Hey friend, I’m a fan of electroplating and have been studying it for about 6 years. What you want to avoid that creeping copper growth is to do pulsed AC plating, and if possible, to force movement of the fluid through the WIP via. Here’s why: 0) By pulsed AC plating, I mean that instead of a constant current, you have a periodic cycle of “plate, plate, erode, rest”, for example. You have, say, 1A for 0.5s, then -1A for 0.25s, then 0A for 0.25s, and this repeats on and on. 1) Pulsed AC plating will, on average, deposit material (deposition is proportional to (average current) x time). In the eroding step, however, it will selectively erode the pointiest surfaces, because that’s where the current prefers to flow through. What this means for your VIAs is that you’ll erode the spurious dendritic growth that creeps towards the electrodes. 1.1) Pulsed AC plating will also help make a smoother, more consistent surface. 2) Forced movement will help you a lot here, because you need copper ions to successfully reach the surface and get plated. In normal circumstances, ion movement is mostly limited by diffusion. When ions get plated, new ions fill replace them by coming from some random direction in the fluid. In the case of the thin 0.3mm VIAs however, diffusion will have a hard time replenishing the plated ions inside the bore, because there is almost no direction where new ions can come from. In your setup, as it stands, that space gets starved of copper ions and stops conducting altogether. Get a good flow of electrolyte solution through the bore, and you’ll see how much better the process behaves. There’s a book on this from the German book shop Leuze Verlag, the book’s called Pulse Plating, it’s silver-colored and has a picture on the cover of electroplated vias that were successfully done using this precise approach. Anyway. Cheers, friend. Your project looks epic. Please reach out if you want some help, I’ll gladly give you whatever info I can.
Thank you for that excellent explanation. I am your second sub. You should make content! Seriously, just feed the above into an AI and have it make a conversation and a video, if you are shy and don't want to use your face or voice. You have excellent information! It would be awesome to see you work together.
It's a good PR move and basically has no financial downsides. It's an alternative, not really a competitor. PCBWay is convenient, easy, and cheap unless you make a lot of boards, most hobbyists will pick them instead just for the convenience factor. Besides that, I can't imagine that the hobby electronics market makes up more than a fraction of their revenue. A nice fraction probably, but not a significant one. It's win-win for them.
This is an ideal video for them to sponsor! Guaranteed that almost 99.9% of viewers are interested in producing PCBs, and also interested in fast turnaround, which is something PCBWay is good at. At the same time, "how to DIY your PCBs" videos thus far have universally shown what a laborious, difficult process it is leading to mediocre results. Against which, online PCB vendors look like a better option. Of course, the best of luck to Levi advancing the state of DIY art!
If your prototype works you still need production boards in modest quantities. That's what they are good at. I order 50 to 200 boards at a time and they turn it around in 3 days.
Oh definitely, it's one of those "journey, not destination" things that make these sort of videos so interesting. Presenting only the success would be a shame to hide the effort and thought process that went into it.
Me listening to this as a mechanical engineer: "Why does he know that ?" "How does he know about that?" "Wait where is his encoder for his position accuracy?' "Why does he know anything about pumps ?". This was awesome.
@@MNSweetDebugging with a proper tool would be much better though? Using the console to print variables is a terrible way of debugging a project. I know you have no option when you are stuck on arduino and stuff like that, but please, if you can, use a hardware debugger. Its much, much better.
@felipevellasco6526 I was just continuing the trend and didn't want over complicate it with too much detail or jargon. Hence why I didn't mention what type of dev. As much as I want to learn about IOT I'm more web application and haven't touched an ardinuo though I did just pick up a ESP32 starter kit and I'm looking forward to making some lights go blink on a bread board.
Btw. The design you came up with is called H-Bot. It's pretty common in industries. But you have to use proper guide length to force distance proportions
The issue with H-Bots is that they add a racking force to the gantry. With solid enough rails this will be less of a problem but it's the tradeoff thats being made and there's a reason printers go for core-xy even though the belt path is a little more annoying.
I second scruffy3121. H-bot have been long known to be the nemesis of cheap linear rails. Plus it requires precise sync of your motors braking timings if you don't want to end up with a lot of skipped steps with open-loop stepper drivers.
Hey Levi, I'm a process design engineer that deals specifically with electroplating metal into nanopores. Is the solution you are using a commercial solution with additives? You need to look into pulse plating, this will dramatically reduce the surface overplating you're experiencing.
I’m assuming with pulse plating you give time for the electrolytes to reach equilibrium in the liquid resulting in a more even coating that isn’t dictated by a variance in surface charge that probably pulls ions to certain regions
@@jonathana.1802I don’t recall why, but the industry settled on plating a long time ago and have done it that way since. I’m not exactly sure why though. It’s a good question.
@@jonathana.1802soldered vias have a few problems. Firstly, they stick up out of the board, getting in the way of hotplate reflow. Secondly, any hot-air, infrared, or hotplate reflow is going to melt the vias too unless you want to mess about with different alloy melting points. Thirdly, automatically soflering from two sides would be a pain, especially with no solder mask to keep the flow in check. Etching a soldered board has its problems too. You’d be better off trying to make a machine to rivet copper wires into holes that have been counter-punched, in order for the riveted heads to be flush. I did that by hand for a while.
You can fix the electroplating problem. This is actually an issue which had to be solved in semiconductor etched trench filling. The diffusion blocking layers needed to coat the bottom and walls evenly and different coating processes are more directional than others. Extend an insulation covered needle with only the tip uncoated into the center of the via, through the center of the tube. This will decrease the distance to the surface and the current path through the electrolyte solution. The proximity to the plating surface will also improve the directionality and placement accuracy. With the current flowing from the needle horizontally to the walls it is a straight path giving more even coverage. You could feed an insulated solid core wire from your reservoir through the tube to avoid any risk of leaks. 3d print an X shaped electrode holder for the center of your tip and use it to center the electrode in the XYZ direction so when the tip seals against the PCB the tip of the wire is centered in the via. I think for the final version you will still want to change the design to a multi-head or headless electroplating bath because even 90 sec per via could take days for a design like a raspberry pi.
@@tomaszradomski8994 Yes. I think he is pulsing the flow if I understood him correctly, which should be good to prevent depleting the copper ions in solution.
I just can't express how much I appreciate your 1. trying new things 2. reporting the failure. Failure is critical, leads to all advances, and sharing those failures destigmatizes this (critical) process. You've done a huge service to the community even if you never get anything working. And you HAVE gotten so much working. I'm a huge fan.
I am also working on something similar hoping to make it work and share it on the internet, Your work and videos on the pnp and laser based PCB manufacturing are super helpful. Thanks to both of you for inspiring and making the knowledge open source and available to all.
For some reason I expected to find you here 😅. A collab would be awesome 😬. Your 2 projects are like jam and peanut butter 🤭. Thank you both for sharing your knowledge with us ❤.
That was cool! Chemist and machinist here, hear me out: What if, instead of chemically depositing copper, you drilled the double layer board, slid in a piece of copper capillary tubing that is slightly longer than the board is thick, and riveted/pressed it in from both sides with custom dies that have centered guide pins + a concave face? Locking it in place and making contact on both sides, like a hollow rivet, you could even solder them on each side I bet you could pull it off
Came here to say this too. Even copper wire mechanically inserted and his existing process could perform easier surface plating of the copper instead of a solder or riveting for a chemical instead of mechanical joining. Tons of opportunities "good enough" for a prototype PCB.
i use it for my home pcb production. The problem with it is that for the top layer (from which the drilling starts) to make the contact with via, rivet must be pressed hard enough, to spread to copper. That creates bump, and most smd components could not be properly placed above it. And applying soldering mask is harder while vias are so much (up to .15mm) above copper. It became the limitation for my designs. My method is not autometed BTW. It might be OK solution for hollow rivets for thru holes tho/
@@ВоваЕршов-й2ьwhat if you drill a slot in the PCB where the rivet would sit against the PCB a few millimeters/micrometers beneath the top surface, like a valley?
Great stuff! Would love to have a desktop PCB printer like that. My method of making vias was always: 1. Drill a small hole 2. Stick a thin wire in the hole 3. Solder on both ends to the copper Caveman method, but works pretty well for prototyping :)
@@E-dartIt’s the “Single Belt Core XY” kinematic system demonstrated in this video. It’s just that Levi didn’t know that is already existed and had a name - which is no great surprise. They were a bit of a flash in the pan about 7 years ago or so. They were a cool idea, but as demonstrated in this video, require a substantial frame to actually be reliably accurate.
Great work, and thanks for sharing. This is the promise of the internet -- each of us learning from others, asynchronously. Looking forward to seeing this project progress.
Unless you are aiming to do multi-layered boards, PCB rivets is the answer. They are old school and extremely robust. If the end goal is multilayered boards, then god speed good sir. I would love to be able to do 4 layer boards at home.
I think you are talking about eyelets. Many years ago, one of the pieces of military equipment I worked on used eyelets rather than plated thru holes. I don't think they trusted PTH at the time. The eyelets gave us a lot of trouble with corrosion around the eyelet ruining the connection. Reflowing the solder around the eyelet would fix the problem, at least for the mean time. This was not a cheap piece of equipment, it was used to synchronize cryptography equipment.
The major problem with PCB rivets are that they aren't actually made to do that- or practical for complex circuits. They might work for this use case, but still have a lot of drawbacks. Instead they are made to mount a heatsink or other non-solderable component via the connection instead. Many circuits that use high amperage op-amps or rectifier blocks use this method for cooling to a passive aluminum heatsink. Toasters/toaster ovens use them to connect the nichrome or steel heating elements to the board, where that much heat would melt the solder of a regular connection. Most of them are thicker where they would be mounted requiring milling a small channel into the board around the via (to account for it's chamfer) for it to sit flush with the board itself. There are also only 3 ways to to connect one half of the rivet to another, the button method, where one of the hole openings has the mount point, but is larger & a few mm thicker than the other, the rack method, where small square channels are cut into each side of the center of the rivet allowing them to slide into each other like the guide for a drawer in a dresser. This allows for the thinnest via's but has the worst resistance & tend to separate under many heat/ stress/ weight cycles, sometimes breaking the connection completely. Lastly is the flange method, which causes one side of the tube to expand outward in the center of the hole. Not only does most of these methods usually require oversizing & extra milling time for the hole & the lap space between them, but the smallest rivets are 0.6mm in size, compared to modern whole board CPM which can have via's as small as 0.15mm in size.
@@elvendragonhammer5433for circuits with any kind of complexity, 2 layers don't cut it in any case, and you can have a professionally manufactured prototype shipped to your door in under a week.
I'm a total noob at this, but couldn't you sandwich boards together to make multilayer PCBs? It obviously won't be a nice as something from a manufacturer, but it might work.
I work at a recycling center as a bus driver for the employees. They recycle plastic, paper / cardboard Etc. They shredded around 30 - 40+ big boxes Packed full of repairman schematics on old televisions. This was either a shop or manufacturer that was from Cincinnati Ohio at one time. Just about every old CRT, vintage television set you could think of was in there. Some televisions were pre-1950. It broke my heart as a Electronics hobbyist seeing all of that engineering hard work & information destroyed!
In my mind, this process makes sense: (This process begins after all via holes are drilled) 1. Insert a strand of wire into the hole(alignment hell + you need multiple wire gauges) 2. Align or trim the end section of wire flush to the board. (Difficult to prevent marking or damaging the surfaces) 3. Deposit a thin film of flux and potentially high temp solder. (Brazing metal?) 4. Spot weld the bottom face. (this is probably the biggest stretch. The difficulty in calibrating the pulse energy and duration is beyond me) 5. Cleanup/maybe surface the weld 6. Repeat 2-5 for top surface As with most engineering ideas, this is much easier said than done. I like this process more because all of these steps have high time optimization potential. If you read this, please let me know if you have any thoughts or critiques on this process. Regardless, this video is badass.
@jammi__ I just don't think plating is optimal for this project's goals. Chemistry clearly scales well. This is why it is so dominant in industry, but if you do 1 via at a time total fabrication time will suffer. Also, having to worry about electro plating consumables seems way more complicated on its face than just buying wire spools. I must say, though, I am not an engineer. That is why I wanted input from others with real experience to tell me where I am wrong/nieve if possible so I can learn.
well i think it would be viable to use riveting. So bottom head would have anvil and top one would have anvil and wire feeder with cutter and fter that guide tube. Steps to do via after drilling: 1 position bottom head under hole 2 position guide tube over the hole 3 extend bit of wire (thickness of pcb + something for rivet heads) 4 use wire feed to push cut piece down the tube 5 lift top head (feeding bit more wire while lifting should prevent tiny piece sticking) 6 move top anvil over hole 7 lower bottom anvil bit to allow wire to poke trough (optionl step can be integrated with inserting wire) 8 squish from both sides Possible problems: cutting 0,25mm copper wire without it being stuck somewhere, possibly not tight enough squeeze, but but if it could apply solderpaste then it should be possible to smear little bit over vias and they would self solder during reflowing
@michatrzcinski8741 This seems more doable but would lead to a lower quality via. First, even if you squish the head super flat, there will still be a raised section that could interfere with smd parts that go over top of the via. Secondly, for pretty much any mechanical method other than maybe stir welding, you will end up with a microscopic gap. This would have higher resistance, could cause intermittent contact issues, could decrease signal quality, and would have lower mechanical strength. I will say, though, that your idea seems much easier to implement.
I appreciate you uploading this despite having run into a roadblock. You're attempting something that nobody has done before. I'll stay tuned because this is something I myself have always wanted as well.
I started doing research for this over 25 years ago. I too needed to rapid prototype boards done in-house. This was way before 3D printing. Additive manufacturing wasn't even known then. I tried using a color desk jet with conductive ink. This allowed different types material; conductive, non-conductive, etc. I also tried color laser printer using metallic powder. Both designs used a Mylar sheet that fit standard printers. For multi-layer type boards, the printer laid down a non-conductive trace.Finished sheets could be mounted on a rigid board, or left flexible. Eventually I decided to wait until technology caught up.
First I couldn't afford to , then I didn't know how to . Finally, after I had overcome those obstacles, pcb manufacturers came out with low-priced fast turnaround high-quality pcb's to order . Now , once I finish my gerbers, I have to avoid the accusing stares of my accumulated pcb equipment and chemicals as I send my order.
Incredibly cool project, and looking forward to seeing it progress. One suggestion I would make is instead of going for a CoreXY or H-Bot architecture, it might be easier to go for (in 3d printing terms) a "Bed slinger" where you would move the PCB instead, meaning your via and trace apparatus (the heavy bits) can be as secure as necessary and don't need to move, and the PCB (which barely even gains mass like the issue is for 3d prints) can whip around as fast and secure as you like. For future developments, if multilayer boards are the goal, it might be possible to make sets of 2 layer boards and then stack them with an extra layer in between, but making everything line up perfectly might get tricky.
For the vias you could try doing like other diy pcb solutions have done: tiny copper rivets. Make the gantries rigid enough and the Z axis strong enough to "crimp" the rivet in place and it should be pretty reliable.
One potential idea I had for simplification of the mechanics is a secondary outer o ring which provides a second sealed area that you could connect to a cheap 12v vaccum pump to suction down onto the board for a solid seal with a less rigid gantry required. You could also probably build magnets into the heads for registration and use an approach where some loose strings pull the heads roughly into position, the magnets cause them to snap into alignment, and then the suction engages providing a solid seal to the board
Dude, great project, you got my sub! Here are some suggestions: 1: Be more specific about your electroplating solution. The PCB industry use PEG-JGB-SPS additives, they have a huge impact on quickly plating small areas with high accuracy. SPS (Bis-(3-sulfopropyl) disulfide) is probably the most important since it's job in the additives is accelerator, it increases Cu deposition in low current density regions, that's the problem you described. Also decreasing the PH aids in conductivity so consider adding H2SO4 (sulfuric acid). 2: Use AC Pulsed plating. The on/off cycle allows for faster and higher quality plating without creating dendrites. 3: Ensure a stable temperature of about 50c-60c, this is the optimum temperature for ion exchange in a typical CuSO4 plating solution. 4: Agitation! An ultrasonic inducer (20-80 kHz) should be placed in the tool head, they're very small and would fit your compact design preference (mine too!). Electrolysis creates a lot of hydrogen bubbles which interrupt ion flow, especially in tight spaces, ultrasonic agitation helps burst the bubbles quickly. The agitation also breaks up surface tension which slows ion movement, and if the ultrasonic inducer were pointed at the via hole it would actively drive fresh Cu ions where you want them. Hope that's helpful. If you made a KickStarter I'd love to back it as I'm sure would many others!
3:35 if you're gonna have multiple tools, just use a rivet gun. Rivets conduct, just drill you hole, put a tiny rivet through, and pop it to bridge the traces
That would work for larger holes, but like he said towards the end of the video, most vias are 0.3 mm. I could be wrong, but I don't think they make rivets that small
This is great stuff. You've got the right idea on the requirements side. I never have a need to make a single sided board and nasty chemicals are out of the question. So many similar projects fail to clear those bars. Definitely going to follow this closely.
Would it be feasible to just do the ink deposition per via, and then switch to another "static" setup with custom-made (hurray to 3d printers!) liquid channels for all the vias? Or even just sealing against the board edges. So electroplating would be parallel for all holes, but still with a small amount of liquid and well sealed and desktop friendly? After all, we are not talking panels but single, small PCBs To my understanding, your idea (and success) of applying conductive ink as the base for the electroplating is already amazing!
Awesome, great overview of the entire process and end results. I'm really looking forward to see this come along and eventually be developed in some form as I want to tackle the same problem myself too. Thanks
You'll need a method of refreshing the electrolytes. I would use a tube that feeds copper wire into your bath so the copper wire becomes the sacrificial electrode. You'll also want to regularly measure the conductivity of the plating bath between runs and have the ability to add solution to make up for lost electrolyte/liquid. Might look into faster drying plating solutions, one option is to use resistive heating to "bake" the resistive element, between the deposition and resistive testing phase. would let you deposit->test resistance->bake->test resistance but would require a current regulated power supply.
Very humble to introduce yourself as a maker. To me you are an engineer. Thanks for sharing your project. The way you looped the XY axis was really smart.
Great video! Slight correction. As a former microfluidics researcher, you’re not using this term correctly. Microfluidics is not the term for small scale fluidics. Microfluidics is specifically about manipulating liquids in laminar flow state using microdevices. Usually the “tubes” (channels) in these devices have diameters below 0.5 mm or 500 microns (hence the term “micro”).
I'd love to see this work! A few suggestions come to mind: * what pressure/feedrate are you operating at? Faster will almost certainly be better; as cybyrd alludes, ions are pulled out of the solution by the process and this will interfere with how even the coating is * use a copper electrode centered in the hole, with electrolyte forced past it at pressure; if electrode erosion is sufficiently fast, move it (a la wire EDM or microhole EDM) ** having something stick through the hole will also make alignment easier; the bottom can center itself by clamping onto the electrode * heating the electrolyte will improve reaction speed (but might result in other trade-offs) * try pumping electrolyte _towards_ the bath that has the electrode in it (or alternate directions) * only ground the copper on the side away from the electrode
If the goal was to have a quick prototype, this fulfills that: prep a board full of vias in a 6.35x6.35 mm grid pattern (all at once, complexity isn't an issue). Design the pcb to hit or avoid (at least on one side) the pre-exising vias. Some can make profit on the prepped via boards (e.g. pcbway or you) - and the desktop machine is much much simpler (a laser and/or engraver).
The goal is a quick prototype that is very similar in electrical properties and layout to a production board, so that's not exactly what I'm going for. But that's a very intriguing idea, not one that I've ever thought about. You could have different pre-prepared boards with different via pitches to accommodate different power/trace density needs. Have you seen this approach used?
As others have said in the comments, your failures and processes are very educational. Perhaps more to me than to most of the engineers on here as I'm a beginner. Thank you!
I like this interesting approach to diy vias. When I was in college we did diy vias with a bit of wire soldered on both ends. It couldn't be as small as this but, it got us off the ground. I think this has a lot of potential, but I wouldn't underestimate the complexity of the rest of the process. I'd love to see this idea progress forward, and the only big thing I'd warn against is the "never ending personal project".I think you've demonstrated a viable approach to this step in process. Some people have even commented potential improvements. Really, 3 minute vias is a pretty arbitrary requirement. Even with 10-20 minute vias, you could still make pretty complex and useful boards much faster than you could order them. I actually always thought it was an interesting challenge/puzzle to minimize vias in a design. Honestly, I think you'd be best served by pushing forward to a minimal viable product. Then, you can come back and improve the sub-processes later. That'll also give you better insight into the actual limiting constraints of the overall process, which might not be what you expect. Either way, I'm definitely interested to see where this goes.
This is a very important problem so I'm really glad you're tackling it! I didn't quite understand what part of the electrochemistry changed between your initial test that got you good results and the high resistance issue at the end.
The chemistry didn't change, but the solution was more "constrained" in the latter prototypes. At the end, the board and electrodes were separated by a thin tube, increasing the resistance. Once I increased the applied voltage, the electrochemistry was exactly the same.
As a maker that enjoys electronics, your concept is extreemely good. but nethertheless a hudge issue came to mind : the electroplating works efficiently because enough current can flow from the oxydizing anode (ie the copper plate in the bath) towards the reducing anode (ie the via being made). To maximize current, you will need to decrease as much as possible the resistance of the electrolyte path, by several factors : 1) put the oxydizing electrode after the pump (the peristaltic pump will make an open circuit with the pinched silicone tube). you can make just a little chamber for the electrode that would stay filled with electrolyte filled on one port by the pump, and the electrolyte would shoot out towards the nozzle. 2) You need to decrease as much as possible the length of tubing beetween the electrodes as the elctrolyte path resistance is proprotional to it 3) You need to keep a wide path to decrease even more resistance (R = c*S/L with C the conductance proportionnal to your electrolyte's concentration, L the pipe's length and S it's surface). I know you will need a thinner end to avoid air comming back up and keep the path filled tanks to surface tension, It will be a matter of compromise. and maybe plating a pcb with a photo-resist layer would avoid getting copper build up on the outer surfaces of the pcb and only in the hole to help it stick ? it's just an idea but there seems to be an issue similar to dendrite groth in lithium batteries. avoid letting copper grow where it shouldn't may help
Sounds like it would be easier to move the board between stationary toolheads rather than moving two toolheads around a stationary board. And why plate holes one at a time? It does make sense to ink them one at a time since it's quick and needs to stay off the surface of the board, but make the plating toolhead bigger. Maybe 25x25mm and move it around in a grid pattern, or make it require 100mm wide board material and have a toolhead 100x10mm or something move across continuously depositing copper.
Yep. Solder it in place and smooth the surfaces. Not sure why he is doing this so complicated way. This machine will cost a fortune, so must be targetting big business, not a diy solution.
@@murraymadness4674 I'd reckon that most people who tried to do the same would say that doing this in any kind of quantity is quite the chore. I did. It is. Picture yourself hunching over the board for hours soldering wires that are not significantly thicker than a strand of hair. The real difficulty lies in getting the solder connection reliable 100% of the time especially after you sanded back the protruding wire ends, so when you try to solder a BGA package on top, some of your joints come loose because there's just a few dozen square microns of solder joint areabetween the trace and the wire.
@@ceoyoyo I'd gladly spend a week building a CNC machine that plates a perfect via in 90s. That's 40 vias per hour, 320 vias a night. That's like 3D printing speed, in perspective. It's okay considering that I don't have to do something that really hurts my upper back and what a machine is supposed to do. :D
Super cool project. One of the issues I have with making PCBs at home using an engraving mill, is the lack of mature and well featured, GUI based CAM software to convert a gerber file into gcode. All currently available software applications suffer from one or more huge flaws, that ultimately make them either unreliable, or time consuming and fiddly to use. I think for your project to be successful and truly achieve the "fire and forget" functionality we all want, it needs good software too. On my own setup, I've modified and optimized the hardware and was able to cut down the setup time by an order of magnitude compared to its off the shelf version, and now configuring and setting up the CAM software, with all its bugs and kinks, is by far the most time consuming and frustrating part of the process.
It's just amazing to see the RUclips engineering community in the comments coming together to make this project work. Seriously, you guys are awesome 😎
This is fascinating! I thought you might like to hear my journey into a pcb mini mill. I was working in R&D in the mid 1990s and transitioning from through hole to surface mount so I asked the company to buya T-Tech pcb mill. Was told it was too expensive so I built my own from surplus motion control mechanisms, steppers, limit switches etc. I controlled the steppers by sending binary sequences over the pc's parallel port. I wrote a gui in Visual Basic and was able to move all 3 axis. I needed a way to automate pcb milling so I wrote code to convert the gerber files generated by my then pcb layout program, Winboard, into movements that would isolate the pads and traces instead of just milling lines as the photoplotters do. This was tricky but I managed to make it work. Later on I got some real stepper controllers made by Phidgets and got more steady control. I ended up milling hundreds of single and double sided pcbs. Since then I've retired but have the machine and still use it.
As earlier posted: copper rivet. 1) Wire through needle to push through via 2) solder paste squirt on back, and solder tip heat 3) solder paste top, solder heat. Yes whiskers will be there a bit, but hand cut possible. FDM shows you can get 80W to a head, so 6 mil is easy. Can even do it by hand w/ cheaper model.
The guys over at Bad Obsession motorsport did a video last month where they went over the whole process to make their own circuit boards by printing the metal tracks.
Maybe more cone shaped tip instead of a tube to concentrate the resistance to mostly to the via. Also what would happen if the electro plating current would be ac. Or pulsing the plating fluid back and forth to keep fresh fluid in the via.
Oooh! This is intriguing! The ability to fabricate two-layer circuit boards in-house could be game-changing. I like that you are keeping the requirements relatively modest to start with to ensure that the goals are attainable. I often make the mistake of making my goals too lofty. The device being able to self-calibrate and self-correct would be an excellent "automatic" capability.
I admire your efforts to do this at home. As an armchair PCB designer, an alternative comes to mind to have your servo and guide run a very thin wire through the hole and bend at the top to lock it in place. You could then hand-solder or add a soldering tip to your inserter. I look forward to your next steps!
Why not go with the THT component approach for creating vias? A very thin copper wire could be inserted into the hole using something similar to a 3d printers extruder, then get secured to the bottom using solder paste and a laser or hot air, then cut the wire at the top and solder it in place there too. Maybe it would be easier to flip the board in the machine instead of putting a solder paste application and heating tool on both sides of the board. Very cool project!
All homemade vias options were explored long time ago. For starts, you have to use good carbide drills at prescribed 30-40k RPM to get VERY smooth holes. Carbon paint is kinda easy at the first glance but leaves VERY rough and uneven plating and very hard to achieve sufficient conductivity in the first place. Silver nitride is better, but too complex chemically and still leave pretty rough and eneven plating. They only real, sustainable primer for homemade vias is palladium chloride - the very thing everyone use in the industry. It's easiest and best in literally every aspect. It seems expensive, but it's really not, 1g will be enough to electroplate approximately your whole house.
The talent you bring to the table is unmatched, i think you should work for nasa or something. I understand the specs you want to be perfect and that is toraly u understandable. I think you made success as a 1.0 version and many people could use it, i also think with enough thought you are going to solve this riddle to get it exactly where you want it. Theres almost always a way with physics and i feel you are right on the cusp of finding your answer. Sometimes the brain works so hard it squeezes but needs yo relax and the answer will come faster than lightning when enough bloodflow is alowed to be rushed in during relaxation periods like durring sleep is sometimes when i come up with the answers. I remember you talking about these concepts in the discord a long time ago and its super fun to see all the progress you have made and stuck with it. Your dream will trully help many folks, please never give up on it. Thank you for all the time and efforts you have committed. If more people were this comitted its possible most of the worlds problems would have been solved by now
Wow what an impressive project ! To make vias on single face prototype board (made with laser or cnc mill) the tricks I found is to drill the via the size of copper solid core wire, then insert a small piece of it and solder it on both sides. It's a quick hack but it works a charm when making pcb prototype at home without equipment
Here's my dumb idea for a machine: Have a machine that can drill a hole into the pcb, then have a tool head with a silicone pencil-eraser sized plug that can seal the bottom of the via hole in the pcd. You can then have a tool above the via hole inject higher temperature solder into the via hole. Then once the solder cools, you can have the bottom tool remove the silicone pencil eraser from pressing itself against the bottom of the pcb, and you're left with a small cylinder of solder connecting each sides of the board. You don't even need the bottom tool to line-up with the top tool very well, you just need it to be wide enough to make a flat bottom for the solder to be held up against. It'll probably have way too much resistance, and be at risk of melting from other things you solder onto the board, but it's technically a cheap and easy solution that'd probably take like 10-30 seconds per via.
This project in very interesting to me. I started a few years ago building a pulsed plating power supply, but got side-tracked and have yet to resume it. This project was in response to the fact that commercial PPP gear is very expensive. The advent of PCBWay and others like it reduced my motivation somewhat, but I still resist the idea of paying for 10 boards when I need only one. I hope you will continue with this research, so I can have a reason to resurrect my own project. Maybe I'll post my work on that and link it to yours so more people will climb onboard and maybe get enough crowd-sourced engineering to create a new solution to a long-time and widely vexing problem.
I tried solving the via problem with handmade PCBs. I take a pin from a berg strip, put it through the hole and solder it from both ends. If you're trying to make PCBs at home for testing or small projects, you can use this method easily. If you want to miniaturize your design, you go for commercial production I think you can also find a way to automate this and scale it down
Have you tried soldering through the hole... On the bottom side you have a flat neoprene butt with a cavity in it to line up with the bottom side of the hole and a hot 3D printing head (with a 0.2mm hole) on the top side and a syringe on the top of the 3D printing head to squeeze solder paste in through. The hot liquid solder would be pushed in through the hole to the other side where it would solder the under side with the top side in a quick precise squirt. If you used a Thermal Electric Cooler or TEC to cool the neoprene to a colder state, the liquid sold would harden quicker and should not drip when moving to the next hole. Because the TEC cold side and 3D Print Head hot side will only touch the PCB at the moment of injection, the board and printing head will not be cooled enough for the solder to harden on the print head nozzle. Would be an interesting video to see someone make that.
Before you toss in the towel just yet, consider looking into electroplating additives that would change the surface characteristics of the plating, making it less dendritic. Using a much faster fluid flow (from the anode to the cathode) may also prevent the dendrite growing. Was your plan to slowly feed copper wire into the anolyte? Or just to use a big anode and refill it manually on occasion? I’d also consider having a third tool on the same head, a warm air duct to dry the ink more quickly, if that becomes a bottleneck. In future, consider making the PCB move in the X-Y plane and keeping the toolheads stationary. As for solder mask, it really helps with SMD components, especially where traces may be beneath an IC. My method was to use a laser to burn off a solder mask layer. Also look into electroless silver plating, it’s much easier than electroless copper.
I am only up to the ad break, but I've always gotten great results from drilling all the vias, adding the same kind of conductive ink, a single electroplating bath, then cutting the traces. You're really missing out by not doing solder mask and silkscreen. Solder mask with silk screen fabric and a cheap sewing frame in white, cure, then another mask color in the intended mask color, laser cut the parts you want to be "silk screen" so it only removes the top layer, profit. I don't even bother buying from fab houses anymore and a cheap CNC with an added laser head next to the spindle is all you need.
Simplicity is key. No need to overcomplicate test procedures. Iterate and fail rapidly. Though I get it. Design is a fun process and those prototypes look rad. So many little troves of information where you’d least expect them to be. Also, if you’d like to accelerate the project community involvement would make this process much more feasible. Even if indirectly. There are many of us tinkerers and the whole desktop machine space is so awesome. Nonetheless, great video. Thanks for sharing :)
“Perfection is the enemy of good enough.” Don’t lose sight of the fact that this is for prototyping. Even something like automated wire insertion into the via, with wither manual or automatic soldering or electro-welding might be adequate. Regardless, great work!
Sorry I was MIA the last time you asked for help, life hit me with something and it took all my attention. Let me know if I can help you do more cool shit in the future.
I learned so much from your mistakes, like the common CoreXY belting. Incredible video, great work, and absolutely loving all the ideas. It's now at the top of my bucket list to make one. FR🙏🤩
About 40 years ago (mid 80's) I wanted to build a laser based printed cct board device, my ultimate idea was every electronic store would have one of these like libraries had photocopiers. The main process was a laser with a feedback loop such that it could cut through the copper but as soon as the reflection changed it would cease as it had penetrated the copper foil. This was purely a prototyping service so there was no attempt to do via's other than drilling the hole. However I knew it was beyond my capability and resources at the time so never went anywhere with it. The maker community has so many more resources available to them now :)
Reading the comment about using wire to make via. The main argument I read is that you need to solder them and it would melt if you heat them back when installing component. But what if you electroplate them. No soldering but creating a copper link to the wire and the copper plate. Once the wire is connected to the board you then remove the copper to make the traces. By CNC or other process.
You can burn the holes with a laser (may be after deleting the copper layer) so the internal surface would be intrinsically coated with carbon from burning the resin.
Some random ideas... 1. instead of plating, insert 0.25mm copper wire and solder to annular rings 2. insert 0.25mm copper wire and under pressure fill it with low temperature solder. you can use like 135C melting point solder, + some heat resistant silicone "nozzle" to form annular ring around via to fill it with solder... Really nice project, subscribed.
And maybe it's possible to use similar process like in point welding used for creating battery packs from ion cells, (not sure about doing that on copper with copper), but if it's possible would be nice to try. Drill hole, insert copper wire, point solder it to annular ring...
Exciting project! I must confess that the lead time of PCB procurement doesn't bug me _that_ much as a hobbyist because for me, the real problem is parts placement - my hands shake, and getting parts placed once the paste is stencilled is a real challenge. That's why I subscribed to Stephen Hawes's channel - the work he's doing on small-scale pick&place is intriguing. But I'm subscribing to you too, to see how this turns out! (PS: I'm not altogether sure I want the chemistry for etching and plating in the house. And if I do it in the garage, it becomes a seasonal affair; the climate here has COLD winters.)
If you find yourself placing challenging components by hand (like ICs with dozens of fine-pitch pins), you might consider putting together a "manual pick-and-place" apparatus. You can use a vacuum pickup (easy to make with a cheap modified aquarium pump and syringe), mounted to an aluminum extrusion arm, in turn on vertical slide (3D printer technology parts), sprung loaded to pull up, but can be lowered by hand. I like to control the vacuum with a small valve and foot pedal. You can then position the PCB by hand below this "crane", and slowly and carefully lower the IC nearly onto the board, maneuver the PCB a little bit to perfect position, then finish placing the part. Ie: no need for the apparatus to have any fancy X-Y mechanism, because moving the PCB around under the crane by hand works fine. You just need the crane to give you a way to lower the IC into place consistently and without shake, with an initial "rehearsal" to get the position just right. This kind of simple apparatus makes all the difference between shakily positioning the IC and messing up the solder paste, versus placing it just right with confidence. Of course there are fancier manual PnP machines too... search for them to see.
Hey friend, I’m a fan of electroplating and have been studying it for about 6 years. What you want to avoid that creeping copper growth is to do pulsed AC plating, and if possible, to force movement of the fluid through the WIP via. Here’s why:
0) By pulsed AC plating, I mean that instead of a constant current, you have a periodic cycle of “plate, plate, erode, rest”, for example. You have, say, 1A for 0.5s, then -1A for 0.25s, then 0A for 0.25s, and this repeats on and on.
1) Pulsed AC plating will, on average, deposit material (deposition is proportional to (average current) x time). In the eroding step, however, it will selectively erode the pointiest surfaces, because that’s where the current prefers to flow through. What this means for your VIAs is that you’ll erode the spurious dendritic growth that creeps towards the electrodes.
1.1) Pulsed AC plating will also help make a smoother, more consistent surface.
2) Forced movement will help you a lot here, because you need copper ions to successfully reach the surface and get plated. In normal circumstances, ion movement is mostly limited by diffusion. When ions get plated, new ions fill replace them by coming from some random direction in the fluid. In the case of the thin 0.3mm VIAs however, diffusion will have a hard time replenishing the plated ions inside the bore, because there is almost no direction where new ions can come from. In your setup, as it stands, that space gets starved of copper ions and stops conducting altogether.
Get a good flow of electrolyte solution through the bore, and you’ll see how much better the process behaves.
There’s a book on this from the German book shop Leuze Verlag, the book’s called Pulse Plating, it’s silver-colored and has a picture on the cover of electroplated vias that were successfully done using this precise approach.
Anyway. Cheers, friend. Your project looks epic. Please reach out if you want some help, I’ll gladly give you whatever info I can.
Omg this need a like or pin from Levi 😆
Thank you for that excellent explanation. I am your second sub. You should make content! Seriously, just feed the above into an AI and have it make a conversation and a video, if you are shy and don't want to use your face or voice. You have excellent information! It would be awesome to see you work together.
High end input!
ok this is a dope explanation! ty
can you explain why electroplating is necessary to make the via? why not just press a tiny rivet through the hole?
Nice of PCBWay to support their competition!!
Ballsy move LOL
I respect it.
It's a good PR move and basically has no financial downsides. It's an alternative, not really a competitor. PCBWay is convenient, easy, and cheap unless you make a lot of boards, most hobbyists will pick them instead just for the convenience factor. Besides that, I can't imagine that the hobby electronics market makes up more than a fraction of their revenue. A nice fraction probably, but not a significant one. It's win-win for them.
This is an ideal video for them to sponsor! Guaranteed that almost 99.9% of viewers are interested in producing PCBs, and also interested in fast turnaround, which is something PCBWay is good at. At the same time, "how to DIY your PCBs" videos thus far have universally shown what a laborious, difficult process it is leading to mediocre results. Against which, online PCB vendors look like a better option. Of course, the best of luck to Levi advancing the state of DIY art!
PCBWay is a manufacturer in China. This is how China collects your know-how.
If your prototype works you still need production boards in modest quantities. That's what they are good at. I order 50 to 200 boards at a time and they turn it around in 3 days.
I absolutely loved seeing your negative results. I honestly learned more from this "failure". Great work !
don't worry you can learn soo much from failure - heimerdinger
Fail faster and fail frequently
Indeed ❤
Agreed, and the way it was presented was good as well.
Oh definitely, it's one of those "journey, not destination" things that make these sort of videos so interesting. Presenting only the success would be a shame to hide the effort and thought process that went into it.
Me listening to this as a mechanical engineer: "Why does he know that ?" "How does he know about that?" "Wait where is his encoder for his position accuracy?' "Why does he know anything about pumps ?". This was awesome.
Me listening as a newly degreed electrical engineer: “Cool.” “Interesting.” “Cool.” “I need one.” “Cool.”
Me listening to this as a developer, why doesn't he just open the console and output the variables as it's running to find the problem.
@@MNSweetDebugging with a proper tool would be much better though? Using the console to print variables is a terrible way of debugging a project. I know you have no option when you are stuck on arduino and stuff like that, but please, if you can, use a hardware debugger. Its much, much better.
@felipevellasco6526 I was just continuing the trend and didn't want over complicate it with too much detail or jargon. Hence why I didn't mention what type of dev. As much as I want to learn about IOT I'm more web application and haven't touched an ardinuo though I did just pick up a ESP32 starter kit and I'm looking forward to making some lights go blink on a bread board.
Mechanical - Aerospace - Kinematic - E&TC - Robotics
MAKER
Yep it checks out
Btw. The design you came up with is called H-Bot. It's pretty common in industries. But you have to use proper guide length to force distance proportions
The issue with H-Bots is that they add a racking force to the gantry. With solid enough rails this will be less of a problem but it's the tradeoff thats being made and there's a reason printers go for core-xy even though the belt path is a little more annoying.
I second scruffy3121. H-bot have been long known to be the nemesis of cheap linear rails. Plus it requires precise sync of your motors braking timings if you don't want to end up with a lot of skipped steps with open-loop stepper drivers.
Hey Levi, I'm a process design engineer that deals specifically with electroplating metal into nanopores. Is the solution you are using a commercial solution with additives? You need to look into pulse plating, this will dramatically reduce the surface overplating you're experiencing.
I’m assuming with pulse plating you give time for the electrolytes to reach equilibrium in the liquid resulting in a more even coating that isn’t dictated by a variance in surface charge that probably pulls ions to certain regions
Why not use copper wire instead of electroplating? Seems more realistic and achievable. Automatically soldering small bits of precut wires maybe?
@@jonathana.1802I don’t recall why, but the industry settled on plating a long time ago and have done it that way since. I’m not exactly sure why though. It’s a good question.
@@jonathana.1802soldered vias have a few problems. Firstly, they stick up out of the board, getting in the way of hotplate reflow. Secondly, any hot-air, infrared, or hotplate reflow is going to melt the vias too unless you want to mess about with different alloy melting points. Thirdly, automatically soflering from two sides would be a pain, especially with no solder mask to keep the flow in check. Etching a soldered board has its problems too.
You’d be better off trying to make a machine to rivet copper wires into holes that have been counter-punched, in order for the riveted heads to be flush. I did that by hand for a while.
What if we apply a flow of fluid during the process? I’m wondering if it would affect the deposition speed…
What do you think about that?
You can fix the electroplating problem. This is actually an issue which had to be solved in semiconductor etched trench filling. The diffusion blocking layers needed to coat the bottom and walls evenly and different coating processes are more directional than others.
Extend an insulation covered needle with only the tip uncoated into the center of the via, through the center of the tube. This will decrease the distance to the surface and the current path through the electrolyte solution. The proximity to the plating surface will also improve the directionality and placement accuracy. With the current flowing from the needle horizontally to the walls it is a straight path giving more even coverage.
You could feed an insulated solid core wire from your reservoir through the tube to avoid any risk of leaks. 3d print an X shaped electrode holder for the center of your tip and use it to center the electrode in the XYZ direction so when the tip seals against the PCB the tip of the wire is centered in the via.
I think for the final version you will still want to change the design to a multi-head or headless electroplating bath because even 90 sec per via could take days for a design like a raspberry pi.
What if we apply a flow of fluid during the process? I’m wondering if it would affect the deposition speed…
@@tomaszradomski8994 Yes. I think he is pulsing the flow if I understood him correctly, which should be good to prevent depleting the copper ions in solution.
This is genius. If only I had known this before shipping faulty product to north america
Yes, the semiconductor industry had the same problem. I think I saw at least one vid on YT.
I just can't express how much I appreciate your 1. trying new things 2. reporting the failure. Failure is critical, leads to all advances, and sharing those failures destigmatizes this (critical) process. You've done a huge service to the community even if you never get anything working. And you HAVE gotten so much working. I'm a huge fan.
AWESOME vid, thank you so much for sharing! really excited to see you tackling this!
As soon as I heard “vias” I knew you’ll be in the comment section
I am also working on something similar hoping to make it work and share it on the internet, Your work and videos on the pnp and laser based PCB manufacturing are super helpful. Thanks to both of you for inspiring and making the knowledge open source and available to all.
You two need to collab
For some reason I expected to find you here 😅.
A collab would be awesome 😬.
Your 2 projects are like jam and peanut butter 🤭.
Thank you both for sharing your knowledge with us ❤.
That was cool! Chemist and machinist here, hear me out:
What if, instead of chemically depositing copper, you drilled the double layer board, slid in a piece of copper capillary tubing that is slightly longer than the board is thick, and riveted/pressed it in from both sides with custom dies that have centered guide pins + a concave face? Locking it in place and making contact on both sides, like a hollow rivet, you could even solder them on each side I bet you could pull it off
Came here to say this too. Even copper wire mechanically inserted and his existing process could perform easier surface plating of the copper instead of a solder or riveting for a chemical instead of mechanical joining. Tons of opportunities "good enough" for a prototype PCB.
i use it for my home pcb production. The problem with it is that for the top layer (from which the drilling starts) to make the contact with via, rivet must be pressed hard enough, to spread to copper. That creates bump, and most smd components could not be properly placed above it. And applying soldering mask is harder while vias are so much (up to .15mm) above copper. It became the limitation for my designs. My method is not autometed BTW.
It might be OK solution for hollow rivets for thru holes tho/
Commented the same thing, this just makes sense
Very similar to the "old school" tiny rivet method.
@@ВоваЕршов-й2ьwhat if you drill a slot in the PCB where the rivet would sit against the PCB a few millimeters/micrometers beneath the top surface, like a valley?
Great stuff! Would love to have a desktop PCB printer like that. My method of making vias was always:
1. Drill a small hole
2. Stick a thin wire in the hole
3. Solder on both ends to the copper
Caveman method, but works pretty well for prototyping :)
even a 2.54mm header pin would be easy to handle with tooling.
"CoreXY vs H-bot".
What is an h bot
@@E-dartIt’s the “Single Belt Core XY” kinematic system demonstrated in this video. It’s just that Levi didn’t know that is already existed and had a name - which is no great surprise. They were a bit of a flash in the pan about 7 years ago or so. They were a cool idea, but as demonstrated in this video, require a substantial frame to actually be reliably accurate.
Came to comment this
Great work, and thanks for sharing. This is the promise of the internet -- each of us learning from others, asynchronously. Looking forward to seeing this project progress.
Unless you are aiming to do multi-layered boards, PCB rivets is the answer. They are old school and extremely robust. If the end goal is multilayered boards, then god speed good sir. I would love to be able to do 4 layer boards at home.
I think you are talking about eyelets. Many years ago, one of the pieces of military equipment I worked on used eyelets rather than plated thru holes. I don't think they trusted PTH at the time. The eyelets gave us a lot of trouble with corrosion around the eyelet ruining the connection. Reflowing the solder around the eyelet would fix the problem, at least for the mean time. This was not a cheap piece of equipment, it was used to synchronize cryptography equipment.
The major problem with PCB rivets are that they aren't actually made to do that- or practical for complex circuits. They might work for this use case, but still have a lot of drawbacks. Instead they are made to mount a heatsink or other non-solderable component via the connection instead. Many circuits that use high amperage op-amps or rectifier blocks use this method for cooling to a passive aluminum heatsink. Toasters/toaster ovens use them to connect the nichrome or steel heating elements to the board, where that much heat would melt the solder of a regular connection.
Most of them are thicker where they would be mounted requiring milling a small channel into the board around the via (to account for it's chamfer) for it to sit flush with the board itself. There are also only 3 ways to to connect one half of the rivet to another, the button method, where one of the hole openings has the mount point, but is larger & a few mm thicker than the other, the rack method, where small square channels are cut into each side of the center of the rivet allowing them to slide into each other like the guide for a drawer in a dresser. This allows for the thinnest via's but has the worst resistance & tend to separate under many heat/ stress/ weight cycles, sometimes breaking the connection completely.
Lastly is the flange method, which causes one side of the tube to expand outward in the center of the hole. Not only does most of these methods usually require oversizing & extra milling time for the hole & the lap space between them, but the smallest rivets are 0.6mm in size, compared to modern whole board CPM which can have via's as small as 0.15mm in size.
As his goal is making prototyping PCBs, I think they might just do the job
@@elvendragonhammer5433for circuits with any kind of complexity, 2 layers don't cut it in any case, and you can have a professionally manufactured prototype shipped to your door in under a week.
I'm a total noob at this, but couldn't you sandwich boards together to make multilayer PCBs? It obviously won't be a nice as something from a manufacturer, but it might work.
I work at a recycling center as a bus driver for the employees. They recycle plastic, paper / cardboard Etc.
They shredded around 30 - 40+ big boxes Packed full of repairman schematics on old televisions. This was either a shop or manufacturer that was from Cincinnati Ohio at one time. Just about every old CRT, vintage television set you could think of was in there. Some televisions were pre-1950.
It broke my heart as a Electronics hobbyist seeing all of that engineering hard work & information destroyed!
In my mind, this process makes sense:
(This process begins after all via holes are drilled)
1. Insert a strand of wire into the hole(alignment hell + you need multiple wire gauges)
2. Align or trim the end section of wire flush to the board. (Difficult to prevent marking or damaging the surfaces)
3. Deposit a thin film of flux and potentially high temp solder. (Brazing metal?)
4. Spot weld the bottom face. (this is probably the biggest stretch. The difficulty in calibrating the pulse energy and duration is beyond me)
5. Cleanup/maybe surface the weld
6. Repeat 2-5 for top surface
As with most engineering ideas, this is much easier said than done. I like this process more because all of these steps have high time optimization potential. If you read this, please let me know if you have any thoughts or critiques on this process. Regardless, this video is badass.
Why not wire + chemistry? Anyhow, wire vias in DIY circuits are common.
@jammi__ I just don't think plating is optimal for this project's goals. Chemistry clearly scales well. This is why it is so dominant in industry, but if you do 1 via at a time total fabrication time will suffer. Also, having to worry about electro plating consumables seems way more complicated on its face than just buying wire spools.
I must say, though, I am not an engineer. That is why I wanted input from others with real experience to tell me where I am wrong/nieve if possible so I can learn.
well i think it would be viable to use riveting. So bottom head would have anvil and top one would have anvil and wire feeder with cutter and fter that guide tube. Steps to do via after drilling:
1 position bottom head under hole
2 position guide tube over the hole
3 extend bit of wire (thickness of pcb + something for rivet heads)
4 use wire feed to push cut piece down the tube
5 lift top head (feeding bit more wire while lifting should prevent tiny piece sticking)
6 move top anvil over hole
7 lower bottom anvil bit to allow wire to poke trough (optionl step can be integrated with inserting wire)
8 squish from both sides
Possible problems: cutting 0,25mm copper wire without it being stuck somewhere, possibly not tight enough squeeze, but but if it could apply solderpaste then it should be possible to smear little bit over vias and they would self solder during reflowing
@michatrzcinski8741 This seems more doable but would lead to a lower quality via. First, even if you squish the head super flat, there will still be a raised section that could interfere with smd parts that go over top of the via. Secondly, for pretty much any mechanical method other than maybe stir welding, you will end up with a microscopic gap. This would have higher resistance, could cause intermittent contact issues, could decrease signal quality, and would have lower mechanical strength.
I will say, though, that your idea seems much easier to implement.
@@michatrzcinski8741 just edm the wire cut clean, would take, less than 2 seconds, squirt some water
I appreciate you uploading this despite having run into a roadblock. You're attempting something that nobody has done before. I'll stay tuned because this is something I myself have always wanted as well.
I started doing research for this over 25 years ago. I too needed to rapid prototype boards done in-house. This was way before 3D printing. Additive manufacturing wasn't even known then. I tried using a color desk jet with conductive ink. This allowed different types material; conductive, non-conductive, etc. I also tried color laser printer using metallic powder. Both designs used a Mylar sheet that fit standard printers. For multi-layer type boards, the printer laid down a non-conductive trace.Finished sheets could be mounted on a rigid board, or left flexible. Eventually I decided to wait until technology caught up.
First I couldn't afford to , then I didn't know how to . Finally, after I had overcome those obstacles, pcb manufacturers came out with low-priced fast turnaround high-quality pcb's to order .
Now , once I finish my gerbers, I have to avoid the accusing stares of my accumulated pcb equipment and chemicals as I send my order.
Incredibly cool project, and looking forward to seeing it progress. One suggestion I would make is instead of going for a CoreXY or H-Bot architecture, it might be easier to go for (in 3d printing terms) a "Bed slinger" where you would move the PCB instead, meaning your via and trace apparatus (the heavy bits) can be as secure as necessary and don't need to move, and the PCB (which barely even gains mass like the issue is for 3d prints) can whip around as fast and secure as you like.
For future developments, if multilayer boards are the goal, it might be possible to make sets of 2 layer boards and then stack them with an extra layer in between, but making everything line up perfectly might get tricky.
17:07 "A few of them could actually be via-ble"... Good video *and* Dad jokes!
Not to mention the couple of times he "couldn't come up with a *solution*".
For the vias you could try doing like other diy pcb solutions have done: tiny copper rivets. Make the gantries rigid enough and the Z axis strong enough to "crimp" the rivet in place and it should be pretty reliable.
could even control the subsequent soldermask steps to be as thick as the rivets' heads so the final pcb is flat
One potential idea I had for simplification of the mechanics is a secondary outer o ring which provides a second sealed area that you could connect to a cheap 12v vaccum pump to suction down onto the board for a solid seal with a less rigid gantry required. You could also probably build magnets into the heads for registration and use an approach where some loose strings pull the heads roughly into position, the magnets cause them to snap into alignment, and then the suction engages providing a solid seal to the board
When the RUclips algorithm finally works as intended 😅
Welcome back! Great video and a very interesting approach to via generation - thanks for sharing.
You can post short videos to keep us up to date, you never know from where innovation could spark from, thanks for sharing your progress!
Dude, great project, you got my sub!
Here are some suggestions:
1: Be more specific about your electroplating solution. The PCB industry use PEG-JGB-SPS additives, they have a huge impact on quickly plating small areas with high accuracy. SPS (Bis-(3-sulfopropyl) disulfide) is probably the most important since it's job in the additives is accelerator, it increases Cu deposition in low current density regions, that's the problem you described. Also decreasing the PH aids in conductivity so consider adding H2SO4 (sulfuric acid).
2: Use AC Pulsed plating. The on/off cycle allows for faster and higher quality plating without creating dendrites.
3: Ensure a stable temperature of about 50c-60c, this is the optimum temperature for ion exchange in a typical CuSO4 plating solution.
4: Agitation! An ultrasonic inducer (20-80 kHz) should be placed in the tool head, they're very small and would fit your compact design preference (mine too!). Electrolysis creates a lot of hydrogen bubbles which interrupt ion flow, especially in tight spaces, ultrasonic agitation helps burst the bubbles quickly. The agitation also breaks up surface tension which slows ion movement, and if the ultrasonic inducer were pointed at the via hole it would actively drive fresh Cu ions where you want them.
Hope that's helpful. If you made a KickStarter I'd love to back it as I'm sure would many others!
This is the first comment to mention the ultrasonic suggestion, hope he sees it!
Please finish this, it would be a blessing to the world. I might even be interested in exploring the same problems once I get my engineering degree.
Why wait until you have your degree?this could be a capstone for you that could be handled before you even need to do it
^ What they said. Why wait?
3:35 if you're gonna have multiple tools, just use a rivet gun. Rivets conduct, just drill you hole, put a tiny rivet through, and pop it to bridge the traces
That would work for larger holes, but like he said towards the end of the video, most vias are 0.3 mm. I could be wrong, but I don't think they make rivets that small
PLEASE KEEP GOING WITH THIS. My hobbyist ideas die with lead times and shipping costs. The world needs what you're building.
Welcome back, missed your uploads. Great work as always
This is great stuff. You've got the right idea on the requirements side. I never have a need to make a single sided board and nasty chemicals are out of the question. So many similar projects fail to clear those bars. Definitely going to follow this closely.
Would it be feasible to just do the ink deposition per via, and then switch to another "static" setup with custom-made (hurray to 3d printers!) liquid channels for all the vias? Or even just sealing against the board edges. So electroplating would be parallel for all holes, but still with a small amount of liquid and well sealed and desktop friendly? After all, we are not talking panels but single, small PCBs
To my understanding, your idea (and success) of applying conductive ink as the base for the electroplating is already amazing!
This is very exciting. Very much looking forward to the next video.
It’s always a good day when a new Levi video drops into the world. Thanks for sharing your projects, and your knowledge!!
Very excited to see where this is going! Great work and thanks for showing your process!
Awesome, great overview of the entire process and end results. I'm really looking forward to see this come along and eventually be developed in some form as I want to tackle the same problem myself too. Thanks
You'll need a method of refreshing the electrolytes. I would use a tube that feeds copper wire into your bath so the copper wire becomes the sacrificial electrode. You'll also want to regularly measure the conductivity of the plating bath between runs and have the ability to add solution to make up for lost electrolyte/liquid.
Might look into faster drying plating solutions, one option is to use resistive heating to "bake" the resistive element, between the deposition and resistive testing phase.
would let you deposit->test resistance->bake->test resistance but would require a current regulated power supply.
Very humble to introduce yourself as a maker. To me you are an engineer. Thanks for sharing your project. The way you looped the XY axis was really smart.
Great video - looking forward to seeing updates!
Awesome. Really excited to see your next idea on via generation.
Great video! Slight correction. As a former microfluidics researcher, you’re not using this term correctly. Microfluidics is not the term for small scale fluidics. Microfluidics is specifically about manipulating liquids in laminar flow state using microdevices. Usually the “tubes” (channels) in these devices have diameters below 0.5 mm or 500 microns (hence the term “micro”).
4 minutes in, you've put this together really well. I hope you pull it off, loving the video so far!
I'd love to see this work!
A few suggestions come to mind:
* what pressure/feedrate are you operating at? Faster will almost certainly be better; as cybyrd alludes, ions are pulled out of the solution by the process and this will interfere with how even the coating is
* use a copper electrode centered in the hole, with electrolyte forced past it at pressure; if electrode erosion is sufficiently fast, move it (a la wire EDM or microhole EDM)
** having something stick through the hole will also make alignment easier; the bottom can center itself by clamping onto the electrode
* heating the electrolyte will improve reaction speed (but might result in other trade-offs)
* try pumping electrolyte _towards_ the bath that has the electrode in it (or alternate directions)
* only ground the copper on the side away from the electrode
If the goal was to have a quick prototype, this fulfills that: prep a board full of vias in a 6.35x6.35 mm grid pattern (all at once, complexity isn't an issue). Design the pcb to hit or avoid (at least on one side) the pre-exising vias. Some can make profit on the prepped via boards (e.g. pcbway or you) - and the desktop machine is much much simpler (a laser and/or engraver).
The goal is a quick prototype that is very similar in electrical properties and layout to a production board, so that's not exactly what I'm going for. But that's a very intriguing idea, not one that I've ever thought about. You could have different pre-prepared boards with different via pitches to accommodate different power/trace density needs. Have you seen this approach used?
As others have said in the comments, your failures and processes are very educational. Perhaps more to me than to most of the engineers on here as I'm a beginner. Thank you!
FYI 4th Gen MakerBots used the H belt design for their xy motion, it was heavily criticized for the reasons you identified.
I like this interesting approach to diy vias. When I was in college we did diy vias with a bit of wire soldered on both ends. It couldn't be as small as this but, it got us off the ground. I think this has a lot of potential, but I wouldn't underestimate the complexity of the rest of the process.
I'd love to see this idea progress forward, and the only big thing I'd warn against is the "never ending personal project".I think you've demonstrated a viable approach to this step in process. Some people have even commented potential improvements. Really, 3 minute vias is a pretty arbitrary requirement. Even with 10-20 minute vias, you could still make pretty complex and useful boards much faster than you could order them. I actually always thought it was an interesting challenge/puzzle to minimize vias in a design.
Honestly, I think you'd be best served by pushing forward to a minimal viable product. Then, you can come back and improve the sub-processes later. That'll also give you better insight into the actual limiting constraints of the overall process, which might not be what you expect. Either way, I'm definitely interested to see where this goes.
I like how you found out that the mechanism isn't used for a good reason the hard way and took it as a lesson. Great project!
This is a very important problem so I'm really glad you're tackling it! I didn't quite understand what part of the electrochemistry changed between your initial test that got you good results and the high resistance issue at the end.
The chemistry didn't change, but the solution was more "constrained" in the latter prototypes. At the end, the board and electrodes were separated by a thin tube, increasing the resistance. Once I increased the applied voltage, the electrochemistry was exactly the same.
@@LeviJanssen Ahhh, I see thanks for the clarification. So do you think the initial prototype also wouldn't work with small vias?
@@LeviJanssen I'm repeating myself from a comment a few minutes ago, but I really want to see what happens with a high flow rate :)
As a maker that enjoys electronics, your concept is extreemely good. but nethertheless a hudge issue came to mind : the electroplating works efficiently because enough current can flow from the oxydizing anode (ie the copper plate in the bath) towards the reducing anode (ie the via being made). To maximize current, you will need to decrease as much as possible the resistance of the electrolyte path, by several factors :
1) put the oxydizing electrode after the pump (the peristaltic pump will make an open circuit with the pinched silicone tube). you can make just a little chamber for the electrode that would stay filled with electrolyte filled on one port by the pump, and the electrolyte would shoot out towards the nozzle.
2) You need to decrease as much as possible the length of tubing beetween the electrodes as the elctrolyte path resistance is proprotional to it
3) You need to keep a wide path to decrease even more resistance (R = c*S/L with C the conductance proportionnal to your electrolyte's concentration, L the pipe's length and S it's surface). I know you will need a thinner end to avoid air comming back up and keep the path filled tanks to surface tension, It will be a matter of compromise.
and maybe plating a pcb with a photo-resist layer would avoid getting copper build up on the outer surfaces of the pcb and only in the hole to help it stick ? it's just an idea but there seems to be an issue similar to dendrite groth in lithium batteries. avoid letting copper grow where it shouldn't may help
Sounds like it would be easier to move the board between stationary toolheads rather than moving two toolheads around a stationary board. And why plate holes one at a time? It does make sense to ink them one at a time since it's quick and needs to stay off the surface of the board, but make the plating toolhead bigger. Maybe 25x25mm and move it around in a grid pattern, or make it require 100mm wide board material and have a toolhead 100x10mm or something move across continuously depositing copper.
You could also just drill a hole and solder a wire through it for the vias.
That would work for a low number of vias. Think of a BGA package with a few hundred microvias between the pads... not so much.
Yep. Solder it in place and smooth the surfaces. Not sure why he is doing this so complicated way. This machine will cost a fortune, so must be targetting big business, not a diy solution.
@@sniperasys That would be a bit of a problem for an electroplating process that takes 90 seconds to 3 min per via too.
@@murraymadness4674 I'd reckon that most people who tried to do the same would say that doing this in any kind of quantity is quite the chore. I did. It is. Picture yourself hunching over the board for hours soldering wires that are not significantly thicker than a strand of hair. The real difficulty lies in getting the solder connection reliable 100% of the time especially after you sanded back the protruding wire ends, so when you try to solder a BGA package on top, some of your joints come loose because there's just a few dozen square microns of solder joint areabetween the trace and the wire.
@@ceoyoyo I'd gladly spend a week building a CNC machine that plates a perfect via in 90s. That's 40 vias per hour, 320 vias a night. That's like 3D printing speed, in perspective. It's okay considering that I don't have to do something that really hurts my upper back and what a machine is supposed to do. :D
Love your content SO MUCH! & HAPPY NEW YEAR, Levi and everyone here 🎆
Super cool project. One of the issues I have with making PCBs at home using an engraving mill, is the lack of mature and well featured, GUI based CAM software to convert a gerber file into gcode. All currently available software applications suffer from one or more huge flaws, that ultimately make them either unreliable, or time consuming and fiddly to use. I think for your project to be successful and truly achieve the "fire and forget" functionality we all want, it needs good software too. On my own setup, I've modified and optimized the hardware and was able to cut down the setup time by an order of magnitude compared to its off the shelf version, and now configuring and setting up the CAM software, with all its bugs and kinks, is by far the most time consuming and frustrating part of the process.
It's just amazing to see the RUclips engineering community in the comments coming together to make this project work.
Seriously, you guys are awesome 😎
Excellent work, description, self-critique, dedication, keep going !!!
This is fascinating!
I thought you might like to hear my journey into a pcb mini mill. I was working in R&D in the mid 1990s and transitioning from through hole to surface mount so I asked the company to buya T-Tech pcb mill. Was told it was too expensive so I built my own from surplus motion control mechanisms, steppers, limit switches etc. I controlled the steppers by sending binary sequences over the pc's parallel port. I wrote a gui in Visual Basic and was able to move all 3 axis. I needed a way to automate pcb milling so I wrote code to convert the gerber files generated by my then pcb layout program, Winboard, into movements that would isolate the pads and traces instead of just milling lines as the photoplotters do. This was tricky but I managed to make it work.
Later on I got some real stepper controllers made by Phidgets and got more steady control. I ended up milling hundreds of single and double sided pcbs. Since then I've retired but have the machine and still use it.
As earlier posted: copper rivet.
1) Wire through needle to push through via
2) solder paste squirt on back, and solder tip heat
3) solder paste top, solder heat.
Yes whiskers will be there a bit, but hand cut possible. FDM shows you can get 80W to a head, so 6 mil is easy.
Can even do it by hand w/ cheaper model.
The guys over at Bad Obsession motorsport did a video last month where they went over the whole process to make their own circuit boards by printing the metal tracks.
Maybe more cone shaped tip instead of a tube to concentrate the resistance to mostly to the via. Also what would happen if the electro plating current would be ac. Or pulsing the plating fluid back and forth to keep fresh fluid in the via.
Oooh! This is intriguing! The ability to fabricate two-layer circuit boards in-house could be game-changing. I like that you are keeping the requirements relatively modest to start with to ensure that the goals are attainable. I often make the mistake of making my goals too lofty.
The device being able to self-calibrate and self-correct would be an excellent "automatic" capability.
I admire your efforts to do this at home. As an armchair PCB designer, an alternative comes to mind to have your servo and guide run a very thin wire through the hole and bend at the top to lock it in place. You could then hand-solder or add a soldering tip to your inserter. I look forward to your next steps!
This is cool. Can’t wait to see the end result!
Why not go with the THT component approach for creating vias?
A very thin copper wire could be inserted into the hole using something similar to a 3d printers extruder, then get secured to the bottom using solder paste and a laser or hot air, then cut the wire at the top and solder it in place there too. Maybe it would be easier to flip the board in the machine instead of putting a solder paste application and heating tool on both sides of the board.
Very cool project!
All homemade vias options were explored long time ago. For starts, you have to use good carbide drills at prescribed 30-40k RPM to get VERY smooth holes.
Carbon paint is kinda easy at the first glance but leaves VERY rough and uneven plating and very hard to achieve sufficient conductivity in the first place.
Silver nitride is better, but too complex chemically and still leave pretty rough and eneven plating.
They only real, sustainable primer for homemade vias is palladium chloride - the very thing everyone use in the industry. It's easiest and best in literally every aspect. It seems expensive, but it's really not, 1g will be enough to electroplate approximately your whole house.
The talent you bring to the table is unmatched, i think you should work for nasa or something. I understand the specs you want to be perfect and that is toraly u understandable. I think you made success as a 1.0 version and many people could use it, i also think with enough thought you are going to solve this riddle to get it exactly where you want it. Theres almost always a way with physics and i feel you are right on the cusp of finding your answer. Sometimes the brain works so hard it squeezes but needs yo relax and the answer will come faster than lightning when enough bloodflow is alowed to be rushed in during relaxation periods like durring sleep is sometimes when i come up with the answers. I remember you talking about these concepts in the discord a long time ago and its super fun to see all the progress you have made and stuck with it. Your dream will trully help many folks, please never give up on it. Thank you for all the time and efforts you have committed. If more people were this comitted its possible most of the worlds problems would have been solved by now
Seriously in love with this. I would love a video on your chemistry too.
Wow this is a awesome idea, and a few years ago I had a similar project but never finished it, so this kinda makes me happy
Wow what an impressive project ! To make vias on single face prototype board (made with laser or cnc mill) the tricks I found is to drill the via the size of copper solid core wire, then insert a small piece of it and solder it on both sides. It's a quick hack but it works a charm when making pcb prototype at home without equipment
Very nice project! Hope to see more soon! 🎉
Here's my dumb idea for a machine:
Have a machine that can drill a hole into the pcb, then have a tool head with a silicone pencil-eraser sized plug that can seal the bottom of the via hole in the pcd. You can then have a tool above the via hole inject higher temperature solder into the via hole. Then once the solder cools, you can have the bottom tool remove the silicone pencil eraser from pressing itself against the bottom of the pcb, and you're left with a small cylinder of solder connecting each sides of the board. You don't even need the bottom tool to line-up with the top tool very well, you just need it to be wide enough to make a flat bottom for the solder to be held up against. It'll probably have way too much resistance, and be at risk of melting from other things you solder onto the board, but it's technically a cheap and easy solution that'd probably take like 10-30 seconds per via.
I'm excited to see where this project goes!
Wow! Incredible video, thanks for sharing!
This project in very interesting to me. I started a few years ago building a pulsed plating power supply, but got side-tracked and have yet to resume it. This project was in response to the fact that commercial PPP gear is very expensive. The advent of PCBWay and others like it reduced my motivation somewhat, but I still resist the idea of paying for 10 boards when I need only one. I hope you will continue with this research, so I can have a reason to resurrect my own project. Maybe I'll post my work on that and link it to yours so more people will climb onboard and maybe get enough crowd-sourced engineering to create a new solution to a long-time and widely vexing problem.
Awesome man!! I subscribed right away! Very interesting
I tried solving the via problem with handmade PCBs. I take a pin from a berg strip, put it through the hole and solder it from both ends.
If you're trying to make PCBs at home for testing or small projects, you can use this method easily. If you want to miniaturize your design, you go for commercial production
I think you can also find a way to automate this and scale it down
Have you tried soldering through the hole... On the bottom side you have a flat neoprene butt with a cavity in it to line up with the bottom side of the hole and a hot 3D printing head (with a 0.2mm hole) on the top side and a syringe on the top of the 3D printing head to squeeze solder paste in through. The hot liquid solder would be pushed in through the hole to the other side where it would solder the under side with the top side in a quick precise squirt. If you used a Thermal Electric Cooler or TEC to cool the neoprene to a colder state, the liquid sold would harden quicker and should not drip when moving to the next hole. Because the TEC cold side and 3D Print Head hot side will only touch the PCB at the moment of injection, the board and printing head will not be cooled enough for the solder to harden on the print head nozzle. Would be an interesting video to see someone make that.
Fellow hobbiest maker. This is awesome. Thanks for posting.
If not anything else I at least learned about you, your channel and your projects. Subscribed
Before you toss in the towel just yet, consider looking into electroplating additives that would change the surface characteristics of the plating, making it less dendritic. Using a much faster fluid flow (from the anode to the cathode) may also prevent the dendrite growing.
Was your plan to slowly feed copper wire into the anolyte? Or just to use a big anode and refill it manually on occasion? I’d also consider having a third tool on the same head, a warm air duct to dry the ink more quickly, if that becomes a bottleneck.
In future, consider making the PCB move in the X-Y plane and keeping the toolheads stationary.
As for solder mask, it really helps with SMD components, especially where traces may be beneath an IC. My method was to use a laser to burn off a solder mask layer.
Also look into electroless silver plating, it’s much easier than electroless copper.
I am only up to the ad break, but I've always gotten great results from drilling all the vias, adding the same kind of conductive ink, a single electroplating bath, then cutting the traces. You're really missing out by not doing solder mask and silkscreen. Solder mask with silk screen fabric and a cheap sewing frame in white, cure, then another mask color in the intended mask color, laser cut the parts you want to be "silk screen" so it only removes the top layer, profit. I don't even bother buying from fab houses anymore and a cheap CNC with an added laser head next to the spindle is all you need.
Simplicity is key. No need to overcomplicate test procedures. Iterate and fail rapidly. Though I get it. Design is a fun process and those prototypes look rad. So many little troves of information where you’d least expect them to be. Also, if you’d like to accelerate the project community involvement would make this process much more feasible. Even if indirectly. There are many of us tinkerers and the whole desktop machine space is so awesome. Nonetheless, great video. Thanks for sharing :)
“Perfection is the enemy of good enough.” Don’t lose sight of the fact that this is for prototyping. Even something like automated wire insertion into the via, with wither manual or automatic soldering or electro-welding might be adequate. Regardless, great work!
This is such an amazingly good and entertaining video, thanks bro
The irony that this is a video of home-made circuit board maker, yet, sponsored by PCBWAY
Sorry I was MIA the last time you asked for help, life hit me with something and it took all my attention. Let me know if I can help you do more cool shit in the future.
I learned so much from your mistakes, like the common CoreXY belting. Incredible video, great work, and absolutely loving all the ideas. It's now at the top of my bucket list to make one. FR🙏🤩
How about inserting a conductive pin that gets punched flat on both sides like a rivet?
Everything about this was ridiculously awesome. New subscriber ✌🏻
Dude check Robert Murray Smith, he deposited copper on a pcb with laser, the same laser could dig vias and coated them
Awesome work! This is just what I come to RUclips for 😊
thanks a lot showing what designs don't work and exactly why. Despite a seeming dead end, it is very educational
About 40 years ago (mid 80's) I wanted to build a laser based printed cct board device, my ultimate idea was every electronic store would have one of these like libraries had photocopiers.
The main process was a laser with a feedback loop such that it could cut through the copper but as soon as the reflection changed it would cease as it had penetrated the copper foil.
This was purely a prototyping service so there was no attempt to do via's other than drilling the hole.
However I knew it was beyond my capability and resources at the time so never went anywhere with it. The maker community has so many more resources available to them now :)
Rivets?
Incredible work 🎉
Reading the comment about using wire to make via. The main argument I read is that you need to solder them and it would melt if you heat them back when installing component. But what if you electroplate them. No soldering but creating a copper link to the wire and the copper plate. Once the wire is connected to the board you then remove the copper to make the traces. By CNC or other process.
You can burn the holes with a laser (may be after deleting the copper layer) so the internal surface would be intrinsically coated with carbon from burning the resin.
Some random ideas...
1. instead of plating, insert 0.25mm copper wire and solder to annular rings
2. insert 0.25mm copper wire and under pressure fill it with low temperature solder.
you can use like 135C melting point solder, + some heat resistant silicone "nozzle" to form annular ring around via to fill it with solder...
Really nice project, subscribed.
And maybe it's possible to use similar process like in point welding used for creating battery packs from ion cells, (not sure about doing that on copper with copper), but if it's possible would be nice to try. Drill hole, insert copper wire, point solder it to annular ring...
Exciting project!
I must confess that the lead time of PCB procurement doesn't bug me _that_ much as a hobbyist because for me, the real problem is parts placement - my hands shake, and getting parts placed once the paste is stencilled is a real challenge. That's why I subscribed to Stephen Hawes's channel - the work he's doing on small-scale pick&place is intriguing. But I'm subscribing to you too, to see how this turns out!
(PS: I'm not altogether sure I want the chemistry for etching and plating in the house. And if I do it in the garage, it becomes a seasonal affair; the climate here has COLD winters.)
If you find yourself placing challenging components by hand (like ICs with dozens of fine-pitch pins), you might consider putting together a "manual pick-and-place" apparatus. You can use a vacuum pickup (easy to make with a cheap modified aquarium pump and syringe), mounted to an aluminum extrusion arm, in turn on vertical slide (3D printer technology parts), sprung loaded to pull up, but can be lowered by hand. I like to control the vacuum with a small valve and foot pedal. You can then position the PCB by hand below this "crane", and slowly and carefully lower the IC nearly onto the board, maneuver the PCB a little bit to perfect position, then finish placing the part.
Ie: no need for the apparatus to have any fancy X-Y mechanism, because moving the PCB around under the crane by hand works fine. You just need the crane to give you a way to lower the IC into place consistently and without shake, with an initial "rehearsal" to get the position just right. This kind of simple apparatus makes all the difference between shakily positioning the IC and messing up the solder paste, versus placing it just right with confidence. Of course there are fancier manual PnP machines too... search for them to see.