285
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
this post was submitted on 20 Oct 2023
285 points (86.3% liked)
Technology
59983 readers
2282 users here now
This is a most excellent place for technology news and articles.
Our Rules
- Follow the lemmy.world rules.
- Only tech related content.
- Be excellent to each another!
- Mod approved content bots can post up to 10 articles per day.
- Threads asking for personal tech support may be deleted.
- Politics threads may be removed.
- No memes allowed as posts, OK to post as comments.
- Only approved bots from the list below, to ask if your bot can be added please contact us.
- Check for duplicates before posting, duplicates may be removed
Approved Bots
founded 2 years ago
MODERATORS
Gee... telling the engineers to getting precision to below 10 microns would cause production challenges.
I've been doing PCB-board design recently. Here's the manufactuering specs: https://www.digikey.com/en/resources/dkred
So that's 0.13mm tolerances to my printed-circuit board. Or 130 microns.
Current leading theory is that Elon Musk is such an ignorant dumbass that he doesn't know the difference between mils and microns, despite running a car company / manufacturing firm. Give that a thought. Even then, 10-mils tolerance is near this PCB design, an object that's only a few inches in size. Cars are much larger and normally should be built to much wider tolerances than a fucking PCB board.
It's almost like Elon Musk is a complete fucking moron and not an Engineer. The wanker has never actually designed a thing in his life. He just tells other people to design something, or buys an existing company, then struts around like he thinks he's the smartest thing around.
He was fired for being incompetent. Only got rich because rich daddy and because he got lucky with stocks
If he said <10 mils, I'd might have bought the explanation that Elon actually meant millimetres. Micron is a very specific metric-based unit which to Elon might have been trying to use like a buzzword.
The moral of the story is don't say stupid engineering stuff if you don't want engineers to laugh at you.
And 10 microns at what temperature? Because on something the size of a car, made of mixed materials, thermal expansion of less than a degree is going to blow that figure.
They couldn't apply paint to a tolerance of 10microns.
Ten mils is .010" or .25mm if I'm not crazy.
It's a very standard tolerance for aircraft parts.
Yeah and that wouldn't be too bad either... still expensive but not completely unrealistic for ALL parts of a car.
Kinda unreasonable for the number of cast parts most cars use, but for machined surfaces it shouldn't be too bad.
I recently listened to a podcast about musk which was more on the anti side. The podcast had some parts about spacex and musks own work ethics, which told more of a story that he actually has some insights and knowledge and was a insane workoholic. Which shifted my perception of him. He isn't dumb, he is a really good conman.
Lmao “if Lego and soda cans can do this, so can we.” At least he found materials similar to his existing vehicle build quality
Hey, don't disrespect Lego like that
Serously folks have snapped steel rods using them.
"At this point I think I know more about manufacturing than anyone currently alive on Earth."
Flashbacks to Steve Jobs and the NeXT Cube.
What an idiot
I have what are basically toy 3d printers capable of 1 micron precision...
Cool. Build 10 pieces and fit them together with sub 10 micron precision.
Ok, and once that's done with the high level of repeatability and quality I've done it a thousand times, then what?
Seems like a reasonable next step!
Scale that up to a 4 ton production-ready consumer vehicle without introducing defects, I imagine
The heaviest Tesla is around 2.5 tons...but that's beside the point.
The scale doesn't matter. In fact, taking a desktop process and industrializing it makes it more accurate and repeatable - especially in a large-item manufacturing setting.
The best argument against the 1 micron requirement is that it's unnecessary from a practical standpoint. I completely agree with that, for sure.
Which printer is giving you 1 micron precision?
It depends on the model being printed.
1 micron is easy to achieve on the Z axis with almost any printer, using any 3d printing technology (FDM/SLP/SLA/etc).
In the X/Y axis, the vast majority of FDM printers are off the table for 1 micron - increasingly so as the Z height of the model increases, especially on bed-slingers like the Ender 3. The taller the model, the harder it is to maintain accuracy on X and Y.
In SLA/DLP printers, it's all about the motor, controller, and whatever shields excess light/beam diameter.
To answer your question directly: I have no trouble getting highly reliable micron accuracy in a $99 Creality Halot. The key is to understand your model in relation to the pixel density of the screen. Some calibration prints tell you where the steps are, describing the relationship between input and output - which will most certainly be different from printer to printer at the consumer level. Once you have that data, some simple math tells you exactly how to design your object in a way that takes these natural constraints into consideration.
Yes - at any scale, even at the size of a car...or a battleship. When you accept the constraints of the hardware into the design of the object being produced, you can get micron repeatability out of just about anything.
Oh I see.... You've confused microns with Thou (aka mils)... much like Elon has.
Nope.
So does that mean that you can get 1 micron accuracy as long as the part is sized to a multiple of the pixels width on the x and y? Is that just for aligned straight edges or can that be done for curves too?
The pixel alignment is a good place to start, but no 2 printers will produce an identical result at that level. That's why it's important to tune the model to the printer - not the other way around.
Can you get it on curves? Yes, certainly. For 3d printers, even the position you choose for the model within the build area makes a difference.
The question is really about executing the process of engineering in the correct order.
The most common mistake is to design the thing you want to build first. In reality, you start with what is essentially a sketch of what a functional end product looks like. Then, you buy/build tools/manufacturing equipment. Finally, you refine your sketch into a manufacturable product based on assessment of the most reliably repeatable results available from the actual machinery as it is installed.
Then take them to a different environment and measure them and then put a Lil water on them and measure them. Then get a mechanic to fix them and then shove it up your butt because any real engineer designing a car would tell you that kinda precision is just fucking stupid at production level for a car
I've already said a dozen times: no shit Sherlock, this is absolutely unnecessary for this application.
It's also completely irrelevant to change up the environmental parameters when that is not a constraint set at the outset of this conversation.
I swear to God, it's like there's an entire subspecies of moron non-engineers who exist for the sole purpose of arguing where no argument exists.
A car doesn't go through environmental changes? Because we are talking about engineering a car. You are talking about your 3d printed dildo of musks cock or something that you got into microns or some shit. It's almost like there's a entire subspecies of moron non-engineers who exist for the sole purpose of arguing about why musks cock is microns... you asked what to do next
Are you still arguing with a ghost? Nobody said a car doesn't go through environmental changes... It just has nothing to do with this discussion.
Bullshit.
The thermal coefficient of expansion of say... Aluminum is 23.
That means that when a 1 meter piece of Aluminum rises from 20C to 21C, just one degree Celsius, it grows by 23 microns.
Your 3D printer is not a temperature controlled precision instrument. Your tolerances are no where close to 10 microns let alone 1 micron.
There are micron-level precise instruments in the engineering world. They all come with temperature characteristics because thermal expansion is a bitch. 3D printers that literally heat up hundred degrees and cools down regularly literally can't be this precise, the heat alone wrecks your precision.
Bullshit.
You're changing the premise of the question.
Pick a temperature - design your model to be whatever you need it to be at that target temperature - just like every other engineer with 26 years of experience, such as myself.
(by the way, my UV resin printer is quite temperature stable.)
Your resin printer does not have the resolution of 1/15th the size of a damn white blood cell. Your blood is 15 microns per cell or so. Red blood cells are smaller at around 7 microns.
You, and Elon, have confused your units quite significantly. I've given you the opportunity to see your error by reminding you the difference between thou, mils, mm, and microns. But apparently you haven't gotten the hint yet.
My university created micrometer-sized balls, gears, and other such devices. They're called MEMS and are really cool. They're not made with 3d printers but instead lithography (same technology as computer chips, because they're so small its easier to make through lithography). You've confused your units and its clear based off of how you've been talking. Take a step back, and double-check the difference from thou, mil, mm, and microns.
Sigh.
Please stop being confused about the capabilities of modern SLA/DLP processes.
What is the diameter of A PHOTON? And don't forget to answer ignorantly, in a condescending i-know-more-than-my-betters tone. No, seriously - look it up. You clearly don't know.
How many blood cells wide is light? By all means - tell me what the particulate size of photocuring resin might be? Could it be....2 microns maximum even for the largest ceramic-infused resin slurry - with sub-micron particulate sizes easily available even from Amazon? Yes - far, far smaller than blood cells. This is why you can't let this shit get on your skin. It will literally traverse the blood-brain barrier. The fumes can get particles of this size into your bloodstream.
Well, what on earth can I do to get 1 micron accuracy from zero-width photons and 2-micron UV cured particles? Perhaps - design my model to be LITERALLY ANY SIZE that happens to be divisible by 2 on X, Y, and Z axes, then center it on the build plate?
Can you please get with the program, here? I've likely been doing this work longer than you've been alive. I'm not even supporting Elon's demand, which is almost certainly unnecessary for this application.
Thanks for giving me more absurd examples. Your UV light is a 400nm wavelength or so. Or roughly 1/3rd a micron. The fucking wavelength of your curing light.
Now get outta here with your attempts at pretending that your $200 UV printer has the level of accuracy of three fucking wavelengths of the light it's using.
The size of the color red? That's 700nm wavelength, or 0.7 microns. Your resolution that you're printing here is no where close to the size of red-light photons.
Oh, so you don't know how light works either. Good grief man... light has a size and you're running up against the size limitation of the light itself. Especially because I know for a fact that these UV Printers are NOT using lasers, so you have no way to actually line up all the photons to hit the same location since their wavelengths are all unaligned.
In any case, car parts are not made at scales comparable to the wavelength of infrared light (ie: the "size" of a infrared-light photon).
Shocking narcissism, yet again.
You clearly have no intent to learn from your mistakes and pollute your entire argument with ad hominem and every other logical fallacy in the book - all wrapped up with a bow of dismissal of the original premise.
Oh, and by the way, light is both a particle and a wave. "light has a size" - indeed, the wave of light we aren't talking about has a size - called wavelength. Don't be afraid to use the most accurate terminology for your irrelevant responses! The wavelength has zero influence on the dimensions of the final product, as any wavelength more than a couple nm out of spec won't cure the resin at all, and there will be no object to measure. I think you know that and are just trying to win an argument by talking over the heads of the average readers on here - which doesn't help your case.
It's extremely likely you legitimately have a personality disorder, and are capable of much more if you clean up your act.
Wavelength has a very direct impact on the resolution you can print because it's an optical system. Under perfect conditions, it'll be diffraction limited, which is typically anywhere from several hundred nm to tens of microns. That's an ideal system though, you're actually going to be getting a dimensional accuracy somewhat above that in practice, probably tens to hundreds of um.
Please make sure Form Labs, and the rest of the companies I have worked for, are aware of how wrong their engineers are.
I'm not sure what the obsession is with arguing against the list of objective facts I have provided - each argument peeling away into a deeper and more obvious lack of understanding...
Please feel free to run anything other than the recommended wavelength of UV light through a photoreactive polymer resin and let me know the dimensions of your resulting print. Remember to do it twice, so you can compare the results - the point of this discussion. Repeatable results, to 1 micron of accuracy.
(your repeatable result will be a measurement of thin air, as your print will not cure - but feel free to try, as I've said.)
I couldn't find official dimensional accuracy specs for any formlabs machines except the 1, which lists 150um. Perhaps you're talking about the 3, which has a specified minimum spot size of 85um according to this paper. Where did they claim micron dimensional accuracy?
I don't care what the specs of the printer are. This is a discussion about repeatability.
I can get sub micron repeatability out of playdough with the right extruder and hardener. Every single one of you has been arguing from an entirely false premise to begin with.
The problem isn't that it can't be done. The problem is that it is unnecessary and very expensive.
This is definitely true in terms of industrial production. (super cheap for "makers" - taking the hardware constraints into consideration with the design, of course).
Can you share a link? I'd love to buy me some of these.
https://www.creality3dofficial.com/products/ld-002r-lcd-resin-3d-printer
Don't forget - it's not as simple as just buying the printer. You need the right resin, several iterations of test models, and the right modifications of the model to work within the constraints of the printer and material.
With the right tuning, you will be able to create parts which measure within 1 micron tolerance reliably. (don't forget to use an indicator and good reference blocks - you aren't going to measure microns with a caliper or micrometer)