one machine for PCB assembly

The LumenPnP can now do it all: deposit paste, then immediately place parts! Blank board goes in, working board comes out!
Check out the LumenPnP: bit.ly/3U5m0vc
Paste Extruder repo: github.com/opu...
Leash repo: github.com/opu...
-------------------------------------------------------------------
-- FOLLOW --
Twitch bit.ly/stephen...
Instagram bit.ly/stephen...
Twitter bit.ly/stephen...
-- GEAR --
These are affiliate links so if you buy anything through them I get a small percentage which helps out the channel. Thank you!
LumenPnP Desktop Pick and Place: bit.ly/3BHFksc
PCB Holder (Omnivice): amzn.to/344qhVD
Microscope: amzn.to/348aSnw
Soldering Iron: amzn.to/2KlHDGs
Wire Strippers: amzn.to/2IESgDV
Cutting Mat: amzn.to/3oNCwOz
Flush Cutters: amzn.to/3ncFwnc
Digital Calipers: amzn.to/3gIzZSL
Leatherman Sidekick: amzn.to/3a4sQuY
Part Organizer: amzn.to/3a9uDyV
Soldering Iron Tip Cleaner: amzn.to/3a3wZiD
Oscilloscope: amzn.to/2W4zrNt
Multimeter: amzn.to/2JVp6Ba
Crimping Tool: amzn.to/37WymwP
SpaceMouse: amzn.to/3gCWR64
Tweezers: amzn.to/3a1auL5
Solder Wick: amzn.to/347kUVI
Files: amzn.to/37bMXp0
Dremel: amzn.to/3gIADj9
Headlamp: amzn.to/3gE5VYw
Pelican Case: amzn.to/3ncFpbg

[@jaredharvey1511]

The most accurate and reliable way to dispense is to use air pressure pulse. Move tip to something like 0.005in from PCB. Then provide a brief puff of air pressure. Once a very accurate dab is dispensed, the air will stop moving the paste. This is how we do thousands of caps each day for the last 30 years. You should experiment with that approach.

[@stephen_hawes]

i might even be able to repurpose the vacuum pump's outlet to do the pressure pulse! thanks for the idea!

[@leggysoft]

Do you do the old solenoid rod pokes the air pipe method?

[@RaoulEvilD]

​​@@stephen_hawes Edit: I am most probably wrong given that the volumetric control requirement is actually less stringent here than for 3D printing. Output speed requirement is way more important here, thus compressed air makes sense. Original: Working with Liquid Silicone Rubber 3D printing (paste extruders), air pulse is probably NOT the way you want to go due to lack of volumetric control. This syringe contraption is crude but actually good (consistently predictable volume dispensing). Better even would be a non-pulsating metering pump. But there is already so much more that can be done with your simple syringe with clever control algorithm that I wonder whether it is worth spending crazy money on a high accuracy metering pump for a "side project".

[@xl0xl0xl0]

@@stephen_hawes Back in my Shenzhen days, I pasted a lot of one-off prototypes with a "983A Auto Glue Dispenser" ($60 on Aliexpress). It has a knob to control the pressure, and then you can either dispense manually, or with controlled length bursts. And it even does retraction with some vacuum with a venturi tube. Highly recommend it even for prototype assembly - for small one-offs it's faster than using a stencil, and for large one-offs, you can paste and place half the board, then paste the other half, and not have issues with the paste drying while you place.

[@xl0xl0xl0]

But yes, you need a small air compressor for it.

[@eshkrab]

I think lumen wants a hot plate attachment. Just cook the solder paste in place. Shirt on point 👌

[@joshuarileymagic]

that's been what I've been thinking about ever since they built it, it's like the Voltera which I've been following

[@staticvoid1337]

Maybe with an infra-red heating bulb - I seen a guy printing PEEK on an ender with this method

[@TheCebulon]

I am so happy to have the energy and happiness of the videos in the beginning back again. Love this development and engineering content back.

[@JaenEngineering]

When it comes to precision dispensing using syringes the trick is using air pressure (pneumatics) on the back of the plunger.

[@RestartVandelay]

There is a paper called "Photo curable resin for 3D printed conductive structures" in the Additive Manufacturing journal. "In this study, researchers used acrylic as the photocurable resin, AgCu flakes, and Ag NP for the metal fillers. They prepared the mixed resin solution/paste by adding 15 wt% of ethylene glycol, 7.428 g of AgCu flakes, 5 wt% single-walled CNTs, and 8 mg of photo-initiator Irgacure-819 into 80 wt% of commercial acrylic resin followed by ultrasonication and magnetic stirring for an hour." Not sure if these are feasible on your time scale but what you are trying is definitely possible.

[@epremeaux]

assuming you have (or are willing to get) a shop compressor, I STRONGLY recommend getting a 983a digital auto paste dispenser. It attaches to the syringe with an airtight cap via tubing. So, far less mechanical load on the lumen. Basically NO mechanical stuff to add, other than the syringe holder. The controller gives you .. control .. over the dispense air pressure, dwell time and back-off pressure. So you can set how much pressure is needed to push the paste (solder paste needs much more than ink or glue for example). Then the wait time for dispensing (controls how much actually comes out), and finally it sucks the piston back a bit to control drool. for interfacing, it has a foot peddle switch input (common 3.5 mm headphone style plug). As a digital input, it reduces the code on the lumen as well (no step driving.. just trigger the logic pin and wait whatever you set the paste time for. Its really helpful for manual pasting to prevent cramping or just hand exhaustion from pressing the plunger over and over.

[@epremeaux]

its really awesome you are exploring this stuff..

[@dfgaJK]

I'm so excited to see it going with solder paste... you could have a reflow plate on the Lumen bed so it automatically reflows after... then maybe even a flying test probe to test the board once soldered, all in the one fixture!!!

[@GoughCustom]

That little stepper motor is friggin adorable! 🤩

[@ericlotze7724]

Now I’m intrigued what the “weird stuff” is! (Granted it’s been a while since I’ve checked the discord :3 )

[@cummibear69]

:3

[@cerberes]

Me too... do tell more. We like weird.

[@bastiat691]

I suggested doing this with solder paste a while back on the lumenpnp discord and got shot down :p

[@Icuosnas7]

Coming from the polymer world here, maybe if you tune the extrusion rate and filler percentage to optimize the conduction, a "film" is all you'd need if you let the UV resin cure for long enough. Alternatively, you could experiment with a reflective filler instead of graphite that conducts but honestly at that point solder paste with a re-flow is the mechanically better option.

[@Roobotics]

One thing that would be super clever is if you had a way to dock the nozzle into an airtight sheathing so it's capped back closed as soon as the operation it's needed for is done, prevent drying out and keep it performing as best it can, since it wants to actively evaporate, so basically treat it like isopropyl. Also I don't think the UV resin is going to pan out easily TBH, you sort of need a conductive grain suspended in a liquid that will remove itself from the equation and promote bonding between the grains to form conductive chains/blobs, because most conductive liquids are actually pretty bad at it! Resin hardly looses any mass when it cures, it won't promote those bonds, just freeze the grains in place(and if it was able to reasonably flow, that means they are spaced apart, so they are now frozen in that spaced apart state since next to no resin was lost), and as you found out, also get inhibited curing by the grains etc. You'll also have to deal with settling over time if the suspension isn't stable. Nitpick: 'dead reckoning' is used in relation to finding an *unknown future position* by using previous known positions, headings, and speeds. You know the intended future positions already, so there is nothing to 'reckon' here. This would be referred to as running an 'open loop' placement method, instead of the 'closed loop' of the vision system and active refinements it makes. Though technically speaking your steppers themselves are also 'open loop' without encoder feedback from skipping. Pedantic things aside, I think it's amazing you got this going already, I just knew it was coming after the prior video. For your extruder system you might need to make sure the plunger never starts spinning, else that will cause unmetered results. I'd at least put marker dots on the edges of internal things it so you can visibly observe to see if that ever does occur or the friction keeps it running straight.

[@EndlessRaidApp]

Try switching to indium tin oxide powder, It is both transparent and conductive, so potentially will allow more UV light to pass though, a bit expensive for the powder though.

[@stephen_hawes]

oooh good tip, thank you! I'll try that!

[@UnexpectedMaker]

Huzzah! That's an awesome result, well done! Yeah, forget conductive ink and get the solder paste application going. That's the golden ticket here :)

[@ross123540]

Your excitement is contagious, thank you.❤

[@1234567890CAB]

There's a manufacturing technique called wave soldering where a board is basically dragged across a pool of liquid solder. This works well with through hole components that are all on the same side. However this process can easily be adapted to SMD components by adding a drop of adhesive between the pads of an SMD component. Often it's two tiny drops along the nonconductive edges of the SMD component. Many power boards are made this way where all the through hole components are on one side and all the solder joints and SMD components are on the other side. Anyway, this syringe setup is perfect for this process; it could add the adhesive, then place the components in the adhesive, and then the board would be ready to solder.

[@maximilianmustermann7140]

Awesome, just imagine we are coming in the right direction to create something like a desktop version of the "Essemtec - All-In-One" Machine which can Solder Paste Jet Dispensing and Pick-and-Place... thats the dream!

[@mrembeh1848]

Please do not eat nema motors

[@stephen_hawes]

but it looks so delicious

[@ericlotze7724]

*If not food then why food shaped?*

[@carlmen6567]

Nema 8*

[@mrembeh1848]

@@carlmen6567 neither

[@MistahHeffo]

It's not eating it that will be the pain in the backside.

[@bytesizedengineering]

I can't stop smiling after watching this!

[@TheCmdCousteau]

Congrats on this iteration ! Very inspiring. Maybe you can try putting more ink and some retraction to suck back some of the ink back in the syringe. I think the pad should then fully covered without having too much of ink. Also could add a sort of retractable cover that moves out when the ink head goes down and comes back under the syringe to prevent any unwanted drop on the table. Keep up the good work !

[@siberx4]

If you could interleave the paste application and component placement (ie: paste the pads for a single component _immediately before_ placing the component, even after the part is picked up potentially) then you don't need to worry as much about having longer curing times for the paste. It only needs to stay workable on the board for a few seconds, although you do want to ensure it won't dry out too badly on the dispensing tip at least.

[@asynts]

NightHawkInLight made a video on KZbin called "Test Multiple Variables at Once to Optimize Anything" recently. If you have trouble finding the right composition of ingredients for the ink, that method might help.

[@HelgeKeck]

so cool. couldnt you make the traces as well out of the ink? You could add a 3d print head to the machine and then 3d print a plate with some cutouts for the traces, fill it with ink and then add some simple circuits on it 😁😁

[@pipahuwua7579]

I CAN FEEL YOUR HAPPINES, THIS IS FREAKING AWESOME

[@SpencerPaire]

I was aboit to suggest Titanium oxide (super white, so better UV transmission, but maybe not as conductive. Maybe a mix with graphite?) But ITO may be a better suggestion, haha. Also, try looking into graphite/carbon strands instead of graphite flakes; I bet having long members of conductive material will give much better bulk conductivity than hoping the flakes achieve sufficient contact. I think chopping up carbon fiber weave would be a good place to start! Though, maybe worse workability than flakes susppended in a medium. Another idea is something in the same family as superglue; an adhesive that only sets when another chemical is added. Superglue doesn't "dry", it actually sets when exposed to trace amounts of (I think) oxygen or hydrogen ions. There may be a better chemical that won't cure just when exposed to atmosphere, but instead when hit with a kicker or something. But failing the chemical pathway, the solder paste applicator would be a HUGE next step. To remove all the human action of masking and applying paste would be such a big step forward

[@feilko2170]

This guy : Builds an exotic pcb manufacturing machine that works on the first tries and is absolutely stocked of his performance. Me : Tries to print a benchy on my crappy Ender 3 and proceeds to cry and fly it out the window (the printer).

[@widelx1259]

Algorithm boy here. You could also make a little shape pattern on the back with multiple settings (back and forth on the serynge or air with different pressure). You can calibrate it quickly with it if you have the good tips (in your case, you're the first...). Python function could generate gcode like for 3d printers pattern and z axis alignement. You can calibrate a lot !

[@Chino-dy4ib]

i hope it gets an experimental option so we can all use it, would make prototyping so much faster and easier, cause then you can say goodbye to hot plates or baking, and get the prototypes made and tested way faster.

[@seanmcdonald656]

Great video! One suggestion would be to have a little area where it could purge and wipe before each job. Sort of like a 3D printer

[@Kmnri]

there are small geared DC motors with M3 shaft (3-6V that do 30-60 RPM) they work perfect with this. I have made a paste dispenser with it, has plenty of torque. And due to its size it can be inline with syringe.

[@leggysoft]

Have you heard of this method? Start with bare copper PCB, apply solder to entire PCB(it comes in a thin foil although some places just dip the board), CNC out traces, pick&place dips parts on flux, places on PCB, PCB is heated and all parts soldered in place, no moving PCB means no parts get moved.

[@nicholasryanscroggs8761]

The python library sounds awesome. Great idea.

[@CodeOptimism]

Incoming video that optimizes formulations by changing multiple variables at once like that NightHawkInLight video with Taguchi arrays? 🤩

[@FilamentFriday]

My first thought is could you use it to make a quick circuit board by laying out traces with conductive ink. It would be quicker than a PCBmill.

[@andymouse]

That product already exists, can't remember the name but it's expensive and begins with 'V'

[@FilamentFriday]

@@andymouse - Voltera V-one. I meant on his LumenPnP.

[@andymouse]

@@FilamentFriday Thanks ! and yeah I'm sure someone will try it with the Lumen :)

[@azyfloof]

Solder paste is the next logical step :D Gonna need much more torque on the output though :O

[@cxob2134]

This could eliminate the need for a stencil if you fill that thing up with regular solder paste. Paste printers do expensive but are very expensive. Very cool project!

[@agasa_workshop]

It won't be a surprise but it was quite predictable that the first drops of conductive ink woulnd't go onto the board, it is quite clear that you left a lot of play on the gear connected to the threaded rod, you can see it from the video, you can predict it if you know how you're doing this, you could close the gap a bit, use different materials to avoid friction, or even add a spring probably. Also if you wanna play around with UV resin you might find some helpfull stuff in proper printing's attempts at an FDM resin printer, it's not the same thing but he was indeed extruding resin with a syringe, you might find helpfull stuff there. That said I see many people suggest an entirely different route in any case good luck with that!

[@DaveCharnock]

I was watching this thinking the point of UV resin was to seal the whole board after populating. So don't put it in the conductive compound but somehow get a layer on after the ink has dried to secure components in place.

[@isthisuseravailable]

Have you seen the self soldering PCBs that Carl Bugeja was working on with a built-in heating element layer? It'd be pretty cool to do this with solder paste and have the board solder itself without a reflow oven

[@dontwanttousemyrealnametol6765]

Black UV inks for industrial inkjet printing require sufficient curing, and are probably formulated with photoinitiators that are different to those in less opaque inks of a different color. Different photoinitiators initiate crosslinking at different wavelengths, and not all of them work by means of free radicals, but sometimes with a cation or anion... honestly that's the extent of my knowledge, I merely operate such a single-pass printer, and there sure are lots of difficulties when it comes to high speed curing.

[@chironbramberger]

Amazing! If you built it with solder paste could you then have a heated bed to automatically flow the solder paste when it's done?

[@johnnyfp01]

Try adding 12hr setting resin as well as the uv. So that you get a skin set which holds the part with UV then the 12hr setting resing hardens the insides and sets the rest.

[@WilliamTatum]

you want to use thermally cured resins - there are initiators that are thermally reactive instead we use them in our Hercules resin so that its biocompat and skin safe. so you would hit it with a bit of uv to get the surface skin and then bake it. if you are in the US can get you a sample sent out.

[@Corebb]

About the UV resin, did you try different wavelengths UV?

[@andymouse]

Why are you not doing this with the solder paste please ? makes more sense, but great fun with the ink !

[@donaldritter7413]

At the rate you’re making progress when should we expect a spindle attachment to mill out traces on blank boards? I’ve thoroughly enjoyed all your videos and it inspired me to start learning PCB design.

[@Ziraya0]

Can you elaborate on Weird Stuff? I can imagine having a really good 2.5D motion platform like this would be useful for all sorts of things, like a university near me is frequently selling off old lab devices that automatically pipette small amounts of whatever into dozens of test tubes, they're often basically a LumenPnP but uglier, extremely single purpose, non-interactable. I'm sure it wouldn't be tricky to make an autopipette toolhead.

[@Sa3vis]

Did you calibrate the ink amount per rotation? This is so cool

[@xanokothe]

Dispenser sodering and bake the pcb with lumen then it perfect

[@AlexanderCooper-n1j]

"You wouldn't download a pick and place" LOL

[@stephen_hawes]

👀 www.opulo.io/products/you-wouldnt-download-a-pick-and-place

[@MarkGhebrial]

Have you considered using the same syringe toolhead to apply solder paste?

[@sumguysr]

Next you need an led and lens to expose traces. Then you need a sponge head to etch it.

[@39XenonD]

Is that wobbling table interfering with precision?

[@joepartlow]

Still think you should try off-the-shelf conductive epoxy.

[@lucasc5622]

Sounds nice. What guitar are you using?

[@markjones9180]

What’s the weird stuff people are thinking of doing with the Lumen?

[@lucianchapar]

One guy is somehow using a Lumen to feed aquatic animals 😅

[@andymouse]

One guy has found a bizarre method of satisfying his wife.

[@larryslobster7881]

love the tsmc t-shirt

[@fearlabsaudio7815]

I'm not at all interested in conductive ink, uv resin, or solder paste. However the real use case for this kind of thing is adhesive to hold parts upside down for a double sided smd pcb.

[@stevenkeller386]

I agree. Glue dispenser for double sided boards has immediate application. For large parts the current setup is probably good enough. Solder paste is a cool idea but I would be surprised if you could get the accuracy necessary for fine pitch parts.

[@mikesimons1544]

yep would have skipped the ink and went right to well known paste.

[@pascalh2417]

Carbon is too good at absorbing a broad spectrum. Commercial solutions use silver a lot so maybe that's a better option as it might reflect the light deeper into the blob. Loctite has a hole range of conductive silver inks like LOCTITE® ECI 1006 E&C

[@tommyfisher1435]

This machine is totally awesome!

[@hikingpete]

A heat curing resin may make more sense than a UV curing one.

[@Pixelcrafter_exe]

Why not use a pressure fed dispenser. I think it could have saved a big amount of weight and complexity and you already have pneumatics and a compressor on the lumen.

[@testsubject318no6]

Use metallic micro spheres they have a cool tape that works like that you just need to put them in a glue

[@sumguysr]

Doesn't that resin also cure when it's heated?

[@mkamphof2648]

Why not dispense the ink on the part when picked up by the lumenPNP?

[@gsestream]

you could even put a blank pcb plate in, and cnc/laser the lines out. yep one layer. or multiple layers. not trying to pick on you, but.. you only have to do the alignment calibration once. before the jobs. how about silver instead of graphite. yep you can 3d print multi-layer pcbs. well laser the copper traces in between layers.

[@CraigBurden1]

Surely there is some clear conductive material from the touch screen or heated glass industry that could be mixed into yhe UV resin rather than the lamp black? Actually what about chatting to the Positron 3D printer folks, their heated bed is clear!

[@Knatte_Anka]

Hope it comes a version for solderpaste so no stencil is needed 😉

[@maciejwierzbowski4511]

im sorry but i dont think its practical yet, it works with big foot prints, but what about lqfp qfn or maybe bga. Hope u tone it up soon, Big hopes from Poland in siplace PNP machine there is option for automatic paste hight controll, u can implement this feature soon Bro!

[@randycarter2001]

Increase the gear ratio. A small motor will drive a bigger load. Instead of <a href="#" class="seekto" data-time="61">1:1</a> go to <a href="#" class="seekto" data-time="121">2:1</a> or <a href="#" class="seekto" data-time="181">3:1</a>. You don't need speed here.

[@DUIofPhysics]

Have you seen NEMA 6 motors tho? They're basically Jellybeans!

[@mbtrimpe]

You can also try two component conductive epoxy. You’d only need to switch out the mixing tips between runs. Henkel Adhesives sells some commercial ones as well you could try starting with.

[@RJ_Eckie]

This is so freakin aweseome. Also: called it 😎

[@FruchtcocktailUndCo]

Yeah, so I'm not the only one thinking solder paste, right?

[@RJ_Eckie]

I was literally thinking like: the perfect mixture of conductive and sticky... so solder paste?

[@ResistanceLion]

Yeah, definetely don't mix it with peanut butter from your previous tests or it'll end up in your stomach :D

[@shash321]

why not just use the solder paste spreader metal paste thing ?

[@CallSignFallen]

Ngl I love this.

[@lamnguyenvan-de4xo]

It will be difficult to apply in industrial environment because Cycletime is too high for 1 circuit board.

[@VinoVeritas_]

You should contact fellow KZbinr Robert Murray-Smith. His carbon based ink is much more conductive than the other carbon based inks on the market. His KZbin channel has several videos regarding the use of the ink and even shows you how to manufacture it yourself.

[@fyfoh]

Has anyone reviewed the V4 Lumen? Thinking about getting one, but I want a tool not a hobby.

[@zachbrown7272]

LEASH MENTIONED

[@jaroslavtocek6026]

First of all, what weird stuff? If you want your resin to cure, I would suggest to not use carbon... with something reflective like silver maybe even copper you might have a chance. Look into transparent conducting films if it fails.

[@TheStuartstardust]

Nme update: LumenSNP - squirt n place (system name) or new head only squeeze n paste 😄

[@slendi9623]

<a href="#" class="seekto" data-time="186">3:06</a> me on the toilet

[@jonathantribble7013]

0202 electro Hershey kisses made by the lumenPnP

[@KacperPotoczny]

CNC kitchen have a video about curing UV resin with heat in addition to light, and it turns out, it can just cure with heat itself. Maybe heat applied to a partially cured part would allow continuing crystallization process inside the blob? kzbin.info/www/bejne/bmmZiWWdmqiFqKM But hey, I'm not an expert But yeah, to focus on soldering paste might be a way to go - even if it wouldn't achieve as high accuracy, but would be fucking awesome. Thank you for experiments! I hope you will succed!

[@3nt3_]

what is the point of this?

[@THEescapebanana]

First

[@joshuavincent7884]

I'm a little bit gay for you tbh

[@meiskam]

for your paste experimentation, there's a multivariate experimental method called orthogonal (Taguchi) arrays, that significantly speeds up testing when you want to test lots of variables. (YT video describing how it works: 5oULEuOoRd0 )