That's very nice. The nut dispenser is very effective. Small, and feeds well.
The screw dispenser is starting to run into jamming problems. It will probably start to jam more as the acrylic gets scratched and friction becomes worse. But it's manual and low volume, so jamming isn't a big issue.
He's discovered that dispensing is easy, but order from chaos is harder.
There's a whole theory of feeder design.[1] There are clever tricks to orient strangely shaped parts using feeders made from passive components. A basic trick is to get parts aligned in one axis, then arrange it so that the ones that are backwards or upside down hit some obstacle or are not supported, so they fall back down for another try.
It occurs to me that the screw counter's main difficulty is in orientating the screws.
The machine does solve that (as a product of all the shaking and jostling and doubtless unjamming), but judging by the length of the feeder tube it's not a very fun step. And the end goal isn't to have screws that are each oriented in exactly the same way, but instead to have a specific quantity of screws placed in each of a series of containers.
All of that effort to orient them so precisely does make them easy to count using the nut dispenser mechanism, but that effort is otherwise ultimately discarded.
I'm lead to wonder if the process of dispensing 6 screws could be accomplished more simply (ie, with less fiddling and shaking) by reducing the amount of orientation necessary.
Perhaps by using a sorter that puts the screws in a line, axially, without a preference for heads-first or threads-first orientation?
So that explains why the smallest parts often have spares in ikea and lego builds. Is this done because of the error in weighing the smallest parts, so they have a margin for error by allowing for an extra 1 or 2?
> Is this done because of the error in weighing the smallest parts, so they have a margin for error by allowing for an extra 1 or 2?
This is a secondary benefit, the primary benefit is if the end user loses/breaks one. That part very well could be show stopper (Ikea 110630 anyone?). Now the end user is stuck - has to call, you have to ship, do you charge? do you give for free? they have to wait. they're annoyed, you're annoyed.
No one is happy.
The supply chain headaches for giving exact number of tiny parts is terribly expensive, relatively speaking. So you give spares because in the long run it's way cheaper.
Being aware of this, I am waiting for a solution to what to do with the leftovers besides chuck them into a landfill. The problem, of course, is scale. No one is mailing 3 screws and an Allen wrench anywhere. Maybe once you hit 5 pounds of spare Lego . . .
Just tacking on to mention the smallest parts are most likely to be lost, they’re the ones that - if dropped - seem to bounce and roll under a refrigerator or into the ether. They don’t give extras on the larger parts because they’re not likely to be lost. Frequently enough all it takes is a violent/careless bag opening to send the small pieces flying.
When we count pills, we use a simple device that has a flat part and a channel to count into. A hockey-stick-like spatula moves pills into the channel. The excess is poured back into the bottle the opposite way and the channel is then emptied into the bottle.
I count in multiples of 3 so I don't lose my place. The last number is unique for every 30.
If you count by threes, the ones place is unique until you pass each multiple of 30 - 3, 6, 9, _8, _1, _4, where was I? I hadn’t made it to 30, ends with a 4, must be 24. 27, 30, 33, _6…
Serves as a sort of checksum, as long as you know roughly how many you have and just the last digit.
> You can probably guess my opinions on it though, the software is very good but the cloud-based vendor lock-in is grating, and the free tier is hobbled beyond the point of usefulness. On the plus side, being browser-based, it works perfectly on Linux.
> The 1.1 release of FreeCAD should be soon. I really want FreeCAD to succeed, but blimey they have a big hill to climb. My fingers are crossed.
Since these parts mostly seem to be laser cut acrylic (so mostly 2D), it seems like solvespace would do a good job at cranking out the designs. I haven't used it for a larger project like this though, maybe it was already considered.
For what it's worth I've been using FreeCad 1.1RC2 lately, and for me it's the first FreeCad version worth bothering with. It's now a tool I actively reach for over OpenSCAD and Blender for various projects. Previously I couldn't make the simplest part with it.
I can't wait for the release proper, but I can heartily recommend everyone try the release candidates as well. I've got a feeling this is the tipping point for FreeCad like 2.5 was for blender.
I mean I'll be honest, it's still a car crash of a program, but at least it's now a usable car crash. I've found the following workflow to be pretty good, using the part design workbench:
- create a part
- create a body
- create a sketch
- pad/pocket/revolve/etc
- repeat with additional sketches to taste
I've also been using the proxy object thing, I forget the name, the button is a green blob, to "import" geometry from a master sketch into more specific sketches.
> I mean I'll be honest, it's still a car crash of a program
I'm glad you said that so I feel a little less mean...
I gave it another try but it still feels pretty dire. FPS is bad on a macbook pro with a 120Hz screen on simple models and sketches. I explicitly selected "touchpad" as the navigation scheme, but I still can't figure out how to rotate, and even figuring out panning took me longer than almost every other 3D program out there (blender, PrusaSlicer, macos quick look STL viewer, solvespace).
It still has a splash screen and takes quite a long time to load, like an application from the 90s.
Buttons and actions that are completely irrelevant to me are shown, but disabled, which gives a really cluttered feel.
There's still "part design" and "part" benches. No idea what distinction is being drawn there.
Obviously part of this is from me being inexperienced with the tool, but as a new user all these issues add up to something that doesn't feel approachable or enjoyable to use.
Solvespace has its own issues, but at least it opens instantly and is generally a joy to use.
I'll watch some others slog through FreeCAD 1.1 though so I don't have to, and maybe I'll learn something.
You definitely need to use a mouse with a middle button/scroll wheel, IME. With this, there are presets which work just like other CAD programs e.g. Solidworks. Without it, it’s very difficult to use, but that’s not unique to FreeCAD.
The base UI is quite bad but there are ways to improve it - either through settings and better organisation [0] or via plugins.
I’d suggest to watch a couple of tutorials specifically on 1.1 ([1] was my entry point) as every CAD program had quirks and frustrations at first. I’d say that for hobby-level creations, 1.1 now has ~80% of the usability of Solidworks, once you figure out how.
I’m not sure what’s going on with the performance on your system; I’ve used various 1.1 versions on a Windows laptop and a MacBook Pro and they’re both sufficiently performant. (I usually use a development or RC build from GitHub [2])
Thanks for the links, especially [0]. That one was the most compelling for showing how an experienced user actually models a part. It's a shame that the UI defaults need to be tweaked so much to make things usable but I know there's no single UI that works for everyone.
The trouble with Solvespace is that using it quickly turns into randomly nudging points in hope to avoid kernel failures. Lately I have been using Dune 3D which uses much more reliable kernel.
> It still has a splash screen and takes quite a long time to load, like an application from the 90s.
Lots of it is single-threaded, which is an endless frustration on a machine with umpteen cores. Especially frustrating given that it means compute happens on the UI thread.
> It still has a splash screen and takes quite a long time to load, like an application from the 90s.
The splash screen can be disabled and it takes 3 seconds to start on my mac. Fusion however has two splash screens (first a regular one, then one that covers the whole app window) and it takes 32 seconds to load! (to be fair, once loaded it's much better than FreeCAD).
Oh yeah, you won't find me defending Fusion. I can understand when you're loading a heavy scene or recomputing _everything_ in a complicated model, but I can't understand multi-second loading times and splash screens for loading...the start screen.
Thanks, I eventually discovered it after a ton of trial and error. It's a shame though because the whole point of a touchpad is multitouch gestures which actually make navigating CAD applications pretty nice. I'll use a touchpad or a combination of touchpad and mouse in other apps like KiCad and it works quite well. Seems to me like all these open source programs should be stealing/sharing the best implementations of some of these basic things like 2D/3D input controls with each other.
For 2D, sometimes I find it nearly easier to write the gcode by hand (or make a python program that writes the gcode for you). It really isn't as complicated as it sounds, especially if you can tolerate doing 3D in openSCAD.
I'm very into code-based CAD, I actually gave a small talk on it a year or two ago. A longer term goal of mine is to make some sort of hybrid CAD modeling tool which is mostly code-based, but has a GUI for certain things like defining sketches/constraints, and selecting particular geometric features that are hard to describe in code.
Here's a link to the talk if anyone can bear to listen to me for an hour:
It is sad that FreeCAD gets all the attention. If Solvespace had some of it, and the development time following from it, it could get improvements and some of the cool stuff in their pipeline. That would IMO make it a much better CAD program than FreeCAD could ever become.
I know, it's just that in this particular blog post, the designs mostly seem to be extruded 2D sketches which solvespace is particularly good at with its sketch interface.
Solvespace can also do a lot of useful 3D stuff, but it's also missing a lot so I can't in good faith recommend it for any arbitrary CAD work.
I don't know any better, but the screw counting mechanism seems awkward. Imagine the set has 10 components..
I'm surprised there is no standard solution to this - like a tape and reel solution? A counting and dispensing gun that works for different sizes? But how much more would anyone pay for M3 bolts on a tape?
Helmke had a tube feeding his dispensers in one of the videos, with bolts lengthwise. That tube idea could be used for a manual dispenser - imagine a drink dispenser, but giving 3 bolts. Maybe easier to store away, but just as awkward to load.
> the (OnShape) free tier is hobbled beyond the point of usefulness.
The free tier is identical to the standard tier except you can not create private documents and it has a no commercial use clause. This has been the case for many years, so I'm not sure where "hobbled beyond the point of usefulness" is coming from.
I love this so much. Such simple machines, for human-scale problems. I often get pulled down rabbit holes of machines and automation - this is a nice reminder that you can solve a lot of problems without reaching for an arduino or a servo.
>> I have wasted a significant chunk of my life counting out small numbers of parts into bags and posting them to people.
So, small parts like this are always counted by weight, and I'm wondering why you would spend so much time on a counting solution when "buy a scale" is right there.
He's counting out like 6 at a time. He needs a fast way to pick small quantities precisely, not a fast way to check large quantities. Once they're picked they're easily verified by eye.
In volume, small parts are dispensed by carefully designed machines, and then the result is counted by weight. You still need control of the dispensing, and as he's putting in small numbers of items the counting is the easy bit.
He needs 6 screws at a time, and the goal is to save time compared to counting manually. I'd guess that 7 would probably be fine occasionally -- maybe even 8 from time to time if the process is fast enough. I'd further guess that 9 screws is a non-starter (screws are inexpensive, but 9 represents 50% waste, which is quite a lot).
The lower limit is hard-set at 6 because the kits that he's producing and selling require exactly 6 of these screws for end-user assembly.
A small cup that would reliably scoop out at least 6 screws and no more than 7 or 8 screws sounds like a simple and elegant concept.
What does this cup look like? Is it faster to use this cup than counting by hand is? (Is it faster than the reproducible screw counter that he's already built?)
Up to roughly 100 bills it's pretty much bang on - even with a cheap $10 scale (American Weigh Scales Digital Pocket Scale has a bunch of different options). Each bill weights roughly 1 gram. So - accurate to within 1% - and presumably the banks have better scales.
I suspect at scale (moving either a lot of batches or large batches), you also need to take variance into account more. Some bills might be dirty or have stuff stuck to them, some bills might be damaged and have bits missing? And other things that occur in practice that I can't think of from the comfort of my armchair in 30s.
Awesome solve!!! Lasers and 3d printing is my side hobby business and is what keeps my sanity intact. I love seeing the practical creations that are realized by them! One of my core tenets is being self-sufficient and achieving efficiencies. This post is exactly that. Well done.
He's discovered that dispensing is easy, but order from chaos is harder.
There's a whole theory of feeder design.[1] There are clever tricks to orient strangely shaped parts using feeders made from passive components. A basic trick is to get parts aligned in one axis, then arrange it so that the ones that are backwards or upside down hit some obstacle or are not supported, so they fall back down for another try.
[1] https://www.youtube.com/watch?v=ZlyuHIxSC-A
It occurs to me that the screw counter's main difficulty is in orientating the screws.
The machine does solve that (as a product of all the shaking and jostling and doubtless unjamming), but judging by the length of the feeder tube it's not a very fun step. And the end goal isn't to have screws that are each oriented in exactly the same way, but instead to have a specific quantity of screws placed in each of a series of containers.
All of that effort to orient them so precisely does make them easy to count using the nut dispenser mechanism, but that effort is otherwise ultimately discarded.
I'm lead to wonder if the process of dispensing 6 screws could be accomplished more simply (ie, with less fiddling and shaking) by reducing the amount of orientation necessary.
Perhaps by using a sorter that puts the screws in a line, axially, without a preference for heads-first or threads-first orientation?
This is how pretty much every IKEA, LEGO, etc works with very small, cheap parts.
End users benefit because it's easy to drop/lose/break one.
This is a secondary benefit, the primary benefit is if the end user loses/breaks one. That part very well could be show stopper (Ikea 110630 anyone?). Now the end user is stuck - has to call, you have to ship, do you charge? do you give for free? they have to wait. they're annoyed, you're annoyed.
No one is happy.
The supply chain headaches for giving exact number of tiny parts is terribly expensive, relatively speaking. So you give spares because in the long run it's way cheaper.
For stuff bought online, e.g. Amazon, not much you can do.
Easier with heavy objects, and needs the variation on weight to be low for the number of items you're dispensing.
I count in multiples of 3 so I don't lose my place. The last number is unique for every 30.
https://share.google/VshUpiSioUh6rLg4q (Image link)
I interpreted it as: For every three pills I put into the channel, I add +1 to my internal count.
"The last number is unique for every 30" means... if you have a mental count of 13, you have 39 pills channeled? I didn't quite follow...
Serves as a sort of checksum, as long as you know roughly how many you have and just the last digit.
> The 1.1 release of FreeCAD should be soon. I really want FreeCAD to succeed, but blimey they have a big hill to climb. My fingers are crossed.
Since these parts mostly seem to be laser cut acrylic (so mostly 2D), it seems like solvespace would do a good job at cranking out the designs. I haven't used it for a larger project like this though, maybe it was already considered.
I can't wait for the release proper, but I can heartily recommend everyone try the release candidates as well. I've got a feeling this is the tipping point for FreeCad like 2.5 was for blender.
- create a part - create a body - create a sketch - pad/pocket/revolve/etc - repeat with additional sketches to taste
I've also been using the proxy object thing, I forget the name, the button is a green blob, to "import" geometry from a master sketch into more specific sketches.
I'm glad you said that so I feel a little less mean...
I gave it another try but it still feels pretty dire. FPS is bad on a macbook pro with a 120Hz screen on simple models and sketches. I explicitly selected "touchpad" as the navigation scheme, but I still can't figure out how to rotate, and even figuring out panning took me longer than almost every other 3D program out there (blender, PrusaSlicer, macos quick look STL viewer, solvespace).
It still has a splash screen and takes quite a long time to load, like an application from the 90s.
Buttons and actions that are completely irrelevant to me are shown, but disabled, which gives a really cluttered feel.
There's still "part design" and "part" benches. No idea what distinction is being drawn there.
Obviously part of this is from me being inexperienced with the tool, but as a new user all these issues add up to something that doesn't feel approachable or enjoyable to use.
Solvespace has its own issues, but at least it opens instantly and is generally a joy to use.
I'll watch some others slog through FreeCAD 1.1 though so I don't have to, and maybe I'll learn something.
The base UI is quite bad but there are ways to improve it - either through settings and better organisation [0] or via plugins.
I’d suggest to watch a couple of tutorials specifically on 1.1 ([1] was my entry point) as every CAD program had quirks and frustrations at first. I’d say that for hobby-level creations, 1.1 now has ~80% of the usability of Solidworks, once you figure out how.
I’m not sure what’s going on with the performance on your system; I’ve used various 1.1 versions on a Windows laptop and a MacBook Pro and they’re both sufficiently performant. (I usually use a development or RC build from GitHub [2])
[0] https://youtu.be/LKq7hgbu7ks
[1] https://youtu.be/VEfNRST_3x8
[2] https://github.com/FreeCAD/FreeCAD/releases
Lots of it is single-threaded, which is an endless frustration on a machine with umpteen cores. Especially frustrating given that it means compute happens on the UI thread.
The splash screen can be disabled and it takes 3 seconds to start on my mac. Fusion however has two splash screens (first a regular one, then one that covers the whole app window) and it takes 32 seconds to load! (to be fair, once loaded it's much better than FreeCAD).
Here's a link to the talk if anyone can bear to listen to me for an hour:
https://youtu.be/0wn7vUmWQgg
It is sad that FreeCAD gets all the attention. If Solvespace had some of it, and the development time following from it, it could get improvements and some of the cool stuff in their pipeline. That would IMO make it a much better CAD program than FreeCAD could ever become.
Solvespace can also do a lot of useful 3D stuff, but it's also missing a lot so I can't in good faith recommend it for any arbitrary CAD work.
I don't know any better, but the screw counting mechanism seems awkward. Imagine the set has 10 components..
I'm surprised there is no standard solution to this - like a tape and reel solution? A counting and dispensing gun that works for different sizes? But how much more would anyone pay for M3 bolts on a tape?
Helmke had a tube feeding his dispensers in one of the videos, with bolts lengthwise. That tube idea could be used for a manual dispenser - imagine a drink dispenser, but giving 3 bolts. Maybe easier to store away, but just as awkward to load.
The free tier is identical to the standard tier except you can not create private documents and it has a no commercial use clause. This has been the case for many years, so I'm not sure where "hobbled beyond the point of usefulness" is coming from.
So glad my ex didn’t have one of these
https://mitxela.com/shop/clock4
a medium scale problem = a better assisting device
a large scale problem = hire people
The lower limit is hard-set at 6 because the kits that he's producing and selling require exactly 6 of these screws for end-user assembly.
A small cup that would reliably scoop out at least 6 screws and no more than 7 or 8 screws sounds like a simple and elegant concept.
What does this cup look like? Is it faster to use this cup than counting by hand is? (Is it faster than the reproducible screw counter that he's already built?)
It wasn't very precise but you could move a lot of money in ball park with this method. Atleast internally across branches.