The Index Pick and Place Can Build Itself

well i just finished doing the kickstarter for the glow tie last year and i had to solder over 3 000 components by hand for it and frankly it sucked so i want to pick in place i'm going to build my own hello my goblins and ghouls my name is stephen it does it two years ago i started working on an open source pick and place project and as of about two hours ago [Laughter] and as of about two hours ago it populated its own motherboard here it is this is the first panel made on the index picking place a lot has happened in two years we've gone from doing a little bit of hacking some existing products into persuading them to pick neopixels very very roughly into a fully working machine that can populate its own boards i'm a mushy person so i'm going to try and rein it in a little bit here but this is oh no this is really cool anything that you have done to participate in this project thank you i am so grateful for your interest in this and for your help in getting it to where it is so how do we get to the point where we're able to populate this entire panel first off it was a lot of calibration in open pmp there's this awesome trick where if you put a little bit of double-sided tape onto a board you can effectively go through the entire picking and placing process and your chips will stay exactly where they're placed by the machine and you don't have to deal with all the messy solder paste in it going bad from sitting out and you can just test over and over and over with that board and it's really good for getting all your calibrations dialed in without committing to soldering it so step one was spending about two weeks configuring openpmp to work with the index and getting it to populate parts on this tacky sticky tape board this is such a good trick and it helps so much getting your configuration dialed in before you're ready to commit to actually populate boards for useful work for actually getting it to reflow so i spent a bunch of time getting all of the feeders configured the part heights and sizes and vision pipelines all calibrated getting fiducial homing set up getting the automatic nozzle exchange working pretty much everything tuning calibrated to the point where i could hit run and it would go through and populate a board with minimal intervention mostly me understanding that it's going to do what i wanted to do with small calibration issues really being the only thing that's keeping me from just letting it run after i felt really good about populating every single part on the motherboard onto sticky tape it was time to actually try it with a board and solder paste so i brought a blank panel over to the solder paste station put the stencil down got some of my solder paste and wiped it across i used this tool here to actually apply the solder paste i don't know if it's quite the right thing to use i have seen that there are some actually off-the-shelf tools that are specifically meant for applying solder paste if you know of one of these tools that you really like or you have a brand that you recommend please drop it in the comments i'd love to know what you use because this did work but it didn't feel like it was quite optimal the base went on super well and i'm really excited with the alignment of it and generally how it looks especially the pads on something like the tq fp 100 or lqfp100 the stm32 chip oh man it just looks so good i really really like the look of paste applied especially when you mount it onto the machine you can look at it with down vision it's really cool and of course all the solder face is stored in my adorable little makeup fridge that i have sitting back there i'm almost positive it uses a peltier element which is like one of the least efficient possible ways to refrigerate something but it's small and it does the job so it works for now once i had it pasted it was time to mount it to a staging plate now i could use the little board holders that i designed that have magnets in the bottom that are meant to hold a pcb at the right height for vision and placing but this is a really big board and i want something that's going to be pretty repeatable every time that i can just screw it in and it'll be right in place and also my configuration here with one index is really tight i am packing this thing full with as much as i possibly can so i don't really have room for extra space on the outside to kind of clamp it from the edges so i just designed a really simple spacer effectively that takes in a few different heat set inserts and it bolts to the staging plate from the bottom and you can bolt this panel in from the top and that works super well then it was time to just press go i had everything configured all the parts in the right place all of the vision pipeline stuff tuned as much as i could possibly make it sometimes open pnp doesn't quite remember what the camera settings should be for each camera so sometimes i'll have to reset that and then everything gets all jumbled up and i have to reset all the vision pipelines and that aside it was all pretty well tuned and off it went i made a few silly mistakes with setting up the configuration like some of the diodes were oriented 180 degrees around so there were some things like that that i realized only after the fact that they were placed incorrectly but it effectively went without a hitch after i spent all of that time setting up and calibrating with the sticky tape board i pretty much knew what it was gonna do i knew what it was capable of i knew which nozzles were gonna work for which parts based on what vacuum suction i could get on them after it was all populated i unscrewed it from the jig and put it in the reflow master and that's it this is the first panel made on the index i think part of the reason why this process of getting this panel populated and having really good consistent results with placement of things is partially because of a lot of upgrades that were made very recently that improve the rigidity of the machine a lot of these are the work of the dev stuart who made four rollers on the x-gantry which greatly reduces the amount that it rotates about the x-axis this is a lot of forces that we're actually putting on the gantry when the nozzle comes down and it pushes a little bit farther into a board or a feeder it's applying an upward force and that causes a bit of a torque around the x-axis so the stronger that is the better adding that fourth roller and a whole new tensioning system makes it so that that is rock-solid on that rail that goes across the excantry rail it's dead on there's also some upgrades i'm testing right now with the y gantry the place where the aluminum extrusion fits in just kind of goes over the sides and the top but ultimately we're trying to completely enclose it and hopefully that will help reduce rotation as well but even without that we're getting incredibly consistent picking and placing positioning over and over and over again with the camera and the nozzle it's doing really good lucian and i also suspect that adding this umbilical swivel might eliminate the problem that we had with the xy positioning a few months back when we ran a whole bunch of validation tests on the machine part of the reason we may have been getting that weird cyclical motion wasn't the air conditioner but it was the umbilical the fact that the x gantry wasn't as rigidly affixed to the rail as it could have been combined with the fact that there is a weight of the umbilical moving all over the place as it moves those things probably had something to do with the fact that we saw that sinusoidal response from the positioning i haven't rerun this test with the improvements we've made to the cad but i have a feeling it's going to eliminate it i would definitely notice a fluctuation of 200 microns over the course of using this machine and it just keeps coming back to the same position so i have a feeling it won't but i want to run the same script on this design and see what we get and because of these rigidity updates i decided to try something and pick some 0402s and it works okay so we are picking and using vision on 0402 successfully [Music] all right we're gonna test alignment see what it looks like there it is that's a no 402 on the index go to the discard location and we'll get another one there it is another freaking a 402 it just does it all day long all right we're just gonna do one more here that's so cool [Music] there it is i was able to go through and pick about two dozen three dozen o402 components in a row over and over put them over revision and align them without a single problem this wasn't even part of the original plan the original specs that i set for this machine was that it needs to be able to at least pick and place o603 so this is really cool the fact that it might actually be able to do this is awesome now we don't know for sure yet whether or not it's also able to very reliably place them we're able to pick them super well and align them with vision but i don't have any boards that use o402s i only ever use a weight of five for my designs so as soon as i'm able to get in some of a 402 boards and really put it through the paces of running it and like running jobs with it then we'll be able to validate and see whether or not it can do it but with all the signs we've seen so far it's looking pretty good another thing that this means for the project is we are almost at milestone 2 on the github wiki milestone 2 is the minimum viable product and we're just about there it's pretty cool so now that we've made one it's time to make a lot this is going to be a process of continuing to fine-tune the open pnp configuration tune the test jig and make sure that all the tests are running the way that i think they will based on my initial tests that i ran optimizing things like cycle time for solder paste application and how quickly can we run the index so that it can make boards as fast as we can make it go along with setting up another index to do ring lights and making a test jig for those as well all right that's it for this one you won't see me in two weeks i'm going to take the week off for thanksgiving but the week after that i'll be back with another video it's one you should definitely tune into because i think we're gonna have some news i have a patreon so if you'd like to help support me working on this project there's a link in the description where you can become a patron thank you so much for watching and i'll see you next time but before i go and i thank this video's sponsor pcb way we've been using pcb way for all the boards in this project so far and they always come out absolutely beautiful especially now that i'm actually going through and placing full panels of the motherboard i'm really getting to see the precision and the alignment of all of the footprints on the boards and it's really exceptional and super super cool to see how precise they can get it i've had nothing but good luck with the boards from pcb way if you're looking for a board shop i highly recommend them thank you so much to pcbway for sponsoring this video [Music] that's so cool number one serial number one and then it boops

Channel: Stephen Hawes

Views: 84,983

Keywords: stephen hawes, steven hawes, stephen hawes pick and place, pick and place, index pick and place, pick and place machine, stephen the robot, diy pick and place, index pnp, index pnp machine, pcb assembly, pcb, smt, pcb production, pcb production process, pcb production line, making pcbs at home, pick and place diy, reprap, reprap pick and place, index pnp kit, index pnp github, reprap pnp, index machines, index machines pnp

Id: DrwvorYolqU

Length: 10min 51sec (651 seconds)

Published: Sat Nov 13 2021