This channel is reminiscent of Stuff Made Here, seemingly appearing overnight with really impressive content. I predict it will experience similarly rapid growth once the videos start reaching their target audience. Great job, I will be following closely!
@@andycrask3531 I checked SMH, and it seems to be mostly crazy, semi-dangerous projects that while the boy in me loves, the adult in me sees no little practical value. I don't see egotism in SMH, just his own style.
I've been following a handful more robotics and automation design channels over the past few years, but this is different. This is explaining concepts and ideas I could only get in my Mechatronics engineering courses at uni. I hope to see a lot more videos from you, and I know I'll make use of some of these concepts in my career designing and implementing automation processes.
Key takeaways: * Pressing a timing belt pulley onto 3d printed parts is a great way to transmit torque. Thanks! * Standoff and through bolts is a great way to assemble parts that interface with other assemblies. No need for precision, it self adjusts as it clamps. In order to simplify programming you could add an optical or mechanical sensor on the individual tines of your counting sprocket. That way the stepper angle and gear ratio get factored out and don't have to be considered. No encoder, no calibration, all you need is a number of "pulses" to count. Very clean presentation, thanks for spending your time putting this together, I look forward to your future videos!
Great idea on utilizing the exiting sprocket tines for flags. I have used "u-shaped through beam photoelectric sensors " for a very similar position encoding project. How would you decellerate the load before stopping with your design?
@@aaronholwick8659 In this use case I would assume that the mass of the load is small compared to the power of the motor, and thus it is capable of decelerating well within the required positioning accuracy. Dispensing 10 SHCS took 17 frames (~.5s) with no perceptible acceleration/deceleration
if you put a worm gear you get breaking when trying to back drive it, so you don’t need to worry about the holding torque of the stepper or it unloading when powered down. you also might be able to make it more compact and mount the stepper 90°.
If anything I would say 3D printing has turned out only to be useful if you already have a hobby that involves building things - drones, rc cars, guns, crossbows, telescope mounting, etc. So far it seems that when 3D printing itself is the hobby, the only thing they ever print that's useful is more printers.
@@smorrow it can be a starting point to get into those interests. Although I do feel like it's hard to find something actually productive to do with 3d printing. I bought mine and set it up and did a few essential projects and it sat for a year lol.
Thanks for the specific shout-out; wasn't expecting that. XD If this prototype is anything to go by, the sprocket method works even better than I thought it would! I agree with avoiding a Geneva drive. Even before tolerance and speed concerns, I worry that a 3D printed version of that mechanism would have a pretty short service life.
Came here from HackaDay, and subbed. Really great engineering, and to the point without ego-massaging intro or lots of hemming and hawing. This seems to be a niche space at first glance, however in reality its the underpinning for almost all future manufacturing processes going forward.
Love your engineering! I had a similar problem once: portioning plastic granules. I had a leadscrew prototype and thought about doing it by weight. But in the end everything was too complex and I wanted an easy ans fast solution. I just did it by volume. I measured four cylinders from drainage pipe to the approximate size, attached them under a board with holes the diameter of the pipe. They were closed by flaps on the bottom. To these flaps I added dead volume to fine tune the amount but left enough slack/over-volume for inconsistencies due to compacting variability etc. We would slide a bag over each tube, close the flap, dump in the product, take a ruler to wipe off excess granules (think sintering 3d printers). Then open each flap take out the bag filled with a good approximation of the target weight.
Love the project, and sharing your progression of ideas! Yet another armchair designer here… I see there are two mechanical problems to solve: a mechanism to stop the screws from falling out, and another to overcome that mechanism for one piece. Solution could be as simple as a leaf spring to stop the flow of parts and then an actuator (say, a wedge on a solenoid, or a flap on a motor) to push one piece past that passive barrier. Something more complex like a toothed gear seems prone to getting jammed because it depends on the precise spacing of multiple parts when really you only need to engage the one part being fed. Consider that even a one-piece actuator will need its position adjusted to accommodate different parts.
This is awesome! This feels like the same type of channel that the founder of Opulo started before his company went public. Really cool mid-scale manufacturing type of machines!
Excellent stuff. I would also brace that stepper with a bracket from the rear. Having a moving mass (servo) on an arm can cause trouble down the line with play in the gears and fatigue of that arm.
Im watching all your videos in one go, just discovered the channel, I wondered why you aren't using something continuously spinning in the last vid, and now this one starts with it, amazing.
CAN bus is likely what you want to use if each module will have its own MCU. However, you could just as easily use stepper motor drivers with a I2C interfaces and just use one MCU
if you wanted to use the weight of the screws, you could replace the stepper motor and gear assembly with a brake disk connected to an encoder. PID control could slow and stop the disk and so the sprocket pretty accurately either with force from a servo motor or a 'variable voltage'/pwm driven solenoid to pinch the disk. A separate solenoid could rest with a pin through a hole in the brake disk, and turn on to pull the disk out, so its 'off state' holds the system still. It might be overly complicated for the task but could work. As someone else said, it could be good (in any system), to have a light gate or optical sensor to validate the number of screws. Also really cool video! Just found your channel.
Pretty clever Chris. Keep working for yourself if you can, you'll pollute your love for this work by subjugating yourself. I speak from personal experience.
1) you can use a line of sight sensor between the sprocket teeth to count the no of screws dispensed 2) or you can calculate what degree of rotation does one sprocket rotation needs from the DC motor and just multiply it with the number of screws you wanna get (You also need a digital pad or a screen to input the number of screws you need) You can also add an automatic packaging mechanism at the end and add a feature to make n packets of m screws and bolts
Man, respect to you from Russia, Chelyabinsk. There are few of us on the planet) Sawed-off engineers (I can see it in your eyes) If you suddenly become sad and lonely, know that somewhere in the heart of Morder is an oasis of adequacy and you will always be welcomed with love) Really good for you, just know it.
I like your way of bringing your design decisions and the respective thought process across a lot! Especially if you hint at a cool feature or point out how much you like a part - I feel that...
in this prototype your trusting that the magazine will be always full i suggest you add a TCRT5000 Infrared Reflective Sensor or something similar to count the screws !! you cant really be sure that the sprocket mechanism really grabbed a screw , maybe the motor skipped or the magazine is empty for some reason !! it should be mounted as close to the sprocket as possible , it will also act as a magazine empty signal i think that will be cool
if you're concerned about dumping the magazine if the stepper loses power, do a ratchet system with the pawl on a continuous duty solenoid that is actively being held open so if power is lost it automatically snaps forward and locks the dispenser, dropping hopefully only one or zero screws if it's fast enough. or you can have a separately powered circuit that the detects a failure on the main system and snaps the solenoid out, there's a couple ways to achieve that sort of thing. another way is also a worm drive, but that's not conducive to 3D printing due to the high contact and wear plus it'll greatly reduce the max dispensing speed, but it is 100% mechanical and you can find fairly cheap small brass gear sets online.
einfach cool wie du das umsetzt. Sobald ich mal ein paar schrauben verpacken oder sortieren muss kaufe ich deine Produkte !! Super stark bitte weiterhin so geilen Content
My PLA spline sprockets pretty reliably last 500k M6 bolts when directly connected or 4+m when using a radial spring shock absorber (basically a fancy flexible spring coupling designed to deal with radial shocks instead of vibration) I have no idea where yours will end up but i feel like i've gotten some ideas on how to improve my own
@@bamboo9840 not externally, i am a consultant and some parts of it are my IP to ensure that my clients pay He has however updated his design and i believe copying it should more than suffice for any needs you may have
The square teeth on the sprocket are going to wear early, specifically the last edge holding back the screw (and specifically screws with their sharp threads), but this is actually good and I think you should do it on purpose in the design. If you round over the trailing edge of the sprocket then the point in rotation where the screw is released will happen earlier, decreasing the latency between command and delivery, letting you move the rest position of the next screws back to potentially put them in a stronger position to bear the force of stopping the stack after dispensing, or a combination of these. It will also reduce filament use a marginal amount and eliminate almost all of the shift in latency you'd expect over the lifetime of the square tip as it rounds over. An alternate way to think about this is the two form a rack and pinion, with the screws being the rack and the pinion being the pinion. The ideal shape of the trailing edge is the involute shape for a pinion, not perfectly circular but with the point of contact lying on an approximate line (of engagement) rather than the circle of the pinion.
A worm on the stepper will also not allow the parts dispenser gear to run in reverse because all of the force would be axial on the stepper. Very nice work.
If you go with TMC stepper drivers you could use stallguard for detecting the load on the stepper. It's also being used on 3D-printers for sensorless endstops / homing.
I like your project. From an engineering point of view the project has some ingenious solutions, but if we're talking about efficiency, I think it would be more efficient to find solutions whereby you weigh them or measure the length. I think your solution is more accurate. I really like what you've done.
You can use a esp32 that uses WIFI to control a stepper TMC2208/ or DRV8825 to control the stepper motor, really simple and cheap (a pair of $ if you get the wroom) and you need low complexity to wire it. Then controll each esp32 by a raspbery pi or anycomputer with wifi by sending json to them with how many bolts you want (this is cool because you can have a response after the filling too). All boxes will have the same code and same schematics, so no complexity, just a different ID for each esp32. This also can be made by wire by controlling them with the i2c protocol or others but having wifi makes them much more modularity. Cheers! And if you want accuraccy, i suggest adding a light sensor to count the bolts as they fall if you want a second measure security, so you can count the bolts with this system and dont worry about the gear not having bolts in the buffer or the stepper lossing steps. Cheers! also a weight calculator in a hopper with all the bolts as final check will be great, a tensor sensor measure them in miligrams, very precise. And final tip, if you are scared for the stepper motor to fail during offpower, there are some motors with autobreaking feature with no power, and also there are other simple ways like blocking the gears with the actuator or a servo during offline and probably your first approach so you can run this with batteries instead of maingrid. And because I edited this message several times, I can give you my last solution to the problem, instead of having a machine box for each type of bolt, will be better to have just 1 machine with several buffers in pararell (and focus on making the magazine with bigger capacity, for example by zigzagin the bolts inside side to side) and feed the different bolts to the machine manually as the magazines gets lower or with a robotic arm that can grab them from the cartonbox reading a qr code for example. So you can save space and just build 1 machine and relay on the magazines instead of boxes like a pick and place machine.
You can also look at the ESP8266 boards which can be a little cheaper for the modules and an esp32 should have enough resources to be the server and interface depending on how fancy you choose to get. Also note, using the ESP processors gives you access to their own ESP-Now protocols for sending data between the boards significantly faster and less overhead than sending TCP/UDP packets. There is the option of wiring the boards together as serial devices which eliminate a lot of issues regarding stability/hackability, however requires additional thought on how to pass the wires/signal through modules. Wired would give you the advantage of letting the boards go into deep sleep, waking up to a pulse/message from the server as opposed to having to keep the radios on constantly waiting for a command.
I have a couple suggestions. First, include an optical sensor for detecting the rotation of the final gear. This could be tied into calibration logic to detect the rotation ratio between the stepper and the output without hardcoding (more modular). Second, install the middle gear in a spring-loaded sliding carriage with the spring pushing the B gear into both the A and C gears. This will allow for the gear to pop out of place should a jamb occur (sensing this condition would then rely on the first suggestion). Third, offset the gear contact planes, so that the B and C gears can be readily swapped for different gearing ratios. (The 2nd suggestion will aid in this process, particularly if the rotational axis is positioned above that of both the A and C gears.) Fourth, secure the stepper to the mounting frame from both the back and front in at least two opposite holes on each side. This will minimize rotational movement of the stepper body in all three axis. Fifth, include an optical sensor under the output to detect when no screw/part has been dispensed despite dispense gear rotation.
Maybe mounting an optical gate below the dispenser to count the number of dispensed screws. Basically, a tight feedback loop to cross-check reliability.
Why count them when you can weigh them? Decant straight from main supply into your smaller box or zip bag? Don't have to build/load/run machine? Faster with bigger amounts but it is a cool machine my friend 😉. 🇺🇦
you could use closed-loop stepper motor drivers, these come in boards that clip on the back of the stepper motor, and then send the step signals to the closed-loop driver with esp32 controllers. these can then be clustered together over wifi, or over cable with preprogrammed number of steps per screw, and a host controller that just sends a signal to each dispenser of how many screws it wants dispensed.
I wonder if you can throw in a couple horizontal or even uphill swings in the loading hopper to minimize the weight of the screws instead of having to load it with a separator. Some of that excess force maybe will be lost in the turns?
Just found your channel and binged all three videos. Loving what you have so far. These screw packaging videos feel like I have joined you in the middle of a project and missing the larger context. Are you planning on a video explaining the idea for the complete system? Or maybe just a short description at the beginning of the next video?
One problem I faced with stepper motors and printed parts is the heat of the motor. Steppers can easily exceed the glassing point of plastic, softening the gear and the holder.
Very cool design! Maybe I missed it, but why do you need to singulate the screws in the dispenser? Couldn't the screws be counted before they enter the dispenser using a light gate? Then you'd run the pickup wheel until the right number of screws has been picked.
You don’t need closed loop control for stepper motors. Because of how they work, they will not slowly drift with time. They might be off +/- from where they should be, but that will stay consistent over time, unlike an open loop DC motor which would very quickly loose track of its position. This is assuming that the stepper doesn’t skip any steps from being overloaded (which I doubt it would because it has such a large gear reduction already)
I like this method better than the solenoid one, seems to be a better way of doing this. You asked for suggestions, the only one I can think of is possibly using a DC motor, and a light sensor after the the sprocket 🤔 you could turn the sprocket on until a specific number of triggers of the light sensor has been completed, and then turn off the sprocket.
I’m also curious about the vertical stack design, and how universal it will be. I’m picturing that it works uniquely because you’re modeling with hex cap screws and hex nuts. Other screw types like a countersink head would not keep the screw shafts horizontal and lead to a different feed solution. What I’ve seen along these lines tends to be a mostly horizontal + sloped design that’s still gravity fed but uses a vibrator to overcome stiction and keep the parts flowing. Is this something you eliminated in your design decisioning, perhaps because you don’t have this requirement?
Another way to reduce the load on the motor from the falling screws would be to add a small offset in the path of the screws, a bit like a sloped cartoonish lightning bolt.
Off the shelf stepper with a planetary gearbox on its face attached to a larger wheel would simplify things a lot. I think the binary solenoid approach might be advantageous for certain hardware types, especially larger items.
just one thing to add to your future plans, make sure the rotation sensor is absolute, or has some way of homing, or restating the system could ba a real pain
You can put optical sensor in the end of the magazine (bottom) to count screws and use simple DC motor. Then you can make it spin until some amount of screws is counted
You could use an opto interrupter at the sprocket to track the actual rotation of it and to count the number of screws dispensed. You could also use a second one after the dispenser to make shure that the right amount of screws were actually dispensed (maybe the magazine has run dry). And to counteract the screws falling out when power is cut you could use a solenoid to clamp the screws down with some sort of rubber or foam via spring tension (i.e power applied -> solenoid unclamps the screws, power cut -> solenoid releases, spring clamps down screws)
in fact, that is exactly my plan. :) I also thought about the solenoid clamp, but I'm not sure yet if it is necessary. But I like the idea to combine sprocket and solenoid
Use a serial servo in stepper mode. They are small, have good torque and many can be stringed together. Waveshare sell a 30kg/cm one for $22ish usd. I use them for a lot of projects.
I would look into an escapement mechanism, so you don’t need to set the stepper to a specific rotation, but you can something like a solonoid again to send seperate pulses for each part ( no matter the amount of teeth the gear has). Cool project!
Hi. Is there a reason splitting bracket for the two solenoids was not considered? The top solenoid could then be positioned on a rack attached to the magazine, with incremental fix points for every fastener count. Position top solenoid at "10" for a drop of 10 fasteners. Downside is different racks/fixture brackets for every fastener type -> but already an inherited design choice from the fully constrained magazine itself.
Once you start setting up multiple dispensers, you could look at using TMC2208/TMC2209 stepper drivers like what's used on 3d printer mainboards. TMC2209s can be driven over a UART connection, so your main controller can manage them all simultaneously with few issues.
I'm wondering if you are planning a video on the hopper and the feed mechanism in the first segment of the video? I have a similar application, but entails sorting rather than counting (but counting as well). This means that the parts in the bulk hopper will vary, and I would like to feed them singly into a location where a camera can establish a bounding box to sort the part by size. I am really enjoying this video series, BTW. Great designs.
Very nice project and I really like the drawing board. I wonder if you could connect all the stacked boxes to one vertical feeder and combine this with the solenoid approach to do the mixing right there. There would be a buffering issue there which could be solved by having the geared counter turn backwards and eject the unneeded buffer. Which brings me to my second thought. Place the geared counter directly below the feeder and you can have two horizontal exits and feed two different packaging lines. detail and bulk for example. Have fun!
If you can manage to fit a worm great gear in, that would really help with holding strength I have no idea if the mechanism would become slower as a result though
With a future weight sensor, you could add a small scale inside of the counting module where you just add one of the item you'll be counting. That way, you can have the calibrated weight ready for any item you want to count without having to adjust anything. Just add one of whatever you want to count, and set the count number to calculate weight and count # with whatever mechanism you want to use, and have a solenoid to just drop everything at once. Also, why are you not counting the items as they come off of the orienting plate?
I enjoyed your video presentation because I'm interested in using stepper motors and 3d printed gears. One thing you can look into if you want a cheap setup is using 28byj stepper motors and modifying them to be unipolar to make them stronger. You get the controller board and 5 steppers for like 15$!
I would be interested in the process of 3D modeling some of the parts. From the requirements and a drawing to the finished model and the decisions like orientation, thickness etc. Considering strength, looks, printability, tolerances and so on.
I don’t know if you considered it yet, nor how well it would work in comparison to this method, but something akin to a “Rotary Air Lock” / “Rotary Hopper” could be used as well.
@@akkudakkupl The thing i am thinking of, rather than being a gear with a hole near tangent to the chute, has a Circle in-line with it. Rather than having a gear it has one (or potentially more, although I don’t know much about this) indentation in a cylinder. So every rotation it drops one part. It may be slower, although it would keep the parts better aligned potentially. May also be able to make the motor control easier, but that’s debatable. Also it may be easier to adapt to other parts/materials, you can make a different indentation for each part/material, unlike this which (at least as of now) is quite dependent on the screws being able to have that screw notch fit in there. GRANTED whatever works works, and i feel like the rotary hopper may honestly have quite a few disadvantages. But i just wanted to throw the idea out there and make sure it was considered.
Before you make 40 of them, I recommend using an optical sensor you see on 3D printers to count the teeth on the sprocket so you can always verify the dispensed amount of screws, while being able to compensate for any index drift that might accumulate over time. Stepper motors are very reliable, but not perfect, so solely relying on sending an X amount of steps every time without feedback from an external sensor or encoder is unwise. Benefit is that this will probably make your system 10x more reliable while it only costs about 3 dollars to add to the system.
If you have multiple dispensers that do not need operate simultaneously, you could make a single stepper motor run all of them via a switching transmission.
I'd add an optical sensor on the output. This allows a few things if you program them in. Off the top of my head, preventing drift, self training the number of steps for each screw, and validating the number output to either correct or flag an error state. Pi Pico is a MCU I'd suggest looking at, it's not cheap at $4 a board ($6 for the W version with WiFi and now Bluetooth), but it's a nice environment for using CircuitPython to program it, as it just shows up as a USB stick with Python files on it. The programmable IO (PIO) allow for some pretty powerful low level control, which may or may not be useful. If you don't like Python and want to do C/C++, a second Pico can be used as a debug probe and dig into it properly. You can do a wired bus, like CAN BUS or something, or if you opted for an MCU with WiFi, could opt for that. WiFi does give you the ability to do things like tying into a dashboard, or a web page that lets you control it. Could even do things like OTA updates to the code if you wanted to.
while you’re at it why not make it internet connected and use block chain technology, also slap a bit of micro services and you can use edge servers to count steps. I somehow think you can put NFTs somewhere also 😂
@@victornpb Granted having a good industrial system and being goofy buzzword vaporware are a bit different, but that is a valid criticism to an extent.
Very interesting and well accomplished. With you, 3D printing is no more for prototyping. Personnaly, I prefer the use of worm gears and not sprockets. Because it is not reversible and so it can protect your motor. But the friction is higher and so your current flow will be higher too. Need of lubrification. So you have advantages and disadvantages. Go on, you are very entertaining (nerds need you)
This channel is reminiscent of Stuff Made Here, seemingly appearing overnight with really impressive content. I predict it will experience similarly rapid growth once the videos start reaching their target audience. Great job, I will be following closely!
Yes I agree but not with the same arrogance.
@@andycrask3531 Hard to not be arrogant when you're a genius - I don't think he's arrogant either, more confident than arrogant
@@andycrask3531 I checked SMH, and it seems to be mostly crazy, semi-dangerous projects that while the boy in me loves, the adult in me sees no little practical value. I don't see egotism in SMH, just his own style.
Have never seen one of this channels videos before, but this comments is probably one of the best advertisements for a channel I’ve ever seen
Ooooh, I’m intrigued now 😊
I've been following a handful more robotics and automation design channels over the past few years, but this is different. This is explaining concepts and ideas I could only get in my Mechatronics engineering courses at uni. I hope to see a lot more videos from you, and I know I'll make use of some of these concepts in my career designing and implementing automation processes.
Thank you.
Key takeaways:
* Pressing a timing belt pulley onto 3d printed parts is a great way to transmit torque. Thanks!
* Standoff and through bolts is a great way to assemble parts that interface with other assemblies. No need for precision, it self adjusts as it clamps.
In order to simplify programming you could add an optical or mechanical sensor on the individual tines of your counting sprocket. That way the stepper angle and gear ratio get factored out and don't have to be considered. No encoder, no calibration, all you need is a number of "pulses" to count.
Very clean presentation, thanks for spending your time putting this together, I look forward to your future videos!
Great idea on utilizing the exiting sprocket tines for flags. I have used "u-shaped through beam photoelectric sensors " for a very similar position encoding project. How would you decellerate the load before stopping with your design?
@@aaronholwick8659 In this use case I would assume that the mass of the load is small compared to the power of the motor, and thus it is capable of decelerating well within the required positioning accuracy. Dispensing 10 SHCS took 17 frames (~.5s) with no perceptible acceleration/deceleration
if you put a worm gear you get breaking when trying to back drive it, so you don’t need to worry about the holding torque of the stepper or it unloading when powered down. you also might be able to make it more compact and mount the stepper 90°.
I was about to say the same!
Same
3d printing is such a great hobby - so many ideas and not enough time!
If anything I would say 3D printing has turned out only to be useful if you already have a hobby that involves building things - drones, rc cars, guns, crossbows, telescope mounting, etc. So far it seems that when 3D printing itself is the hobby, the only thing they ever print that's useful is more printers.
@@smorrow it can be a starting point to get into those interests. Although I do feel like it's hard to find something actually productive to do with 3d printing. I bought mine and set it up and did a few essential projects and it sat for a year lol.
The Solution is MORE PRINTERS!! "Laughs Maniacally!
not a hobby, it's a tool
Only 178 subscribers?? Wtf??? You are a genius!
This is excellent! Glad to see the 20T pulley hub idea is spreading into other applications.
Thank you! I just love your work!!!
Thanks for the specific shout-out; wasn't expecting that. XD
If this prototype is anything to go by, the sprocket method works even better than I thought it would!
I agree with avoiding a Geneva drive. Even before tolerance and speed concerns, I worry that a 3D printed version of that mechanism would have a pretty short service life.
Wow, I’ve never subbed so fast to a channel. Seriously, this is awesome. Keep it up man!
Your skill level and attention to details will take you far.
Thanks for sharing.
Came here from HackaDay, and subbed. Really great engineering, and to the point without ego-massaging intro or lots of hemming and hawing.
This seems to be a niche space at first glance, however in reality its the underpinning for almost all future manufacturing processes going forward.
Nice to see that at least someone is still working successfully on Wintergatan Marble Machine X.
Please keep making videos!!! This is great. I'm so glad I found this channel. Thank you!
This is exactly the kind of diy industrial engineering I enjoy. Keep it up!
Love your engineering! I had a similar problem once: portioning plastic granules. I had a leadscrew prototype and thought about doing it by weight. But in the end everything was too complex and I wanted an easy ans fast solution. I just did it by volume. I measured four cylinders from drainage pipe to the approximate size, attached them under a board with holes the diameter of the pipe. They were closed by flaps on the bottom. To these flaps I added dead volume to fine tune the amount but left enough slack/over-volume for inconsistencies due to compacting variability etc. We would slide a bag over each tube, close the flap, dump in the product, take a ruler to wipe off excess granules (think sintering 3d printers). Then open each flap take out the bag filled with a good approximation of the target weight.
Love the project, and sharing your progression of ideas!
Yet another armchair designer here… I see there are two mechanical problems to solve: a mechanism to stop the screws from falling out, and another to overcome that mechanism for one piece.
Solution could be as simple as a leaf spring to stop the flow of parts and then an actuator (say, a wedge on a solenoid, or a flap on a motor) to push one piece past that passive barrier.
Something more complex like a toothed gear seems prone to getting jammed because it depends on the precise spacing of multiple parts when really you only need to engage the one part being fed.
Consider that even a one-piece actuator will need its position adjusted to accommodate different parts.
Excellent video, looking forward to the next one!
Very enjoyable to watch. Would love to see more about the sorting and orienting of the screws and nuts.
Instant happiness when the screws are so precisely dispent🎉❤ my autism craves just that!😅
Got my sub 💪
This is awesome! This feels like the same type of channel that the founder of Opulo started before his company went public. Really cool mid-scale manufacturing type of machines!
Excellent stuff.
I would also brace that stepper with a bracket from the rear. Having a moving mass (servo) on an arm can cause trouble down the line with play in the gears and fatigue of that arm.
Im watching all your videos in one go, just discovered the channel, I wondered why you aren't using something continuously spinning in the last vid, and now this one starts with it, amazing.
Smart, using toothed pulleys as spline adapters. I shall steal this idea :)
CAN bus is likely what you want to use if each module will have its own MCU. However, you could just as easily use stepper motor drivers with a I2C interfaces and just use one MCU
if you wanted to use the weight of the screws, you could replace the stepper motor and gear assembly with a brake disk connected to an encoder. PID control could slow and stop the disk and so the sprocket pretty accurately either with force from a servo motor or a 'variable voltage'/pwm driven solenoid to pinch the disk. A separate solenoid could rest with a pin through a hole in the brake disk, and turn on to pull the disk out, so its 'off state' holds the system still. It might be overly complicated for the task but could work. As someone else said, it could be good (in any system), to have a light gate or optical sensor to validate the number of screws.
Also really cool video! Just found your channel.
Super cool project, and really well explained. Also great use of 3D printing! Keep it up :)
Pretty clever Chris. Keep working for yourself if you can, you'll pollute your love for this work by subjugating yourself. I speak from personal experience.
1) you can use a line of sight sensor between the sprocket teeth to count the no of screws dispensed
2) or you can calculate what degree of rotation does one sprocket rotation needs from the DC motor and just multiply it with the number of screws you wanna get
(You also need a digital pad or a screen to input the number of screws you need)
You can also add an automatic packaging mechanism at the end and add a feature to make n packets of m screws and bolts
Man, respect to you from Russia, Chelyabinsk.
There are few of us on the planet) Sawed-off engineers (I can see it in your eyes)
If you suddenly become sad and lonely, know that somewhere in the heart of Morder is an oasis of adequacy and you will always be welcomed with love)
Really good for you, just know it.
I like your way of bringing your design decisions and the respective thought process across a lot! Especially if you hint at a cool feature or point out how much you like a part - I feel that...
in this prototype your trusting that the magazine will be always full
i suggest you add a TCRT5000 Infrared Reflective Sensor or something similar to count the screws !! you cant really be sure that the sprocket mechanism really grabbed a screw , maybe the motor skipped or the magazine is empty for some reason !!
it should be mounted as close to the sprocket as possible , it will also act as a magazine empty signal
i think that will be cool
in fact, that is exactly my plan. :) Infrared reflective sensor also spot on.
You are genius. Nice job dude ❤
Best Drawingboard ever.
if you're concerned about dumping the magazine if the stepper loses power, do a ratchet system with the pawl on a continuous duty solenoid that is actively being held open so if power is lost it automatically snaps forward and locks the dispenser, dropping hopefully only one or zero screws if it's fast enough. or you can have a separately powered circuit that the detects a failure on the main system and snaps the solenoid out, there's a couple ways to achieve that sort of thing. another way is also a worm drive, but that's not conducive to 3D printing due to the high contact and wear plus it'll greatly reduce the max dispensing speed, but it is 100% mechanical and you can find fairly cheap small brass gear sets online.
einfach cool wie du das umsetzt.
Sobald ich mal ein paar schrauben verpacken oder sortieren muss kaufe ich deine Produkte !!
Super stark
bitte weiterhin so geilen Content
Modulsystem out of Euroboxes. Very clever. Nice work with the gear and the stepper. 👍🏻
My PLA spline sprockets pretty reliably last 500k M6 bolts when directly connected or 4+m when using a radial spring shock absorber (basically a fancy flexible spring coupling designed to deal with radial shocks instead of vibration)
I have no idea where yours will end up but i feel like i've gotten some ideas on how to improve my own
Did you share your design somewhere?
@@bamboo9840 not externally, i am a consultant and some parts of it are my IP to ensure that my clients pay
He has however updated his design and i believe copying it should more than suffice for any needs you may have
The square teeth on the sprocket are going to wear early, specifically the last edge holding back the screw (and specifically screws with their sharp threads), but this is actually good and I think you should do it on purpose in the design. If you round over the trailing edge of the sprocket then the point in rotation where the screw is released will happen earlier, decreasing the latency between command and delivery, letting you move the rest position of the next screws back to potentially put them in a stronger position to bear the force of stopping the stack after dispensing, or a combination of these. It will also reduce filament use a marginal amount and eliminate almost all of the shift in latency you'd expect over the lifetime of the square tip as it rounds over.
An alternate way to think about this is the two form a rack and pinion, with the screws being the rack and the pinion being the pinion. The ideal shape of the trailing edge is the involute shape for a pinion, not perfectly circular but with the point of contact lying on an approximate line (of engagement) rather than the circle of the pinion.
A worm on the stepper will also not allow the parts dispenser gear to run in reverse because all of the force would be axial on the stepper. Very nice work.
When the module fitted into the crate ... ♥
Wirklich sehr gut, und Ihr Englisch ist perfekt!
If you go with TMC stepper drivers you could use stallguard for detecting the load on the stepper. It's also being used on 3D-printers for sensorless endstops / homing.
Fascinating concept and content. Amazing what can be accomplished and your explanations are excellent. Excited to see what comes next.... brilliant!!
I like your project. From an engineering point of view the project has some ingenious solutions, but if we're talking about efficiency, I think it would be more efficient to find solutions whereby you weigh them or measure the length. I think your solution is more accurate. I really like what you've done.
glad you like it!
Bravo. I don't need one. But somehow I want one:-)
You can use a esp32 that uses WIFI to control a stepper TMC2208/ or DRV8825 to control the stepper motor, really simple and cheap (a pair of $ if you get the wroom) and you need low complexity to wire it. Then controll each esp32 by a raspbery pi or anycomputer with wifi by sending json to them with how many bolts you want (this is cool because you can have a response after the filling too). All boxes will have the same code and same schematics, so no complexity, just a different ID for each esp32. This also can be made by wire by controlling them with the i2c protocol or others but having wifi makes them much more modularity. Cheers! And if you want accuraccy, i suggest adding a light sensor to count the bolts as they fall if you want a second measure security, so you can count the bolts with this system and dont worry about the gear not having bolts in the buffer or the stepper lossing steps. Cheers! also a weight calculator in a hopper with all the bolts as final check will be great, a tensor sensor measure them in miligrams, very precise. And final tip, if you are scared for the stepper motor to fail during offpower, there are some motors with autobreaking feature with no power, and also there are other simple ways like blocking the gears with the actuator or a servo during offline and probably your first approach so you can run this with batteries instead of maingrid. And because I edited this message several times, I can give you my last solution to the problem, instead of having a machine box for each type of bolt, will be better to have just 1 machine with several buffers in pararell (and focus on making the magazine with bigger capacity, for example by zigzagin the bolts inside side to side) and feed the different bolts to the machine manually as the magazines gets lower or with a robotic arm that can grab them from the cartonbox reading a qr code for example. So you can save space and just build 1 machine and relay on the magazines instead of boxes like a pick and place machine.
You can also look at the ESP8266 boards which can be a little cheaper for the modules and an esp32 should have enough resources to be the server and interface depending on how fancy you choose to get. Also note, using the ESP processors gives you access to their own ESP-Now protocols for sending data between the boards significantly faster and less overhead than sending TCP/UDP packets. There is the option of wiring the boards together as serial devices which eliminate a lot of issues regarding stability/hackability, however requires additional thought on how to pass the wires/signal through modules. Wired would give you the advantage of letting the boards go into deep sleep, waking up to a pulse/message from the server as opposed to having to keep the radios on constantly waiting for a command.
Jeez i love the use of those shoulder screws for fitment! :)
I have a couple suggestions.
First, include an optical sensor for detecting the rotation of the final gear. This could be tied into calibration logic to detect the rotation ratio between the stepper and the output without hardcoding (more modular).
Second, install the middle gear in a spring-loaded sliding carriage with the spring pushing the B gear into both the A and C gears. This will allow for the gear to pop out of place should a jamb occur (sensing this condition would then rely on the first suggestion).
Third, offset the gear contact planes, so that the B and C gears can be readily swapped for different gearing ratios. (The 2nd suggestion will aid in this process, particularly if the rotational axis is positioned above that of both the A and C gears.)
Fourth, secure the stepper to the mounting frame from both the back and front in at least two opposite holes on each side. This will minimize rotational movement of the stepper body in all three axis.
Fifth, include an optical sensor under the output to detect when no screw/part has been dispensed despite dispense gear rotation.
Maybe mounting an optical gate below the dispenser to count the number of dispensed screws. Basically, a tight feedback loop to cross-check reliability.
Why count them when you can weigh them? Decant straight from main supply into your smaller box or zip bag? Don't have to build/load/run machine? Faster with bigger amounts but it is a cool machine my friend 😉. 🇺🇦
Wow. Great work. I see you become big in the 3D printing community. Keep it up mate!
This is seriously impressive. A lot of "out of the box" thinking which is very hard
Really impressive stuff, great work man!
you could use closed-loop stepper motor drivers, these come in boards that clip on the back of the stepper motor, and then send the step signals to the closed-loop driver with esp32 controllers. these can then be clustered together over wifi, or over cable with preprogrammed number of steps per screw, and a host controller that just sends a signal to each dispenser of how many screws it wants dispensed.
I wonder if you can throw in a couple horizontal or even uphill swings in the loading hopper to minimize the weight of the screws instead of having to load it with a separator. Some of that excess force maybe will be lost in the turns?
First time watching, very impressed by your engineering and your video creating skills!
Looks smooth, I like it
Just found your channel and binged all three videos. Loving what you have so far. These screw packaging videos feel like I have joined you in the middle of a project and missing the larger context. Are you planning on a video explaining the idea for the complete system? Or maybe just a short description at the beginning of the next video?
Glad you like it and great idea! I think I will make a separate video about the hole system.
One problem I faced with stepper motors and printed parts is the heat of the motor. Steppers can easily exceed the glassing point of plastic, softening the gear and the holder.
Very cool design! Maybe I missed it, but why do you need to singulate the screws in the dispenser? Couldn't the screws be counted before they enter the dispenser using a light gate? Then you'd run the pickup wheel until the right number of screws has been picked.
New subscriber here, I work in automation and manufacturing and believe your onto a great thing here, keep up the brilliant work.
You don’t need closed loop control for stepper motors. Because of how they work, they will not slowly drift with time. They might be off +/- from where they should be, but that will stay consistent over time, unlike an open loop DC motor which would very quickly loose track of its position.
This is assuming that the stepper doesn’t skip any steps from being overloaded (which I doubt it would because it has such a large gear reduction already)
BRO I LOVE THIS
This is some really amazing work, well thought out and very professional!
I like this method better than the solenoid one, seems to be a better way of doing this.
You asked for suggestions, the only one I can think of is possibly using a DC motor, and a light sensor after the the sprocket 🤔 you could turn the sprocket on until a specific number of triggers of the light sensor has been completed, and then turn off the sprocket.
I’m also curious about the vertical stack design, and how universal it will be. I’m picturing that it works uniquely because you’re modeling with hex cap screws and hex nuts. Other screw types like a countersink head would not keep the screw shafts horizontal and lead to a different feed solution.
What I’ve seen along these lines tends to be a mostly horizontal + sloped design that’s still gravity fed but uses a vibrator to overcome stiction and keep the parts flowing. Is this something you eliminated in your design decisioning, perhaps because you don’t have this requirement?
Excellent content, great design and really good and crystal clear explanation. Thank you!
This is genuinely so impressive, really brilliant system idea 👌🏻
Another way to reduce the load on the motor from the falling screws would be to add a small offset in the path of the screws, a bit like a sloped cartoonish lightning bolt.
Off the shelf stepper with a planetary gearbox on its face attached to a larger wheel would simplify things a lot.
I think the binary solenoid approach might be advantageous for certain hardware types, especially larger items.
After I watched the first video my first thought was that a servo driven wheel would be more appropriate :)
Gem Channel! Please keep on producing videos!
just one thing to add to your future plans, make sure the rotation sensor is absolute, or has some way of homing, or restating the system could ba a real pain
You can put optical sensor in the end of the magazine (bottom) to count screws and use simple DC motor. Then you can make it spin until some amount of screws is counted
Why not use a worm gear to both hold and prevent the screws from destroying the bearings by falling?
You could use an opto interrupter at the sprocket to track the actual rotation of it and to count the number of screws dispensed. You could also use a second one after the dispenser to make shure that the right amount of screws were actually dispensed (maybe the magazine has run dry). And to counteract the screws falling out when power is cut you could use a solenoid to clamp the screws down with some sort of rubber or foam via spring tension (i.e power applied -> solenoid unclamps the screws, power cut -> solenoid releases, spring clamps down screws)
in fact, that is exactly my plan. :) I also thought about the solenoid clamp, but I'm not sure yet if it is necessary. But I like the idea to combine sprocket and solenoid
Fascinating. Thanks for this :)
Use a serial servo in stepper mode. They are small, have good torque and many can be stringed together. Waveshare sell a 30kg/cm one for $22ish usd. I use them for a lot of projects.
I would look into an escapement mechanism, so you don’t need to set the stepper to a specific rotation, but you can something like a solonoid again to send seperate pulses for each part ( no matter the amount of teeth the gear has). Cool project!
Great analysis and project.. very well performed...
Now i just need a project of counting screws...
This is very reminiscent of stuff made here crossed over with wintergatan and then a little bit of tom Stanton, perfect combo
Hi. Is there a reason splitting bracket for the two solenoids was not considered? The top solenoid could then be positioned on a rack attached to the magazine, with incremental fix points for every fastener count. Position top solenoid at "10" for a drop of 10 fasteners.
Downside is different racks/fixture brackets for every fastener type -> but already an inherited design choice from the fully constrained magazine itself.
Once you start setting up multiple dispensers, you could look at using TMC2208/TMC2209 stepper drivers like what's used on 3d printer mainboards. TMC2209s can be driven over a UART connection, so your main controller can manage them all simultaneously with few issues.
I'm wondering if you are planning a video on the hopper and the feed mechanism in the first segment of the video? I have a similar application, but entails sorting rather than counting (but counting as well). This means that the parts in the bulk hopper will vary, and I would like to feed them singly into a location where a camera can establish a bounding box to sort the part by size. I am really enjoying this video series, BTW. Great designs.
Very nice project and I really like the drawing board. I wonder if you could connect all the stacked boxes to one vertical feeder and combine this with the solenoid approach to do the mixing right there. There would be a buffering issue there which could be solved by having the geared counter turn backwards and eject the unneeded buffer. Which brings me to my second thought. Place the geared counter directly below the feeder and you can have two horizontal exits and feed two different packaging lines. detail and bulk for example. Have fun!
You could use an optical sensor around the sprocket teeth to sense if it is turning the correct amount
If you can manage to fit a worm great gear in, that would really help with holding strength
I have no idea if the mechanism would become slower as a result though
Amazing video. Can not wait for the next video
With a future weight sensor, you could add a small scale inside of the counting module where you just add one of the item you'll be counting. That way, you can have the calibrated weight ready for any item you want to count without having to adjust anything. Just add one of whatever you want to count, and set the count number to calculate weight and count # with whatever mechanism you want to use, and have a solenoid to just drop everything at once. Also, why are you not counting the items as they come off of the orienting plate?
I enjoyed your video presentation because I'm interested in using stepper motors and 3d printed gears. One thing you can look into if you want a cheap setup is using 28byj stepper motors and modifying them to be unipolar to make them stronger. You get the controller board and 5 steppers for like 15$!
I would be interested in the process of 3D modeling some of the parts. From the requirements and a drawing to the finished model and the decisions like orientation, thickness etc. Considering strength, looks, printability, tolerances and so on.
I don’t know if you considered it yet, nor how well it would work in comparison to this method, but something akin to a “Rotary Air Lock” / “Rotary Hopper” could be used as well.
Interesting, I never thought about that.. will think about it
Isn't this basicaly it, but for single screws at a time?
@@akkudakkupl The thing i am thinking of, rather than being a gear with a hole near tangent to the chute, has a Circle in-line with it.
Rather than having a gear it has one (or potentially more, although I don’t know much about this) indentation in a cylinder.
So every rotation it drops one part.
It may be slower, although it would keep the parts better aligned potentially. May also be able to make the motor control easier, but that’s debatable.
Also it may be easier to adapt to other parts/materials, you can make a different indentation for each part/material, unlike this which (at least as of now) is quite dependent on the screws being able to have that screw notch fit in there.
GRANTED whatever works works, and i feel like the rotary hopper may honestly have quite a few disadvantages.
But i just wanted to throw the idea out there and make sure it was considered.
@@ericlotze7724 Okay, I think I get your idea. The tool needs to be engineered for a specific part anyway ;-)
Before you make 40 of them, I recommend using an optical sensor you see on 3D printers to count the teeth on the sprocket so you can always verify the dispensed amount of screws, while being able to compensate for any index drift that might accumulate over time. Stepper motors are very reliable, but not perfect, so solely relying on sending an X amount of steps every time without feedback from an external sensor or encoder is unwise. Benefit is that this will probably make your system 10x more reliable while it only costs about 3 dollars to add to the system.
If you have multiple dispensers that do not need operate simultaneously, you could make a single stepper motor run all of them via a switching transmission.
I'd add an optical sensor on the output. This allows a few things if you program them in. Off the top of my head, preventing drift, self training the number of steps for each screw, and validating the number output to either correct or flag an error state.
Pi Pico is a MCU I'd suggest looking at, it's not cheap at $4 a board ($6 for the W version with WiFi and now Bluetooth), but it's a nice environment for using CircuitPython to program it, as it just shows up as a USB stick with Python files on it. The programmable IO (PIO) allow for some pretty powerful low level control, which may or may not be useful. If you don't like Python and want to do C/C++, a second Pico can be used as a debug probe and dig into it properly.
You can do a wired bus, like CAN BUS or something, or if you opted for an MCU with WiFi, could opt for that. WiFi does give you the ability to do things like tying into a dashboard, or a web page that lets you control it. Could even do things like OTA updates to the code if you wanted to.
If you wanted to go overboard, you could use DIN Rails and the Arduinos/Raspberry Pi’s designed for those too!
(Granted *for a cost* )
If you want to go wireless you could probably use MQTT to communicate between all the mcu’s
while you’re at it why not make it internet connected and use block chain technology, also slap a bit of micro services and you can use edge servers to count steps. I somehow think you can put NFTs somewhere also 😂
@@victornpb Granted having a good industrial system and being goofy buzzword vaporware are a bit different, but that is a valid criticism to an extent.
very cool concept and execution!
A timing belt pulley as an insert for a plastic part is very clever indeed.
As a part counting goes, nothing beats camera and computer vision. It is fast and modular. Counting the parts one by one will take an eternity.
Impresive ! Great material very good video
Very interesting and well accomplished. With you, 3D printing is no more for prototyping.
Personnaly, I prefer the use of worm gears and not sprockets. Because it is not reversible and so it can protect your motor. But the friction is higher and so your current flow will be higher too. Need of lubrification. So you have advantages and disadvantages. Go on, you are very entertaining (nerds need you)