Pick and place machines are a wonder to behold, as they delicately and accurately place part after part. Unfortunately, they have to have a similarly wondrous price tag. Luckily, they aren’t too difficult to make yourself as they share many properties of a 3D printer with some extra constraints. [Stargirl Flowers] released Starfish, an open-source pick-and-place control board based around an RP2040 to help people make their own.
She purchased a LumenPnP, and the itch to tinker became too much to ignore. The STM32 on the stock controller also happened to get fried, leaving an obvious opening to create a custom board. [Stargirl] chose Trinamic TMC2209 motor controllers to drive the three stepper motors. The power circuit is impressively overbuilt with a 3A fuse, a TVS diode for shunting voltage spikes, a P-channel MOSFET for reverse polarity protection, a low-pass filter for AC ripple, and a large 100μF capacitor.
The RP2040 is a good choice since it’s easy to get and has plenty of digital I/O. USB connects the board to the outside work and includes ESD TVS diodes to protect the board when connecting and disconnecting the USB port. Motors for vacuums are controlled by a 74HC2G34 buffer that drives enable lines to two MOSFETs. Solenoids are similar but with a high current peak and a much smaller current to keep them open. The DRV120 fits the bill as it is a single-channel relay with current regulation. I2C vacuum sensors are the same ones on the Lumen motherboard; they just required an I2C multiplexer.
It’s an extremely well-documented project explaining why each part was chosen and why. If you want to create an RP2040 project that needs to last, we consider this a guiding star. It’s all up on GitHub for you to take a look at.
This isn’t the first time we’ve seen RP2040 as part of a motor controller, and we suspect we’ll see more.
Why is there a car type fuse on the brett?
They’re cheap, widely available, and have the rating written on them.
so you can easily replace it, when it pops. making it user friendly.
It’s been really fun following this project on Twitter. The author (@theavalkyrie) has fairly varied interests and a lot of irons in the fire but pretty much exclusively puts out really quality work like this (see wntrblm products) and most everything is open source. Well worth a follow if you haven’t already.
The design of the board is absolutely impeccable, I really can’t find anything that I would have done differently, both in the schematic and layout. The excellent writeup is very good documentation.
Outstanding work!
Do see the fuse but what happens if you accidentally flip the incoming polarization? Or spike the voltage too quickly? Would it handle too much current or voltage coming over USB? JTAG is nice to see at least. Does look like a pretty solid build just briefly looking at it. Wonder how well it would upgrade to a newer RP2040 improvement and also how it uses both cores.
This might be a stupid question, but how difficult would it be to modify this project for a “pick and pull” instead? The idea being to harvest and sort components from obsolete boards to help reduce e-waste.
I know there would be a myriad of issues due to the variety of ways PCB are constructed and their increasing component density. However, I think some of these issues could be overcome with obsolete boards by using their diagrams/schematics as a map and a camera for confirming the components before their desoldered and sorted.
Where schematics aren’t available the software could use OCR and a multi-meter test head to identify the unknown component. The test head could also identify weather or not the salvaged component is still viable.
This is mostly about reducing the time and number of steps involved in recycling old circuit boards. Even if all you’re after is the raw materials and metals, not having to go through multiple toxic baths to recover them has to be an improvement.
The bigger question, is it worth the effort? Is it better than putting the boards through a shredder and dissolving everything in multi step caustic baths?
I think the standard technique for recycling components is to heat the entire board, and give it an edgewise whack so that the components slide off. I’ve seen videos of this being done over an oil drum fire, even.
How often would you actually build something with desoldered components? Even for one-off hobby projects, it’s rare for me, since you would need to design a board around a component you may have only one off, and it has to do exactly what you need.
For hobby purposes, I’ve used hot air and tweezers in the past, for both SMD and TH components.
I like the idea, but I think it would probably be time and monetarily expensive to implement.
That being said I remember stories about washing machines being bought just to salvage chips out of them during the height of the chip shortage. I think if the need is high enough someone will do it, though in the past it seeems like recycling components has be religated to humans manually recovering the parts, so it might just be cheaper to use meatbag labor period.
I think the best solution if you intend to recycle surface mount parts is to simply keep your old/broken boards on hand and remove the components with a hot air station or soldering iron as needed for your current project.
The whole washing machine thing is an urban legend. There is no chip usable in a washing machine, apart from specialized micro controller, dedicated to run a washing machine (or dishwasher), there is only a (crap) SMPS chip.
The original legend begins with two separate event:
Russian use consumer grade chips in their weapon (Orlan drone for example) used in Ukraine.
Russian ruffian steal washing machine in Ukraine.
Both were concatenate by stupid journalists and buzz makers:
Russian steal washing machine to produce weapon used in Ukraine.
Considering the story has been around since the 90s (at least) I don’t think that’s the origin.
FWIW similar things can happen. When Blu-ray was new it was easier to buy cheap Blu-ray players to harvest the lasers than to buy the lasers alone. When the PS3 was new it was cheaper to build a cluster of them then a comparable cluster of x86 machines. Today it is easier to buy a pack of Arduinos than to buy the same number of ATMEGA328Ps.
Sounds like something really only useful to try at a more industrial scale but then other solutions for separation after desoldering en masse would probably make more sense, a la the various kinds of implementations used to sort Lego. And even then I think materials recovery would likely be the most feasible goal, not working parts other than perhaps expensive chips.
It would be way easier to just heat all the board upside down and order them after they droped, as Derek says.
Anybody know what firmware is used in Starfish board reprap,marlin ect