alaris2

looking at the price tag, i hope she comes with the handscanner.

I use M221 myself and it works well. be aware that it takes a short time to have effect and that small numbers (ie. close to 100) have very little observable effects on the print quality. I think (not totally sure) that it is accumulative, ie. M221 S200 followed by M221 S100 will not set the values back to the original, but M221 S50 would be required instead. perhaps someone else can comment on that?

for an UM that's an excellent piece. i'd be surprised if you can do better. how accurate were you hoping for?

paid to spam? I'm in the wrong job... I actually have to work to earn a living :(

I read that too - something about GT vs GLX belts or something but I can't find the post now i want it. what you're saying makes good sense - you don't happen to have any links to photos or other info about how commercial printers are doing this do you? none of the ones i've dismantled used belts so I have no comparison point. The method of tensioning seems flawed at present - I quite liked what I saw in corexy which seems to make tensioning easier and uses less belts (= less problem?)

that's because the majority of the interesting technical discussion is still happening on the google forum (which is also prone to spam). this forum has been forgotten, despite being the official one, on account of most of the posts either being spam or people complaining that this that or the other doesn't work. can we have more emphasis on technical discussion please? (maybe Ian can think of a way to arrange the categories to make this so?) more about making UM the better and best, making new parts, trying new materials. great post from Bernhard Kubicek on the google forum about an UM in brass. can i see anything like that on here? no ;(

So we can print at about 0.1mm resolution. that's good enough right? but no, many of us want better, especially for repeatability (accuracy!=repeatability). so I thought I'd ask the obvious question - just where is the backlash and inaccuracy in the current design? what prevents us printing at 10 microns repeatably? which parts are most to blame and should be upgraded /replaced my starter for 10 is the belts. several of us have observed these stretching over time and continuously needing re-tightening (and how tight is tight anyway?). would re-inforced belts help perhaps? what do others think (or can anyone categorically say exactly where the inaccuracies arise from?)

disclaimer: Daid works for UM

whoa! bowl of water! now that's fancy I didn't see the belt sander in there anywhere and I'll add a pillar drill too..

I confess, I've seen everyone's getting wood lately but I'm not sure what the excitement is. who wants to print with wood? I could understand metal, electronics and *real* plastics and carbon fiber..

Troy is spot on there and that prediction may turn true. Fuchs' idea is much like mine as an ideal HPB, but the cost and availability are the issue, that's before we even address safety concerns.

I've commented on this many times and usually get ignored, so it's nice to see someone finally agreeing with me steps per E into thin air is different from steps per E when extruding through a hot nozzle (ie. as the machine should be used). however, be careful with the second technique - if you take too long warming the nozzle, the material will leak out and will affect your measurements - I like to prime, mark, extrude, measure as fast as possible myself to give most accurate values. I too have found the number is always larger - I think i'm running on 940 or so right now which fits with SG's observations nicely. any less and you will get underextrusion and poor surface quality. but those z-scars you were seeing are best fixed by changing the slicer. use a modern slicer like kisslicer, which makes a much better job of surface quality than all the other slicers at present and won't have that z-scar. been there, got t-shirt, etc.

yeah that's the problem - solving as much of the problem without completely redesigning everything. you can get re-inforced belts which might be better? I read something about the belt type - GT (I think) as opposed to LX (or whatever the current one is) is supposed to be more suitable? I haven't really spent the time to look into this enough but it's highish on my list of things to do. direct drive and heated bed are next up.

a heated bed is desperately needed. let's hope it's done properly.

um. why does it matter? is this so the history books record the momentous occasion of the rise of UM to world domination correctly?

1) you'd have to draw a picture I think. it sounds like you added another axes which may lead to binding or other problems? ball screws are generally better yes, but also expensive (main reason people try to avoid them). 2) again, probably helps to draw a picture - it sounds like you're trying to eliminate wobble across the Y axis on the bed itself? I've seen little evidence of this and wouldn't the sprung bed counteract any improvement? 3) lots of work, but this might help slightly. perhaps a better (altho still lots of work) solution is to adopt corexy motion instead. the belts can (I think) be replaced with steel wire or similar to avoid stretching or at least be upgraded to re-inforced belts. I think I'd like to see improvements in the corexy direction if possible - the firmware even supports this so it's the route of least pain for most gain.

that means it's no good - you should send it to me Ian

that's one more article than I ever saw then. there are plenty of claims about resolution to 40 microns, but be wary, what you can do in Z you cannot do in X and Y, and what is claimed in any is often for a Yoda head, not a precision calibration object. If I had to write a spec. for UM I'd say you can do 100 microns on a good day if you know what you're up to and set the machine up nicely. I'm hoping someone will challenge me and say, no, UM can do so much better, at which point I'll ask them to print and show us all a calibration item that proves it, and then I'll buy them a beer, or maybe even two, depending on how generous i'm feeling. but now, back to reality..

sort of. at least 6 months of tweaking and fine tuning has led me to the conclusion that SG's direct drive mod is an important step forward and removing those belts is the next big step. that's most of the error (probably still be some left in the bearings, woodwork etc.) you can get to 0.1mm repeatability but don't expect better unless you tweak the g-code. fitting a big gear would change the gear ratio and might improve step size, yes, you'd want a gear without backlash of course, or it won't help as much. the microstepping is a feature of the motor driver and it's not entirely guaranteed or accurate either but that's a discussion in itself. I hear what you're saying - I want it to. finding ways to replace the belts is the best solution right now.

how could he (assuming speed is not an issue) change to say 4x that - ie. 3 micron per step. it would give him a fair chance of achieving the accuracy at the expense of speed? gearing on the steppers was my first thought.. any other options?

you printed that without support? tis very impressive, the microdumper i liked too.

oh i did something more stupid in the early days. I took the silly default plastic shroud off the fan and decided i'd keep it in case it was useful later. so i put it under the bed. standing upright. then i printed a 9hr print which was max height and one side of the bed hit this shroud, which bent but didn't give way - the bed actually tilted by a degree or two on one side before the print was ruined. so then i panicked and pulled the power and shutdown - without making a note of where it had gotten to. but i recovered the print. it took 2 a4 pages of math and a lot of trial and error to calculate how to resume a twisted print from an unknown stopping point. so don't give up yet - yours should be easy to fix

ebay, local hardware store, garden center.. but don't try it if you have a V1 or V2 hotend, the plastic parts won't take the heat. you'll want to upgrade to an all metal hotend first.

yes, i think loading it all in memory is the problem. I also noticed when printing something much smaller that it stopped and started a lot during printing. this might work better if i minimize it - i'll try that next time, thanks fluxline. what did you mean by 'cleaned with meshlab' incidentally? is there a magic sequence of buttons i should press there?

wouldn't be spectacular or exciting, except for the fact it's printed on my UM out of NYLON. the first piece (middle) I had too close to the bed, the nozzle was burning the blue tape and I didn't get enough nylon on the first layer for the second to stick to, so it curled off (at which point i stopped the print) 2nd piece (right) I had better adjusted, and by the 3rd piece (left) I had it perfected - check out the flexibility on this fully waterproof one way valve. the slight burning marks are residue of PLA left over in the nozzle which doesn't like being heated to 260C and occasionally splats out and makes a mess. it will be gone soon.