Just as cabinetmakers refer to carpenters as “wood butchers”, at the grinding shop we referred to the traditional lathe machinists as “metal butchers”. Their tightest tolerances were generally our sloppy starting points. Of course when your parts are going out for secondary finishing, you waste zero time holding tolerances you don’t need to.
Cool! At our shop we precision ground custom screw threads on sets of electrodes for EDM machining of die molds. I learned a lot of trig figuring those threadforms out.
I actually find this deeply unsettling. Is there an opposite of trypophobia?
Oh I get that… from the digital perspective. When I see people don’t use magnetic alignment and they just drag their vertices in the general vincinity of other nodes - of course this is sub-sub-sub-sub-pixel and even sub-sub-sub-ink-dot, but still!
…but actually I was referring to my/our unfamiliartiy with Imperial units. (I’m aware that my feeling for 0.0001cm might be totally off, but i still find comfort in the illusion that)
I just looked at a nice demo of a fairly affordable DIY EDM setup- drilling, small scale- but still an actually useful application.
Call up your local EDM shop and see if they’ll whip up a sample. Make it clear you’re willing to pay and wouldn’t be a full run client and most can manage something.
As someone who learned manual machining a long time ago, those tolerances are already in the realm of mind-blowing. It must be fascinating work, although I suppose like anything you would get quite used to it.
Out of curiosity, those are just your print tolerances, right? Your precision is even higher, is my guess.
I run into this with my Chinese colleagues. They think of tolerances as accept/reject inspection limits and anything that is within tolerance is acceptable. When we get parts made in the USA, those are assumed to be statistical tolerances. So part deviation is less than 1/4 what we see made with the same print tolerances compared to parts made in China.
It’s one of the reasons we make critical parts in the USA.
Should have gone with Barry White for the soundtrack.
Not a bad idea, but I would guess I’d get a “fuck off” number. Which would be expected but you never know, I appreciate the suggestion
Look for one that does prototyping. In my experience machine shops are used to weird requests and are often open to them if you are up front that it is a one off project and leave the time frame open.
Vacuum welding. It’s one of those underappreciated difficulties of spacecraft design. If you have two pieces of the same kind of metal that are supposed to slide over one another (like a hinge), it may turn into one piece in the vacuum of space.
It’s been described as the atoms sort of forgetting that they were supposed to be separate.
Yeah, I don’t think these parts would stay like that for long in the wild. The tiniest amount of dust on the mating surfaces, the slightest wear on the edges, or even fingerprints from handling the parts when separated, would be enough to make a visible seam. They’d still fit together really well, but not visibly more than you could achieve with ordinary milling.
That said, if someone took the time to make demonstration pieces like this with a decent jewellery milling machine, I think people might be surprised by the result.
Making seamless metal couplings – easy peasy!!!
Learning to adjust sound levels on a YouTube video – hard as fuck!!!
3D toolpaths, micrograin minature carbide ballmills from harvey tools and low stepover are your friend.
You would be surprised the kind of finish you could get with a good setup on a crappy CNC mill if you do very elaborate and long programs using those features,
And if you don’t have shrink fit tooling for zero run-out, you indicate your tools and adjust them with careful tapping techniques in a collet holder.
Making parts fit like this is impossible even with CNC just because there will always be tooling marks left that keep surfaces from being perfectly flat and joining like this to where you can’t even see the seams but making parts fit extremely nicely together is totally doable
For certain things that I do including some special manual machining I do hold tenths (0.0002" or so)- and sometimes thats on my prints, but most of the time, no.
I could hold that tolerance with a hand held graver on a watchmakers lathe though, and could for maybe the last 6 years or so. I learned tool and die skills in watchmaking school in Oklahoma that has since closed.
Some of my tolerances are bigger, effectively around +/-0.002" or so, but I can’t elaborate further than that. Forging dies get heated up a lot and smashed, so the tolerance isn’t as tight as you would expect. Reworking them back to spec afterward is harder.
There is a Kern micromilling machine in my building that can do stuff to a micron precision, with a 5 micron accuracy. We use it for making the tiny waveguides in blocks of brass that become submillimeter radio telescope receivers.
The smallest end mills are .002" in diameter.
I believe ball bonding is most commonly used:
You are correct, though in industry, we call it “wire bonding”.
Here’s an example of how fast modern machines are:
Filmed with a high speed camera:
Animation with description of process:
It is also used to bond nodes from silicon dies to PCBs (typically made from copper clad fiberglass). You have probably seen an “epoxy top” die on a cheap blinky something. This incredible video from Bunnie Huang shows how wire bonding is used to achieve this. Note that the operator is not in a clean room, is using regular FR4 PCBs, and how quickly these bonds are completed. Notice the CRT display, and yet this is all done with automatic optical inspection via computer vision (on a computer system that is very obviously very old).
http://bunniefoo.com/ntw/usb1q13_hd.ogv
Bunnie talks about this eloquently in much greater depth here if you are interested:
https://www.bunniestudios.com/blog/?p=2946
If you find this stuff interesting, you should absolutely read his book “The Hardware Hacker”. I’ve never seen a publication that makes high level computer manufacturing processes more accessible.
Regarding the article topic, EDM is truly an impressive manufacturing technique. That said, impressive as these demonstrations are, we have little need for this in any modern devices, and the cost is quite often prohibitive. Other countries, not so distracted by the incoherent babbling of ultra-rich heirs, care about the quality of manufacturing to the extent of filming TV shows that revolve around it. Check this out:
ps, 15 years lurking, first post… I guess wire bonding gets me more excited than I realized
As @ficuswhisperer said, silicon dies are bonded to lead frames or PCB pads via wire bonding, which uses a combination of heat, pressure, and ultrasonic energy to initiate intermetallic bonding between a gold wire and the metal layer of the die. Cold welds are not as “instant” as you might think; they tend not to happen when you want them to without extreme pressure, though often bond spontaneously when unwanted, such as in space…
However, wire crimping is an extremely common form of cold welding - you may have even done it yourself already. Graphic:
More info:
“cheers fam”