Could this be the world's smallest LinuxCNC build? Let me know if there's a smaller one out there. Video about weenyPRU: • Palm-top LinuxCNC (wee... github.com/iforce2d/weenyPRU github.com/iforce2d/linuxcnc-...
Being the "simplest solution" is not all there is to it. Depends what you're interested in and enjoy spending your time on. I have no interest whatsoever in designing, building, owning or using a solder bath and I'm getting tired of seeing dozens of comments assuming I didn't know about them just because I never mentioned it in the video. It's also unlikely I could have put one together for the $81 I spent on this. I like designing and building LinuxCNC machines and this was a fun challenge to make something really small, and validate the PRU I spent a couple months designing and programming. Feel free to approach your own projects however you see fit.
Purely from a soldering perspective i would: Add in a tip cleaning method, move the solder feed so it somes in at ~45 degrees so it pushing into the 90deg corner created by the pin and soldering iron. When the iron is moved onto the pin do a mini solder, just to wet the iron and pin, this will immensly help with getting the heat into the pin and pad and when timed right will wick down through the board, then do a final squirt to bring you up to the correct solder amount. drag vertically off each pin to help prevent the bridging.
I agree with everything here from my own experience soldering, but also would like to add that almost all those solder joints have too much solder imo. I could be wrong.
Vertical dragging is what I immediately thought, shouldn't be hard to implement by retracting the soldering iron + pushing the board against the tip. Also, using a thinner soldering wire could help controlling the amount of solder feeded into the pin...
@@Karshilistics You're 100% correct. He's using about 50% too much solder, which is causing the globby joints & the bridges as the excess builds up on the tip of the iron.
the other thing which is a bit more complex would be to have an insulated conductive ring at the end of the solder extruder then measure conductivity through the solder and you would be able to retract and determine a zero point for the solder after each joint, and actually control how much is extruded into each joint without the assumption.
Yet the machine has potential to survive me, you, 3 or 4 of your descendants generations and keep being unbeatable at whatever many million times it accurately repeated it's gracious routine, completing for the many billionth times what one designed and built while onlookers saw him spending dollars to save pennies.
i realize my exponential emphasis on random number.. it quickly jumps to billions, making sure the salad sells 'à la Musk' a french would surely be yelling to inspire a bit of ... oh damn, that was some extra strenght H20 i subjected myself to. Geeeszz, beezz and louise. Thel-ma what ya stunk!
These sorts of machines are useful for two purposes: 1) soldering prototypes with tiny components too small to reliably do by hand, 2) soldering LOTS of boards if you're running a home fab to sell some kind of electronics.
That "clicky thing" on the back of the extruder anchors PTFE tubing. A larger size PTFE tubing than what you are using that holds 1.75 ID tube for 3D printer filament. It might be a good idea to get a short piece of that tube to make solder wire a smooth transition into the extruder. It looks like you may be running the extruder in reverse because that is usually the exit path of a 3D printer extruder. Fantastic job though. Love what you have made here.
From an agricultural-looking vice and rusty hacksaw, through "DIY CNC", a vast array of projects, balloons, carbon-fibre, anything that navigates itself, anything that flies, anything that navigates itself and flies, a mouth-watering selection of electronic and coding examples, "Another DIY CNC", to this elegant baby... is there no end to it?! That's an evolution path that Charles Darwin would be proud of. More Please, Thankyou. 😜 👍
you do know that when you export the gerber files on easyEDA, there is a drill hole file that you can use for knowing the hole positions. No need to select them one at a time like that.
Yes, I do know that. But I would still need to define the order somehow, and which ones I actually want to solder, and tag them as being ground/signal, and move them so that the first pin is at 0,0 position. DXF is more convenient for this. In QCad you can select elements by type and size, eg. circle of max radius 2, so you don't even need to manually select them in the DXF anyway.
Have you considered syringe dispensing solder flux and low temp solder paste. Two syringe screw dispensers. You can get much better flow control, less material and better adhesion with flux. Just a thought.
I've been watching your videos for 9 years and haven't ever commented, just wanted to say I'm constantly amazed by what you manage to accomplish. Do you work on all these things after your day job? Thanks for all the great videos over the years
Perhaps add a conductivity check from the solder to the headed pin through the iron. I think G-code has a way to do a "trigger action" instead of the down/wait/up and have ground and normal actions. A conductivity check could also check the resistivity and estimate the temperature. An ideal one would have extra pins which checked to ensure they were NOT touching anything else. But very cool.
I LOVE the idea of a conductivity check. Once a connection is detected between the iron and the aluminum solder tube, you then start measuring how much solder is dispensed. Brilliant!
some ideas: - make the vertical move after the joint go higher before moving sideways, to avoid pulling the solder blob onto the neighboring pad - instead of just moving down onto the joint, you could also shove the iron into the joint horizontally by say, half a mm, to make sure the tip is pushed up against both the board and the pin to ensure the solder adheres to both. if you're concerned about guillotining the pins off, you could put some spring between the z actuator and the tool. have the spring preloaded against a rigid connection, so that under normal circumstances it's rigid, but once you're pushing against something the spring can give. that way you can add a small amount of overtravel to guarantee good contact, while limiting the excessive forces that would otherwise generate. since the "z" axis is diagonal, this might also help protect against y-axis overtravel as well. - maybe break the solder dispensing into 2 phases, one small pre-extrusion to help wet out the surfaces and help with heat transfer, and then the main push to actually solder the joint. then you have the ratio of the 2 extrusions, and the 2 different dwell times, so more variables to manage! - maybe start out by just brushing on some flux before loading the board? or have some sort of rotating tip cleaning tool to park the soldering head on periodically - you could pass the solder through an extra pair of rollers with an optical encoder to detect when the extruder is slipping. then you could retract the solder, and wait an extra second or so, and then try again, maybe let out a beep and/or log the event so that you can adjust your settings accordingly. that way if the joint is still cold, it'll adjust automatically. you could set a limit to the number of retries so you don't scorch the board if the solder is stuck for unrelated reasons. - add a power button, so you can shut it down without SSH, the raspberry pi has support for using a GPIO pin as a power-off button. also, you could try booting the system in read-only mode, to reduce the risk of SD card corruption. You'd need to do a bit of work to make sure that nothing's trying to write logs or temp files to disk, and mount a tmpfs to any directories with writes you can't get rid of. of course then it's a pain to add new gcode files, perhaps you could add a USB port and use a USB stick for gcode and any other files you want to be writable. then you either mount that read-only on the pi and edit on another machine, or accept the corruption risk, with the scope of damage limited to just the files on the flash drive, that are backed up and/or tracked in git on another machine anyways - probably overkill: instead of passing in a flag to your g-code to choose between 2 hardcoded dwell times, you could have your planning tool could calculate the dwell times dynamically, you start with a base value that you set by hand for signal and ground pins, and then for the last N (maybe 5-10?) pins, divide the dwell time of that pin by the linear distance from the current pin and length of time from it, sum them up, multiply by some constant, and subtract that from your base value to approximate heat contribution from neighbors
Homeboy could've soldered more boards than he's ever going to need in the time it took to build this with a K tip, but where's the challenge in that :D Awesome build, worked way better than I initially expected!
Am back to finish this video, it's amazing what you have put into this one, love the details and all the additional tricks involved. Looks like a powerful programming language, very procedural. Also that you made yourself such a tool adds to the excitement. Thanks Chris for sharing. Cheers, Robert
Impressive project, really appreciate documenting all of the steps, especially the nitty gritty things like kinematics. I also appreciate you explaining the whole thought process, from the initial idea, its concepts, to your final design. A small-scale example of a real engineering project.
Not sure if you though about it already, but adding a PTC heater module to bottom of the bed to pre-heat the board and the pins to about 120°C I feel like that would make it quicker, with better joints and would produce nicer joints since it wouldn't burn all of the flux out of the solder (possibly also more reliable too) Prob the easiest thing you can do to improve it dramatically
You may get better results if you add a routine to lightly tin the iron before it makes contact with the pins. I often see this in soldering robots. It helps to preheat the pin before the final solder is added, and will give you higher quality joints that are less likely to bridge.
I've got a feeling that for many of these Cartesian movement platforms that don't require high stiffness, it's probably easier to just modify a cheap/broken existing 3d printer/laser cutter than it is to design and build something entirely new. Plus, you could probably abuse the bed heater and hot-end heater circuitry for something too.
I may be a bit late to comment as others might have already suggested this but you will need to add flux for this type of soldering process. I would suggest going with some water soluble flux + solder with water soluble flux inside(for example Kester) manually applied before soldering step then you will need an extra cleaning step which involves distilled watter and then drying with compressed air. Flux can also be integrated on your robot, to dispense it automatically along the pads in straight lines. And some form of hot air blown over the PCB will help preheat everything and will help with those GND pads.
@@voltlogFlux core solder is enough in most applications. Many commercial solutions don’t use additional flux. Cost to benefit. Also, compressed air is an opportunity for ESD damage.
You can turn a raspberry pi on and off safely by grounding one of the GPIO pins (requires adding a line dtoverlay=gpio-shutdown to /boot/config.txt). I think it is GPIO 3 on RPi 3 but not sure, you can google it. Note that if you disconnect the power after pi is shut down using this method, when you reconnect power pi will turn on.
Two things that I think would help a lot. The first one's easy. Flux the board before you start soldering. That will help keep the solder where you want it and make much better solder balls. The second one would be a little more difficult but if you can figure out a way to clean the tip of the iron every few joints that would help reduce contaminants in the solder joints and make for cleaner soldering as well. Also if you could bring the soldering iron straight up as quickly as possible at the end of the soldering action then move it to the next location it would probably help reduce bridging as well.
Yep, yep, yep. Flux could be added just before soldering with an atomising spray or foaming hand soap dispenser or a simple peristaltic pump into a small brush that visits the next joint just before soldering the current joint.
Exactly. When it comes to soldering, too much flux is never enough, and doubly so when it's a machine doing the soldering. And the contact must cease before all the flux burns up.
@@Phil-tl6ln ehh. It depends. Flux core solder is good but added flux definitely makes things better. This is especially true in less than ideal circumstances like where the iron isn't perfectly aligned or when the tip isn't getting cleaned frequently enough.
SOLUTION TO BRIDGING: Because the Z axis is diagonal, the Z+ motion (iron up) includes a vicarious X- motion (iron left). Add a function that accepts the Z axis angle as an argument/context (because you may alter this angle at some point) and have it return a X+ value that will be perfectly vertical to the PCB's reference frame. Once you have this ability, you might modify your "iron lift" gesture to be a more intricate "iron dismount" gesture in the shape of a capital J rotated 180 degrees. To clarify, the way you would draw a J on paper in the X,Y plane. Here, we are dealing with the X,Z plane. And we are drawing the handle of a down-pointed umbrella. 🌂 Amazing work regardless, my brother! ☮❤🌈
Sooo clever. You're fast becoming one of my favorite TH-camrs! A decent quality flux applied by syringe first should make a big difference for not much effort if you haven't already tried that. Awesome project, thanks for sharing.
So many interesting aspects to this project. Great random video montage of observations from the CNC in operation. The belt system had me intrigued before you began explaining how ti worked. Great explanations of the overall design. For better consistency, (~23:45) ability to automatically clean tip will help, as would pre-applying solder paste to the pcb pads (not so bad if have a solder mask, and pcb maker often provide as an option). Expect this was a real educational journey, as so many smaller problems to solve and design around along the way. A very impressive project.
Nice project! Impressive! FYI, you can also setup a pushbutton to power off the raspberry pi, no SSH needed. If you also end up with different boards to solder, you could add an lcd to select which file to run. I have a couple of projects that could use the board that plugs into the pi and steppers. Definitely looking into it more.
It looks like the main issue is that the solder's surface tension pulls an unpredictable volume of solder in. You could abate this with finer solder wire perhaps, but I'd also try it at a higher temperature. I usually do through hole soldering at 370C personally. How I would approach this is a PTFE tube that goes vertically down over the pin and solder pad, then a measured shot of powdered solder gets fed from a screw conveyor and falls down the tube onto the pad. As long as the acceleration of the machine isn't too fast, the powdered solder would stay put and then the entire thing could be popped into a toaster oven.
I had a few comments, but I see that other folks have mentioned most of them. I noticed the unsoldered pin happened because the iron tip didn't make good contact with the pin, so the solder just pushed against it instead of melting. Tweaking the appropriate parameter might help. As far as motion systems with stationary motors, you might look into H-bot, which is easiest to build, or core-XY, which is considered superior but is more complex and probably unnecessary for a small setup like this. Your setup is like a modified H-bot, but with more parts and without the symmetry.
Some suggestions: (1) Use variant of PCB as fixture: If you are iterating a new PCB, milling and drilling a new aluminum pin-holder fixture each time is very tedious and wasteful. Instead, you could use the PCB itself as the pin holder. With not much effort, while creating the actual PCB design, you can add outlines that the PCB fab can mill out to create windows where your already-mounted SMD components can fit (corresponding to the pockets you would have made in the aluminum fixture.). (2) part of the rationale for the pin-holder fixture is so you can load up all the headers at once. You can't place them on the PCB itself because when you flip the board over they all fall out. Even if you did get the board flipped with headers in place, the headers tend to wiggle around in the hole clearance, and end up slightly non-uniformly aligned. You can solve that problem by modifying your header footprints in the PCB software, so that some of the holes are slightly offset in different (but symmetrical) directions. This creates some friction that (a) registers the header uniformly and (b) stops it from falling out when you flip the board. You can then use a much simpler non-custom fixture that just pushes on the tips of the headers (assuming uniform height) and doesn't need to have holes for each pin specifically. That also avoids having the aluminum fixture acting as a heatsink on the pins.
Cool project! The strange nut on the extruder is for the bowden tube. You have kind of mounted the extruder backwards, not that it matters much, but it would probably hold the tube more securly than the rubber thingy. You could keep it as is and run an bowden tube to the spool instead of moving the spool.
I think it would not hold the PTFE tube because this one is smaller than the standard PTFE used for 1,75mm filament. But, yes, it seems the extruder is mounted backwards.
Wow! I needed headers on my JLC boards too, but I just placed SMD ones and had it done as part of the build. They will hand-solder TH pins too of course but I wanted to use the economical PCB-A option :) Hi again from Nelson
Inspiring build! Another way to keep the solder wire from slipping off the pulley is to file a crown over the silicone tube sleeve. It's how bandsaws keep the blade stationary on the wheel. There's also a manual tilting adjustment mechanism to angle the wheel in and out in order to keep it centered. Your solution for keeping the wire captive is simpler.
I would never think about designing a soldering robot but must say it came out better than would I would expect. With some small enhancements like suggested by others could be pretty good
I work with a professional Ersa Ecoselect 4 machine First a nozzle sprays flux on all solder joints, then the PCB gets preheated by IR heaters (like soldering ovens do) only then do we actually solder To get reliable consistent solder joints
Amazing project, cool time saver! A few suggestions - try clamping the solder in line with the iron tip, not to the side, which is what I tend to do by hand. Possibly try making the solder come in from a slightly higher angle. Looks like you definitely need a smaller solder tip, and when it comes to lifting the tip, should probably be purely vertical. Think about reordering the job to always solder left to right from camera perspective, and yes looks like you definitely need some kind of tip clean system as others have suggested. Otherwise, yeah amazing that it does so well - just thinking about how it saves me from keeping my head in a perpetually smokey area, a real lifesaver too.
If you're interested, you can buy a cheap 3D printer, possibly even cheaper as an Amazon return or similar, and you get all the frame parts, motors, controller to do some pretty good experimenting. You can send G-Code commands to the main board as supplied to make it perform the movements you want. If that firmware is Marlin or similar, then the board can be completely reprogrammed in the commonplace Arduino environment to do something completely different. Bottom line: The "on-ramp" to this sort of fun is not very steep!
In the extruder section, the little clicky bit (the first thing the solder contacts on its way to the extruder) is for holding the bowden tube (basically, the equivalent of the little white piece that came in the package for your bowden tube.) It's presumably sized for a 1.75mm ID bowden (for 3d printing.) You could either leave it as-is, or if the teeth inside (which are there to grip the tube) interfere with the solder, you could either remove it or replace it with a smooth piece of aluminum (using a die to engage the threads that the bowden holder is screwed into.
Soooo impressed with your skills, have watched you for years, you are the one that got me tinkering with arduino GPS and GSM modules! But, it kinda depresses me that I’m not capable of achieving something like this! lol! For a better solder joint and finish, try putting extra flux on the board and drag iron up the the pin to finish. Also, firstly, try adding a tiny bit of solder to the tip before moving to the pin, it will give a greater heat transfer area between the iron, pad/pin and with the extra flux, the solder will be attracted to the pad/pin and should require less time for the iron to heat it up and make it more efficient. Hope this helps with your masterpiece!
3d printer extruders with two driven rollers are a relatively new thing that Bondtech in Sweden popularized with the BMG extruder. Plenty of printers still use a single driven roller and a spring loaded idler wheel. A newer Bondtech innovation is the LGX with two larger diameter (20mm ish) driven rollers to let more teeth engage the filament and travel in a shallower arc. Two larger diameter, lighty textured or rubber covered aluminium or brass rollers might be optimal for solder dispensing. Love the gimbal jog controller and the pass through X axis belt drive.
The bridging is due to moving away at an angle - you need the iron to lift completely vertical, up the pin to ensure a good "conical" solder. Also, slow down the retraction of the iron and keep the iron in contact with the pad for a few milliseconds more to allow the solder to flow. You non-solder points are most likely due to a slightly uneven bed - I'd manually solder some of the pins in the corners to begin with and clamp down the board (using mount holes if you have them) or just some clamps / alligator clips might help Your V2 upgrade should be leadscrews - belts loosen over time and your accuracy needs to be within 10's of microns ;)
Completely awesome! You might want to try to add the solder a bit earlier and dwell a bit longer in exchange for that. Will help with the heat transfer and you may even get less bridging. Also regarding bridging: In 3D printing it often helps to move faster in order to prevent stringing - you might want to try moving the iron away from the pins with a quicker/jerkier movement in order to break the tension of the molten solder an (hopefully) get less bridging. Also: Engineer a way to automatically clean the tip between runs! ^^
Very impressive! It looks like the latent heat from nearby joints makes the process rather unpredictable. If you adjust the G code to make sure the heat can dissipate before it is soldered you should have a much more predictable result. Another idea i had when watching is have a "per-pin lingering offset" that you can adjust to account for differences in heat dissipation on a per-joint basis.
it's beautiful, looks to me like a wedge tip about half as wide as the one you are using would solve your bridging problem. I have one on my Hakko about that size, 0.6mm. I'm guessing it must exist for the ts100, if not you could turn a small slip on sock/tip adapter from some copper rod. Lovely stuff, you must be pretty pleased with it. Would it be possible to sense the temp drop from the TS to determine soldered vs non soldered pin given the larger thermal mass? I don't know if that data is on the TS usb line? Maybe sense current draw? That way you could do a smart second run. PS how do I get my hands on an iomixer board?
If you change the order of the pins in such a way that it the solder is on the side with an un-soldered pin, it should reduce bridging. And maybe lifting the iron up vertically instead of at an angle?
amazing project - i could only dream of such level of expertise! Little hint for "Linux stuff" perhaps: just wire another switch to a GPIO of the Pi, and enable "gpio-shutdown" in config.txt... and for starting linux-cnc, a basic systemd-service should do the trick...
You published code I was looking at it last night. Nice work thanks a lot. I will try to use it as starting point for other chip and to understand how to interface LCNC. I have your project with SPI and other for UDP to learn from.
27:07 personally i think it's the tip size that's causing the bridging problem since it's heating up other pins also some cooling would be nice also like a fan blowing it so it cools off the pins
Solder positioning nuance: Your machine acts like it read a soldering tutorial and is following it diligently :-). First it applies the iron to the front side of the pin, pauses to get pad and pin heated, then applies the solder to the back side of the pin allowing the temperature of the pad and pin to heat the solder and get it to flow. This respects typical tutorial's admonition against applying solder to the iron tip. However, try this alternate technique manually, and see if it can be adapted to this machine. First come in with the solder at say 30 degree from horizontal so that it lands at the _front_ side of the pin at its base. Then bring the iron in, similar to the machine's current motion, mashing the solder into the intersection of pin and pad. In quick succession, this melts the flux out of the solder, flowing it around the pin and pad, preparing them. Meanwhile it also melts the solder into a "pillow" that quickly conducts heat from the iron to the pin and pad. Finally the solder flows around the joint onto the heated and prepared pad and pin. This has the prospect of applying flux more consistently, and heating the joint more consistently and more locally, since it gets the heat flowing faster via the pillow. I certainly achieve much faster and more consistent manual soldering this way.
Hey that's pretty cool. Probably already mentioned in the comments some where but maybe a tip cleaning stand, like a vertical plate with some cleaning pad attached and running the tip down it between PCBs. Also might be good to brush some flux on the PCB before running the solder bot might help. Flux core solder doesn't seem to have quite enough flux for automated use.
As a PInecil V2 owner i'd like to mention i can utilize either bluetooth or usb-ttl/serial in order to get a live-reading of the iron's temperature and other things, so i think using a pinecil might be able to give the ability to read out and utilize your iron temperature :) alternatively utilizing a pinecil breakout board you could modify iron os to output certain pins to high or low depending on its temperature :)
Also a common thermistor might do the trick. He mentioned he’s not actually familiar with 3d printing in general, but measuring tip temperature is a common parameter for initiating the job and for validating that temperature remains as expected throughout the script.
Help with solder flowing with flux in a syringe, just needs pushrod to dispense out of the cartridge down a fine tube. Good luck with the build it's coming along great.
With the extruder you have it backwards. The clicky thing where the solder goes in is a pneumatic fitting designed to get a PTFE tube pushed into it, that PTFE tube then goes from the extruder to the hot end.
The clicky thing at the back is for a Bowden on that side, it will lock in and you need to push in the grey thing to remove it, this might be worthwhile to allow the solder to enter at a little less aggressive of an angle. You can solder a switch to the GPIO of the pi zero that are programmed to perform a safe shutdown so you don’t need the Wi-Fi. I also think a spray of flux on the board before soldering will give you more consistent results and you will just need to throw them in Isopropanol to clean them after.
Amazing project. I see a few improvements in the soldering. Slightly shorter heating of the leg, slightly less solder added and add a half a second of after heat to allow the solder to flow better. Some form of cleaning protocol of the solder tip at intervals would remove too much solder from the tip. a tad of flux could help as well. Once more, amazing project, loving the shit out of it!!
I suspect the remaining inconsistencies in joint quality could be helped by rubbing a flux pen across the pins with the board mounted while the iron is warming up.
quite an amazing project my man. I would recommend if you are looking for better throughput to get a solder pot where you dip solder the boards and making fixturing for that.
Actually using real flux might help out too. Human soldering might work perfectly fine with just flux core, but actual liquid flux pre-applied to the board might help the robot.
I love a good Rube Goldberg machine as much as the next guy, but wouldn't dip soldering these boards be the simplest solution?
Being the "simplest solution" is not all there is to it. Depends what you're interested in and enjoy spending your time on. I have no interest whatsoever in designing, building, owning or using a solder bath and I'm getting tired of seeing dozens of comments assuming I didn't know about them just because I never mentioned it in the video. It's also unlikely I could have put one together for the $81 I spent on this. I like designing and building LinuxCNC machines and this was a fun challenge to make something really small, and validate the PRU I spent a couple months designing and programming. Feel free to approach your own projects however you see fit.
@@iforce2d I hope your day gets better.
Grow up
@@JohnJones-oy3md
@@brycecrichfield1934 What do you mean by that?
@@JohnJones-oy3md
Purely from a soldering perspective i would: Add in a tip cleaning method, move the solder feed so it somes in at ~45 degrees so it pushing into the 90deg corner created by the pin and soldering iron. When the iron is moved onto the pin do a mini solder, just to wet the iron and pin, this will immensly help with getting the heat into the pin and pad and when timed right will wick down through the board, then do a final squirt to bring you up to the correct solder amount. drag vertically off each pin to help prevent the bridging.
Fully agreed. Some great tips here
I agree with everything here from my own experience soldering, but also would like to add that almost all those solder joints have too much solder imo. I could be wrong.
Vertical dragging is what I immediately thought, shouldn't be hard to implement by retracting the soldering iron + pushing the board against the tip. Also, using a thinner soldering wire could help controlling the amount of solder feeded into the pin...
@@Karshilistics You're 100% correct. He's using about 50% too much solder, which is causing the globby joints & the bridges as the excess builds up on the tip of the iron.
the other thing which is a bit more complex would be to have an insulated conductive ring at the end of the solder extruder then measure conductivity through the solder and you would be able to retract and determine a zero point for the solder after each joint, and actually control how much is extruded into each joint without the assumption.
An engineer is somebody who will spend 5 years building a machine to avoid doing 5 minutes of manual labour.
Yet the machine has potential to survive me, you, 3 or 4 of your descendants generations and keep being unbeatable at whatever many million times it accurately repeated it's gracious routine, completing for the many billionth times what one designed and built while onlookers saw him spending dollars to save pennies.
i realize my exponential emphasis on random number.. it quickly jumps to billions, making sure the salad sells 'à la Musk' a french would surely be yelling to inspire a bit of ... oh damn, that was some extra strenght H20 i subjected myself to. Geeeszz, beezz and louise. Thel-ma what ya stunk!
These sorts of machines are useful for two purposes: 1) soldering prototypes with tiny components too small to reliably do by hand, 2) soldering LOTS of boards if you're running a home fab to sell some kind of electronics.
Try adjusting the movement vertical before horizontal instead of both at once and that may stop that bridging.
To do that he would have to move the bottom board at the same time as he raises the soldering iron, but yes, same thought I had.
@@AlJay0032 It would add a bit more complexity to the motion, but doable albeit a bit steppy in motion.
Came here to say this 🙂
I'd bet more flux would help. Tends to keep things round and shiny. Maybe you can just get a solder reel that has more in it.
Or a wet sponge function after so many pins.
That "clicky thing" on the back of the extruder anchors PTFE tubing. A larger size PTFE tubing than what you are using that holds 1.75 ID tube for 3D printer filament. It might be a good idea to get a short piece of that tube to make solder wire a smooth transition into the extruder. It looks like you may be running the extruder in reverse because that is usually the exit path of a 3D printer extruder. Fantastic job though. Love what you have made here.
From an agricultural-looking vice and rusty hacksaw, through "DIY CNC", a vast array of projects, balloons, carbon-fibre, anything that navigates itself, anything that flies, anything that navigates itself and flies, a mouth-watering selection of electronic and coding examples, "Another DIY CNC", to this elegant baby... is there no end to it?!
That's an evolution path that Charles Darwin would be proud of. More Please, Thankyou. 😜 👍
you do know that when you export the gerber files on easyEDA, there is a drill hole file that you can use for knowing the hole positions. No need to select them one at a time like that.
Yes, I do know that. But I would still need to define the order somehow, and which ones I actually want to solder, and tag them as being ground/signal, and move them so that the first pin is at 0,0 position. DXF is more convenient for this. In QCad you can select elements by type and size, eg. circle of max radius 2, so you don't even need to manually select them in the DXF anyway.
@@iforce2d okay. Nice progress by the way, loving the videos. I always learn something new from you.
Cant wait for the solder benchy.
Have you considered syringe dispensing solder flux and low temp solder paste. Two syringe screw dispensers. You can get much better flow control, less material and better adhesion with flux. Just a thought.
I've been watching your videos for 9 years and haven't ever commented, just wanted to say I'm constantly amazed by what you manage to accomplish. Do you work on all these things after your day job? Thanks for all the great videos over the years
I've been watching for quite a few years. But I've still been wondering the same thing, how do you manage work time and hobby time
@@BuzZ.agree. I built the cheep ass quadcopter long ago. My favorite TH-camr.
I'm guessing he is retired. Not sure though
@@rowannadon7668 I don't know if he is retired but for sure he is single!
Perhaps add a conductivity check from the solder to the headed pin through the iron. I think G-code has a way to do a "trigger action" instead of the down/wait/up and have ground and normal actions.
A conductivity check could also check the resistivity and estimate the temperature. An ideal one would have extra pins which checked to ensure they were NOT touching anything else.
But very cool.
I LOVE the idea of a conductivity check. Once a connection is detected between the iron and the aluminum solder tube, you then start measuring how much solder is dispensed. Brilliant!
This video/project was such a wonderful treat. Thank you for taking use along your journey!
some ideas:
- make the vertical move after the joint go higher before moving sideways, to avoid pulling the solder blob onto the neighboring pad
- instead of just moving down onto the joint, you could also shove the iron into the joint horizontally by say, half a mm, to make sure the tip is pushed up against both the board and the pin to ensure the solder adheres to both. if you're concerned about guillotining the pins off, you could put some spring between the z actuator and the tool. have the spring preloaded against a rigid connection, so that under normal circumstances it's rigid, but once you're pushing against something the spring can give. that way you can add a small amount of overtravel to guarantee good contact, while limiting the excessive forces that would otherwise generate. since the "z" axis is diagonal, this might also help protect against y-axis overtravel as well.
- maybe break the solder dispensing into 2 phases, one small pre-extrusion to help wet out the surfaces and help with heat transfer, and then the main push to actually solder the joint. then you have the ratio of the 2 extrusions, and the 2 different dwell times, so more variables to manage!
- maybe start out by just brushing on some flux before loading the board? or have some sort of rotating tip cleaning tool to park the soldering head on periodically
- you could pass the solder through an extra pair of rollers with an optical encoder to detect when the extruder is slipping. then you could retract the solder, and wait an extra second or so, and then try again, maybe let out a beep and/or log the event so that you can adjust your settings accordingly. that way if the joint is still cold, it'll adjust automatically. you could set a limit to the number of retries so you don't scorch the board if the solder is stuck for unrelated reasons.
- add a power button, so you can shut it down without SSH, the raspberry pi has support for using a GPIO pin as a power-off button. also, you could try booting the system in read-only mode, to reduce the risk of SD card corruption. You'd need to do a bit of work to make sure that nothing's trying to write logs or temp files to disk, and mount a tmpfs to any directories with writes you can't get rid of. of course then it's a pain to add new gcode files, perhaps you could add a USB port and use a USB stick for gcode and any other files you want to be writable. then you either mount that read-only on the pi and edit on another machine, or accept the corruption risk, with the scope of damage limited to just the files on the flash drive, that are backed up and/or tracked in git on another machine anyways
- probably overkill: instead of passing in a flag to your g-code to choose between 2 hardcoded dwell times, you could have your planning tool could calculate the dwell times dynamically, you start with a base value that you set by hand for signal and ground pins, and then for the last N (maybe 5-10?) pins, divide the dwell time of that pin by the linear distance from the current pin and length of time from it, sum them up, multiply by some constant, and subtract that from your base value to approximate heat contribution from neighbors
Homeboy could've soldered more boards than he's ever going to need in the time it took to build this with a K tip, but where's the challenge in that :D
Awesome build, worked way better than I initially expected!
This looks absolutely great. You are a real inspiration.
A tricky task to automate, very impressive. Well done.
Am back to finish this video, it's amazing what you have put into this one, love the details and all the additional tricks involved.
Looks like a powerful programming language, very procedural.
Also that you made yourself such a tool adds to the excitement.
Thanks Chris for sharing.
Cheers, Robert
Impressive project, really appreciate documenting all of the steps, especially the nitty gritty things like kinematics. I also appreciate you explaining the whole thought process, from the initial idea, its concepts, to your final design. A small-scale example of a real engineering project.
Not sure if you though about it already, but adding a PTC heater module to bottom of the bed to pre-heat the board and the pins to about 120°C
I feel like that would make it quicker, with better joints and would produce nicer joints since it wouldn't burn all of the flux out of the solder (possibly also more reliable too)
Prob the easiest thing you can do to improve it dramatically
Maybe also shield the machine from drafts. The pins on the edges might be cooler so dont wet properly.
You may get better results if you add a routine to lightly tin the iron before it makes contact with the pins. I often see this in soldering robots. It helps to preheat the pin before the final solder is added, and will give you higher quality joints that are less likely to bridge.
I tried that but the solder just flows down under the rear of the iron and does nothing useful.
I've got a feeling that for many of these Cartesian movement platforms that don't require high stiffness, it's probably easier to just modify a cheap/broken existing 3d printer/laser cutter than it is to design and build something entirely new.
Plus, you could probably abuse the bed heater and hot-end heater circuitry for something too.
I may be a bit late to comment as others might have already suggested this but you will need to add flux for this type of soldering process.
I would suggest going with some water soluble flux + solder with water soluble flux inside(for example Kester) manually applied before soldering step then you will need an extra cleaning step which involves distilled watter and then drying with compressed air.
Flux can also be integrated on your robot, to dispense it automatically along the pads in straight lines.
And some form of hot air blown over the PCB will help preheat everything and will help with those GND pads.
I think he's using some flux-core solder -- board gets pretty shiny with what is likely flux.
@@MatthewWalker0that is not enough.
@@voltlogFlux core solder is enough in most applications. Many commercial solutions don’t use additional flux. Cost to benefit. Also, compressed air is an opportunity for ESD damage.
Freaking awesome! Bumped into your channel earlier this year. Great content.
You can turn a raspberry pi on and off safely by grounding one of the GPIO pins (requires adding a line dtoverlay=gpio-shutdown to /boot/config.txt). I think it is GPIO 3 on RPi 3 but not sure, you can google it. Note that if you disconnect the power after pi is shut down using this method, when you reconnect power pi will turn on.
Two things that I think would help a lot. The first one's easy. Flux the board before you start soldering. That will help keep the solder where you want it and make much better solder balls. The second one would be a little more difficult but if you can figure out a way to clean the tip of the iron every few joints that would help reduce contaminants in the solder joints and make for cleaner soldering as well. Also if you could bring the soldering iron straight up as quickly as possible at the end of the soldering action then move it to the next location it would probably help reduce bridging as well.
Yep, yep, yep.
Flux could be added just before soldering with an atomising spray or foaming hand soap dispenser or a simple peristaltic pump into a small brush that visits the next joint just before soldering the current joint.
@@KallePihlajasaari or just brush it on before the process starts. It would literally take 5 seconds.
Exactly. When it comes to soldering, too much flux is never enough, and doubly so when it's a machine doing the soldering. And the contact must cease before all the flux burns up.
you shouldn't need extra flux when using flux core solder
@@Phil-tl6ln ehh. It depends. Flux core solder is good but added flux definitely makes things better. This is especially true in less than ideal circumstances like where the iron isn't perfectly aligned or when the tip isn't getting cleaned frequently enough.
SOLUTION TO BRIDGING:
Because the Z axis is diagonal, the Z+ motion (iron up) includes a vicarious X- motion (iron left). Add a function that accepts the Z axis angle as an argument/context (because you may alter this angle at some point) and have it return a X+ value that will be perfectly vertical to the PCB's reference frame. Once you have this ability, you might modify your "iron lift" gesture to be a more intricate "iron dismount" gesture in the shape of a capital J rotated 180 degrees. To clarify, the way you would draw a J on paper in the X,Y plane. Here, we are dealing with the X,Z plane. And we are drawing the handle of a down-pointed umbrella. 🌂
Amazing work regardless, my brother! ☮❤🌈
Sooo clever. You're fast becoming one of my favorite TH-camrs! A decent quality flux applied by syringe first should make a big difference for not much effort if you haven't already tried that. Awesome project, thanks for sharing.
An impressive result, thanks for sharing, it looks great.
how you've never heard of a solder pot just boggles my mind, inventing solutions for problems that have already been fixed AGES ago
GREAT project. I can full understand the pleasure in overcomimg each of the issues step by step. More power to you, and thanks for sharing your work.
Fabulous! Thanks for making this into a clear and entertaining video.
Projects like these are super inspiring for me! Thanks for documenting it!
Very interesting project, nice result !
Glad to See Your palmtop LinuxCNC working. Very nice!
It took me 15 seconds to fall in love with this!!!
Just perfect.
What a lovely job! Well done m8!
So many interesting aspects to this project. Great random video montage of observations from the CNC in operation. The belt system had me intrigued before you began explaining how ti worked. Great explanations of the overall design.
For better consistency, (~23:45) ability to automatically clean tip will help, as would pre-applying solder paste to the pcb pads (not so bad if have a solder mask, and pcb maker often provide as an option). Expect this was a real educational journey, as so many smaller problems to solve and design around along the way. A very impressive project.
Great job, I am glad I found this project. I am always happy to learn from others.
Nice project! Impressive! FYI, you can also setup a pushbutton to power off the raspberry pi, no SSH needed. If you also end up with different boards to solder, you could add an lcd to select which file to run. I have a couple of projects that could use the board that plugs into the pi and steppers. Definitely looking into it more.
It looks like the main issue is that the solder's surface tension pulls an unpredictable volume of solder in. You could abate this with finer solder wire perhaps, but I'd also try it at a higher temperature. I usually do through hole soldering at 370C personally.
How I would approach this is a PTFE tube that goes vertically down over the pin and solder pad, then a measured shot of powdered solder gets fed from a screw conveyor and falls down the tube onto the pad. As long as the acceleration of the machine isn't too fast, the powdered solder would stay put and then the entire thing could be popped into a toaster oven.
I second the idea of a solder paste machine as mentioned in the comments somewhere.
This is one extremely cool machine! TY for sharing!
Amazing talent and skill
I had a few comments, but I see that other folks have mentioned most of them. I noticed the unsoldered pin happened because the iron tip didn't make good contact with the pin, so the solder just pushed against it instead of melting. Tweaking the appropriate parameter might help. As far as motion systems with stationary motors, you might look into H-bot, which is easiest to build, or core-XY, which is considered superior but is more complex and probably unnecessary for a small setup like this. Your setup is like a modified H-bot, but with more parts and without the symmetry.
Wow super cool. If you used a pine64 soldering iron you could get the status from it
Nice machine. Obviously you never sleep judging by all the things you manage to design and build.
Some suggestions:
(1) Use variant of PCB as fixture: If you are iterating a new PCB, milling and drilling a new aluminum pin-holder fixture each time is very tedious and wasteful. Instead, you could use the PCB itself as the pin holder.
With not much effort, while creating the actual PCB design, you can add outlines that the PCB fab can mill out to create windows where your already-mounted SMD components can fit (corresponding to the pockets you would have made in the aluminum fixture.).
(2) part of the rationale for the pin-holder fixture is so you can load up all the headers at once. You can't place them on the PCB itself because when you flip the board over they all fall out. Even if you did get the board flipped with headers in place, the headers tend to wiggle around in the hole clearance, and end up slightly non-uniformly aligned.
You can solve that problem by modifying your header footprints in the PCB software, so that some of the holes are slightly offset in different (but symmetrical) directions. This creates some friction that (a) registers the header uniformly and (b) stops it from falling out when you flip the board. You can then use a much simpler non-custom fixture that just pushes on the tips of the headers (assuming uniform height) and doesn't need to have holes for each pin specifically. That also avoids having the aluminum fixture acting as a heatsink on the pins.
Cool project! The strange nut on the extruder is for the bowden tube. You have kind of mounted the extruder backwards, not that it matters much, but it would probably hold the tube more securly than the rubber thingy. You could keep it as is and run an bowden tube to the spool instead of moving the spool.
I think it would not hold the PTFE tube because this one is smaller than the standard PTFE used for 1,75mm filament. But, yes, it seems the extruder is mounted backwards.
Really cool! Improvement... maybe add some sort of fluxing before soldering? Try a different soldering pen/point? Amazed how you made this!
Wow! I needed headers on my JLC boards too, but I just placed SMD ones and had it done as part of the build. They will hand-solder TH pins too of course but I wanted to use the economical PCB-A option :) Hi again from Nelson
Awesome work.
Inspiring build! Another way to keep the solder wire from slipping off the pulley is to file a crown over the silicone tube sleeve. It's how bandsaws keep the blade stationary on the wheel. There's also a manual tilting adjustment mechanism to angle the wheel in and out in order to keep it centered. Your solution for keeping the wire captive is simpler.
I probably shouldn't be surprised you made this, but I am, It's just so damn cool.
Excellent work
I pioneered that slipping extruder method in 2015 - Shrink Tubing works great too for providing "pressure" rather than gearing for flexible filament!
I would never think about designing a soldering robot but must say it came out better than would I would expect. With some small enhancements like suggested by others could be pretty good
I work with a professional Ersa Ecoselect 4 machine
First a nozzle sprays flux on all solder joints, then the PCB gets preheated by IR heaters (like soldering ovens do) only then do we actually solder
To get reliable consistent solder joints
Tickle me impressed! Very nicely done!!!
This is exceptional. Incredible work!
really excellent work!!!
Amazing project, cool time saver! A few suggestions - try clamping the solder in line with the iron tip, not to the side, which is what I tend to do by hand. Possibly try making the solder come in from a slightly higher angle. Looks like you definitely need a smaller solder tip, and when it comes to lifting the tip, should probably be purely vertical. Think about reordering the job to always solder left to right from camera perspective, and yes looks like you definitely need some kind of tip clean system as others have suggested. Otherwise, yeah amazing that it does so well - just thinking about how it saves me from keeping my head in a perpetually smokey area, a real lifesaver too.
Really wonderful project and video! Great work and I wish I had one!
3:41 now, if on purpose or not, just amazing looking footage!
Wow! This is amazing. Love that you just casually designed and built a custom CNC machine to solve your problem.
If you're interested, you can buy a cheap 3D printer, possibly even cheaper as an Amazon return or similar, and you get all the frame parts, motors, controller to do some pretty good experimenting. You can send G-Code commands to the main board as supplied to make it perform the movements you want. If that firmware is Marlin or similar, then the board can be completely reprogrammed in the commonplace Arduino environment to do something completely different. Bottom line: The "on-ramp" to this sort of fun is not very steep!
Re: Moth riding on snail. Why? Some kind of parasitic moth? Re: Solder bot. Beautiful.
In the extruder section, the little clicky bit (the first thing the solder contacts on its way to the extruder) is for holding the bowden tube (basically, the equivalent of the little white piece that came in the package for your bowden tube.) It's presumably sized for a 1.75mm ID bowden (for 3d printing.) You could either leave it as-is, or if the teeth inside (which are there to grip the tube) interfere with the solder, you could either remove it or replace it with a smooth piece of aluminum (using a die to engage the threads that the bowden holder is screwed into.
Really impressive.
Bro, das ist really really cool!
Soooo impressed with your skills, have watched you for years, you are the one that got me tinkering with arduino GPS and GSM modules!
But, it kinda depresses me that I’m not capable of achieving something like this! lol!
For a better solder joint and finish, try putting extra flux on the board and drag iron up the the pin to finish.
Also, firstly, try adding a tiny bit of solder to the tip before moving to the pin, it will give a greater heat transfer area between the iron, pad/pin and with the extra flux, the solder will be attracted to the pad/pin and should require less time for the iron to heat it up and make it more efficient.
Hope this helps with your masterpiece!
Excellent work.!! I recon if you lift the iron straight up higher the bridging wiil go.
3d printer extruders with two driven rollers are a relatively new thing that Bondtech in Sweden popularized with the BMG extruder. Plenty of printers still use a single driven roller and a spring loaded idler wheel.
A newer Bondtech innovation is the LGX with two larger diameter (20mm ish) driven rollers to let more teeth engage the filament and travel in a shallower arc.
Two larger diameter, lighty textured or rubber covered aluminium or brass rollers might be optimal for solder dispensing.
Love the gimbal jog controller and the pass through X axis belt drive.
The bridging is due to moving away at an angle - you need the iron to lift completely vertical, up the pin to ensure a good "conical" solder. Also, slow down the retraction of the iron and keep the iron in contact with the pad for a few milliseconds more to allow the solder to flow.
You non-solder points are most likely due to a slightly uneven bed - I'd manually solder some of the pins in the corners to begin with and clamp down the board (using mount holes if you have them) or just some clamps / alligator clips might help
Your V2 upgrade should be leadscrews - belts loosen over time and your accuracy needs to be within 10's of microns ;)
Completely awesome!
You might want to try to add the solder a bit earlier and dwell a bit longer in exchange for that.
Will help with the heat transfer and you may even get less bridging.
Also regarding bridging: In 3D printing it often helps to move faster in order to prevent stringing - you might want to try moving the iron away from the pins with a quicker/jerkier movement in order to break the tension of the molten solder an (hopefully) get less bridging.
Also: Engineer a way to automatically clean the tip between runs! ^^
Very impressive!
It looks like the latent heat from nearby joints makes the process rather unpredictable.
If you adjust the G code to make sure the heat can dissipate before it is soldered you should have a much more predictable result.
Another idea i had when watching is have a "per-pin lingering offset" that you can adjust to account for differences in heat dissipation on a per-joint basis.
that control panel is so pretty. You could probably run this on esp32, fluidnc is great.
it's beautiful, looks to me like a wedge tip about half as wide as the one you are using would solve your bridging problem. I have one on my Hakko about that size, 0.6mm. I'm guessing it must exist for the ts100, if not you could turn a small slip on sock/tip adapter from some copper rod. Lovely stuff, you must be pretty pleased with it. Would it be possible to sense the temp drop from the TS to determine soldered vs non soldered pin given the larger thermal mass? I don't know if that data is on the TS usb line? Maybe sense current draw? That way you could do a smart second run. PS how do I get my hands on an iomixer board?
If you change the order of the pins in such a way that it the solder is on the side with an un-soldered pin, it should reduce bridging.
And maybe lifting the iron up vertically instead of at an angle?
Pretty cool content you have man!
Incredible as always ! .. have you thought about solder bath ? .. smd connectors ?
amazing project - i could only dream of such level of expertise! Little hint for "Linux stuff" perhaps: just wire another switch to a GPIO of the Pi, and enable "gpio-shutdown" in config.txt... and for starting linux-cnc, a basic systemd-service should do the trick...
You published code I was looking at it last night. Nice work thanks a lot. I will try to use it as starting point for other chip and to understand how to interface LCNC. I have your project with SPI and other for UDP to learn from.
The original Remora project would probably be useful too, that's where I learned from.
27:07 personally i think it's the tip size that's causing the bridging problem since it's heating up other pins also some cooling would be nice also like a fan blowing it so it cools off the pins
such a great project!
This looks great! Any chance you plan on releasing the CAD alongside the code?
Excellent!
Solder positioning nuance: Your machine acts like it read a soldering tutorial and is following it diligently :-). First it applies the iron to the front side of the pin, pauses to get pad and pin heated, then applies the solder to the back side of the pin allowing the temperature of the pad and pin to heat the solder and get it to flow. This respects typical tutorial's admonition against applying solder to the iron tip.
However, try this alternate technique manually, and see if it can be adapted to this machine. First come in with the solder at say 30 degree from horizontal so that it lands at the _front_ side of the pin at its base. Then bring the iron in, similar to the machine's current motion, mashing the solder into the intersection of pin and pad. In quick succession, this melts the flux out of the solder, flowing it around the pin and pad, preparing them. Meanwhile it also melts the solder into a "pillow" that quickly conducts heat from the iron to the pin and pad. Finally the solder flows around the joint onto the heated and prepared pad and pin.
This has the prospect of applying flux more consistently, and heating the joint more consistently and more locally, since it gets the heat flowing faster via the pillow. I certainly achieve much faster and more consistent manual soldering this way.
Hey that's pretty cool. Probably already mentioned in the comments some where but maybe a tip cleaning stand, like a vertical plate with some cleaning pad attached and running the tip down it between PCBs. Also might be good to brush some flux on the PCB before running the solder bot might help. Flux core solder doesn't seem to have quite enough flux for automated use.
That is so smart 👍🏼
As a PInecil V2 owner i'd like to mention i can utilize either bluetooth or usb-ttl/serial in order to get a live-reading of the iron's temperature and other things, so i think using a pinecil might be able to give the ability to read out and utilize your iron temperature :)
alternatively utilizing a pinecil breakout board you could modify iron os to output certain pins to high or low depending on its temperature :)
Also a common thermistor might do the trick. He mentioned he’s not actually familiar with 3d printing in general, but measuring tip temperature is a common parameter for initiating the job and for validating that temperature remains as expected throughout the script.
Help with solder flowing with flux in a syringe, just needs pushrod to dispense out of the cartridge down a fine tube.
Good luck with the build it's coming along great.
Muito prazeroso ver seu Video amigo! Abraços.
With the extruder you have it backwards. The clicky thing where the solder goes in is a pneumatic fitting designed to get a PTFE tube pushed into it, that PTFE tube then goes from the extruder to the hot end.
The clicky thing at the back is for a Bowden on that side, it will lock in and you need to push in the grey thing to remove it, this might be worthwhile to allow the solder to enter at a little less aggressive of an angle. You can solder a switch to the GPIO of the pi zero that are programmed to perform a safe shutdown so you don’t need the Wi-Fi. I also think a spray of flux on the board before soldering will give you more consistent results and you will just need to throw them in Isopropanol to clean them after.
Amazing project. I see a few improvements in the soldering. Slightly shorter heating of the leg, slightly less solder added and add a half a second of after heat to allow the solder to flow better. Some form of cleaning protocol of the solder tip at intervals would remove too much solder from the tip. a tad of flux could help as well. Once more, amazing project, loving the shit out of it!!
Nice job. Thanks for sharing. Cheers from a fellow NZ er
I suspect the remaining inconsistencies in joint quality could be helped by rubbing a flux pen across the pins with the board mounted while the iron is warming up.
quite an amazing project my man. I would recommend if you are looking for better throughput to get a solder pot where you dip solder the boards and making fixturing for that.
A: change to thinner solder
B : increase the angle of the solder to the pin.
If I could do what you've done here, I'd think I was the coolest person in the world! Amazing!
Actually using real flux might help out too. Human soldering might work perfectly fine with just flux core, but actual liquid flux pre-applied to the board might help the robot.