Feeders just got a whole lot smaller
Vložit
- čas přidán 12. 08. 2022
- The LumenPnP feeders are trimming down! After testing REV05 of the feeder design, we have three big improvements we want to make: 1) smaller feeder width, 2) more secure mounting, 3) better positioning feedback. We're on a really good path to hitting all of the goals for the feeder, and once some parts come in, I'll be able to do some comprehensive testing!
You can get a LumenPnP here!: opulo.io/
A video about how Encoders work:
• How Rotary Encoder Wor...
-------------------------------------------------------------------
Check out my Patreon:
/ stephenhawes
Join the STR Discord Server:
discordapp.com/invite/TCwy6De
LumenPnP Github Repo:
github.com/opulo-inc/lumenpnp
-- FOLLOW --
Twitch bit.ly/stephen-twitch
Instagram bit.ly/stephen-instagram
Twitter bit.ly/stephen-twitter
-- GEAR --
These are affiliate links so if you buy anything through them I get a small percentage which helps out the channel. Thank you!
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 - Věda a technologie
Another good reason for having the rail retention part be removable is that over time plastic springs wear out, so you can easily replace a single part I stead of the entire housing
Also look into the latching mechanisms of DIN rails in the electrical industry, their design principles may be incorporated into the latching mechanism you're trying to develop for the feeders
Gotta say, this is some very creative engineering. Retaining small, precise repeatability like that is no small feat, especially in 3D printed fixtures.
It ought to be possible to get the original wheel encoder to work - for a reflective one it's important to get the angle of the emitter and sensor right. Other alternatives would be to put a reflective surface behind it, or maybe use capacitive sensing to detect the copper on the board. As you are only moving on one direction you can just look at transitions, which can also help eliminate variations in static reflectivity.
You may also be better off not relying on a shiny surface, as that is rather dependent on surface finish and angle - may be better to use diffuse reflection of matt white vs. gap, which produces less output but likely to be more consistent
Another option would be to take a note from mice wheels. The wheel is already set up for it with the slots that are cut into it. This would also give much more accurate and more than likely thinner.
I would use optical gates for 3D printers. They pass the light through the slots and shine directly on sensor. That is way more stable than reflections.
Nah. Magnetic orientation is far better and precise. Optical has more failure points and requires the wheel. You can get magnetic sensors now that not only detect a magnets fields strength, but its orientation. So, it can detect if the magnet is being rotated very precisely. So much better and easier than optical solutions. All you have to do is stick a magnet onto the end of the drive shaft and align the sensor to it and then store the sensor output compared to the drive wheel's orientation and you can calculate the wheels position to a higher precision than you need. Much easier to do.
Or using AS5600 magnetic encoder, it so powerfull for this kind of application. It have 12bit resolution also
Especially it is important for dual density tape (two components per sprocketl
Kool. The new design does solve many of the problems I had with the old design. I like the idea shown during the factory tour a couple of weeks ago where the feeder was passive. It only needed power and would detect the nozzle came by and advance automatically. Perfect for a simple machine.
Great job of getting the feeders slimmed. This project just gets better all the time, you deserve all the success you've all worked so hard for!
I’m happy for you bro. This looks awesome. I can’t wait to see the final machine.
Stephen -- a great step forward on the feeder design! Well done! This design is parameterized for different tape widths, right? (Or could be easily when you're happy with the design?) I'm really hoping you guys come up with some bright ideas for reducing the wasted components from getting the cover film started. Wasted components are normally negligible for a big production run, especially if started with a complete reel that has a leader. But your PnP is so well suited to small runs that wasted components are a more significant issue. Especially with ongoing chip shortages and prices! Is there some kind of "leader extender tape" one can buy to stick on to a short length of peeled cover film? And if so, that might introduce a lumpy section of leader that might challenge the film pulling gears.
You'll only need to extend the cover film - so only one or two parts are "wasted"
It's awesome, I hope you'll make it parametric for wider / thicker tapes
One thing I've done on photointerrupters... My MCU had a DAC output so I used that but you could do it with PWM and a LPF. Feed that into an op-amp wired with the positive input on your programmable voltage reference and the negative input to the voltage on a low side current sense resistor. Now you can vary the current through the LED from 0-20mA.
Sweep from 0-20mA with the photo interrupter occluded and also with it not occluded. Subtract the graphs to find the ideal current which maximizes the difference between occluded value and unoccluded value. Now just drive the LED at that fixed current.
You could do this as an automatic calibration step at boot up by moving the motor back and forth slightly.
One other change I would make would be to make the sprocket pcb all black, with a white surface behind it (or vice versa) the reflective metal surface would be like looking in mirror where you can see yourself rather than a flat colour.
@@MistahHeffo would ambient light changes be a problem?
@@TheRainHarvester typically with a photo interrupter you want to keep as much external light away as possible
@@MistahHeffo i wrote some software to first sample ambient, then sample ir led with ambient. Then some Taylor series expansion to create a curve so that i could transform then subtract to find ir-only. ( I didn't know if he did something like that. )
Another suggestion, use the same right angle motor, but with a rubberized wheel and tension mounted idle wheel to remove the protective film (think like the way filament is fed in a 3D printer), it'd decrease your need for a separate motor style, letting you increase your buying power, and be one less item of inventory
nice work on the feeder! You could also use a slotted optical sensor for the feeder gear. This would remove the motor gearbox slipping as a potential failure point that the feeder would not sense. With so many feeders on the machine at a given time it may be nice for individual feeders to detect such problems :)
I wish I had your positive energy. Great video.
This is a solid improvement. The new tape peeling setup is a huge improvement, and so is the mounting. I've been working a bit on a CL integration but this has the potential to be way better!
Always love when another video from you drops, Stephen - your enthusiasm is just so infectious! 👍🙃🙂 Great how you can quickly refine and prototype with the 3D printing too!
Looking forward to this when I'm finally ready for production.
really, it is SO nice to see you so happy with your project and excited to make more and evolve, I would love to work with ou dude, for real
I swear to God, from the thumbnail I thought this was some sort of DIY videogame based off of the feeders VHS film from 20+ years ago.
Anywho gotta love random CZcams channel feed recommendations.
If you ever do a part 3 of this series you should use a intro clip from the movie "feeders"
This video was on my suggested, I have no idea what's going on but I subscribed to see how whatever this is progresses.
Oho! 54 feeders is getting into the territory of meeting my projected needs for sometime-next-year. I'll be watching with considerable interest. (Decision time is maybe next Spring, depending what turns life & business take in the meantime.)
The cutaway with the headphones and audio at 2:52ish is very nice. 👍
Awesome video!
I don't have a clue what's going on here or how I got here but you're passionate about this, keep it up : ]
Super excited for this. I have a project with about 35 unique components and auto feeders are an absolute must.
I’m curious how my current PNP works to advance the reel and peel the tape. I’ll have to go pull that apart to look.
Very excited for this to hit the beta stage so I can try it all out
I recently watch your video from a couple years back and was surprised you were still using an optical sensor. The magnetic sensors make things so much easier. Not sure if they are the same type as you are using, but I learned of magnetic sensors that can detect not only a magnets field strength, but also its orientation. So you can align the sensor to a magnet and just spin it and the sensor will feed out changes in position as it spins. Yours looks like it may have magnets in a wheel and sense when they pass. Either way, it seems so much easier than the wheel. Optics always had issues with dust accumulation or surface oxidation or a host of other issues depending on the setup.
Heyyyooo, it's that song that I blew up your ticket system about.
Appreciate that y'all replied ... even though I ended up finding what I was looking for literally the moment after I sent that email (which is pretty much what happens to me every time I reach out and ask for help ... the answer presents itself the minute I reach out ... oof).
Thanks again!
I’ve absolutely no idea what the real purpose of the feeder is or what a PNP is, but this is really cool.
So, imagine industrial PCB manufacturing. ALL of the components need to be placed on the boards, right?
That's what a pick-and-place machine, or PNP does. And the feeder just feeds fresh components from tape so that the head of the PNP can pick it up from a known position and place it on the board.
Have you considered incorporating the retention bracket into the underside of the motor housing cover, then relocating the mounting holes on the cover to coincidence with ones used for the retention bracket, that'd potentially eliminate the need for one, or both, of the housing screws, decreasing assembly time and part usage.
I see that 3d printed hinge on your headphones. My roommates, my brother and I all have that same pair of headphones, and mine is the only one whose hinge hasn't broken yet! When I saw yours in this video, I paused it and took a moment of victory, like "I'M NOT CRAZY!"
Awesome work Stephen. When I'm settled in Barcelona in January I'll be ordering one. Oh by the way you have 100% gear interference in your CAD at 11:18. :^)
For the feeder flor: just use short cabels and a jst connector - not pretty but cheap
Congrats! That looks amazing
You could drill holes in the aluminum profile and have a spring loaded metal pin that goes in the hole. On the other side you could have pogo pins to connect the feeder.
This is super cool..
I worry about plastic creep in the film drive flexture. While I think it would work for a while, eventually the plastic is going to creep under the constant load and you will end up losing the tension needed to keep it reliable.
It should be possible to design it to accept a small coil spring for the tension. I would look at how DIN Rail mountings are done, this problem has been solved with a dozen different designs.
I hope this design will be extended to be compatible with different tape thickness too. There are so many different tapes, especially for bigger capacitors or inductors. It would kinda suck if you need to have a whole bunch of different feeders laying around, just to have enough thickness variations to set up one job.
You will probably have most high volume components on 8mm tape, but once you go forward with the wider tapes, the thickness issues become even more apparent as bigger components tend to have even more height variation.
use belts on a swing arm so that you can position motor in many locations and you can have feeders near width of reel
Another worm gear and a regular gear attatched to the PCB gear for the tape pulling motor might do the trick. Then you'd have the encoder pcb inside the feeder and you'd maybe even have an extra bit of tourque. If the Prints for the Gears are too rough you should consider making an SLA print or buying off the shelf parts for them.
9:54 No, THREADED INSERTS ARE
(Just joking, whatever works works and that method is probably a good bit cheaper, I’m just a threaded insert nut (pun?) myself. )
(You can also make threaded insert presses, basically drill press with soldering iron)
That's so cool!
Can't wait to see what you come up with next.
Also, there's a way you can eliminate the second motor entirely. Let me know if you're interested...
You could simplify the design of the feeders by using a single drive shaft running the full width of the frame that is driven by a single motor and then use a solenoid to connect the feeder to the rod as needed. It would remove the need for custom motors, simplify the electronics and reduce costs considerably. As each feeder doesn't need a motor to drive or take up the slack, it can be made as thin as you need it to be.
Your enthusiasm for this is infectious. Its pretty fun watching the builds come together even though i have exactly 0 interest in owning a pnp machine lol
Neoden has just fitted an 'upgrade ' to the K1830, that is a pneumatic raised PCB Stop for positioning the PCB on the transport rails. A filure of the Pin on the head unit failure cused this 'upgrade'
Just at the start of this, but you can use a laser or light woth a sensor. The encoder wheel can block or allow light through, now have your high and low. Laser will allow for much finer encoding. Computer mice scroll wheels sometimes use this feature.
Edit, can also use hall effect sensors. Vehicles use this alot. And go through alot more physical and heat punishment. Honestly vehicle grade stuff goes through hell and back then asks for more.
NOW we are talking!
Have you considered doing a dual tape variant? There appears to be enough room for two sets of motors offset with each other.
If you want to save more space, do 2x8 or 3x8mm feeder
Looks great.
I find PLA denatures over time though, especially unused filament get brittle. I built my first 3D printer including motherboard, around 13 3 years ago. a couple of years after that, filament became easier to source and I redid my printer with a matterhackers DELTA 3D printed parts kit. over the next few years I reprinted every part of that kit due to breakages and my redesigns. Better to print with sprues and parting planes, locator bumps and then mould and make your parts with polyurethane 2-part plastic. Or CNC them. ABS may be better but I've very rarely used it. PETg is very fussy about moisture.
Your enthusiasm is more contagious than COVID. The fact that you basically designed a tape extruder in the back is pretty bad ass. It was such a pleasure meeting you guys at MRRF and getting to experience some of the aforementioned enthusiasm IRL. Keep it up!
Huge improvement! The film puller assembly looks very fragile.
The fixing petal just needs to be turned over. Thus, it will form a rigid emphasis. Sometimes a similar one is used in the mechanism of fixing a knife.
Cool stuff! I can't help but wonder if you're not overcomplicating everything for yourself with that amount of nuts and bolts. You can make the flexure installable just with a friction fit instead of two nuts and bolts, and I'm sure you can save some hardware elsewhere too.
awesome! stephen this is a great project not so many people in my country Kenya can buy these big machines due to capital am willing to work with you and bring this affordable pick and place machine product to africa.am gonna followup on your youtube tutorials though can't afford your patreon subscription but am committed to build mine too coz I have passion on this tech
Hey Stephen i've been following for some time now but this is the first time I've knowingly seen your space mouse setup, it looks like you've modded it into a custom base. I was wondering if you could give a smidge more info or link to this design cos it looks much more ergonomic?
I guess the encoders are super cheap anyhow so maybe there is no need for a different solution. But I guess the cheapest solution would be just having two spring contacts or brushes for detecting the position of the tape wheel. Also, there are some inductive sensor ICs which are basically made for this kind of application, like the LDC0851, super simple to use and very reliable.
Now you can probably print the indexing wheel in SLA, maybe with Rigid 10k !
What about using gold plated contacts on the wheel it should be pretty reliable on low speeds and there should be no calibration
use a grab screw to tighten the feeder to the rail.
Left you a Discord on the episode, congrats on the size reduction and worm gear…
Could you put a big screw across from the back of the feeder to the rail slot? That will pushes against the rail grove
Or a slide piece with a strong spring that enters the groove and you can pull from the back
If you can get an IR LED to shine through the tape from above and an optical sensor below, you may be able to make the feeder self-adjusting on some tapes - look for a sudden change in translucency (either suddenly darker if the tape is translucent or suddenly lighter if the tape is opaque) and register off that. Or, for tapes where the drive pin holes align with the parts, sense the holes and time off that (e.g., every second or third hole is a part). Obviously this will be severely dependent on the tape, but since you're running a programmable feeder controller with bidirectional comms capability, making it possible to tell a feeder how to index itself off the tape it's feeding should be fairly trivial.
Someone else mentioned having a separate encased-magnet roller driven off the tape that can be used as an encoder, and IMO this would also be worth considering as it would remove having to have an encoder on the motor and could be used to index position off the tape itself. You could put an encoder right onto the control board.
Some tapes use clear plastic. Unfortunately this makes the tape hole sensing with IR impossible. Paper tapes yeah it works great
@@outgoingbot Clear tapes would still be detectable - the part will block IR and the tape will pass it. You'd simply have to flip the detection, and some simple initial-feed sense logic could discern whether a clear or opaque tape was being loaded.
With so much more motor torque, I'd be tempted to retry direct-displacement of the tape - since it could possibly eliminate a motor.
You should position the encoder on the wheel that engages to the tape
That was the plan.
Can you put the encoder on the drive sprocket itself ? That way you can PID the backlash of the gear train ?
Stephen please add the links to the motor and it's encoder, Thanks for sharing your work.
I have used little magnets in my manual pick and place to lock the manual feeders in the structure of the machine. 2 magnets in the feeder and 2 on the machine, they will stick very good but if you pull they come out easly in order to swap. PS: I want to develop and sell a product just so I can get a pick and place from you :)
The tape wheel with it's holes doesn't work itself like an optical encoder?
I wonder if a IR reflection sensor will be able to look up and see the holes in the edge of the tape?
Or how about an idler gear that engages with the holes like the drive gear does, and have an encoder on that?
An encoder driven by the tape itself would instantly remove all issues regarding drivetrain backlash/play, and could be used to automatically index the drive system off the tape itself if the parts pockets have any mathematically direct relationship with drive holes, e.g., if a part pocket is aligned with every third drive hole. Plus, the encoder could be built right into the control board - stick two hall-effect sensors at a 90° angle and have quadrature encoding built in. Having the encoder driven off the tape would also eliminate having to get the motor with the encoder on it.
@@OddlyIncredible Yeah that's what I'm thinking. There's a direct relationship between the hole pitch and the placement of the pockets, so it'll be the most accurate way to do it. And if you used an encoder on an idler wheel, you could match the teeth count of that wheel to the resolution of the encoder
good job reducing the size.
By the way I was looking at the lumen mobo on your website but the shop does not give much spec about it. For example what are the three USB port for on it? It is to connect the cameras? ANy documentation I missed about the lumen mobo please send a link because the shop does not point to anywhere technically informative.
3:26 What's the workbench you have? I assume it's not in the gear list because you can't buy them on amazon lol
I'm wondering if you could gold plate the reflectors to standardize the output without risking corrosion.
relying on shiny reflectors can be counterproductive as the angle needs to be right. A diffuse reflector - white/cutout is likely to be more consistent
Have you looked into "PCB motors"? May it is possible to use one of these and get even less width.
Perhaps consider a large motor powering a shaft located in the rail. Then the feeders use a clutch to engage the shaft.
Can the whole side plate be a pcb so the motors simply plug in to the board. It may save a lot of time having to wire up the motors
neat idea to use an encoder on a tiny motor. Why not use a single servo and a ratcheting pawl wheel?
Why not an optical quadrature encoder on the pcb to measure tape movement instead of a motor encoder?
He was having sensor calibration issues. Reflective sensing is a bit more finicky than through sensing. Also for some strange reason he left a reflective metal surface instead of a white surface, the shiny surface is more sensitive to the beam reflection angle and just painting it white might resolve the issue.
Wanna make it smaller? Combine two feeders into a single device. Use different orientations for motor 1 and motor 2, both engaging the drive wheels on the same axis.
That looks like a really smart solution, but why couldn't your sensor auto-calibrate with the wheel? Was it hard to spin the wheel and look for bright and dark spots? It can be done, Zebra label printers calibrate themselves in a similar way. It seems more elegant. Spin the wheel, add min and max brightness levels, divide by half and that's your threshold. Unless I'm missing something, in which case I apologize.
Yes, the calibration should not be much more than some code.
Can you make a playlist just for the feeders?
Wouldn‘t it be better to put a sensor on the indexing wheel and let the wheel move 1 pin for every ejection? Because the rotary encoder might be inaccurate after a few hundred turns
In real situations the humidy of tapes storage also affects the peel back adhesion andd some cover tapes actually break during peeling so I am not sure how this should be dealt with?
Any reason why you can’t use a set screw instead of a latching mechanism to bind the feeder to the rail? It would be super simple to incorporate.
faster swap of feeders
If you need inspiration to handle perforated film you can disassemble a old movie projector they had a lot of ingenuity and experience built in there - it's fun trust me
Whoa - chamfered holes in the pcb? Did the house do that for you or did you do it? Never seen flat screws used with a pcb like that.
Based on the captions of that section, it makes me think he did that himself, because the caption said “the next ones will be pre-drilled”
Yeah, pretty cool! Many board houses offer counterbore and countersink drilling. A smaller subset also offer Z-axis milling, too.
An alternative to countersinking is to simply oversize the holes.
9:10 Locking Feature in PETG: make a dovetail with friction fit instead of bolting
Why dont make two designs where in one designe the motor is moved backward so, from the front, the feeder electronics overlap.
what sw do u use for 3d model?
Algorithm brought me here. I have no idea what you are building but I hope to find out soon.
this is how the robot overlords while procreate......
thoughts before he show how the encode on the before the gear box works. I would say that since the gear ratio is so high on the m20 motors gearboxes, you "backlash" is going to be to the millionths degree e.g. not even close to a degree off to worry about. hehehe yea I thought so
Mount the Encoder sensor on your PCB, so you have only the Magnet on the Motor...
This is lateral thinking.
CZcams dropped you into my suggested feed.
It was interesting to watch your video but I still don't quite understand what the end product is supposed to do.
Can someone explain why the $4 passive design was dropped? (The feeder without motors)
I think the main engineering issues are that it is slow in volume production and it can misalign the pick-up nozzle from the side loading. I do not know why it is not the default or at least mentioned here.
The cynical me says that there is better profit margin in a motorised feeder but that is not a strong argument with open source designs.
@@KallePihlajasaari i guess a solenoid-extended pin to pull the tape (instead of the nozzle) would be too heavy?
@@TheRainHarvester It would take quite a solenoid and that defeats the object of the simple feeder again. It is very tempting to use the available XYZ motion to power the feeders but those axis have been designed for accuracy and speed, not heavy work.
It's a shame you can't just rotate the gearbox on the motor 90° so that the connector for the encoder pcb lines up with the main body of the feeder.
8:05 love the wallpaper, anyone knows where I found?
Just a dumb question, what are you using this feeder for. Nice concept, well executed.
Lumen PNP I think.
I would like to collaborate with you to design and build the feeder
I don't design boards I just like the engineering involved so I have no idea if 54 tape dispensers is overkill or still not enough. anyone care to chime in?
Did you ever consider using stepper motors? Not saying anything is right or wrong, just curious if there's some disadvantage to them that ruled them out.
They would still need gearboxes so much of the convenience is lost in small sizes.
Ideas for next version:
* instead of a motor with 90degree angle you could probably used regular motor with a worm gear that drives a gear, (would make it alot thinner)
* it should be possible to only use one motor by basically just adding the two gears to mesh with a a gear attached to the feeder pcb . the feeder feeds a the same speed as the tape should be removed anyway.
(edit, like you already showcases: czcams.com/video/Gm1oQjoRitc/video.html )