Fidget Spinners

[ehughes]

It was only a matter of time before a LED version was coming out.

https://www.kickstarter.com/projects/1339026152/electron-fidget-spinner?ref=nav_search

[ZeTeX]

At least they don't claim something unrealistic like "IT WILL SPIN FOREVER WE SWEAR!!"

[SL4P]

Sorry, but the sell price is ridiculous.

$50 selling price for your first unit is about 10 times the low-volume  manufacturing cost.
As a novelty cost saving, i'd lose the ceramic bearing and blow-molded case, which would bring it down to $2-$3 each.

Users will tinker with it for a couple of weeks, give it to the kids, then it will sit in a drawer until the battery leaks and catches fire, or the dog eats it.
(Provide clear lithium safety & disposal instructions, or your own house will also be taken - not by the dog!)
The tee shirts will cost more to make than the spinners.

Sure, there's R&D time/costs but your financial model is broken if you have to charge Lexus prices for a VW beetle.  Say 3-months of spare-time hw & sw R&D, plus $1000 prototyping costs.

'Actual' costs $3000 (generous), vs $50 x 1000 unit sales if you're lucky !

Reduce the retail to less than $10 (your sell around $5-6), and you may have a chance to sell 10K units before the big novelty manufacturers walk away with it...
https://www.google.com.au/search?q=LED+spinner&rlz=1C9BKJA_enAU630AU630&hl=en-GB&prmd=sivn&source=lnms&tbm=shop&sa=X&ved=0ahUKEwjQs7GYjevSAhWBi5QKHSncAMEQ_AUIBygB&biw=768&bih=909

Stick with the tee-shirts.  The spinner is old technology being re-hashed.
Sorry.

[f4eru]

I think you have no idea of the investment going into that kind of product.
Take for example the case. You claim it's a blow molded case. This is not possible, given the shape they present.
You need injection molding to realise the struts, and the tight fit.

So you need a very complex polished mold ( or two, if you cannot make a shape that matches both sides.
Now to have clear plastic, this mold need a very very careful hand polishing.
Mold cost : 25000 Euros.

I don't think their pricing is exaggerated. People like you overseeing the huge NRE costs for making physical things are one of the reasons why kickstarter often fail.....

[SL4P]

I agree if that's what they're doing, and also maintain all the other points.
Why spend 25K to tool for a 35K sales proposition ?
The only way they'll get anywhere near a realistic cost price is simple kiddy-style assembly.

[janekm]

(snip)
Now to have clear plastic, this mold need a very very careful hand polishing.
Mold cost : 25000 Euros.
(snip)

Possibly 25k if you're getting the mold made in Germany. This is a pretty tiny two-cavity (most likely, unless the pieces can be symmetrical as you point out) mold, maybe 5k$ in China.

[kosine]

Injection moulding is expensive like maintaining your car is expensive.

If you pay someone else to do it, expect a big bill. DIY, not so much.

Check out ebay for the price of a secondhand moulding machine. Few thousand dollars. Blank mould base? Less than that. CNC machine? Hobbyist have been making them for years.

A suitable case mould for this thing could be made in a couple of weekends - all polished up and everything. OK, it might not be quite as glamourous as a fully spark-eroded and polished tool-steel mould, but it'd be better than a quality 3D print job, and certainly good enough quality for an initial batch of a few thousand units. (If they sell, then you have enough gold to gild the lily...)

I expect they'll get a cheap mould done in China. Won't cost a lot if they know the right people, but it's a shame they're not doing it themselves.

[kony]

Injection moulding is expensive like maintaining your car is expensive.

If you pay someone else to do it, expect a big bill. DIY, not so much.

OK, show me one sucessfull product done with mould set produced this way. 

[kosine]

Here's four off the top of my head:

https://shop.pimoroni.com/products/micro-metal-gearmotor-to-lego-axle-adaptor
They had 5000 purple adapters made using a cheap hobby CNC milling machine and secondhand injection moulder. Took less than 2 weeks. They're now looking at setting up their own little mould-shop as a result. The black raspberry pi lid in the attached photo is probably next on the list.


https://www.kielowatts.co.uk/shop-1/
All the "Contact" range casings were produced from ABS in a similar fashion. The touch plates were laser cut and glued on. He's got a couple of other versions in the pipeline as well.


http://scratchgpio.blogspot.co.uk/
The red plastic end caps for the aluminium case were done on a mould you can hold in your hand. (Think he's made a couple of hundred so far.)


http://www.rapstrap.com/
They shift about 20 million of them every year. First 100,000 units were done on two 6mm plates of aluminium. It was on TV and everything.


I could show you more, but check out this guy on You Tube if you're still not convinced:
https://www.youtube.com/channel/UCJT6JcjZCttm5PHnW9kWluQ/videos


I've just submitted a research paper with the local university comparing the technique against 3D printing. The cross-over point is about 50 units. After that, it's quicker to make a cheap mould. If you can get access to small moulding machine it's not actually all that hard. You can easily do it in your garage. (And there's a lot of small moulding firms out there who'd be able to run your mini-mould on one of their machines.)

[kony]

I'm aware of benchtop units like babyplast for micromoulding in small volumes, and there is plenty of companies doing prototype AW7075 moulds for low volume/design validation scattered across europe.
However the inhouse prototyping does not bode well for larger or more complex plastic parts - just the sole design of the mould is becoming cumbersome, often involving actvie mould preheat and cooling, and is hard to get right without several iterations if you don't have access to flow simulation tools. Some part features mandate use of multipart moulds where you cannot skim on machining precision (so unless you have EDM wire & sinker in your garrage, bad luck), clamping pressure of the press might come in as a serious limitation as well. Generally you are best off farming it out to people specialising in this field, unless the part is really trivial or not time/performance critical.

What I would recommend is to take very close look at local CNC sheet metal processing companies and their capabilites, and browse trough few catalouges of press-fit mounting HW for sheet metal. Sooner or later realize that >80% of your issues in mechanical design can be solved by properly designed sheet metal structure (worst case with few welds), and remaining bits tackled with conventional machining. At least that is how it works out for me (chemical laboratory instrumentation & research tools).

Mind adding DOI of the article later?

[kosine]

Don't think the article will be published for a while. It's for a conference that isn't until September. I think there are plans for more research, but I don't know when. (I'm just a co-author / outside consultant.)

It's very true that there are limitations on the size and complexity of what can be done on the cheap, but it's not as limiting as you may think. Certainly a case for these spinners could be done. A 20-40 tonne clamp would be more than adequate - and that's really small by industrial standards. (It's car bottle jack territory.)

Temperature control is not really a problem either. Any industrial moulding machine made in the last 50 years has all the necessary heating and cooling systems built in. It's not something you model. You just fiddle with the settings until it runs. (Seriously, that's how it's done! I used a Sandretto machine about a year ago that had everything labelled in percentages. Percentages of what?? It didn't say! And this was at a well known moulding company. If you're in the UK, you will have certainly seen their work.)

Hot runners are a little more complex, but if you're only making one unit at a time then you don't need those. (They're only PID controllers inside, anyway.)

Tolerances of 50 microns are routine on hobby-level CNC equipment, and that's good enough to make Lego parts. Producing multi-part moulds is therefore not a problem either - they do fit together exactly. And because home-brew soft tooling is quick, it's also easy to iterate the design. (You can't do that with conventional "professional" moulds. They can easily take 10-12 weeks to complete and are a right PITA to modify. And no matter how much computer modelling you do, they very rarely work first time. I've seen 1 tonne "paperweights" on many occasions over the years.)

Hobby-level EDM is certainly something that would be nice. I think AvE was working on one, but you still have to CNC the electrodes for mould making, so it's not a show stopper. (Wire EDM is more useful for extrusion.)

[Corporate666]

Tolerances of 50 microns are routine on hobby-level CNC equipment,

I don't know about that.  I am not saying you're lying, but the people making those claims very likely are full of shit. I am not sure why, but it seems there is a competition among the hobby CNC guys to beat each other on specs.  I've never seen more ridiculous claims of performance than with hobby level CNC or 3D printers. 

I wouldn't doubt that some hobby CNC machines are capable of machining some parts to 50 micron tolerances, but I would be very skeptical about whether someone could create a mold on the vast majority of hobby level machines and hold 50 microns where it counted.  And by very skeptical, I mean, I'd probably call bullshit.  Now, it depends on what the machine is and what the operator skill level is - but I am envisioning some 80x20 extrusion contraption with acme screws and a glorified rotary tool motor with an ER collet holder as a spindle and no coolant cutting chunks of 6061 aluminum... and in that case, I definitely call bullshit

But that said, I have seen people doing cool stuff with Miniplasts and such.  But there's definitely parts that require "real" injection molds (features, finishes, etc).  Most of the hobby parts you linked above, while cool, seem more functional and have a home made feel to them.  Nothing wrong with that, though.

[Koen]

There's a whole other world of hobby-CNC machines out there. People with experience, knowledge and access to tools reproducing industrial machines.

Anyway, the prices above for injection molding are way too high.

[kony]

Problem is that for soemone, hobby level means HAAS machining center, and for other it is 500$ ducttaped together chineese engraver. The two orders of magnitude price difference makes its mark on the performance. 2000kg of cast iron vs. 30kg of aluminium extrusions makes helluva difference in stiffness and vibration dampening.

[frozenfrogz]

Anyway, the prices above for injection molding are way too high.

For such a simple and small part, mold making is 5000€ max. It depends on the machine, but since you don't need cores to be pulled, you don’t have to build a full tool – just simple mold plates and you are good to go. The tooling depth is only a couple of millimeters and without texturing required even 5000€ seems overstated.
Cost per injection depends on the plastic used and cycle time. The latter can be estimated as being in the lower range because the part has rather small wall thickness – what results in lower cooling times – and injection pressure phase will be quite short, because the part is simple and there are no crazy geometries that the plastic needs to flow around. We are talking 20 to 50 cents here I guess
Quite possible you could go with both halves of the case being identical with snap-together connectors. In that case the tooling gets even cheaper. I am thinking more like 3000€ for the mold and 30 ct per complete case – drawings / engineering done in Germany, tooling and molding made in Asia, as it seems to be common these days.
All done in Germany, maybe double the cost. Then again there might be companies that just would not care about an inquiry for such small volume orders, or charge you 20.000€ for the tooling...

[kosine]

My CNC machine is an older version of this : http://www.axminster.co.uk/axminster-engineer-series-zx30m-mill-drill-505109

Nothing special, but good enough for my purposes. 50 micron repeat positioning may sound small, but I use a lot of dental burs down to 0.7mm diameter, and under a jewelers loupe you can definitely see if it's out by 10%, even 5% with a bit of practice and a good eye. So yes, 50 micron is totally doable on modest equipment. (And I wouldn't dismiss claims of 10 micron on a newer machine and well-setup backlash compensation.)

As for the moulds and mouldings, yeah, they're still homemade, but as a proof-of-concept they're better than a 3D printed alternative.

And you can churn out hundreds of them in an afternoon.
And test out different materials on the fly.
And work out moulding issues like where to feed and where to vent.
And learn what size machine and what settings you need for mass production.
And impress potential customers and investors.

And it's also quite satisfying!

[Koen]

You are talking about repeatability but Corporate666 about accuracy. Anyway, hobby-level dial gauges are cheap, useful and would confirm/infirm your observations.

[kony]

I'd say the community of DIY CNC machinists is split into two groups - those who choose to believe, and those who measure.

What often comes up after a while is that people claiming such good numbers do not have in fact any measurement tools capable of verification nor knowledge about usign them. Numbers derived from stepper motor angular resolution multiplied by HW interpolator is really not what decides here (stating obvious, but not for everyone as it turns out).

10um positioning accuracy claim on such category of machines, especially with backlash compesnating system without absolute position feedback is nonsense - not even the ball leadscrew will have enough linearity to guarantee it - as cheap ones are usually rolled, not ground. That being if the bulid even uses ball leadscrews. Also axis loading from cutting side-forces, differences in axis speeds and accelerations make backlash magnitude strongly variable. About the only case where I'd agree that SW compensation makes sense on such machine frame is coordinate dill with constant positioning speeds and no loads on the compensated transversal axis.
If you said 0,2/100mm, then yes - on well maintained machine.

BTW: does anyone here have expirience with doing polycarbonate optic elements on small (benchtop) presses? The refractive surface mould insert is not that big of an issue for me, but having to do embossing in vacuum oven really puts it on the edge of non-viability due to troughput and manual labour per part required.

[frozenfrogz]

What often comes up after a while is that people claiming such good numbers do not have in fact any measurement tools capable of verification nor knowledge about usign them.

Properly calibrated instruments and a suitable measuring room will presumably be more than the whole machine setup for the proclaimed 10 micron Range
Real life accuracy of 10µm is something you are talking about in the field of micro-milling and spark-erosion.
For general milling, take that number and up it by at least one order of magnitude.

Here is a nice read on tolerances for starters.

[kony]

Yeah, we can agreee on the costs side, not to forget the implicit temperature (and humidity if plastics are involved) controlled enviroment.
But then again, we are talking about toolsmaking, not tractor repairs. 

However it is not that of a remarkable achievement to work down to IT6 grade in homeshop when it comes to conventional machining (well, defacto grinding) processes when doing critical features. The arbitary/complex shape requirement (release tapers included) of moulds is what makes it very hard to DIY.

[kosine]

I'd need to see your optic elements to pass comment, but I've successfully produced lenticular card lenses in the past (https://en.wikipedia.org/wiki/Lenticular_printing). Hardest part was actually getting a good enough printer - 1200dpi means there are 1200 dots scattered within the inch, just not in nice orderly rows..!

As for accuracy, there are many things that affect it, even thermal expansion. (Steel is about 10 microns per meter per degree C.) A spindle can easily warm up 10-20C over a couple of hours, which can affect the Z axis by 25-50 microns.

And I think most engineers worth their salt would always check their work with decent measuring equipment. If I make a cut that's supposed to be 100mm wide, I expect to get closer to 100.0mm than 99.9 or 100.1mm. So 100.00mm +/- 50 microns is what I typically aim for - and generally get. (Though cutter compensation and spindle runout do start requiring attention.)

If you want a real world example, have a close look at the attached photos. The pink things were some rubbery mounts I made a few years ago for an action camera case a friend was making. It's a two-sided mould, so if either side is out by even a fraction it doubles up and you get a big overhang where they meet. Can you see the joining line down the centre of the one I'm rotating? It's barely perceptible even when magnified.

And if you have a look at some random plastic parts you have lying around you right now, you'll see much, much worse. This scrubbing brush I use to clean up swarf came to hand. It's ghastly! (You can tell I didn't make it!!)

[kony]

If I make a cut that's supposed to be 100mm wide, I expect to get closer to 100.0mm than 99.9 or 100.1mm. So 100.00mm +/- 50 microns is what I typically aim for - and generally get. (Though cutter compensation and spindle runout do start requiring attention.)

Sorry to nitpick, but +-0,05 is not 50 micron tollerance, that is 100 micron one! 

Yes, my case is actually quite close to lenticular multiplex array, at leas in sense that it is repetetive in the pattern and planar on one side. I'd suggest to continue this over PMs not to spam this thread any further (oops).

[Simon]

What is the actual point of the device ?

[Don Hills]

What is the actual point of the device ?

As the name implies, you fidget with it.

As well as being something to be seen playing with, they're useful for people with some anxiety disorders. Focusing on manipulating the device helps to disrupt the anxiety feedback loop. One common device for this was a simple rubber band around the wrist - when anxious, stretch and release the rubber band.

[Corporate666]

My CNC machine is an older version of this : http://www.axminster.co.uk/axminster-engineer-series-zx30m-mill-drill-505109

Nothing special, but good enough for my purposes. 50 micron repeat positioning may sound small, but I use a lot of dental burs down to 0.7mm diameter, and under a jewelers loupe you can definitely see if it's out by 10%, even 5% with a bit of practice and a good eye. So yes, 50 micron is totally doable on modest equipment. (And I wouldn't dismiss claims of 10 micron on a newer machine and well-setup backlash compensation.)

As for the moulds and mouldings, yeah, they're still homemade, but as a proof-of-concept they're better than a 3D printed alternative.

And you can churn out hundreds of them in an afternoon.
And test out different materials on the fly.
And work out moulding issues like where to feed and where to vent.
And learn what size machine and what settings you need for mass production.
And impress potential customers and investors.

And it's also quite satisfying!

Who cares about positioning?  That means nothing.  The mold you're cutting doesn't care if you got the endmill in the vicinity of the toolpath you were trying to cut.  All that matters is the finished part off the machine. 

I don't mean this in reply to you at all - so please don't take any offense.  But I always laugh when I see these ridiculous claims (usually from hobby 3D printer makers/owners) about how their machines will hold 50, 20, 10, 5 microns or whatever.  It used to be that they would perhaps take a stepper resolution and divide by a gear tooth count and quote some theoretical number, but now it seems they just pull numbers out of thin air, or perhaps they just use a 32 bit number instead of a 16 bit number and quote the theoretical accuracy based on how many bits they are sending to the motor driver?  But I digress

For making a mold, all that matters is what gets cut.  I've seen hobby machines make parts with tooling marks that are way, way over 50 microns.  Holding 50 microns is pretty difficult, even for a 'real' CNC machine.  It varies depending on what material you're cutting, what sort of cutting operation, how you're holding the material and all of that.  I'd definitely agree that some hobby machines can hold 50 microns under some circumstances.  But then again, I can throw a dart and hit a bullseye under some circumstances... but the only thing that counts if I wanted to be a professional would be my ability to do that reliably again and again.  I'd say "some circumstances" would probably be some small percentage of the time when conditions are right and the operator knows the machine well and so on.

No offense to the hobby machines intended or their owners.  But 50 microns most definitely isn't easy.

EDIT:  I just checked out the machine you have.  Looks neat!  Very similar or maybe even the same as the Grizzly G1007 I started out with years ago.  But if yours is similar to my Grizzly, it uses acme screws - and in that case, that is not machine that will hold 50 microns unless you go to lengths to work around its limitations - like putting measurement instruments on its axes. And there's enough backlash in the screws that cuts are generally a "set it up and make the cut, then stop it and set it up for another one" rather than "make a program, hit go and wait for your part to be done" affair.  My G1007 had to be re-setup every time I raised and lowered the head too, due to the cylindrical column.  And the spindle bearings and R8 collet setup were very limiting and contributed to enough runout that anything other than quite short tooling was already pushing the 50 micron boundaries just due to that issue alone.  And I'd say your machine is most definitely on the very high end of hobby level... IIRC a lot of people CNC those things (or did back in the day).  People were putting laser pointers with mirrors on the head to get the head re-aligned when they moved it.   No knock on your or the machine - sounds like you are getting fantastic results but would you say you could make an injection mold in aluminum on that machine and be able to hold 50 microns on the finished part?  Including curves and transitions and radii and such?  I think that would be too tall of an order, IMO.