|
It would be much simpler for you to just report how far off your calibrated e-steps are from your theoretical e-steps?
|
#
?
Aug 11, 2020 23:17
|
|
|
|
| # ? Jun 11, 2024 22:24 |
|
|
Sagebrush posted:It would be much simpler for you to just report how far off your calibrated e-steps are from your theoretical e-steps? Why? The point is, setup your printer mechanically, and then check it. Why is the idea of checking it controversial? The idea of going "I have 200steps per revolution, and a 20 tooth pulley I have no need to check it in the end" is... well enough for a lot. But the question that started this was "what should someone do", and you absolutely check. Or am I missing your argument?
|
|
#
?
Aug 11, 2020 23:34
|
|
|
Looks like my department may be able to add a Prusa MK3S. Are there any accessories/addons that are considered essential? I may not get another chance to order from them, the way budgets are looking.
|
|
#
?
Aug 11, 2020 23:36
|
|
|
Listerine posted:Looks like my department may be able to add a Prusa MK3S. Are there any accessories/addons that are considered essential? I may not get another chance to order from them, the way budgets are looking. Textured plate for PETG and such, I guess, but otherwise they are pretty complete.
|
|
#
?
Aug 11, 2020 23:51
|
|
|
Nerobro posted:Why? The point is, setup your printer mechanically, and then check it. Why is the idea of checking it controversial? The idea of going "I have 200steps per revolution, and a 20 tooth pulley I have no need to check it in the end" is... well enough for a lot. But the question that started this was "what should someone do", and you absolutely check. The biggest problem is that what you're describing (printing a 20x20 cube and then measuring it) used to be common practice and recommendation, and then it was abandoned because it causes more problems. Many of the errors you'll measure on that 20x20 are not movement proportional errors, and simply attributing them to movement will result in bang on 20x20 cubes, and worse calibration at other sizes. If you want to measure how far an axis moves, buy an indicator and actually measure it.
|
|
#
?
Aug 11, 2020 23:52
|
|
|
This is feeling more and more like people are talking about different things. I mean at first it sounded something like "you should definitely and obviously change your e-steps to match belt stretch as a normal printer setup thing" but I can't keep track of it since.
|
|
#
?
Aug 12, 2020 00:04
|
|
|
Nerobro posted:Why? The point is, setup your printer mechanically, and then check it. Why is the idea of checking it controversial? The idea of going "I have 200steps per revolution, and a 20 tooth pulley I have no need to check it in the end" is... well enough for a lot. But the question that started this was "what should someone do", and you absolutely check. Your argument is that a properly assembled, well-designed printer will still end up printing noticeably misshapen parts, because of the belts stretching past their nominal dimensions, unless it has its X and Y motors recalibrated right off the bat. I'm saying that I don't believe that is the case, and I am curious to know exactly how far off your calibrated steps are from their theoretical figure to understand the magnitude of the problem you're talking about. If you had said that they were 0.2mm off sure I totally believe that. Though it's going to be hard to tell whether that error is from positioning issues or extrusion issues. 1.3mm of positioning error though? Across a normal sized bed? No, something else is wrong.
|
|
#
?
Aug 12, 2020 00:24
|
|
|
The Eyes Have It posted:I mean at first it sounded something like "you should definitely and obviously change your e-steps to match belt stretch as a normal printer setup thing" but I can't keep track of it since. You should DEFINITELY check the size of stuff your printer is printing. That is what i'm saying. I suppose me saying "adjust your esteps accordingly" was being ahead of things and assuming understanding of the user. From a noobie point of view? Fuckit, build it, assume it's ok, and we can talk about parts fidelity when you need parts that fit well. Aurium posted:The biggest problem is that what you're describing (printing a 20x20 cube and then measuring it) used to be common practice and recommendation, and then it was abandoned because it causes more problems. I've not seen how it was abandoned. I see it all the time ~now~. You do the cube (or in my case, a 10x20x40 shape..) after you've verified your wall dimensions. Yes, and many of the errors you see on a 20mm cube tell you what else is going on with your printer. Up to, and including loose belts, slipping or sticky Z screw, ringing, low temp issues, and even more radical extruder issues. Indicators that are useful over more than 30mm or so are very, very expensive, and hilarious to read over more than one revolution. Calipers that get you within a tenth of a milimeter.. aren't so bad. Getting something arranged so you can easily, and repeatably measure with calipers is hard. Printing a thing you can measure, isn't. Sagebrush posted:Your argument is that a properly assembled, well-designed printer will still end up printing noticeably misshapen parts, because of the belts stretching past their nominal dimensions, unless it has its X and Y motors recalibrated right off the bat. I'm saying that I don't believe that is the case, and I am curious to know exactly how far off your calibrated steps are from their theoretical figure to understand the magnitude of the problem you're talking about. My argument is that you should measure what your printer does, so you know what it does. In order of reasons that printers end up making mis-shapen parts, starts with not getting the slop taken care of, not building it square, and finally, it not being the size it's expected. (read: belt stretch). In my paritcular case, the difference between X and Y is enough that large diameter threads are.. a bit funny to screw together. And it's due to not having the same tension on each belt. I'll need to hook something up to my ender to see what my e-steps are at. I had to lower it just a touch. My MPMD needed to be between 1 and 1.5% adjustment to each of it's motors to get it printing round and accurate size. Sure it's 1.3mm of positioning error. But it's repeatable in position to a degree that I don't have the patience to measure. The printer is repeatable, which is the first measure of a printer being usable. Things are 0.7% big in the X axis. (as I measured today) It's a thing, it's a real thing. It's enough to gently caress up some threads I print. The question was about making a printer perfect. If you want a perfect printer, this is a thing you should do. I definitely was far to simple in my description of why. A reality check though. My buddy is turning out production peices on his Ender 5s, and Ender 3, and he's never calibrated like.. anything. it's bit him a couple times. And the fact he's never characterized his printers means his prints were taking ~at least~ 25% longer than they need to. (that took a solid week of convincing before he'd try anything other than defualts speed wise..) https://www.thingiverse.com/thing:2871553 Something like that is a good test block.
|
|
#
?
Aug 12, 2020 02:42
|
|
|
Fortuitous that we're in the "belt tension" zone right now. I'm building a corexy with a 3mm volcano for a ~2m x ~2m x ~200mm print volume. I've got some 4m flight control cables that I salvaged from somewhere that I plan on using for XY, with a normal leadscrew for Z. Do I need to worry about cable tension? In the aircraft, these 1/8 cables are tensioned to ~75lb, but the procedure to set from new is to tension to 230lb, run the flight controls 30x times, then set to normal value. Each cable is 1/8" nominal diameter, with yarns, each yarn being 19 strands. Alternatively, instead of 1/8 7x19 stainless steel cable, I could go with the other cable I've got, which is 150lb test SpiderWire(tm) Dyneema(r) fluoropolymer-treated microfiber fishing line. It claims "no stretch." I'd be wrapping the fishing line around a capstan a couple five times instead of just once around a pulley with the aircraft cable. The xy bed is made of 2" rigid metallic conduit welded together, and has delrin rollers, so I can get my axes square, straight, and solid. Actually, do I need to put this whole thing in a temperature-regulated room so that my cable and my frame don't expand at different rates, thereby throwing my dimensional tolerance ridiculously far out?
|
|
#
?
Aug 12, 2020 02:52
|
|
|
If you're building a 3d printer with string you should abandon corexy and just build the most epic HangPrinter ever.
|
|
#
?
Aug 12, 2020 03:04
|
|
|
babyeatingpsychopath posted:Fortuitous that we're in the "belt tension" zone right now. The hangprinter uses fireline which is a dyneema fishing line. Individuals have gotten good results with 40lb weight line, though the default build uses 90lb. Considering one of the typical calls for corexy is speed, I'd be concerned with how much inertial mass is in the steel cables. I'd probably try the fishing line first and see if it is enough. Nerobro posted:
My experience is is that it's gone from the thing to do to calibrate a printer to something that's discouraged. I can't say I'm surprised that it would still comes up though, it'd be easy to independently come up with and compensates for some issues. Calipers may be easy, but they don't actually measure the right thing. A cheap 25mm indicator with a stated .01mm would measure an error down to .04% over its travel.
|
|
#
?
Aug 12, 2020 04:21
|
|
|
Mofabio posted:This is only true when the belt run is perfectly lined up with the axis it powers. The distance equation is (belt distance on x) = sqrt((x position)^2 + (y error)^2 + (z error)^2). When y and z error are zero, then the calculated numbers will work. The error is the sine of the angle made between the belt and the axis. When the error is small, the angle is small. Since the derivative of sine is cosine, and cosine near zero (small error) ~=1, its maxima, the total distance error doesn't change very much over the length of the axis, and is approximately linear, and can be fine-tuned out by the method described. For what it's worth, by spurious means or not the above mathematical argument makes an accurate prediction that the distance error caused by non-parallelism in the belt, is roughly linear when the angle is small, and so also the wisdom of the 10-20-40 cube. With the 40mm dimension along a 200mm the test axis, I believe the sample of the middle 40mm would fit the line well. Thanks for the tip Nerobro!  The function plots error = belt length signal - belt length actual = x position - sqrt( (x position)^2 + (y error)^2 + (z error)^2 ) so I believe that a linear adjustment to esteps/mm comparing (print signal - print actual) would correct for it. No test yet tho just yapping. The x position error is roughly 1/400 of the y and z errors so an adjustment to parallel even by eye would seem be not the biggest problem these printers have with their kinematics. Mofabio fucked around with this message at 08:55 on Aug 12, 2020 |
|
#
?
Aug 12, 2020 08:47
|
|
|
oh, 26 posts let's see wha- y'all are nerds. just inherit a printer that caught fire like i did and slowly replace every part.
|
|
#
?
Aug 12, 2020 13:01
|
|
|
Deviant posted:oh, 26 posts let's see wha- I'm offering to help them do exactly this, but nobody wants my old DaVinci printer.
|
|
#
?
Aug 12, 2020 16:19
|
|
|
biracial bear for uncut posted:I'm offering to help them do exactly this, but nobody wants my old DaVinci printer. e: nvm Caffeinated Jerkoff fucked around with this message at 07:08 on Aug 13, 2020 |
|
#
?
Aug 12, 2020 17:55
|
|
|
Nerobro posted:Why? The point is, setup your printer mechanically, and then check it. Why is the idea of checking it controversial? Somehow it is in this thread. The last time this came up many people were very offended in regards to my saying these things are typically not assembled properly or accurately and the printer manufacturers, even the beloved Prusa, are treating them like consumer electronics rather than the machine tools that they are. Measuring things with the ends of zip ties and SWAG-ing belt tension by deflection somewhere along the belt travel with no mention of pre-load on the side (if one is even mentioned) that you are measuring. No proper squaring using actual tools. lovely bed materials that move around and get fixed "in software" rather than for real. This is what makes these things so freaking unpredictable. They are built to a cost and assembled/tuned in ways that are absolutely designed for "all I have is a screwdriver and set of allen wrenches and I will NOT be buying anything else" methodology. If all you want to to is download poo poo off of thingiverse to put on your knick knack shelf it's fine to treat them this way. But if you're looking for serious dimensional accuracy these consumer level printers aren't likely the right thing. You might be able to make them do it reasonable well, but it's likely to be something you need to continue to calibrate very often. Because when it's not he belt it's gonna be something else. And something else after that. Extremely dimensionally accurate printers exist. They existed long before the consumer variant. They've been used by jewelers and dental appliance places and probably many more for quite a long time. And most of them print multiple materials so they can do soluble supports because these things are getting used as molds and no, slicer generated single material supports aren't good enough and would require too much hand finishing. Motronic fucked around with this message at 18:32 on Aug 12, 2020 |
|
#
?
Aug 12, 2020 18:26
|
|
|
Anybody that seriously wants to use hobbyist 3d printers for engineering-level products can gently caress off, they're toys.
|
|
#
?
Aug 12, 2020 19:05
|
|
|
biracial bear for uncut posted:they're toys. 100% agreed. But that's not how a lot of people see it, against overwhelming evidence to the contrary.
|
|
#
?
Aug 12, 2020 19:12
|
|
|
My ender 5 has been doing just fine on printing custom microscope components for me. Definitely better than toy grade even though I haven't messed with dimensional calibration at all.
|
|
#
?
Aug 12, 2020 19:15
|
|
|
Engineering grade meaning that you can reliably get things like interference fits, locational transition fits, truly parallel and precision surfaces. Then using those surfaces as data for GD&T. You can't reliably stack tolerances, either. You can't get that out of a consumer printer. You can do small parts, but big assemblies and precision parts need precision machines.
|
|
#
?
Aug 12, 2020 19:26
|
|
|
I would say that a good quality desktop printer like a Prusa is comparable in machine tool performance to, say, a benchtop 8x12 mini lathe. That is, actual machinists, having the ego that so many machinists have, will take a dump all over it because it can't hold .0001 and therefore by their measure it's poo poo trash toy for babies and completely worthless. But it was never targeted at them in the first place. The people actually buying it are hobbyists working in their garage, for whom hell .005 is totally fine, and the machine does a perfectly good job for them. If you are a random dude in your garage using a $300 3D printer for your hobby stuff -- including functional mechanical detail down to the sub-millimeter scale, not just lumpy yoda heads -- you will be perfectly happy. If you are a random dude in your garage trying to use a $300 3D printer to achieve the results you'd get from a $100,000 CNC mill you're going to be disappointed. If you are a professional machinist comparing a $300 3D printer to the performance you're used to from a $100,000 CNC mill and then declaring that the printer is a useless piece of trash, you're a moron.
|
|
#
?
Aug 12, 2020 19:31
|
|
|
That is correct. I wanted to elaborate on small parts != precision parts.
|
|
#
?
Aug 12, 2020 19:35
|
|
|
Point is there are tons of professional tasks that don't require the above engineering grade tolerances and for which a consumer 3d printer is more than adequate.
|
|
#
?
Aug 12, 2020 19:42
|
|
|
Sagebrush posted:I would say that a good quality desktop printer like a Prusa is comparable in machine tool performance to, say, a benchtop 8x12 mini lathe. That is, actual machinists, having the ego that so many machinists have, will take a dump all over it because it can't hold .0001 and therefore by their measure it's poo poo trash toy for babies and completely worthless. But it was never targeted at them in the first place. The people actually buying it are hobbyists working in their garage, for whom hell .005 is totally fine, and the machine does a perfectly good job for them. I'm the professional machinist/CNC programmer/engineer that's not expecting what I get from my Prusa Mini and pleasantly surprised when it exceeds those expectations, but I'm not trying to use it for repeatable results either, just first pass design/concept testing before taking a model over to a mass-production method. The idea that someone wants to look at Prusa's machines in an industrial/business look is hilarious to me though. Either look at 3DSystems/Stratasys/whoever is operating in those "No poo poo industrial 3d printer" ranges or go home. Ultimaker was super proud of that automaker factory using their machines to build jigs/fixtures a while back, but I bet if you revisited that case study you'd find that the original machines used in that case study have long since been obsoleted/replaced. Some Pinko Commie fucked around with this message at 19:56 on Aug 12, 2020 |
|
#
?
Aug 12, 2020 19:53
|
|
|
Scarodactyl posted:Point is there are tons of professional tasks that don't require the above engineering grade tolerances and for which a consumer 3d printer is more than adequate. Absolutely. There are plenty of low tolerance things that they're fine for. I've made a ton of brackets/face plates/panel fillers with mine. It just doesn't matter how accurate they are past what the scale on a cheap tape measure can handle. Anything you're going to measure with mics is right out.
|
|
#
?
Aug 12, 2020 19:56
|
|
|
Motronic posted:Absolutely. There are plenty of low tolerance things that they're fine for. I've made a ton of brackets/face plates/panel fillers with mine. It just doesn't matter how accurate they are past what the scale on a cheap tape measure can handle. I think this is still a mischaracterization. There's a range in between "tape measure" and "micrometer" (some might even say...the accuracy of a vernier caliper) where 3D printers are perfectly capable of making good parts. Yeah you aren't going to print a hole with an H7 fit the first time around, or make a precision leadscrew or something. But I can churn out M5x0.8 nuts and bolts that mesh just fine with mass-produced hardware all day long, and it's good for people to realize that the machines are capable of that and not just restricted to shelf brackets and plastic boxes. Sagebrush fucked around with this message at 20:16 on Aug 12, 2020 |
|
#
?
Aug 12, 2020 20:13
|
|
|
Let's say that mine has been perfectly sufficient within the tolerances of a Harbor Freight digital caliper used somewhat haphazardly. Being able to get something like this adapter done all in one day was really nice (m37x1female thread on one end and an 18.5mm inner diameter with set screws on the other).  I don't think I could measure those dimensions with a tape. These things are a bit too annoyingly specific to have an off the shelf solution and having a one-off made would be relatively slow and expensive. This certainly isn't something that requires insane tolerances or anything, but it's saving me money on something I'm going to sell (in addition to the savings on making stuff for my own rigs).
|
|
#
?
Aug 12, 2020 20:16
|
|
|
The first printer we got a work was a Markforged Onyx Pro. Markforged markets themselves as industrial grade printing, and it seriously is. It’s also expensive as gently caress - we got the printer as a deal with our large format printer that does 8x6 ft graphics... I wanna say the selling price is like 7 grand. It prints basically two types of nylon and can also do fiberglass reinforcement. Markforged sells their own material, and it’s like 2x as expensive as anything else. However, I think because the control both the printer and the filament, that allows them to make really good engineering decisions because it’s a controlled environment. For all that, we consistently get great .1mm layers, a beautiful matte finish and I’ve literally never had the machine or their required proprietary slicer gently caress up. However, it also means I need to tension the belts properly every now and then, but even when it’s at its worst, we get great prints. In comparison, we also got a GMAX 2, billed by Gcreate as a “workhorse” with a huge bed. The gmax consitently has issues running the nozzle into the print surface, no matter how hard I try to fix it. We’ve fiddled with settings tons and still haven’t been able to get the same Quality at .1mm. .15 seems to be its happy zone. It’s definitely a different world.
|
|
#
?
Aug 12, 2020 20:17
|
|
|
Got a good print of CF Nylon  
|
|
#
?
Aug 12, 2020 20:57
|
|
|
Yeah yeah, I posted some pics/videos the other day (week?) where I was impressed by the Prusa Mini printing a piston bracket that had acme threads in it. Accurately and relatively quickly in ASA with some bridging I nerded out over. Doesn't mean I'd tell someone wanting to actually use one in a for-profit business they should buy the same machine and have at it with whatever they come up with. Some Pinko Commie fucked around with this message at 21:04 on Aug 12, 2020 |
|
#
?
Aug 12, 2020 21:01
|
|
|
The best objective measurement should be precision-per-dollar 
|
|
#
?
Aug 12, 2020 21:08
|
|
|
insta posted:The best objective measurement should be precision-per-dollar thou for thou
|
|
#
?
Aug 12, 2020 21:11
|
|
|
https://www.amazon.com/gp/product/B07K33DCNW Elegoo Mars (OG version) for 200$ (sold by elegoo/amazon). Dunno how long it'll last, but that's a loving amazing deal for the printer. I love mine.
|
|
#
?
Aug 12, 2020 22:40
|
|
|
insta posted:Got a good print of CF Nylon my favorite material of all time. strong enough to replace light-duty aluminum parts and it looks so good. you'll wanna work on the continuity of your fillets though. goddamn that transition. it's not even g1 is it lmao
|
|
#
?
Aug 12, 2020 23:06
|
|
|
Sagebrush posted:you'll wanna work on the continuity of your fillets though. goddamn that transition. it's not even g1 is it lmao Oof
|
|
#
?
Aug 12, 2020 23:17
|
|
|
It's not my part, just my print.
|
|
#
?
Aug 13, 2020 00:04
|
|
|
Sagebrush posted:I think this is still a mischaracterization. There's a range in between "tape measure" and "micrometer" (some might even say...the accuracy of a vernier caliper) where 3D printers are perfectly capable of making good parts. Yeah you aren't going to print a hole with an H7 fit the first time around, or make a precision leadscrew or something. But I can churn out M5x0.8 nuts and bolts that mesh just fine with mass-produced hardware all day long, and it's good for people to realize that the machines are capable of that and not just restricted to shelf brackets and plastic boxes. Tape measure, steel rule, dial caliper, micrometer.... It's fun introducing people to even their first dial caliper, or thousandths indicator. Teaching them to use a micrometer, then showing them how much just holding a block of aluminum will change the measurement usually leads to a "I can't trust anything!" moment or two. .... I've done that a few times. I keep an old beaten up 0-1" micrometer in my work bag. This all comes down to the right tool for the right job. There's a lot of people who see this 3d printing hammer then... everything must be hit with that hammer. And usually, it's people who can't do ~any~ of the math to decide if a part is suitable. Or even design a useful part. The "hammer" thing also happens with mills. Though the more "cnc" it is, the worse the hammer effect is. I think there's a link between programmers and "I can just tell the machine X" that doesn't jive with material reality. More people need to watch CNC Kitchen. :-)
|
|
#
?
Aug 13, 2020 05:13
|
|
|
Stupid_Sexy_Flander posted:https://www.amazon.com/gp/product/B07K33DCNW Normally don't quote myself but goddamn, it's down to 180 for the OG, 190 for the red/black. I'm tempted to grab another one at that price.
|
|
#
?
Aug 13, 2020 18:37
|
|
|
Clearing inventory for when they release the 7”x10” model.
|
|
#
?
Aug 13, 2020 19:01
|
|
|
|
| # ? Jun 11, 2024 22:24 |
|
|
desktop 3d printers are cheap poo poo garbage toys for babies. useless for real work https://www.youtube.com/watch?v=d8us3OWgS1A
|
|
#
?
Aug 13, 2020 19:40
|
|