Here are links to some of the equipment shown in the video: ESP32 M5Stack - amzn.to/2J67IIg Object Tracking Pixy Camera - amzn.to/2LTMP0y Braided Cable Sleeve - amzn.to/2LedCnb Stainless Steel Cable Ties - amzn.to/2LevZsb Strain Relief Cable Glands - amzn.to/2LdnuO8 Silicone Fusing Tape - amzn.to/2J5RPO8 Locking Casters - amzn.to/2kJ75WB Aluminum Extrusion - amzn.to/2Lh14M7 Cable Clips - amzn.to/2xColG8 M5 T Slot Nuts - amzn.to/2HdnNWP M5 Screws - amzn.to/2J4OSBs LED Ring Light - amzn.to/2sqeIFs Carbon Fiber Tubes - amzn.to/2J5xVqz Suction Cup Nozzle - amzn.to/2LQFrDw

@JeremyCook

6 жыл бұрын

Any idea where I can find the metal buttons you're using? Even better if they have lights around them :-)

@ProtoG42

6 жыл бұрын

Read the product descriptions carefully. Some are latching, some are momentary. Some have solder leads, some have screw terminals. Here is the button from the video: amzn.to/2Mk9wvm Here is one with a led power symbol: amzn.to/2Mm7daV Here is one with a ring light: amzn.to/2MkJkAS

@JeremyCook

6 жыл бұрын

Nice, thx.

@waelwael2869

5 жыл бұрын

Any idea where I can find the ball joints you're using?

@calvinbarajas5144

5 жыл бұрын

Amazing video and project Anthony, thanks very much for sharing. Can you provide the links to purchase the motor driver modules and AC/DC enclosed power supplies please.

you going to pick and place cars?

Those are huge steppers!! You could almost start a car off them. Nice build!

Nice Delta Robot build. Very impressive!

Those are some serious steppers! Great work.

How could anyone dislike this? This is awesome!

It's super cool to see this moving. The speed is definitely impressive.

Awesome project! I wonder if that back plate is contributing to the noise as well. Maybe something like foam acoustic paneling on the back could reduce vibration.

I don't know why, but that thing looks terrifying lmfao

Looking good!

Wow awesome work👍.

I can already see this sold out in the japanese adult market. Boi you going to b rich

Very cool!

The delta robot system and explanation were very good, but I have some doubts: 1. How to define the reach of the delta robot? 2. How to select the rod length based on reach? 3. How to define the payload? 4. How to define the picks per minute? If you guide me on this, it will be very helpful.

COOL project! what about using Something like and old Control panel from some old industrial machine to control the robot?

Very cool.

Microstepping torque does not reduce the torque. Amazing build but can’t believe the noisy !!! Delta robots in industry are virtually silent in operation Cheers from Australia

@ProtoG42

4 жыл бұрын

Industrial delta robots use motors and strain wave gear boxes that are thousands of dollars each. It depends on the motor geometry and shaft position but as far as I know, microstepping does reduce holding torque. 1. www.machinedesign.com/archive/article/21812154/microstepping-myths 2. www.faulhaber.com/fileadmin/user_upload_global/support/MC_Support/Motors/AppNotes/Faulhaber_AN015_EN.pdf 3. hackaday.com/2016/08/29/how-accurate-is-microstepping-really/

@danrudmin6797

Жыл бұрын

Square stepping usually gives slightly higher torque because you are operating in the corners of the phase diagram, but nothing close to the values you were quoting in the video. Microstepping typically circularizes the phase plot so that you get a uniform torque at every microstep position. This reduction is 1/sqrt(2) or about 70% and it doesn't decrease with further microstepping.

Looks damn impressive so far! Can't wait for the finished result :)

Been following the project and love it. What will it be used for?

Do you have any ideas on how you might counter-balance the thing? It seems like it's generating some pretty hefty forces.

wow i cant say iv ever heard motors that loud

Imagine that on a mobile platform reaching around at intruders.

Great project! While you can combine a incremental encoder and an end switch, what you really want for this application is an absolute rotary encoder, they don't lose position even after power loss. Optical absolute rotary encoder does exist, but the magnetic ones are cheaper, easier to use and very accurate. Check out the AS5048 from ams or the MA730 from mps.

@ProtoG42

6 жыл бұрын

Tat Wong thanks! I have been looking for an encoder I’ll look into those! I have found a lot of single turn absolute encoders but they are definitely the most expensive part of this project. The rear shaft is 14mm so if I can’t find anything to match that, I can build a coupler or use gears and/or belts.

@topQuark12

6 жыл бұрын

Nice thing about the magnetic encoders is the ease of use. I just super glue a small disk magnet to the end of the motor shaft, mount the rotary encoder ic 1-2mm apart from the disk magnet in parallel and that's it. They are also super cheap compared to the traditional ones.

@ProtoG42

6 жыл бұрын

Thanks, I'm definitely going to give one of those a try! Do you have a link to the disk magnet you've been using?

@topQuark12

6 жыл бұрын

I bought my magnet locally at a shop, so I don't have a link. Though, it is nothing fancy, just a neodymium magnet with 5mm diameter and about 3mm thick, I think one of the datasheets recommended that size of a magnet.

@ProtoG42

6 жыл бұрын

Tat Wong I spoke with the senior application engineer at MPS and he told me that .25”x.125” n42 magnets I have will work. For custom applications, they can tell you what the optimal distance from the magnet to space the sensor. I have a few evaluation kits on the way! Thanks again for the tip! I think those will be perfect for this application and much more affordable than optical encoders.

Very cool build. My 2 cents on the clunking noise is the backlash in the stepper gearboxes. It's intrinsic to spur gearboxes. You need something like a strainwave gear or a roller worm drive to get zero backlash but they're real pricey. You can probably tune a lot of the clunk out though like you said

@ProtoG42

6 жыл бұрын

You are absolutely right. The gear boxes are planetary spur gears. I could not afford "zero backlash" gear boxes as they were extremely expensive. Even moving the shaft with only a linkage attached and no power, it is still loud and "clunky" but I'll try to minimize it as much as I can. Perhaps I'll build some sound proofing around the motors like the casings on industrial delta robots.

@ProtoG42

6 жыл бұрын

Do you have any recommendations for low end strainwave gearboxes for NEMA 34 stepper motors?

@lawmate

6 жыл бұрын

I've only seen them as discreet gearboxes and they're more than £1k new. I see them 2nd hand on ebay for less but theyre still pricey. From korea and china mainly. If you know anyone in china that can send from taobao they are even cheaper. Harmonic drive is the brand with the patent. I did a project with a reduction worm drive in a large robotic installation that we got lots of harmonic vibration from the worm gear backlash. As a bodgy fix we added a spring to provide a constant torque in one direction and made sure the motor acceleration torque didn't exceed the spring torque but it wastes a lot of overall torque. Since then I've always used timing pulley reductions for zero backlash but they take lots of room if you want a significant amount of reduction. I'm waiting for a project where i can justify getting some strainwave gears though. They're real cool. Some industrial robots have them in. Maybe that could be another cheap source?

@lawmate

6 жыл бұрын

The other option maybe for nema34 is to use the hybrid steppers. I've used them and they have massive amounts of torque. The driver doesn't have current limiting like a open loop stepper driver but will dynamically adjust the current. I think it's in a PID loop and uses an optical encoder for feedback. They're pretty hefty on their own and if there's a temporary stall they will still move to the postion they should be after the stall. They're more like a servo in that respect but way cheaper than a normal 3 phase servo. They're pretty cheap from china.

wonderful!!!!, I have some questions. How did you control this robot?. What is controller? Thank you!

What is resolution of the movement? Can your robot used to place SMD components on pcb?

Greetings. First of all, congratulations on your magnificent and impressive project, it's beautiful. I would like to ask you if the pap motors have a gearbox and what model are those motors. Thank you, I really enjoyed watching your video.

Awesome job - sounds horrible at the moment, looking forward to the next video.

Could you take some weight out of the main linkages? It's rotating mass, you can gain a lot from the reduced moment of inertia. Not like it particularly needs it haha, that thing is fast

@ProtoG42

6 жыл бұрын

I could and may reduce the weight at some point but I don't think it's that necessary right now as the speed is great with plenty of torque left. They weigh 3.7 ounces and I could probably get it down to 2.5 ounces. The motors with the gearboxes have 8,000 in-lbs of torque so they are pretty happy with 3.7 ounces :)

Great work! I am also working on a delta robot, but have primarily been working on the software so far. Are you using gearing on these steppers? If so how did you determine what ratios you wanted?

@murraymadness4674

3 жыл бұрын

What software? Thinking of building one, but it can't use the vertical delta software as the arm movement is in an arc..

Can you cover more on the stepper motor exceleration profile, just curious why they were so loud an would like a more in depth explenation of how to make them quieter.

@kubiedubie

6 жыл бұрын

Finn Suchara the loud noise is due to trapezoidal motion profiles. S-curve (aka cubic) type profiles will prevent this.

Why did you not go with linear actuators? I believe using that configuration would reduce the noise significantly and probably higher resolution, though lower speed? I could be wrong though, please give me your input.

It is awesome!! Could you please share the CAD files?

hi bro, your project it's amazing! I'm trying to build a delta robot as well but i have some isues to fix the spherical joints to the lower bars. with the thread the length of the bars are different. How did you do??

Would like to know what size those steppers are

Please do you have a paid course for project of the Delta robot? I'm really interested in learning from the way you design this it's perfect in my opinion.

Could you please explain how did you manage to connect upper link to the lower one? I've seen you use a sort of rod-end bearings on the lower rods. But how, precisely, you fixed lower links to the upper ones? I'm involved in a delta robot project and I'm trying to find a good solution to this issue. Thanks if you'll decide to answer me. Anyway, good job!

Just as an FYI, your audio is a little low on this one. Great video otherwise, looking excellent!

@ProtoG42

6 жыл бұрын

Thanks! I'll bump up the gain on the microphone a little bit next time.

Very cool. I haven't watched to entire series yet but I do have a question. When the arms reach to the fullest extent then 'reach' down to pick up a part, it doesn't appear to drop exactly perpendicular to the table. It appears to grab the part at a 45 degree angle. Is there a way around that? In the video, you mentioned that a part could get bumped when it's being grabbed. Wouldn't that be the case here?

Does anyone know what type of tie rod ends are those or where to get them?

Have you thought about extending the IO using an SPI IO Port Expander. There are many all over the internet. You can also use SPI and connect to an Arduino SAM to extend the capabilities and make your system more modular. You may want to ditch those POTs (They will not last long as they are carbon based and it also mean your accuracy will be no good as you will have to rely on the ADC while rotatory encoders give more pulses which translate to higher resolution for point when your system has to learn its position). My idea is to think more modular and not to have one controller to do everything. later on when your finalize your design, you can put together one controller with all the different modules if you so please

Hi. Are you planning to release the planes for the robot?

Hello!! So, it`s possible to make a 3D printer without linear rails/rods and timing belts with this type of kinematics, right?

Great What is the software used for driving this ?

That horrible sound is the backlash kick inside those cheap gearboxes. That's why I never use them in these type of applications.

Hi can i ask for lists of component u buyed

Hi friend. What motors did you install? What is the gear ratio? thank

Is there a way you would share the code files you used?

Nice project! Have you already checked if the ADCs from the esp32 are good enough for your use? I think they are pretty inaccurate so youre maybe more happy with encoders for the manual positioning. In my project i used a separate I2C ADC to get better rusults.. its called ADS1115. Cheap and easy to use. I also chained it to an analog multiplexer to read up to 16 pots with it.

@ProtoG42

6 жыл бұрын

randomname thanks! I spoke with an application engineer at monolithic power systems and I’ll be using non contact absolute encoders on the output shafts so I can use that for my teaching mode as well. Once in this mode, you can move the robot by hand to teach steps.

links parts solidwork????/

Do you plan to add a cover to the top? Seems kinda dangerous. Otherwise great build!

@ProtoG42

6 жыл бұрын

Probably not, as I will be the only one using it and don't intend on selling it. If I were to ever install in the manufacturing facility I would definitely need to put a cover on it though. Thanks!

Great project. Are you sure that microstepping reduces motor torque? I learned that it does not

@ProtoG42

4 жыл бұрын

Where did you learn that? It depends on the motor geometry and shaft position but as far as I know, microstepping does reduce holding torque. 1. www.machinedesign.com/archive/article/21812154/microstepping-myths 2. www.faulhaber.com/fileadmin/user_upload_global/support/MC_Support/Motors/AppNotes/Faulhaber_AN015_EN.pdf 3. hackaday.com/2016/08/29/how-accurate-is-microstepping-really/

Hi, Can you tell me how you calculated the torque for each motor of this Delta robot?

@CapApollo

4 жыл бұрын

biggest motor available.. nothing to calculate...

I guess you should wait till you have limited switches or encoders, but I'd love to see you push the speed, acceleration, and jerk till you started losing steps. Edit: God damn auto correct. Limit switches

Hanzhen harmonic gear , robot arm gear , over 30 years experience , joint gear

How much torque do those motors have?

Do you use reductors?

Speed is good, how about precision?

If you wat to call it high speed you need to ramp it up - considerably. 😉

... much to strong motors (in my opinion) for such a light mechanic. NEMA23 with low inductivity will work 5 ... 10 times faster. (and silent) To get the gear backlash complete out of the system you can use a radial spring for a asymeteric load. These spring have to be some times stronger then the max power you need ... and 75% the motor can deliver. Mechanic will last very long if you never use Start Stop Mode with the Motors. Use acceleration ramps .... ! I have such a DeltaRobot. In work you can not see the Delta-Arms.

Why do you need all those PSUs? What’s the total current draw at (and what voltage)?

@ProtoG42

6 жыл бұрын

Aaron Beckman Three 60v 6A power supplies are cheaper and more reliable than one 60V 18A power supply.

@aaronbeckman

6 жыл бұрын

Proto G makes sense, thanks.

Да ты крутой 👍👍👍

what exactly is this for ?

@ProtoG42

6 жыл бұрын

It's a project for me to learn more about machine vision. It will have several uses including sorting different colored objects as quickly as possible.

I'm a mechanical engineer, but I feel like I lack the knowledge currently to do something like this. Aside from mechanical design, what do books or topics do you suggest I look into?

@ProtoG42

6 жыл бұрын

asdfghbvxxxv google patents. Reading public available patents will teach about machines and what obstacles the engineers had to overcome and why. Some patents are very detailed with very specific information. Honestly, the best way to learn is to make a lot of cheap robots and that will help build a good intuition between mediocre parts and great parts. You can always do static and dynamic calculations before choosing parts but don’t over think it. Just dive in!

@JeremyCook

6 жыл бұрын

If you work in manufacturing there's a very good chance you'll end up learning a lot of this type of technology.

двигается то быстро, но инерция такая, что весь каркас ходуном ходит, нужна жесткая, обязательно сварная или лучше литая рама, центр тяжести еще и вверху оказался.

@nurlybekmoldagaliev8920

4 жыл бұрын

Ещё и "некислое" преобразование координат😉

Peter Griffin - "Shipoopi, Shipoopi" 4:10

What is the aim for this robot?

@ProtoG42

6 жыл бұрын

This is purely a hobby project, but I manage a manufacturing facility and something like this could one day easily be adapted for real applications such as sorting metal parts coming off an assembly line. Initially, I will just have it sort different colored marbles.

Why are the motors so? Large

An encoder will catch steps, but it can’t, traditionally, give you an absolute position. A limit switch will do this for you. Also, are you sending a simple hi:lo or are you ramping up speed for trapezoidal motion control?

@ProtoG42

6 жыл бұрын

That depends on the encoder. Since the motor doesn't ever make a full revolution, it's a little easier to give an absolute position. I am accelerating and decelerating for trapezoidal motion control. Still tweaking the best values though.

@RobertSzasz

6 жыл бұрын

Absolute encoders aren't that much more expensive compared to the overhead of relative positioning and limit switches.

@ProtoG42

6 жыл бұрын

Robert Szasz I will be using a non contact absolute encoder on the output shaft where the main linkage is.

@kubiedubie

6 жыл бұрын

Try giving cubic motion a shot. Almost all industrial motion applications use cubic profiles to reduce instantaneous jerk and reduce mechanical wear.

@frtard

6 жыл бұрын

For a delta like this, I'd use precision potentiometers directly on the arms themselves. It might be a simpler, more flexible (and potentially cleaner) solution than limit switches.

peep the oscilloscopes in the background. u gonna talk about them or naw

@ProtoG42

3 жыл бұрын

There is only one oscilloscope in this video. The rest are different types of test equipment.

@v1Broadcaster

3 жыл бұрын

ooo what are those on the shelf?? got me curious

can you show more smooth acceleration?

@ProtoG42

6 жыл бұрын

In the current state of the machine, smoother acceleration sounds worse because it spends more time in the lower frequency zone when there are small movement. All the motors are in sync so they try to reach the desired location at the same time. Because of this when one of the motors travels a much farther distance than the others, the others slow down quite a bit to reach the same destination at the same time. I may just turn that off and have them reach their destination as fast as they want for a smoother motion.

I don't understand why it's so noisy 🤔

😱😉🥰🤩💥💯👍💪👏

#dopetech

Tem que melhora as investigações , e por os políticos ladrões na cadeia pro mundo evoluir

"microstepping reduces torque" is a misunderstanding. it really doesn't. torque is proportional to load angle, NOT to how finely you can control your magnetic field. that's where the misconception comes from.

@ProtoG42

5 жыл бұрын

It really does though. It reduces holding torque most of the time. It depends on if you are aligned with a "full step" or not. Having built my own stepper motors, there was a big difference in the torque of half steps and full steps

@crackwitz

5 жыл бұрын

field oriented control (microstepping), where you move _on_ the unit circle, can't be sensibly compared to full stepping, where you move _off_ it and get 141% force. but you didn't say that in the video. 5:04, "if you do 1/8 microsteps, you're only at 19.5% of the original motor's rated torque. now when you want 1/16 microstep like I am here, you're only at about 9% of the original rated torque" again, that's INCREMENTAL torque, i.e. the same as (or proportional to) load angle. nothing is reduced here. I've been dealing with microstepping drivers for a while. if there's a physically grounded explanation for the "more microsteps reduce torque" myth, nobody was willing to point me to it. I'm very eager to be shown wrong, please explain. I realize I won't be able to convince you that it's a misunderstanding.

@ProtoG42

5 жыл бұрын

​@@crackwitz You're not totally wrong. Your statement that "nothing is reduced" is wrong. The normal stalling torque of the motor while in motion is negligibly affected by the microstepping but the holding torque is indeed affected which can easily cause missed steps when starting and stopping the motor as well as holding position. I've built several machines that work fine at 1/16 microsteps but will start skipping steps at 1/32 microsteps or higher. "The real compromise is that as you increase the number of microsteps per full step, the incremental torque per microstep drops off drastically. Resolution increases but accuracy will actually suffer. Significant too is that any load torque will result in a “magnetic backlash”, displacing the rotor from the intended position until sufficient torque is generated." www.faulhaber.com/fileadmin/user_upload_global/support/MC_Support/Motors/AppNotes/Faulhaber_AN015_EN.pdf

@crackwitz

5 жыл бұрын

thanks for pointing to this appnote. they sound like an authoritative source. but they're wrong. it's rubbish. absolute rubbish. they use terms that sound intuitive but aren't, without explaining them, to imply bad qualities where there are none. anything that doesn't show and argue on the unit circle, but keeps harping on "incremental torque", is useless. incremental torque is a useless term. *load angle* is the angle between the magnetic field and the rotor, i.e. by how much the rotor is out of alignment with the magnetic field. it is defined in degrees or radians, whatever's convenient, but on the *full circle* , i.e. a full electrical or physical revolution. it is *not* defined relative to whatever granularity a drive can manipulate the magnetic field by. torque is proportional to that angle. a motor generates *no force* on its rotor when the rotor is in alignment with the magnetic field (standstill, or rotor following magnetic field without drag/load). a motor generates torque when the rotor is out of alignment with the magnetic field, proportional to the load angle. when you do full stepping, you get 90 degrees (electrically) per step, which the rotor has to catch up with (or oscillate around, if you move faster). you get EXACTLY the same torque, assuming the rotor is fixed or the steps come quickly, from doing one full step vs 32 times a 32th step. exactly. because it's the same displacement. you get less torque ripple however, which is a desirable property to reduce resonances, because you don't do one large step, but many small steps on a straight line. they compare the load angle ("incremental torque") of one *fraction* of a full step ("microstep"), to a full step. it's rational nonsense. it's a mathematical sleight of hand to declare the angle of a microstep to be the same as the angle of a full step. we are always dealing with angles relative to a full revolution on the unit circle, or the motor shaft (times pole count). it's madness to suddenly expect a motor to increase its torque per load angle by a factor of 100. microstepping doesn't increase the number of poles. we are still on the unit circle. that never changes. they just conveniently "forget" that to sound smart or something. I can't fathom the reasons behind their nonsense explanations, but they're still provable nonsense. microstepping improves (reduces) resonance behavior. microstepping obviously gives you finer command over the magnetic field (thus the rotor) than coarser (half/full) stepping. motors and drivers obviously have imperfections and limits from construction and pure physics. the more microsteps a driver can make, the more it follows the ideal unit circle. eventually, let's say around 256th stepping, that current control, if it achieves its design goal, is better than the imperfections of the construction of the motor, so that's where gains disappear. "microstepping is bad" is disingenuous marketing B.S. to help them sell their old full/half stepping drives. please listen. I've been diving into the theory/physics of this for a while. I've come across plenty of instances of this. every single time, the explanation was rooted in this "zeno's paradox" kind of argument of reducing the load angle, and calling that "bad". if they'd apply the same nonsense to BLDC motor drives, it would become obvious. if you have machines that work fine at 1/16 but skip steps at 1/32, I'd be glad to analyze that. it's a common problem with many cheaper drivers that have "fixed on-time", the DRV8825 being an awfully bad example of that. they can't generate small currents near zero, which causes them to "round up" those single steps. to alleviate that problem, one would (throw the driver away or) increase the overall current, i.e. increase the radius of the unit circle (which can imply a larger motor to endure the current). they are *still* not worse than the next coarser stepping.

@crackwitz

5 жыл бұрын

one other thing: steps are only ever "lost" in angles of a full electric rotation (four full steps). any other kind of "missed" step isn't lost, but merely not "stepped on". they're kinda jumped over.

it sounds a bit clunky

@ProtoG42

6 жыл бұрын

Mark E it is because of the acceleration profile. At constant speed it sounds much better but it can’t reach these speeds. A bit more tweaking and I think it will sound better. There is also slop in the cheap rod ends that I’m using that attribute to some of the harsh sounds.

wtf you use that amount of beef for a delta?

i love how professional it looks and moves, i am intrested in the motion planner your coding this is my version of the code and gui delta robot project ( link github.com/tunmaker/Delta-Robot-Project ) if you can share your version with me i d be grateful

@ProtoG42

6 жыл бұрын

Thank you! I am just using basic commands at the moment and have not yet gotten around to integrating the kinematics equations yet or my camera and encoders. The code is not even close to being completed but here is the code used for this video: drive.google.com/open?id=1FvciHjzKPLEELoY_5oH0ebFOXO0InPr4

@TunMaker

6 жыл бұрын

Proto G Engineering well now you have a starting point then good luck and keep it up