start gcode and see how it prints then? I'm not sure if the mesh is saved and therefore an old one is used. Just thinking out loud.....
RE:
Given that the build plate is flexible and the z axis shelf has 16(?) little heaters, this machine kinda requires regular probing and z height compensation unlike a machine with a rigid plate. Frankly, I really doubt that any of the Prusa XLs are going to be as exactly spot on as Nerdofepic seems to think it has to be. The larger your layer heights on first layer the looser the tolerance. That’s why we get a first layer height setting with all slicers. Nerd needs to learn to use this setting.
If it were my machine (and it really was a problem) I would locate exactly where on the bed the compensation is out of tolerance(assuming his is). Then I would try a small print right on that spot to see exactly where the fw probes. It’s entirely possible that the probing as a function of object to heater location is not matching up exactly and as such the fw is not getting an accurate “mesh”(another Prusa misnomer).
That said… I have found that with the rougher textured build plate they gave us I do not get nearly enough stick. As such I refer back to a quick shot of Aquanet (please note for those here who only know Prusa printers… I did not write just Hairspray I wrote Aquanet, as in Super Extra hold, unscented. Nothing else has the patented liquid vinyl).
RE: Having a lot of trouble
Ok, there's no reason to be unpleasant about this Tobycwood. The reason I believe it's required that the first layer be perfect is two-fold. First, because it is literally how Prusa is selling Nextruder on both MK4 and XL. And second, because I've seen the Z variances cause problems. Odd looking bottoms is the least of my concern. I've seen useless brims that are too high up on one side to even form a proper brim and instead just form a group of concentric loops that aren't connected together. I'm obviously also annoyed with prints breaking loose (on either the textured or satin sheet).
I have not yet gotten to the G29 gcode revision, but plan to give it a try tonight.
RE: Having a lot of trouble
I commented out the G29 in the startup gcode to prevent the heat absorb delay as suggested. The single layer whole bed print was considerably worse afterwards. I don't really have enough evidence to identify if the lack of G29 caused it or if it was just statistics and random numbers making things worse than usual because reasons.
The issue is largely focused on the back 1/3rd of the bed for this latest print and was so pronounced that the lines laid down by the printer weren't even touching each other. Similar issue as described about the brim earlier where it's so high off the bed there is no squishing at all. The front of the bed is better off than the back.
RE: Having a lot of trouble
After weeks of back and forth with Prusa support, printing dozens of tests, measuring things with a carpenter square, and photographing everything I could point a camera at, it has been determined that I genuinely have a warped bed assembly. No one has come up with a useful theory about why the bed probing can probe successfully, but then not print successfully. Using the "secret live-z adjust" menu option mid-print demonstrated that after a full probe, I had areas that were over 0.700mm lower than the printer thought. This also indicates that the dynamic z adjusting the firmware should be doing as a result of the probe data would have the z-range it needs to handle it since I could fake it with that setting. The issue is that you can't just fix the live-z for the problem area (on every print, since it doesn't save it anywhere) without also ruining the bed or nozzle somewhere else on the bed since the bed is inconveniently not level.
At this point, Prusa has indicated they would be sending me a replacement bed assembly, but has not done so still. They appear to be wandering around the factory talking about it instead, at least so far.
RE: Having a lot of trouble
As a little bit of follow up...
- Prusa support confirmed that my bed was warped from the factory and shipped me a replacement.
- I installed the new bed last night.
- Due to the delay for the shipping and Prusa deciding what to do, my XL5 was still on the -alpha2 firmware and I was still using the -beta PrusaSlicer.
- Without updating anything, I started a ~3hr print which stuck perfectly to the bed (huzzah!) but USB failed roughly half way in.
- Using flush cutters I hacked off blobs caused by the failure, pulled and reinserted the thumbdrive, and managed to mostly salvage the print.
- Thanks to it sticking to the bed properly (because it's no longer a warped bed) I was reasonably satisfied that I could probably print something successfully now.
- Next, I updated to the full release version of PrusaSlicer and updated my XL to the full release version of the firmware.
- Then I sliced up a large model (~17hrs) using almost all default settings. (I switched to organic supports from the other kind, but otherwise didn't adjust brims or temperatures or anything else.)
- I put it into the printer, and hit go.
- It's been roughly 12 hours now of printing this job and as far as I can tell it's printing perfectly. Seems to have stuck to the bed perfectly, hasn't had the USB failure issue, and appears to have dramatically less stringing than previous prints.
The bad news is that my original factory produced bed assembly was warped and USB issues with the firmware were blowing things up. The good news is that the new bed assembly seems to be working great, the latest firmware seems to have corrected the USB problems, and the latest slicer profiles and version seem to have stopped prints from looking like Afghan Hound.
Thanks everyone for joining me on this wild trip.