We're back to serious robotics again next week with part 2 of my drop-in navigation system for any robot. Patrons and TH-cam Channel Members have it already!
hey James I want to build a (Nao) humanoid robot using a 3D printer a bunch of servos and a raspberry pi but I have no idea how to code or how to use a raspberry pi do you have any advice on how I can learn or get a pre-programed file for a basic humanoid robot?
I think the coanda effect would be more obvious if the balls didn't enter the system with horizontal velocity. As is it looks like they're already moving sideways and the fans just provide lift. Maybe release them whilst stationary via a grabber. This then challenges the viewer to work out why they start moving sideways.
Alternatively, maybe just have them enter enter with horizontal velocity but in a different direction. That sounds like it'd achieve the same thing with much less work
Yes, you got in on the end. You need a central wind-source (I recommend a centrifugal blower instead of a compressor), that can provide reliable constant laminar flow.
@@alexairconditioning1581 assuming you have cura 4.8, in the support section, change the drop down box from normal/standard to tree supports. If you’ve got an older version of cura, make sure you’ve ticked the advanced settings, then it’s found at the bottom under the experimental section
When James does videos like this, he always reminds me of a younger version of Tim Hunkin from the great 1980's series, The Secret Life of Machines. Good stuff.
The feeder slide is too wide and to "U" shaped! This allows the balls to have a random velocity component in the lateral direction, meaning they don't hit the first uplift fan airstream in exactly the same place! reprint the slide as "V" shape to ensure the balls roll down with only a longitudinal component!
Have a look into wool tuft testing. It is normally used for car/plane aerodynamics but is really good for visualising air flow. Taping small threads to the 3D prints will show where the flow is attached/laminar and swing around wildly when it is detached. You could also try a clear acrylic wall behind the prints with small threads taped or glued to it. This will show the airflow above the fans and also help contain the odd rogue ball that flies off that side.
You could add a second cut out on the feeder opposit the existing one. That way balls get fed every 180°. Thus doubling the feed rate using the same amount of energy. Or maybe have the cut outs every 90° so 4 balls are feed in quick succession.
I really enjoyed Tom's Coanda-effect drone and hovercraft vids, even though they ended up not working very good, it was fun that someone documented their pursuits. I recommend them to whoever has not watched Tom Stanton's channel.
@@UhOhUmm It would seem we have different definitions of “high current”. Brushless motors and EDFs can use up to around 5000W per motor in similar sizes to those in the video. That’s up to 25 pc power supplies required for each fan. Granted, I doubt these are using anywhere near that, but still
@@UhOhUmm Ah, never mind. Just watched the 5 minute mark again, and he says they draw around 85 watts each, so three of them could be run off of a 300 watt supply easily. I wonder if it was just easier not to mess with modifying the connections since he had the batteries laying around and they easily plug into the escs 🤷♂️
I assume he wants to bring the entire ball moving setup to shows/schools and it's easier to have a portable power sources instead of going to the mains every time. He can also do a power reliability test while he does the main test for airflow.
When you build the full installation you could add a system or platform to get the balls that falls in any step and redirect back to the 1st step. Like a canal to reposition the balls.
I remember when I was young, there was a kid's tv game show based around building "great ball contraptions" and the like. IIRC it was called the Great Egg Race and was hosted by Professor Heinz Wolff
Awesome show! Helped me decided to design stuff for a living in a similar way to mythbusters probably influenced people. I believe there was an adults version that pitted teams against eachother a bit like an indoor version of scrapheap challenge.
interesting project. I think though that the motor cover sticking up might be having a negative effect. Perhaps the 3d printed step should mount the motor deeper down so that the balls are not hitting the black cone, or the air currents that it might be producing.
Yes I think maybe much deeper down with vertical fins to stop air spinning. That said there are probably many other improvement to make in terms of the noise it make and the power it uses ;-)
I first discovered your channel when I was watching the secret Santa video on the Hacksmith's channel. I'm so impressed with what you've been doing - especially with the Boston Dynamics 'dog'. It's been massively educational for me. Enjoyed this video too - learning about an effect I've never heard of before.
James. A couple things come to mind. The wheel that's on the feeder would be better if it had room to multiple balls. The speed is limited for rotation as to getting the balls in but you can have the balls drop out more frequently, so more places to hold a ball and turning the rotation speed down could help. Also the end of the delivery chute doesn't have to be solid if you're only using ping-pong balls. You could have very little at the end and still deliver the balls. Good to see you're doing well and stay safe 👍😷
About that big amout of support at 2:42, you should've laid flat the tip of the curve on the bed, like a dome, that way, you could've saved quite a lot of filament !
when i was in high school me and my friends built a rather large ball contraption in our robotics club, it was probably one of my favorite projects that we built
"not too unhappy with how that works" Oof man I feel you. It's still really cool and I'm liking this series and the novel concepts it introduces. I hadn't seen a screw used like you did in the previous video. As far as laminar flow goes I'm guessing you might have seen this trick but maybe sink the EDFs down into a tube, and place something like a bundle of straws after the fan to straighten the airflow. You could also maybe try widening the walls on the ramp maybe? A quick way to test this would be to build them up with something like duct or painter's tape instead of having to do a whole new print, just experiment with some different height and maybe splays of the sides of those ramps.
I wonder if you need to run each of the stepping fans at different rates, getting progressively slower up the chain to account for the air coming up from the previous fan? Might make the step hugging more consistent.
I dont think it would work. theres no such thing as sucking really its air pressure blowing and it would be a lot less directional stream of air. Allways worth trying though it might have some interesting effects.
I just realized you are building a real life version of the old "Blue Ball Machine" gif that has been circulating the internet for ages. Liked and subscribed :)
What if - hear me out - after the last step you add a final fan to catch the ball if it goes just right, so that it will float at the end (just like you demonstrated in the beginning). Meanwhile, it could be assembled within an acrylic box to catch the balls and cycle them back up your tower, to be fed back into the staircase. You'd have have air slots in the case of course, but that would give you and extra hand at controlling the airflows direction. Just random thoughts! Great video!
you might be right about motor direction, I was thinking the center ducted fan should be a reverse direction spinning one. I noticed that as some that are flowing higher are spinning in same direction as duct and i would think that w/the second one in same direction would give it just enough to flow out of area. Like torque thrusted that way.
I'm surprised you didn't implement a pulsed motor control timed off the rotation of the feeder, so that each "step" is only active for a moment. The pulses could also be current controlled so the air flow either slightly increases, or decreases as the ball "climbs" the steps. Cool project, can't wait to see the next version of your Rube-Goldberg.
You need to use non-planar slicing. This will get rid of the jaggedness between layers and allow a truly curved surface to be printed. You could also add some clear plastic walls to restrict the air to operating in 2 dimensions. There may be turbulence from elsewhere in the room.
Cool video, I cannot wait for the ball contraption representing: Stoichiometry, or Thermal Expansion. you could use one of the weaker EDFs to represent The Leidenfrost effect for a pathway.
@James - one of the big differences between you knuckleballing the ball in, and using your ramp, is the rotation of the ball when it enters the airstream. The air "sticks" to the ball too - that's why it moves in the first place, right? A spinning object will definitely behave differently from one that is stationary. And if you think about it, in your case the ball is rolling "against" the wind - the front of it is rotating down into the airstream. This likely introduces some inconsistent drag and imparts forces that send the balls off in various directions. It might be cool to make a follow up with a sharpie mark on one of the balls, to see if ot behaves this way in the airstream.
I remember playing with suspending balls of various types on streams of air as a kid - blowing through a straw seemed to make it all easier, so I suspect your idea for a laminar flow of air would make it a lot better. I'm sure you could 3D print something that's basically a bunch of straws/a honeycomb of tunes, to help stabilise your airflow.
You might want to fiddle with the part orientation and bridging section to minimize support material. It's likely inevitable with such large, single-print parts, but you might be able to change the orientation and get away with a lot less support w/ good use of overhangs
I thought dropping the fans down and focus the air through a manifold you could compress the air slightly but with a narrow outlet you could direct the air better, I noticed air hitting the past projects on the shelf behind where you were standing, so their is quite a lot of turbulence being created by the fans, this could cause adverse results. If you controlled where the air was going you can reduce the turbulence and put the air where its needed rather than hope the ball falls favorably for your experiment. Just a suggestion, but an extreamly interesting channel, and you have a new subscriber.
If you are worried about the balls grabbing on the feeder and not on the first "stair", you could also try changing the feeder to two thin lines holding the balls from the lower left and right side as they slide down it. This would reduce the surface area for the air to grab onto and not affect the desired trajectory as much, and also allow the balls to enter the fist stair much more accurately for repeatable tests! (also save on print material/time!)
Others have mentioned the turbulent air from the EDFs - I'm also wondering if the layer lines on the curved pieces might be rough enough to cause flow separation. Sanding them smooth might be worth doing. Cool project though!
Cool Project! The coanda effect only contains the ball well on the edge of the air stream. The larger the difference in airspeed at the interface, the greater the restoring force. Since your blowers are much larger than the ball diameter there's a whole region where the ball isn't contained and is free to float sideways. This only gets worse as the air stream spreads out above the blower. You probably want to narrow the outlet significantly to about the diameter of the ping pong ball (which will increase the airspeed as well which is good)
I think part of your issue is that the second and third stages get the previous stage's air mixed in from the side together with the ball. Making the steps higher and instead of bending over have a trap to let the ball fall through towards the next one and the air to go away (deflect upwards or to t he sides) would probably work better.
I wonder if some of your deflection isn't caused by the balls hitting the center cone of the EDF. what if you dropped the EDF down a couple inches and added the air straighteners? I am new to the channel but loving what your doing! thanks for the entertainment.
You could do with a sensor in the chute that turns on the motors for a desired amount of time when a ball goes down it, so the motors are only on whey they need to be.
Seems like the added energy from gravitational acceleration down the ramp leads to a situation where it's having an effect like a bowling bowl on a trampoline, iow, the balls are " bouncing " off the air stream. They need slowed down to the same speed they were when you placed them there by hand. When you placed them by hand, the effect is more like the ball getting sucked into the airflow instead
I think the issue is that the balls are getting pushed up by the left (from our point of view) sides of the EDFs, where the airflow would be mostly vertical. If you covered that side of the fans and delivered the balls to the right side, it should work much more reliably I think.
I remember having fun suspending a steel ball bearing about 8mm in diameter over the nozzle of an air nozzle fed by a compressor. It didn't hover very high up but it was clearly suspended. It was also held in place surprisingly firmly, and it spun very fast. I had heard of the Avrocar and thus presumably the Coanda effect, but never made the connection. (We called the 'nozzle' an air gun, I think, but you'll see something completely different if you websearch that. It had a very small nozzle machined in steel, and a trigger to let the air out.)
The ducting tubes for the fan should be deeper. The issue with fans for producing air pressure is they produce turbulent flow, and measures should be taken to smooth it out. That's why the effect from a compressed air source works a lot better, air from a pressure reservoir doesn't have anywhere as much turbulence.
Cool project as normal, I’m sure you’ve considered this before but I think it’s time you upgrade your power delivery form projects like this. Either a big bench power supply or some moded server PSUs that go for $20 used would work great for stuff like this.
I suspect that you need to reduce the power (in a specific value) as you go up the staircase. You're adding momentum to the balls & this is cumulative, meaning that unless you reduce the pressure from the secondary fans, you eventually reach a chaotic state that overcomes the adhesion from the coander effect. (Obviously a guess, but just my interpretation of what I'm seeing here.)
I think you should make the ramp that feeds into the staircase out of rails of plastic or metal rather than a contiguous piece to help break up the surface area for the coanda effect to stick to. I would also have the feed ramp up slightly to launch the balls up into the air stream to make absolutely sure that you have nothing interfering with the first coanda "catch".
That brought back a memory of a toy my aunt bought for me one Christmas oh just before first Apollo mission went up. it intailed a balloon that said NASA on it and on the bottom where you would fill it up you would tape a plastic piece of landing gear. I could take this other part that had a fan in it and you could direct the fan fore and aft left and right and the balloon you could raise up in the air make it go further away and then land it on something else. Then bring the ballon back close and land it upright , if you were good with the throttle Thank You for the memory
I wonder how much the air flow from the first curve is feeding in to the air flow in the second. The balls do seem to accelerate on each section, and might explain why it seems to almost skip the second curve half the time. Not sure how you would compensate for that, other than by spacing the steps out more. Alternatively, you could maybe run the second fan slower, and possibly the third fan slower still? I like the laminar flow idea too.
U should also try if a big fan with its output funneled down and split up to power all 3 outputs without any of the aerodynamic EDF housings in the way would improve the reliability and reduce the noise.
I noticed that when you placed the ball, you put it in much higher on the air stream. Also, you didn't put the same forward momentum into the ball as the chute. And finally, the chute imparts a rotation on the ball. All these will impact the flight characteristics of the ball.
it might be that when the drop mechanisms spins the balls to drop them into the hole the balls this slight spin is maintained as they fall towards the fan which might explain why they consistently veer off to right
You need to reverse the motors so the air is not divided by the motor. You want a central core of fast-moving air. A partial constriction after each fan stage would probably help.
Biggist problem you had was that the fan's nose cones were causing whirling winds where the balls were entering air flow. Try recessing the fans down away from ball entry point, so you get a smoother (much more uniform.) air flow. then retry experiment!!!
Yo, James if you want to make it more reliable you just need to move the chutes up and setup the fans below so you blow the balls through the rings. This should work 99% with the right amount of airflow.
Hi... You know its great that you made the device out of red. That way we can clearly see the white balls moving up them. Good thing too that there was nothing in the background that was the same colour as the balls.... Oh wait...
On the feed ramp, couldn’t you perforate it to break the Coanda effect? You could also use some kind of mesh to help keep them from falling off the side of the steps (maybe that’s cheating). I definitely think the inconsistency could be solved by laminar air flow. Maybe you could vary the texture to make air flow slower on the sides of the steps or faster in the middle to help keep the balls on track. Cool experiment!
The "shute" (I dont know the spelling) seems to be the problem. If you notice, they bounce on it causing a bad entry into the first fan, also introducing a lot of randomness. I think it should be more of a reversed J shape, with a more steep fall at the beginning eliminating that first bounce, and gradually changing direction sideways, and it could even give it a little bit of upwards movement as it goes into the first fan.
once you make all the great ball contraptions you should link them all together, then have an RC truck that follows a route to go back to the start in a loop, but it waits for the truck to be full till it goes back to the start, then when it gets to the start that it tips them out into a hopper that directs everything into a single point to put it into the spiral
certainly seems reasonable for that first fan, although for the second and following fans, won't they have a huge sideways rush of air hitting them from the previous tube?
I am wondering how much of the inconsistencies of this apparatus are caused by differences in the ping pong balls. Did you uniquely mark any of them to try to remove the worst behaved ones?
Best way to get the laminar flow would probably be using compressed air instead of using fans. Or atleast some veins. The fans naturally create a votex in the air and can cause issues with stability of the ball as it enters and exits the wake. It also seems like the ball likes to bounce off of the first fan's nose cone a lot so that is probably not helping any. If you were to find a wide and flat nozzle that creats more of a "blade" of air, that may be your ticket since the flow out would be likely more predictable and consistent. Also, check the alignment of the steps. Some seem to be angled slightly towards the feed ramp.
I think that if you changed the entry angle (where the ball leaves the entry chute and enters the first air jet) it might work better. You can see the angle it wants to travel at over the stairs when it works just right whereas the ball comes down the chute at about 90°. Instead of being at or near the angle that it wants to be at, the air has to first change the direction of the ball before it goes up the first step. I think it would work better if there was a little bit of a ramp at the end of the chute that let the ball enter the air jet at around 45°, traveling upward
I wonder if it could work better with a positive curvature surface rather than negative curvature that you got... seems to me that your configuration would tend to split the air flow in two, one going along one sidewall, other going along the other sidewall. Picture it being held upside down and water running along the surface.
We're back to serious robotics again next week with part 2 of my drop-in navigation system for any robot. Patrons and TH-cam Channel Members have it already!
I think the reason why it doesn’t work after it comes out of the shoot is because when it goes out of the shoot it has been spin
I really enjoyed this one, cheers! 😊👍
@@ben_ands Maybe an upturned lip at the end could slow the spin as much as possible, so the ball just barely clears the edge
hey James I want to build a (Nao) humanoid robot using a 3D printer a bunch of servos and a raspberry pi but I have no idea how to code or how to use a raspberry pi do you have any advice on how I can learn or get a pre-programed file for a basic humanoid robot?
Just curious to see if the air would stick more after using filler and sanding for a nice and smooth inner surface for the air to follow.
Trying to not lose balls gives me Wintergatan flashbacks :D
Flashbacks to yesterday?
@@SeanPorterPDX ?
@@SteffenBauer He's posting daily on his second channel, and the past few weeks have been all about marble retention.
@@StuckeyIRL oh good to know. Thanks!
It might have been quite informative to feed some smoke into the intake of the bottom fan, and see what the air flow looks like.
I think the coanda effect would be more obvious if the balls didn't enter the system with horizontal velocity. As is it looks like they're already moving sideways and the fans just provide lift. Maybe release them whilst stationary via a grabber. This then challenges the viewer to work out why they start moving sideways.
Yeah, I was wondering if you could get this system working with the units oriented at an angle, so the balls go up in a spiral or a zig-zag.
@@Hugh.Manatee mayby with extreme zig zag, and compensated L tube in the turbines
Alternatively, maybe just have them enter enter with horizontal velocity but in a different direction. That sounds like it'd achieve the same thing with much less work
Cheap way to make mostly laminar flow is to tape a bunch of straws together. A similar 3d printed part might work but texture might become an issue.
It's a little disturbing seeing rasputin's head staring at us from the background
His beard also waving to us (9:25)
@@seseurian The smiling and beardwave cracked me up, haha!
Yes, you got in on the end. You need a central wind-source (I recommend a centrifugal blower instead of a compressor), that can provide reliable constant laminar flow.
he gets it at the end? I stopped the video halfway to look through the comments to see if this was obvious to anyone else :D
I think with regulators it wouldn't matter what the source of air was.
I like that you can see Estefannie's depressed robot in the background.
Have you tried 'tree support' in Cura? Might quicken some of those long prints!
This was a Simplyfy3D slice, I'll take a look
I'd suggest gradual infil steps in the support instead, slices much faster and is more stable on the bed imo
Why fear when filament sponsor is here!😂 just kidding!
How do u get that? I hate the normal supports in Cura its a pain to get of
@@alexairconditioning1581 assuming you have cura 4.8, in the support section, change the drop down box from normal/standard to tree supports.
If you’ve got an older version of cura, make sure you’ve ticked the advanced settings, then it’s found at the bottom under the experimental section
When James does videos like this, he always reminds me of a younger version of Tim Hunkin from the great 1980's series, The Secret Life of Machines. Good stuff.
The use case difference in ball loss standards between you and Martin of Wintergatan is amusing to me.
The feeder slide is too wide and to "U" shaped! This allows the balls to have a random velocity component in the lateral direction, meaning they don't hit the first uplift fan airstream in exactly the same place! reprint the slide as "V" shape to ensure the balls roll down with only a longitudinal component!
Fingers crossed for a Rube Goldberg machine
I have plans!
Have a look into wool tuft testing. It is normally used for car/plane aerodynamics but is really good for visualising air flow.
Taping small threads to the 3D prints will show where the flow is attached/laminar and swing around wildly when it is detached.
You could also try a clear acrylic wall behind the prints with small threads taped or glued to it. This will show the airflow above the fans and also help contain the odd rogue ball that flies off that side.
You could add a second cut out on the feeder opposit the existing one. That way balls get fed every 180°. Thus doubling the feed rate using the same amount of energy. Or maybe have the cut outs every 90° so 4 balls are feed in quick succession.
* *my brain* "Coanda Effect Staircase"
*_sees that it's by James Bruton_*
* *my brain* "Coander Effect Staircase"
I really enjoyed Tom's Coanda-effect drone and hovercraft vids, even though they ended up not working very good, it was fun that someone documented their pursuits. I recommend them to whoever has not watched Tom Stanton's channel.
Stop using 2020 extrusion we are in 2021 now
Do you REALLY want to use 2021 extrusion? Oh got, i already get angry for it messing with constructions
Haha, great comment
Ha! I was just about to comment that!
having it be slightly longer in one direction would make it really annoying to figure out which orientation it is in
I know its a joke ok
🤣
I'm sitting here going, "Cool, but why use a battery over a power source that wouldn't drain and die?"
Probably the current requirement, and the fact that he had li-ion’s handy
@@davak72 but high current 12v power supplies are readily available in every electronics store and they're mega cheap, because PC's use them...
@@UhOhUmm It would seem we have different definitions of “high current”. Brushless motors and EDFs can use up to around 5000W per motor in similar sizes to those in the video. That’s up to 25 pc power supplies required for each fan.
Granted, I doubt these are using anywhere near that, but still
@@UhOhUmm Ah, never mind. Just watched the 5 minute mark again, and he says they draw around 85 watts each, so three of them could be run off of a 300 watt supply easily. I wonder if it was just easier not to mess with modifying the connections since he had the batteries laying around and they easily plug into the escs 🤷♂️
I assume he wants to bring the entire ball moving setup to shows/schools and it's easier to have a portable power sources instead of going to the mains every time. He can also do a power reliability test while he does the main test for airflow.
Longest Coanda staricase should be a Guniess world record category.
Guiness world records needs to stop existing is what should happen
Anything can be a category for Guiness as long as you pay them
I appreciate you writing Henri Coandă 's name right, He was one of the brightest minds in Romania
When you build the full installation you could add a system or platform to get the balls that falls in any step and redirect back to the 1st step. Like a canal to reposition the balls.
I remember when I was young, there was a kid's tv game show based around building "great ball contraptions" and the like. IIRC it was called the Great Egg Race and was hosted by Professor Heinz Wolff
Awesome show! Helped me decided to design stuff for a living in a similar way to mythbusters probably influenced people. I believe there was an adults version that pitted teams against eachother a bit like an indoor version of scrapheap challenge.
@@owensparks5013 I didn't know they'd done a version for grownups. Awesome that it inspired you
Very proud that your are showing the Coandă effect!
Uh, why?
@@hansdietrich83 Because Coandă was cool!
Next week - build a robot to find and pick up pingpong balls
convert a romba to suck n blow back to starting tray,,lol
As you mentioned the turbulence of the air from the EDFs I agree is a factor.
It will be really cool to see the final contraption demonstrating different scientific experiments together!
I do see tom Stanton and yours for general science knowledge, thankyou for providing such content James Bruton
interesting project. I think though that the motor cover sticking up might be having a negative effect. Perhaps the 3d printed step should mount the motor deeper down so that the balls are not hitting the black cone, or the air currents that it might be producing.
Yes I think maybe much deeper down with vertical fins to stop air spinning. That said there are probably many other improvement to make in terms of the noise it make and the power it uses ;-)
I first discovered your channel when I was watching the secret Santa video on the Hacksmith's channel.
I'm so impressed with what you've been doing - especially with the Boston Dynamics 'dog'. It's been massively educational for me.
Enjoyed this video too - learning about an effect I've never heard of before.
James. A couple things come to mind.
The wheel that's on the feeder would be better if it had room to multiple balls. The speed is limited for rotation as to getting the balls in but you can have the balls drop out more frequently, so more places to hold a ball and turning the rotation speed down could help.
Also the end of the delivery chute doesn't have to be solid if you're only using ping-pong balls. You could have very little at the end and still deliver the balls.
Good to see you're doing well and stay safe 👍😷
About that big amout of support at 2:42, you should've laid flat the tip of the curve on the bed, like a dome, that way, you could've saved quite a lot of filament !
when i was in high school me and my friends built a rather large ball contraption in our robotics club, it was probably one of my favorite projects that we built
amazing project bro
this is a great series please keep making these
At 5:33 I like the way the motors appear to be either on or off, with no speeding up or slowing down.
Really good video quality and production
"not too unhappy with how that works" Oof man I feel you. It's still really cool and I'm liking this series and the novel concepts it introduces. I hadn't seen a screw used like you did in the previous video. As far as laminar flow goes I'm guessing you might have seen this trick but maybe sink the EDFs down into a tube, and place something like a bundle of straws after the fan to straighten the airflow.
You could also maybe try widening the walls on the ramp maybe? A quick way to test this would be to build them up with something like duct or painter's tape instead of having to do a whole new print, just experiment with some different height and maybe splays of the sides of those ramps.
I wonder if you need to run each of the stepping fans at different rates, getting progressively slower up the chain to account for the air coming up from the previous fan? Might make the step hugging more consistent.
Most satisfying video yet
I wonder if you could an upside down version of that
Instead of Pushing Air out, it sucks air in like a vacuum
I dont think it would work. theres no such thing as sucking really its air pressure blowing and it would be a lot less directional stream of air. Allways worth trying though it might have some interesting effects.
You make the most interesting stuff
I just realized you are building a real life version of the old "Blue Ball Machine" gif that has been circulating the internet for ages. Liked and subscribed :)
Neat experiment. Nicely done!
11:55 i like the way the lights are mounted. Just clamped onto the cabinet
What if - hear me out - after the last step you add a final fan to catch the ball if it goes just right, so that it will float at the end (just like you demonstrated in the beginning). Meanwhile, it could be assembled within an acrylic box to catch the balls and cycle them back up your tower, to be fed back into the staircase. You'd have have air slots in the case of course, but that would give you and extra hand at controlling the airflows direction. Just random thoughts! Great video!
you might be right about motor direction, I was thinking the center ducted fan should be a reverse direction spinning one. I noticed that as some that are flowing higher are spinning in same direction as duct and i would think that w/the second one in same direction would give it just enough to flow out of area. Like torque thrusted that way.
I'm surprised you didn't implement a pulsed motor control timed off the rotation of the feeder, so that each "step" is only active for a moment. The pulses could also be current controlled so the air flow either slightly increases, or decreases as the ball "climbs" the steps. Cool project, can't wait to see the next version of your Rube-Goldberg.
That would be a good idea - even with compressed air and electronic valves. There may well be a V2 at some point.
You need to use non-planar slicing. This will get rid of the jaggedness between layers and allow a truly curved surface to be printed.
You could also add some clear plastic walls to restrict the air to operating in 2 dimensions. There may be turbulence from elsewhere in the room.
James Bruton makes the Coanda effect into the conundrum effect... Because Sherlock is a detective ⭐⭐⭐⭐⭐
Cool video, I cannot wait for the ball contraption representing: Stoichiometry, or Thermal Expansion. you could use one of the weaker EDFs to represent The Leidenfrost effect for a pathway.
@James - one of the big differences between you knuckleballing the ball in, and using your ramp, is the rotation of the ball when it enters the airstream. The air "sticks" to the ball too - that's why it moves in the first place, right? A spinning object will definitely behave differently from one that is stationary. And if you think about it, in your case the ball is rolling "against" the wind - the front of it is rotating down into the airstream. This likely introduces some inconsistent drag and imparts forces that send the balls off in various directions.
It might be cool to make a follow up with a sharpie mark on one of the balls, to see if ot behaves this way in the airstream.
I remember playing with suspending balls of various types on streams of air as a kid - blowing through a straw seemed to make it all easier, so I suspect your idea for a laminar flow of air would make it a lot better. I'm sure you could 3D print something that's basically a bunch of straws/a honeycomb of tunes, to help stabilise your airflow.
You might want to fiddle with the part orientation and bridging section to minimize support material. It's likely inevitable with such large, single-print parts, but you might be able to change the orientation and get away with a lot less support w/ good use of overhangs
I got a name suggestion like there is Boston dynamics
'Bruton dynamics ' how's that
Awesome !
I thought dropping the fans down and focus the air through a manifold you could compress the air slightly but with a narrow outlet you could direct the air better, I noticed air hitting the past projects on the shelf behind where you were standing, so their is quite a lot of turbulence being created by the fans, this could cause adverse results. If you controlled where the air was going you can reduce the turbulence and put the air where its needed rather than hope the ball falls favorably for your experiment. Just a suggestion, but an extreamly interesting channel, and you have a new subscriber.
If you are worried about the balls grabbing on the feeder and not on the first "stair", you could also try changing the feeder to two thin lines holding the balls from the lower left and right side as they slide down it. This would reduce the surface area for the air to grab onto and not affect the desired trajectory as much, and also allow the balls to enter the fist stair much more accurately for repeatable tests! (also save on print material/time!)
Others have mentioned the turbulent air from the EDFs - I'm also wondering if the layer lines on the curved pieces might be rough enough to cause flow separation. Sanding them smooth might be worth doing. Cool project though!
Cool Project! The coanda effect only contains the ball well on the edge of the air stream. The larger the difference in airspeed at the interface, the greater the restoring force. Since your blowers are much larger than the ball diameter there's a whole region where the ball isn't contained and is free to float sideways. This only gets worse as the air stream spreads out above the blower.
You probably want to narrow the outlet significantly to about the diameter of the ping pong ball (which will increase the airspeed as well which is good)
I admire the uselessness of the whole contraption, and I envy you for making a living doing things that I wish I had time to do for fun.
I think part of your issue is that the second and third stages get the previous stage's air mixed in from the side together with the ball.
Making the steps higher and instead of bending over have a trap to let the ball fall through towards the next one and the air to go away (deflect upwards or to t he sides) would probably work better.
I wonder if some of your deflection isn't caused by the balls hitting the center cone of the EDF. what if you dropped the EDF down a couple inches and added the air straighteners? I am new to the channel but loving what your doing! thanks for the entertainment.
You could do with a sensor in the chute that turns on the motors for a desired amount of time when a ball goes down it, so the motors are only on whey they need to be.
James eres una inspiración ☺️😎❤️
Seems like the added energy from gravitational acceleration down the ramp leads to a situation where it's having an effect like a bowling bowl on a trampoline, iow, the balls are " bouncing " off the air stream. They need slowed down to the same speed they were when you placed them there by hand. When you placed them by hand, the effect is more like the ball getting sucked into the airflow instead
he will forever be finding random pingpong balls in his home.
That happens to me with Styrofoam XD
I think the issue is that the balls are getting pushed up by the left (from our point of view) sides of the EDFs, where the airflow would be mostly vertical. If you covered that side of the fans and delivered the balls to the right side, it should work much more reliably I think.
80% of the time it works 100% of the time. Coanda!
ur job looks so fun omg
I remember having fun suspending a steel ball bearing about 8mm in diameter over the nozzle of an air nozzle fed by a compressor. It didn't hover very high up but it was clearly suspended. It was also held in place surprisingly firmly, and it spun very fast. I had heard of the Avrocar and thus presumably the Coanda effect, but never made the connection. (We called the 'nozzle' an air gun, I think, but you'll see something completely different if you websearch that. It had a very small nozzle machined in steel, and a trigger to let the air out.)
The ducting tubes for the fan should be deeper. The issue with fans for producing air pressure is they produce turbulent flow, and measures should be taken to smooth it out. That's why the effect from a compressed air source works a lot better, air from a pressure reservoir doesn't have anywhere as much turbulence.
Cool project as normal, I’m sure you’ve considered this before but I think it’s time you upgrade your power delivery form projects like this. Either a big bench power supply or some moded server PSUs that go for $20 used would work great for stuff like this.
I suspect that you need to reduce the power (in a specific value) as you go up the staircase. You're adding momentum to the balls & this is cumulative, meaning that unless you reduce the pressure from the secondary fans, you eventually reach a chaotic state that overcomes the adhesion from the coander effect.
(Obviously a guess, but just my interpretation of what I'm seeing here.)
I think you should make the ramp that feeds into the staircase out of rails of plastic or metal rather than a contiguous piece to help break up the surface area for the coanda effect to stick to. I would also have the feed ramp up slightly to launch the balls up into the air stream to make absolutely sure that you have nothing interfering with the first coanda "catch".
That brought back a memory of a toy my aunt bought for me one Christmas oh just before first Apollo mission went up. it intailed a balloon that said NASA on it and on the bottom where you would fill it up you would tape a plastic piece of landing gear. I could take this other part that had a fan in it and you could direct the fan fore and aft left and right and the balloon you could raise up in the air make it go further away and then land it on something else. Then bring the ballon back close and land it upright , if you were good with the throttle Thank You for the memory
That looks so satisfying lol
I wonder how much the air flow from the first curve is feeding in to the air flow in the second. The balls do seem to accelerate on each section, and might explain why it seems to almost skip the second curve half the time. Not sure how you would compensate for that, other than by spacing the steps out more. Alternatively, you could maybe run the second fan slower, and possibly the third fan slower still? I like the laminar flow idea too.
U should also try if a big fan with its output funneled down and split up to power all 3 outputs without any of the aerodynamic EDF housings in the way would improve the reliability and reduce the noise.
Maybe the feeder is adding some spin to the ping pong balls. Wouldn't that add some curve to the trajectory and possibly add to the inconsistency?
That's pretty neat! Is the Coanda Effect what keeps planes up, or is that a different air-going-faster-means-pressure-differences thing?
I noticed that when you placed the ball, you put it in much higher on the air stream. Also, you didn't put the same forward momentum into the ball as the chute. And finally, the chute imparts a rotation on the ball. All these will impact the flight characteristics of the ball.
it might be that when the drop mechanisms spins the balls to drop them into the hole the balls this slight spin is maintained as they fall towards the fan which might explain why they consistently veer off to right
You need to reverse the motors so the air is not divided by the motor. You want a central core of fast-moving air. A partial constriction after each fan stage would probably help.
Osom project
Biggist problem you had was that the fan's nose cones were causing whirling winds where the balls were entering air flow. Try recessing the fans down away from ball entry point, so you get a smoother (much more uniform.) air flow. then retry experiment!!!
Yo, James if you want to make it more reliable you just need to move the chutes up and setup the fans below so you blow the balls through the rings.
This should work 99% with the right amount of airflow.
Cover half of duct ,it starts going up before getting to the area of Chanda effect. Try making ramp with air holes through sides to help push?
Hi... You know its great that you made the device out of red. That way we can clearly see the white balls moving up them. Good thing too that there was nothing in the background that was the same colour as the balls.... Oh wait...
On the feed ramp, couldn’t you perforate it to break the Coanda effect? You could also use some kind of mesh to help keep them from falling off the side of the steps (maybe that’s cheating). I definitely think the inconsistency could be solved by laminar air flow. Maybe you could vary the texture to make air flow slower on the sides of the steps or faster in the middle to help keep the balls on track. Cool experiment!
The "shute" (I dont know the spelling) seems to be the problem. If you notice, they bounce on it causing a bad entry into the first fan, also introducing a lot of randomness. I think it should be more of a reversed J shape, with a more steep fall at the beginning eliminating that first bounce, and gradually changing direction sideways, and it could even give it a little bit of upwards movement as it goes into the first fan.
Always great stuff :-)
once you make all the great ball contraptions you should link them all together, then have an RC truck that follows a route to go back to the start in a loop, but it waits for the truck to be full till it goes back to the start, then when it gets to the start that it tips them out into a hopper that directs everything into a single point to put it into the spiral
Love the flowing beard, blowing in the breeze. Lol
certainly seems reasonable for that first fan, although for the second and following fans, won't they have a huge sideways rush of air hitting them from the previous tube?
use a whole bunch of straws together to get a cheap/hacky laminar flow in front of the fans
The shape that needed lots of 3d print support looks like it might not need so much if you divide it in half along the axis of mirror symmetry.
I am wondering how much of the inconsistencies of this apparatus are caused by differences in the ping pong balls. Did you uniquely mark any of them to try to remove the worst behaved ones?
Best way to get the laminar flow would probably be using compressed air instead of using fans. Or atleast some veins.
The fans naturally create a votex in the air and can cause issues with stability of the ball as it enters and exits the wake. It also seems like the ball likes to bounce off of the first fan's nose cone a lot so that is probably not helping any.
If you were to find a wide and flat nozzle that creats more of a "blade" of air, that may be your ticket since the flow out would be likely more predictable and consistent.
Also, check the alignment of the steps. Some seem to be angled slightly towards the feed ramp.
I think that if you changed the entry angle (where the ball leaves the entry chute and enters the first air jet) it might work better. You can see the angle it wants to travel at over the stairs when it works just right whereas the ball comes down the chute at about 90°. Instead of being at or near the angle that it wants to be at, the air has to first change the direction of the ball before it goes up the first step. I think it would work better if there was a little bit of a ramp at the end of the chute that let the ball enter the air jet at around 45°, traveling upward
I wonder if it could work better with a positive curvature surface rather than negative curvature that you got... seems to me that your configuration would tend to split the air flow in two, one going along one sidewall, other going along the other sidewall. Picture it being held upside down and water running along the surface.