793 posts on a single topic. That must be a record on Makerworld. Haven’t seen any other topic here, with so much user involvement and so many exchanges… I think BL should reward us (some spare parts or perhaps even an A1 combo would do just fine) for our unmatched and unmeasurable contributions…and sharing of invaluable know how and experience… Don’t you think so too?
1 Like
@Nebur
I dont think its slicer based/model shape because it changes with speed. From what ive gathered, its cause by the belts teeth hitting the surface of non toothed pulleys and thats why it affects core xy printers most. A vibration thats frequency changes with speed. Somehow, the frequency matches the printers resonance frequency at certain speeds (the sweet spot). I could be completely wrong though. If im right, it could be fixed by pulsing the extruder in a way that counteracts it. Like noise cancelling.
It shows up in the straight sections too, but its not as pronounced and its hard to catch both the straight section and curve in the same shot because it requires the lights reflection to really see it. Thats whybit shows more on shiny filament.
Why is that? I would think that when you get as high as a $2K-3K machine, there’d be budget to include some quality parts. X1E’s are priced at what, $2,500? Evidently there do exist people who are willing to open their wallet and pay those sums for even modest incremental improvements.
I’m not saying you’re wrong. People can appreciate better print quality even if they don’t understand the mechanics behind it, so I would think that there’d be strong demand for good hardware if having it translates into better prints.
When I look at my own X1C VFA test prints, that curved section is indeed the worst, but frankly the rest of it is has little that’s visible in most of it. What is it about the curved section that makes it so special? Maybe it is the doing of the slicer, like @Nebur said, trying to make that curve out of very short straight line segments?
On the other hand, maybe the reason doesn’t matter. Maybe only the workaround matters. At least to those of us who just need to crack on and get our printing done rather than try to redesign the printer itself. If the solution has eluded people for over two years now, I’m not exactly optimistic it will be solved on the X1C, at least not for the general population. I’m guessing most of us will probably be moving on to the next printer, whatever it might be, in the next year or two, so a solution at this point in time would be coming rather late in the game.
I wouldn’t even be surprised if it’s an artifact of the microstepping. There’s just not as much torque going into those intermediate steps as there is into the coarser, native steps. Maintaining a curve probably requires a lot more of those tiny microsteps than on a straightaway. It would be easy enough to test: swap in some steppers that have twice the number of steps per rotation and see if the VFA’s change. The Prusa’s use those, and maybe this is why. 400 hard steps per rotation instead of 200. Aurora Tech channel thinks the Mk4 has the best surface finish out of all the printers (well, aparently now tied with the Plus 4).
As everyone is waiting for what happened with the X1C to happen again. The P1S was a good step.
I actually hope that the step doesn’t come from Bambulab. I hope it comes from people who want to further develop ther Printers.
But that will take time since if I walk into a store in China and ask, for example:
Do you have silicone, the answer is 100%
- yes or no.
- If the answer was no and you ask where you can get it (and you have to hack again), the answer is in 100% of the cases, no idea. And that even if the shop next door has silicone.
- Just as you never enter the bus in line - as soon as the door opens, everyone is on their own.
If i walk into a store in Europe and ask the same: Do you have silicone? 3 answers are possible:
- Yes or no.
- No but you can get it from… (whithout asking were I can get it).
- No, but we can orgonice it for you.
As already mentioned, the head is like a muscle, it can be trained and acts as he was trained. When I think of Asian leadership structures, the game of laming always comes to my mind - no one looks back and asks. One head, one opinion, one path through the entire structure from top to bottom. We have it too, but not only. But when it comes to following, i.e. following faster and foregoing profit in order to save money, they are unbeatable.
1 Like
@Hank I 100% agree. And furthermore, colorless green dreams sleep furiously.
3 Likes
only the actual acid part. for the chat part you might get 4 answers:
- yes
- no
- perhaps
- it depends
On the VFA test, thats the section everybody pays attention to(the curve). And it is called the vfa test after all.
Thinking about it now, I can imagine that the curve VFA artifacts might (?) be traced, at least in part, to the original model file. If we could import the VFA test as a .step file (.stp) instead of .stl, it might cast some light on this issue. A .stl will have fixed resolution, whereas a .step will be a cleaner, mathematical representation. A bit like comparing pixel graphics to vector graphics, if you will, or pixel fonts to LaTeX. You would expect the curved section of a .stl file that lacks adequate resolution to be rougher than the straightaways, which is a match with what we are seeing. It would be ironic indeed if we’ve been chasing our tail over something as basic as that.
I hadn’t given any serious thought to VFA’s until the last two days, so maybe those who are deeper into it have already considered this. On the other hand, sometimes a fresh set of eyes might notice a new angle on a topic.
The more eyes the merrier. 
Ok. Now that I have both printers in orca, ill do it again today. Both sliced on the same slicer. lol. We will find a situation that makes the bambu look better, or die trying.
1 Like
If you see you don’t succeed (the 1st, 2nd, and 3rd time), try, try again (until you make it or break it).
That “resonance avoidance” feature sure seems strange. It accepts the first 5mm at the initial 60mm speed, and then it overrules the other speed settings between 75mm/sec and below 150mm/sec to all be exactly 70mm/sec. Then at 150mm and above it no longer does anything:
Does anyone see any sense in that? I don’t. At least I haven’t so far. Therefore, I’m thinking maybe it’s a feature the developers stubbed out but haven’t completed. i.e. it’s currently only half-baked, at best.
Or maybe it will make sense after more pieces of the puzzle fall into place.
There also appears to be a stark bifurcation in the layer times, which coincides with this:
How to make sense of that? The layer times should go down as the speed goes up. Right? And the speed is confirmed as increasing:
How can the speed be increasing, but the layer times remain constant? The math seems broken. Am I wrong? On first blush, it seems self-inconsistent.
Well, you’ve already got a point - First layer and Spaghetti Detection 
But very few value this, so I have no choice but to continue to annoy myself with the X1C since if no one is intressted on it never will get included as an standart 
But again, thank you very much for your approach, I’ve been reading along 
One point next year will also be very interesting: Will QIDI allow hard TPU to run on there may upcoming AMS? If Bambulab only then unlocks the function on the printer for lets say shore 50D, this could may stay in my memory for a damn long time.
or more likely, a couple or more of beers
@StreetSports I added a bit more to my post just above. Is this what you’re seeing as well?
Maybe Qidi Studio is still a work in progress. The results I’m showing above are not unique to Plus 4. I get the same results if I use Qidi Studio to slice for a X-Max3.
Edit: OK, so I think I get it now. Evidently, the slicer thinks that the speeds to avoid are roughly 71mm/sec up to 149mm/sec. I haven’t pinned down the endpoints of the range, but it is smashing a programmed speed of 135mm/sec down to 70mm/sec, even though it would seem that the max speed it should cram down would be 120mm/sec based on the form that we can all see. On this theory, 70mm/sec is in the safe zone.
If my interpretation is correct, then what does seem a bit odd about it is that it is just using magic numbers to make that determination, numbers that were staticly entered by the Qidi Studio programmers rather than based on resonance measurements gathered at the hardware level on each individual machine. But the hardware does appear to run a resonance sweep as part of its setup calibration. Is that just for show (“resonance theater” as it were), or does that information get used somewhere, perhaps as a secondary filter by the firmware? But, if the latter, what’s the point of putting magic numbers into the slicer? Why not just rely on the firmware to do it? Is there some way for the slicer to pull the locally, empirically derived resonances to avoid from the local machine, and over-write the magic numbers with that? I don’t know, because I’m still too new to this particular software.
Thankfully, there are Qidi profiles for regular Orca Slicer, so I’ll be using that going forward. Although there isn’t yet one for Plus 4, I’m sure somebody somewhere will come up with a Plus 4 profile for mainstream Orca Slicer before too long.
I didn’t really expect much from Qidi’s homebrewed software. That’s why I hadn’t touched it until yesterday. I think maybe Qidi should give up the pretense and simply put all their wood behind releasing accurate profiles for mainstream Orca Slicer. Or better yet, hire a consultant to do it for them in a day or two. Then they can stick to their core strength of hardware and manufacturing. That would be the best outcome for all concerned, IMHO. The world doesn’t need a separate slicer from every hardware manufacturer.
Qidi Studio is like a re-skinned Orca Slicer, except stripped down in some ways and apparently with some of their own forks built into it. As you can tell from the screenshots, it looks very similar. I assume that’s what @StreetSports is referring to when he refers to Orca Slicer, because Qidi Studio currently has support for slicing for Plus 4, whereas there isn’t yet a Plus 4 profile that can be imported into true, mainstream Orca Slicer, like there is for Q1 Pro or XMax3 or XPlus3. I expect the Plus 4 profile will probably be released soon, though. I mean, how hard can it be?
1 Like
Most people won’t understand what he’s saying - but that’s exactly where the journey goes.
How wonderful it is when such young people have a future for their own company.
That makes my heart skip a beat.
Over 40 printers, the first employee and already produces his own filament.
2 Likes
@NeverDie I agree. It actually looked better with it turned off also. Orca didnt even include it for the plus4 profile. Your edit does explain what it deos though.
@Hank I dont use first layer or spaghetti ditection. Ive weeded out all the reasons for needing it.
@Nebur the profile was added yesterday, after all my testing lol. Its beta but has both the plus4 and bambu profiles As orcaslicer 2.2.0 Beta2. As said though, the qidislicer is very close to basic orca anyway.
I’m just now getting up, but im going to run a max flow rate test so we can all see some straight sections to compare at all speeds. I work 12 hour night shifts so my sleep schedule is odd and changes depending on if its my weekend or not.
1 Like
Top one is Bambu X1c, bottom is qidi plus4. Elegoo rapid petg black was used. going from 5mm3/s to 40mm3/s @ .25 steps. Both from orcaslicer, identical filament and print settings. 255c on the nozzle and 70c on the bed. doors closed but tops off. This is also without running the frequency compensation from the qidislicer obviously. Both show VFA but the qidi is much smaller frequency. The Bambu is the standard 2mm belt spacing. Also, keep in mind these tests are done at .32 layer heigh i believe. Revo .4mm nozzle on the bambu and the factory nozzle on the plus4
Are these pictures from the flowtest calibration that you’re comparing against one another?
At least for X1C and Qidi X-Max3 on the non-Beta mainstream Orca Slicer, none of the standard orca profiles go above 0.28mm layer height. I don’t have the beta that you have installed, so I can’t confirm or deny what’s in that. Or maybe you created your own 0.32 layer height profiles for X1C and Plus4 yourself?
I’m guessing that’s a typo and you mean 70c on the bed?
It’s hard to gleen the full context from these photos. For instance, did they both fail at around the same height? Is this the full side photo, or is there more below it that we aren’t seeing? The wall on the Plus4 looks (?) as though it overheated and started to buckle under its own weight before the overextrusion kicked in.
I don’t know how to evaluate this. What is it that you are able to conclude? I can see what appears to be VFA’s in both, but I’m unclear how to compare because of the disfigurements. IMHO, it seems like there’s too much else going on for a clean comparison of just VFA’s. For instance, comparing VFA production in prints that were in all other respects successful might be a more fruitful path. Not trying to nitpick. Just giving you my reaction, for whatever it’s worth. Maybe I’m just not seeing what it is that you’re seeing, in the same way that you’re seeing it. Maybe others are seeing it, and I’m just failing.