Dremel make a handheld blowtorch, (it's not meant for the kitchen but that's where mine sees most use) you could totally strap it to an x/y gantry (like people do with Dremel rotary tools for mini PCB/engraving mills) and toast bread.
Hard part would be controlling gas flow and ignition.
As far as I can tell, toasters and blenders don't do 25-50 micron x/y alignment by design. Seems like replacing the hot-end module on a 3d printer with a pick&place module could solve a lot of the project scope. Selling the p&p module and software lowers the barrier to entry sufficiently that you'll get lots of interest and easily available replacement parts for most of the machine.
My point, that I should have couched in less snark, is that the ONLY thing that 3D printers have are the X, Y, and Z axes. The Z axis is also way too slow for a PNP. While it seems on the surface to be a good starting place, it’s maybe 10% of the way there and after adding feeders, computer vision, and all these other things, now all you have is a pick and place built on an unsuitable platform and you’ve saved essentially negligible amounts of time or money.
I mean to say that a toaster and a blender are not that much closer to being a good starting platform than a 3D printer as a fraction of the total design of the machine. It’s superficial.
Because the only usable part would be the X axis. The moving bed on the Y axis would most likely cause the previously placed components to shift around a bit as the build progressed every time the bed changed direction. The Z axis has way more travel than needed for pick n place. The Creality motherboard is not usable for this project and needs to be replaced with a bespoke one. Building the hardware is the simplest part of the project, so why compromise and try to shoehorn someone else’s design into yours? 8020 extrusions are cheap and can be ordered pre-cut to length and pre-drilled/tapped if you don’t have the equipment to do it yourself. Everything else is just stock aliexpress parts.
Commercial pick-place machines have moving 'beds' as well, and they move quite fast. SMT parts have sufficiently low inertia that the paste keeps them put. I'm sure that tuning the stage acceleration rates for max speed in development is a messy process though :p
I agree with that overall, except that it seems to me like the stock Melzi board has enough IO and stepper drivers to run the physical part of a basic PnP. (It doesn't change your overall point, of course.) I think the Ender3 and clones are best viewed as a kit of relevant parts.
You have to be able to pick the part, orient it in the correct direction, and place it within 1/4 of the pad pitch, while taking into consideration variations in board placement, thickness, flexure of everything, and thermal expansion.
And avoid hitting components already on the board, of course.
As others have said, it's the vision thing. You have to have live cameras that feed into a controller to do collision avoidance, verify board placement, part placement, and a whole host of other issues.
3d printers assume a flat empty plane, and build upwards in layers. They're not very stiff, nor very accurate.
And of course, you also have to get the components you're going to place off of a roll of parts, for many types of parts and multiple rolls.
(Please bear in mind this is all the details I learned while observing others attempt this at a makerspace, it's not an exhaustive list of all the fun you'll encounter)
Nice. Similar to the Liteplacer.[1] That project seems to have stagnated, but the hardware still seems to be for sale as a kit. The software is still available.
The Liteplacer is unusual in that it doesn't use feeders. You just put the parts in little trays or tape down tapes of them, tell the software approximately where they are, and it locks on with its vision system and picks up the parts. This makes it useful for assembling prototypes.
Last site update in 2018, which isn't good.
One problem with the Liteplacer is that it comes as a kit, and you have to build a baseplate for it. They don't sell a baseplate with pre-drilled holes, which means getting precision alignment isn't easy. They should have had aluminum plates cut on a water jet to make a solid baseplate.
The Liteplacer had a nice vision system. All open source, so you can reuse it with different mechanics.
Someone wrote an application for it which reads part numbers and sorts parts.
I've been following the build videos on YouTube because it's such an interesting project, but be warned he has a very.. particular style? I have to watch with the remote in hand so I can skip the crash zooms in on his face or inside mouth, the squealy 'self-reactions' (like laughing at your own joke, but for anything), etc.
It obviously doesn't bother everyone, and it's orthogonal to the really cool, interesting, hard, well-executed project. But I do wonder if it might be more popular/more viewed with.. just a little less personality!
That's what makes so many you tube videos so difficult, especially when the purpose of the video is explanation or description (i.e. a lesson). Please just focus on the agenda, and maintain a flow of thoughts.
People mostly just want the info. That said, this looks like a great piece of work.
I wish people would stop using YouTube for technical content and blog instead. Sure, some things need to be video, but you could just have that part in the video and link to it from the blog. (SIGGRAPH papers are a decent example.)
Basically, no talking heads. That stuff should be text.
O(months) ago Dave Jones of EEVBlog had a thing of 'where should I put my head' - 'do you like the green t-shirt so it's just a floating head or is it too spooky' - 'lemmeknowinthecommentsclickthebelltosubscribealgorithmsalgorithms'. Gah, if I had an account I'd have been despairing in the comments. Who cares where your head is if it's not in the way (this was specifically for shots where he's talking over a teardown photo)? There's a really easy way to make sure it isn't!
Ben Eater has it right, all content, voice over video.
It's not that I 'hate people' or whatever, depending on the content/style of the rest of it it can be fine: Clough42 for example is moving around a workshop doing physical things, so it seems quite natural when he turns to face the camera, or for the intros.
It helps that his production quality is insanely good too - for all the time spent talking about new expensive AV gear on EEVBlog, it's not great! Not because of the gear, but the refusal to do a re-take if it goes wrong - cc: sorry about the mic it gets better in a sec - editing mishaps like repetition of a shot or not removing one that was clearly intended to be, etc.
Can't complain though, I enjoy it all for free, and I'd recommend every channel I've mentioned.
I don't know how serious of a name collision this is, but Index is the name of a company that has a very long history[0] of making machining tools. Today they mainly make CNC lathes and mills[1].
I own a post-war era Index-25 automatic lathe and it's a marvel of mechanical engineering. They can make a modern general purpose CNC lathe look really slow by comparison. Here's some pictures[2] of a similar but younger machine that I found out there.
I was about to downvote at the mention of name collision, but this one is actually serious. It's not really our business, but no way he'll be continue operating a business with that name in the long term.
I’ve often been tempted to buy or build a pick and place after an evening with tweezers looking through a microscope, the the cost is a little prohibitive - in the past couple of years it’s now cheaper to get even single boards built
I think the biggest breakthrough in making a smaller pick and place would be to stop copying the design of industrial machines. If you don’t need to populate full sized panels you don’t need a large x-y grid that needs to be rigid and support for reels of components both of which are costly.
Vision has come a long way in the past couple of decades and the reflow process is quite forgiving. I’m sure a robot arm (0.5mm accuracy over 30*30cm) and a decent vision system to find components on cut reels attached to the table would work.
The main problem I have is that I have like 100 different types of component, and each component appears 1 or 2 times only (there are some exceptions of 20 components). This means I will have to mount a reel almost as many times as I have components.
35 comments
[ 0.21 ms ] story [ 89.6 ms ] threadSuch a cool project, regardless of whether or not he sells any.
A related take on the same concept - probably a little cheaper and less hassle to put together: https://youtu.be/Kkc7uRNK-7g
Hard part would be controlling gas flow and ignition.
I mean to say that a toaster and a blender are not that much closer to being a good starting platform than a 3D printer as a fraction of the total design of the machine. It’s superficial.
Then they hit the vision issues and the project stops cold.
And avoid hitting components already on the board, of course.
As others have said, it's the vision thing. You have to have live cameras that feed into a controller to do collision avoidance, verify board placement, part placement, and a whole host of other issues.
3d printers assume a flat empty plane, and build upwards in layers. They're not very stiff, nor very accurate.
And of course, you also have to get the components you're going to place off of a roll of parts, for many types of parts and multiple rolls.
(Please bear in mind this is all the details I learned while observing others attempt this at a makerspace, it's not an exhaustive list of all the fun you'll encounter)
The Liteplacer is unusual in that it doesn't use feeders. You just put the parts in little trays or tape down tapes of them, tell the software approximately where they are, and it locks on with its vision system and picks up the parts. This makes it useful for assembling prototypes.
Last site update in 2018, which isn't good.
One problem with the Liteplacer is that it comes as a kit, and you have to build a baseplate for it. They don't sell a baseplate with pre-drilled holes, which means getting precision alignment isn't easy. They should have had aluminum plates cut on a water jet to make a solid baseplate.
[1] https://liteplacer.com/
I have lots of parts that would take more time to setup in the feeder than it would take to do the board by hand.
A 1" MDF board should make a fine base, aluminum is nice but not necessary.
Looks like the software is still being updated: https://liteplacer.com/software-release-history-and-previous...
The owner seems to be active on the forum: https://liteplacer.com/phpBB/viewtopic.php?f=10&p=10292&sid=...
It obviously doesn't bother everyone, and it's orthogonal to the really cool, interesting, hard, well-executed project. But I do wonder if it might be more popular/more viewed with.. just a little less personality!
People mostly just want the info. That said, this looks like a great piece of work.
Basically, no talking heads. That stuff should be text.
Ben Eater has it right, all content, voice over video.
It's not that I 'hate people' or whatever, depending on the content/style of the rest of it it can be fine: Clough42 for example is moving around a workshop doing physical things, so it seems quite natural when he turns to face the camera, or for the intros.
It helps that his production quality is insanely good too - for all the time spent talking about new expensive AV gear on EEVBlog, it's not great! Not because of the gear, but the refusal to do a re-take if it goes wrong - cc: sorry about the mic it gets better in a sec - editing mishaps like repetition of a shot or not removing one that was clearly intended to be, etc.
Can't complain though, I enjoy it all for free, and I'd recommend every channel I've mentioned.
I own a post-war era Index-25 automatic lathe and it's a marvel of mechanical engineering. They can make a modern general purpose CNC lathe look really slow by comparison. Here's some pictures[2] of a similar but younger machine that I found out there.
[0]: https://www.index-traub.com/en/company/about-us/100-years-in...
[1]: https://www.index-traub.com/en/
[2]: https://www.drehautomaten-pyka.de/Reparatur/reparatur.html
I think the biggest breakthrough in making a smaller pick and place would be to stop copying the design of industrial machines. If you don’t need to populate full sized panels you don’t need a large x-y grid that needs to be rigid and support for reels of components both of which are costly.
Vision has come a long way in the past couple of decades and the reflow process is quite forgiving. I’m sure a robot arm (0.5mm accuracy over 30*30cm) and a decent vision system to find components on cut reels attached to the table would work.