What we learned making a plastic injection mold with a Chinese mold maker
airgradient.com
airgradient.com
"I have a great idea, I'll just manufacture it in China and sell it as a product!"
This is a great overview of the process and complexity for *one part* of the manufacturing process for bringing a product to market and almost all of them are as complex (or more) and have pitfalls of their own. Add on wiring harness mfg, PCB/PCBA, machined parts, compliance testing, final assembly, QA testing, and revision management and then you're about halfway to running a successful product business.
From personal experience, the best advice in the article is to hire a domain expert when moving into a field you aren't intimately familiar with. Employing them for as little as a few months can help you avoid several years of very expensive lessons as you figure a new process out.
Downsides are obvious, they work for the supplier and not for you.
The factory can never help tweak your initial concept, right at the beginning of development (when it's ~free!), to save tremendously on future production cost. That's what you hire the consultants for.*
*Disclaimer: I'm just such a consultant, though I work on the EE side.
Numerous small design choices early on can have drastic impacts on the complexity of the mold and how expensive it will be.
A simple example: designing your side I/o ports to be on the parting line, or move the parting line up through the hole centers. Didn't think of this? Now you need a slide.
Deciding the surface finish you want also impacts hard it will be. You can hide a lot of flow problems and tooling marks with a nice rough textured finish.
Some colors show cooling and flow worse than others. Clever gate placement hidden under something hidden in assembly makes cooling simpler. Etc, etc
Well worth the money spent on a guru to look at your design and provide feedback. (I am not one, I've just hit every wall)
I think it's really a problem of not knowing what you don't know yet.
It happens with both plastic and other type of enclosures (metal etc). The better designers understand the manufacturing process and build their designs thinking about how they will be created in a production line. As others have mentioned, most manufacturers will say 'yes' to anything you request and charge accordingly. They'll just figure it out somehow, but you'll be unaware that you've doubled the cost of your part due to a lack of understanding of construction.
I guess it happens in other fields as well, i.e. architects vs builders in the construction industry.
Yeah it's not something you need a skilled artisan to actively control, but a lot of technical skill goes into making consistently high quality parts.
Which is, honestly, kind of a reasonable assumption. In construction, you hire a general contractor. Why is there no manufacturing equivalent of a general contractor? Because all such people are too busy running their own successful product businesses?
Houses are much more similar to each other than products. If you've built a few houses, you can probably find many more customers who will want similar houses. Most people don't care if their house is essentially the same as somebody's on the other side of town.
But a new product must offer something that's not found in the existing products on the market. If it doesn't, then there's no reason to go to all the effort of getting it manufactured — customers could just buy the existing products where all that investment was made already and is largely recouped.
So an industrial design product is more like the houses you see on the "Grand Designs" TV show. Each of them represents a dream and a vision, and usually there's a lot of trouble and blown budgets on the way to the grand finale where the family finally gets to sit together by the fireplace in their unique creation. ("Grand Designs" doesn't show the dream projects that fail miserably.)
how is it then the case that there are multiple versions of products on the market that differ very little from each other?
Let's say you want to sell plastic brooms. With such a common product, what possible competitive edge could you get by creating your own design from scratch? If your reason for getting in the business is that you want to address a niche or trend (e.g. you want to sell hot pink brooms because it's Barbie summer of 2023), that can be accomplished with an existing design. It only makes sense to create new molds for a broom if you're a corporation like Ikea with the massive sales volume that enables vertical integration of every aspect of the product.
You subcontract to various manufacturing companies — companies also being used by your competitors. Those companies have standard "best practice" ways of doing certain things, and they tell you that your design will be executed a lot more cheaply if they're allowed to nudge it toward conforming with those best practices, so that they can use standard parts and techniques.
Iterate this three or four times — especially in the context of a company that's plateaued in market share and is now all about cost optimization — and eventually you have a version of your product which is indistinguishable from your competitors' products. That's what allows for the best economies of scale up the pipeline, and therefore saves you and your competitors the most money.
Houses that are the same as ones across town are cheaper as the builder has experience with it and knows the tricks to build it with less labor. They also are designed with years of feedback about what other like and dislike. They don't have weird corners that are not usable because after putting things in that is what fit.
In a way it's like forking a software project: the more you deviate from what's done before, the more 'pain' you'll be in.
What's done before can be re-done by any competent builder.
But beyond that, you'll need expertise in different fields, to fit parts of a design together in new ways.
No expertise -> project will fail, or be extremely 'painful' (like: costly, time-consuming, many iterations, result isn't what you wanted, etc).
Enough expertise -> steps in new territory might be uncomfortable or include some mistakes, but overall you'd know where you're at, where you're going & have realistic expectations.
So yeah - nothing wrong with getting some expert help when entering new territory.
That assumes part availability, though. There were kit-built octagon-shaped homes in the 1800s; but good luck asking a builder to build you one today.
Great analogy, seriously. As a big fan of that program (cathartic if you have any real state project going on) let me add that I have seen programs in which the house is not finished after 8 years; some in which rooms projected for kids are not needed because they are going to college; and couples that start together, and by the end of the program are divorced (or one of them passed) by the end of the program. It really puts things into context.
Also, my favourite, when 80% of the participants casually say they want to be in by Christmas, summer vacations, New years next year, and the host smiles and says something like (let's see).
That's the thing that I would expect here: that the firm you'd hire would come with a built-in person to look at your design and say "no, this isn't gonna work, try again."
Reputable product manufacturers actually do have that person.
There absolutely are such places. They're called product development companies, and I work for one. We do just about everything, which is nice for people who like variety!
Can you name some companies in the space? I'm curious to know more.
https://metric-designworks.com/
Looks like a very humble site/company, but then you see that they have Teenage Engineering as one of their clients, in addition to Microsoft, etc.
When I have dreams of having a cool successful kickstarter around some quirky hardware musical instrument I’d invent, I definitely see myself reaching out to a company like that..!
When I imagine a company (rather than an individual or "brand") that sets out to get a product manufactured, what they likely already have — and so don't want to pay for the redundant employment of — is in-house IxD and general hardware and product engineering expertise. This is the type of talent that's easy to hire for in pretty much any developed country; so it's who they've probably hired at the very beginning of the product development lifecycle.
What such companies don't necessarily have, is 1. domain expertise in engineering the specific type of product and the parts that make it up (because this may be the first time they're entering this particular space); and 2. ground-level "systems engineering of logistical pipelines" knowledge of how to best put the plants, mills, factories, shipping, etc. of a particular region to use.
Big companies like Apple "vertically integrate downward", extending themselves into their chosen manufacturing region, getting boots on the ground for months/years to grow in-house expertise these fields, perhaps headhunting engineers and foremen directly from the factories to oversee things "from their side" but still local to production, etc.
But small companies don't have anyone to send to the manufacturing region. They want to draw the line "on the ocean." What they want, is to find a company local to the manufacturing region, that has taken the sort of local-logistics-oversight and part-specific-engineering competencies that were developed locally as a response to the demands of these big clients, and "struck out on their own", white-labelling this capacity as a service to be provided to anyone willing to pay for it.
To go back to the "general contractor" analogy, these smaller companies are like commercial property developers who already have architects and structural engineers working for them; who are looking to get a design of theirs built in an entirely different country; and who are looking for a construction company (that employs its own general contractor, who will be the property developer's point-of-contact) in that country. The property developer's structural engineer can say whether the building is safe in a technical sense, but they will rely on the construction company to evaluate whether the building is up to code for the area of the world it's actually going to be built in; and therefore to push back on the property developer to get things changed, when the building "won't work" as planned. They'll also rely on the construction company to guide them toward materials and parts choices that are most idiomatic for the chosen area, and most efficient and abundant in the construction company's own supply chain.
I do realize that a product development company might be willing to do a lower-level job like this. But the fact that a firm like yours employs its own product and design and high-level engineering staff (at least, I presume), probably means your firm has higher running costs than a firm that didn't have those staff; and therefore needs to charge an amount that might be untenable to smaller clients who only need that lower-level kind of support.
Which, again, leads me to wonder why there's no kind of company specifically targeting that "vertical slice" of the logistics chain — being only the gluing know-how to bring together on-shore "product developers" who have no off-shore experience/relationships, with the off-shore low-level manufacturing they need to use to make their product real. If a company that was only this existed, I would bet that it would be a "best practice" for small product companies to use them, instead of trying to send their designs off to factories themselves!
And again, I'll state my hypothesis: this glue is so core to the value of many of these products — especially lower-BOM ones — that any company that starts out just providing this service, inevitably vertically integrates upward until they are a product development company; or even becomes a product company themselves. At which point they leave the low-margin "cross-shore oversight" jobs behind — or start charging as much for them as they do for the full-spectrum development jobs.
And once you're hiring skilled practitioners, the space you've envisioned for a glue company has been squeezed out. We bill by the hour, and we'll do small jobs. Would you rather hire us to make your PCB or someone who's never soldered?
And as for us hiring that sort of person, they're already employed at the contract manufacturers (and unavoidable, the CMs don't really like to work without it). No reason to duplicate that layer. Understanding the role of CMs is critical to understanding modern volume manufacturing.
What's to like about outsourcing it. It's parallelizable so you can work on your core competency while they work. Core competency, this is your angle that allows you to make money. How to lose money, working on stuff outside that. They get it done faster and without fuckups. And the outputs are something you can just give to a manufacturer. If there is an issue the two of them hash it out not you.
Only thing not to like is the wad of cash you need to cough up. But really if you don't have that you're hosed anyways. And paying to get revenue coming in a couple of months sooner is so worth it.
There are design companies who do exactly this. You can go to them with a very generic problem like “design our fall shoe collection” or a very specific one “this bearing in this gizmo is too noisy, find a better one”. They can cost an arm and a leg but they get the job done.
I consult on product development and I'd say about 50% the time, software majority companies think you are trying to rip them off somehow when you try to explain everything that is NOT software that needs to be done, what it costs and how long it takes (and how important it is to lock things in early as you can't just "push out a patch" for a $50,000 molding tool or that you can't just "slap bluetooth in later" and not redo all of your EMC certs).
You can see how well the "regulation but no enforcement" regime is working by picking up any disposable vape from the street and looking for the WEEE "do not bin" symbol. It will probably be there.
Then Germany rejected again because the "CE" was obviously in the wrong style, on a generic sticker (I'd warned, but been told "It'll be /fine/."). And German Customs said our slip directing customers to our site for setup and operating instructions was insufficient. This time, they held our units until we sent satisfactory documentation.
Finally, we were able to send, but with after a warning from Customs that a third error would result in permanent bannage and seizure at the border of anything we sent to Germany.
Broadly and simplified:
Adding a bluetooth module means that, not only is the device no longer "electrically equivalent" (which means you are re-certing), but you have gone from "unintentional radiator" to "intentional", which means you have a different and/or additional set of standards.
A pre-certified module (as in, it has modular approval) under FCC means you do not have to test the "intentional radiator" part the module PROVIDED you exactly follow the instructions of the module supplier AND you don't do anything "stupid".
Under CE, pre-certified module isn't a thing. It will certainly help you pass from a technical standpoint, but doesn't really let you "short-cut" the testing and paper work.
Hope that makes sense and gives a rough idea.
edit Apologies for rehashing what others have already said. I should have refreshed.
The device is what gets certified, not its parts.
I wonder what causes this hardware/software incompatibility.
Heck even software-only companies have problems with this where the technical leadership is, for example, of backend background and they completely fail in frontend and data analysis fronts while having rock solid security.
What even would that role be called? I suppose CTO but if a company that does HW and SW.
Also there's T shaped skills and there's you with:
_________
/|\
/ | \
/ | \
/ | \
No snark taken, I’ve got plenty of blind spots in my skills so those T stems should be a little shallower probably but I’ve been fortunate to cover a lot of ground in the last 15 years.
Because of the focus on mass manufacturing, hardware is incredibly expensive to change. The industry seems to have centered around doing more planning up front.
But we (software people), learned 20 years ago (from select Hardware companies) that this is backwards. If the cost of change is high, and that's where your pain is, change more frequently. Mitigate that pain and make it a strength. (See: Toyota Production System)
I have a product with an addressable market of something like 300 to 500 units per year. That puts me in the Defense/Aerospace/Medical area, where unit economics are completely insane.
And yes. It does feel like being ripped off when someone puts ~30 minutes ($20 max) of Western labor into something and then turns around and charges me $800 for the privilege of supporting an American company.
Ironically, Toyota is moving to gigacasting (large, single piece casting) to reduce costs [1]
[1]: https://insideevs.com/news/687481/toyota-new-ev-production-l...
I guess these giga casts are not very repairable.
Every collision with body damage fixed by an autobody shop is stealing money from our overlords.
This doesn't detract at all from Toyota's ideal batch size of 1, or from the lessons they learned and shared on the way to such huge scale.
My claim is that we (the West) are failing to succeed in manufacturing (and have been for decades) because our focus is on quantity and the very people who could solve the core problems will be so busy planning out the next 2 years, they will not see the wave cresting above their heads.
So as a senior expert in your field, you make $20/hr? Because I don’t know any field that charges by the half hour, and I don’t know any senior engineers that make minimum wage to do their job but maybe you’re the first.
That’s not including the markup the company would need to add assuming you aren’t dealing with an independent contractor.
Not that it detracts from your overall point, but some of the best paid knowledge workers on the planet - lawyers - regularly bill out in 6-minute increments.
Beyond that, just the average documented phenomenon of task switching is 20 minutes or 0.3hrs, so your "only six minute" project really costs the engineer 0.3+0.1+0.3 hours.
But sure, if you are in a long-term project (here defined as more than an hour or two total), you may see line items on your bill of 0.1 or 0.2 hours in a day for a quick call or something.
Are you such a knowledge worker? Could someone actually describe to you a problem worthy of your consideration, give you time to think/calculate, then describe a response in only six minutes? Sure, there might be some such edge cases, but commonly?
Drive to analyze a broken electrical box at customer. Analyze what is wrong. Read the schematics. Repair it (without getting fried). 15 minute repair. 2 hour drive. Zero glory, just a shitty job.
Or exchange something that was manufactured wrong and test it. Climb to a wind turbine, open the box. Exchange a component. Test it. Climb to another 50 towers to fix same problem.
Testing short batches of custom products. The testers are paid more when compared to blue collars, but still earn shit. Also everyone assumes that they are "factory workers" while they program their elecronic testers in C++. Paid a fraction of what the FAANG intern earns.
Life is unfair to many many people. Looking at your post and above - it feels like you are some guys in IT who dont know how hard it is in other areas.
Other alternative are bookkeepers, who run books for multiple small customers. Get easy to impossible questions about tax law - all day every day. From your 200+ clients to whom you provide advice by phone. For which they dont want to pay.
A team of few accountants tries to book invoices and fill taxes from their clients, while being bombared by questions. Often super tough questions. Literally 0/10 experience for a knowledge worker. Every day.
This is what I find so profoundly offensive when so many Western companies charge insane labor rates. I know for a fact that very little of that money is going to the person doing the work.
While it is very unlikely that you'd be able to get and pay for only a six-minute increment of service from a professional, those pros put in a LOT of time outside the six-minute increments for which they are billed out.
An attny in the family has a specific external billing rate and required yearly billable hours to produce. Yet she is burdened with literally hundreds of hours of extra admin, firm, "culture & innovation", interruptions, etc. work (much of which should be done by paralegals, clerks etc.), for which no credit is given. I've seen similar in mech/elec engineering.
I'm taking about working with someone who repeatedly moves materials through a long established process.
It can be arduous and unpleasant and rarely pays very much. We can debate the ethics of all that, but it's not reasonable IMO to pay pennies and charge 40x.
If you are just looking at it as labor then yeah, you are going to feel ripped off.
But it is not just labor, it is labor + overhead: Office space Tools required for job Paperwork associated with the job
I'm not in SF.
Workshop space here rents for like $1-2/ft.
Paperwork isn't hundreds per hour.
Yes. I agree, machine tools are outrageously priced, but it's not quite that bad.
We've got problems here that need attention.
I have manufactured in that area and while you are not going to be in the mass volume pricing levels, you don't have to be in the defense level either. You just have to build the product with the right trade-offs and design the appropriate manufacturing processes. Few seem to know how to do this.
Edit: let me expand. Don't injection mold if you can avoid it. Do resin castings or thermo-forming, neither require as expensive of tooling. 3D print parts with a good material like PC CF. Metal fabrication is easy and cheap these days. You can get a shop to laser cut, form, insert PEMs and powder coat for very reasonable prices. We are starting to get a lot of machined parts from China. You can source wire harnesses from China also. Do assembly with a small team. Use as much off-the-shelf electronics as possible, but don't be afraid to make small and simple PCBs if it makes your product cheaper and simpler. Leave the complex boards to a vendor initially because while they might seem simple to design, they can be complex to debug and production test properly.
I've burned tens of thousands of dollars with them. They have tons of sales, little to no support. They have zero ability to stick with a vendor who does good work. They're a total crap shoot on quality and no responsibility when it fails.
"The Handyman's Invoice" is apocryphal in the details, but almost certainly true in that something basically like it happens all the time.
" The Graybeard engineer retired and a few weeks later the Big Machine broke down, which was essential to the company’s revenue.
The Manager couldn’t get the machine to work again so the company called in Graybeard as an independent consultant.
Graybeard agrees. He walks into the factory, takes a look at the Big Machine, grabs a sledge hammer, and whacks the machine once whereupon the machine starts right up.
Graybeard leaves and the company is making money again.
The next day Manager receives a bill from Graybeard for $5,000.
Manager is furious at the price and refuses to pay.
Graybeard assures him that it’s a fair price.
Manager retorts that if it’s a fair price Graybeard won’t mind itemizing the bill.
Graybeard agrees that this is a fair request and complies.
The new, itemized bill reads….
Hammer: $5 Knowing where to hit the machine with hammer: $4995 "
Sometimes you just need someone with a little dexterity and a willingness to care about their work.
That person shouldn't be billed out at hundreds of dollars per hour, especially when they're almost certainly being paid closer to $15.
There are plenty of different fabrication technologies to suit different scales. Some listed by other commenters.
I agree iteration is good, and good hardware companies do iterate hardware. It's something I have always been a proponent of as a hardware engineer. But also the reality is that everything in hardware land costs 10-100 times as much and takes 10-100 times longer.
To properly develop and iterate on hardware, you need to do upfront investment. You need:
1) Tools for the hardware engineers (scopes, spec-ans, tool shop, test equipment). [In software parlance, download the IDE, debuggers and libraries. You're not going to ask your developers to work without those. That would be ludicrous.]
2) You need to iterate EARLY in the product cycle process and this costs money (but still much less than changing your final tooling). [Software parlance again: You only ever really edit the source code. You would never ask a developer to binary patch a compiler to fix issues in your source code. This is what you're doing IMO when changing final tooling on a product.]
3) You need to account for lead times for fabrication and shipping of a physical thing. [Software parlance again: Compiling takes days to weeks. QA weeks to months.]
I think software majority companies struggle with these because (when compared to hardware):
1) Software tooling is essentially free.
2) You can stop and go back to the source code to re-iterate at any time and the down stream processes to "finish the iteration".
3) Time for iteration is seconds, minutes maybe hours while you wait for a compiler to run.
Speaking about Toyota Production System: This has many misconceptions as nobody ever seems to actually read the source material. One of these myths is iterations. Toyota iterate often and EARLY. You get the part right and THEN you can THINK about final tooling. They don't iterate on final tooling, that is expensive and time consuming.
I would say with some work and trade offs you can save on your product in the 300-500 range. You shouldn't be in the Defense/Aerospace/Medical area unless you are doing something quite esoteric.
As others have stated on the final point (but I will reiterate) no senior software person in the US would be happy with $20 for 30 minutes.
I suppose it's possible we've reached the end of history and there's just no way anyone is ever going to meaningfully improve on the production systems that we have today.
They're basically perfect, right?
There also really aren't any small markets worth serving. Like people with disabilities as one example.
Perhaps we'll just stop trying to build small batch products inexpensively because it's so hard and also those CEOs need 40x markups on labor to make ends meet.
Maybe I'm wrong about the direction, maybe tool and hardware companies should double down on big design up front and extend the time it takes to do their job.
Maybe, but I sincerely doubt it.
There are also safety concerns, temperature and humidity testing, and so on to consider.
Think of it this way: If current runs through it, you need to cert it. Period.
Edit: I think the Pi might already have FCC so if you are NOT electrically altering the Pi in anyway, then you won't need to re-cert (as you are essentially installing software on it and for a consumer device FCC doesn't have ESD requirements).
For CE, depending on the directive, you'd need to do ESD (adding a box will alter that).
Disclaimer: Not compliance advice.
The depth of knowledge required to do it well warrants full university degrees in the topic. As with most things, you can probably take the autodidact route but be prepared to make a lot of mistakes. Also be prepared for those mistakes to cost 5 or 6 figures. The ease and almost zero cost of exploration and learning from failure that software allows does not always map to other domains. This is one of them.
Source: partner is an industrial designer and geeks out on plastics, materials science and manufacturing techniques way harder than obsessions I pursue. It’s literally her entire world and she still fucks up occasionally too.
Yes, but not in a way that's unique to programmers.
People often underestimate the effort a job entails, be it teaching, sales, programming, graphic design, machining, car repair vet medicine, etc.
Most of my career history, when I've heard someone say "ugh, department X just sits around all day and occasionally does Y", they're assuming that the sliver of their insight into department X is the totality of what is actually done.
It is very easy to convince yourself that you understand something because you read a lot about it, and what you read makes intuitive sense. But that is not at all the same as actually going out and doing the thing, and solving the problem from the blue sky. Dunning-Kruger sort-of describes this, but I believe that there's a multi-dimensionality issue where being on the good side of Mt. Stupid in one domain makes you overrate your capabilities in other domains. Vis the nuclear physicist disputing the official explanation of how the buildings collapsed on 9/11, apparently because he hadn't gotten to the part of the freshman material science course that covered how yield strength decreases with temperature.
So articles like this are important, even as they are kind of obvious to people who have already learned these lessons, because often times it takes learning these things the hard way to understand why (for instance) the software is actually the easy part of an IoT thingy.
How many hw startups start with "we'll just ship an Arduino/RPi with our custom module and that's it" heh, that's very not ideal for a variety of reasons
Highly recommend it because they give a lot of details and if you start from the beginning you can really see how they learned how difficult it was in going into the space. The multiple visits, quality control, the issues with who owns what, and where they lost money.
But those are the entertaining ones innit. What fun is it to read a blog (like OP) that says 'yeah we hired a bunch of experienced professionals and we didn't have any problems'. No, we want to read stories of 'so I forgot about one detail and the manufacturer completely misunderstood us so I had to fly to China and spend 100k to get it fixed and it still was 3 months late'.
And employing all those people doesn't guarantee success, but it does guarantee your failures will cost a lot.
You'd probably want to only attempt making something like this as a vanity project, where it matters less if you fail, and thereby half the fun is navigating the path to success by yoursel[f/ves].
But, just for completeness sake, the list of people who got blindsided by the con artist who was ultimately behind a lot of the crazy included both our experienced local on-the-ground project manager and the factory owner.
Why did you go to China, as a two men shop? And not choosing a local company nearby?
And how much experience do both of you have on the hardware / manufacturing aide of things? And running global supply chains (because that's what you have, regardless of production volume)?
He had to cut milling bits with custom profiles to cut custom tools such as a series of graphite and copper spark eroder forms to cut just some of the corners and shapes in the (paper) technical drawings, then repeat for the next ill-advised bevel or interior sharp angle or even lettering which is too close together to get a milling bit inbetween two letters. Figuring that out was his responsibility alone. Instead of computer control, he'd have to do something like first create a scale model of part of the tool out of a block of plastic, then use a pantograph (a mechanical linkage allowing you to trace a shape with a pen and have the movement magnified/minified) to copy the shapes on a milling machine -- even shapes that seem simple. And then maybe throw away (actually, take home as a souvenir) a month or two's work if you cut slightly too deep -- these were high precision parts. Not because the plastic had to be so precise, but because the completed dies are made out of many steel parts that have to fit tightly together and slide past each other, as you can see in the article.
Considering the ~two months quoted in the article for this mold, I'd think a lot of that complexity is still essential despite modern computerized machining. I'm sure a great deal of ingenuity and many steps are still needed.
I've come to find plastics really quite interesting. It's kind of a magic material.
And at the same time it's a sort of like King Midas in which humanity gets this incredible material but now there's plastic everywhere polluting everything. Everything we touch is turning to plastic.
That also translates to customer cost
Yes, silk parachutes exist. But there was a reason why nylon was revolutionary - it was orders of magnitude cheaper and required a lot less care.
> Not to mention its unrepeatable and tends to get brittle with age. Meanwhile you can have a metal can opener thats 80 years old and works as good as new.
Plenty of metal can openers that are rusted, bent out of shape, or corroded enough to be unusable.
Plastic only usually gets brittle if it is left in sunlight (and not always then). Not sure what "unrepeatable" means in this context, but plastic is usually at least as repairable as wood or metal by using resin. Indeed, both wood and metal is often repaired by using resin based fillers.
While metals are known for their strength and temperature resistance, they are also prone to corrosion.
There are plenty of decades-old plastic parts in continued service everywhere, that you just aren't aware of because they haven't alerted you to their presence by failing.
One example: https://en.wikipedia.org/wiki/File:Plastic_(LDPE)_bowl,_by_G...
Agree mostly unless it's for something that needs to deal with water in some way.
A large number of modern medical treatments would be nearly impossible or prohibitively expensive without plastics, mostly for sterile use reasons. These use cases have saved countless lives. I'm not sure how you'd even begin tallying a count.
Beyond this, metal too has serious resource and energy implications. Think mining, smelting, production of the final shape, treatment against rust, paint, transport, etc. There are many steps, many inputs (notably chemicals and energy), and quite some demanding work going into it.
I have no answer to these questions: is it more or less resource efficient to produce 20 plastic can openers or 1 metal can opener? Is it more/less environmentally harmful? Is it more/less socially harmful?
I equally dislike all the plastic junk that is bought for a couple bucks that are basically use for a moment and become trash. There are also amazingly wonderful things made out of plastic that last a long time when taken care of. Heck I still have toys from 30 years ago that were made out of plastic that are still in pretty good darn shape because they do not hit the sun. My rice cooker has a plastic shell and has lasted a long time. Same with my difference keyboards.
Be aware of your consumption but don't let it drag you down.
If we were to limit ourselves to only wood and metal there would be a lot of products both good and bad, that would most likely never have been invented. Tons of life saving medical devices, items that are hard to shape/mold using metal techniques. Of course everything could be made of wood and I don't know the science but I would think the cost to harvest wood might exceed that of the components of plastic. I might definitely we wrong there too but I just don't think its as easy as saying well we can make can openers and simple wooden implements so therefore its due to corporate greed we use plastic.
I’ve been a big fan most of my life and it was / is my favorite toy product by far, but only as I got older, I realized how the whole system is not just brilliantly designed, but the mechanical _execution_ of the parts is at least just as impressive.
…And that it wouldn’t be possible without plastics. No other type of material could ever achieve such durability, flexibility and precision; and that’s not even considering costs – no other materials _period_.
LEGO is a perfect example of a product designed AROUND plastics – not something turned to plastics to lower costs.
Mokulock may not be identical or cost the same, but it's basically the same product without plastic. They don't snap together or come apart as easily, but I like the idea of it. I'd love to see LEGO done in metal too.
I wonder if you could do it with horn. I'm not sure how you'd make the hollow bits of blocks. But it has the right sort of properties. Relatedly, see:
This is known as "draft":
https://en.wikipedia.org/wiki/Draft_(engineering)
I'm more familiar with metal casting, but plastic injection molding seems to have a lot of similarities, such as shrinkage and flow considerations.
Recounted in an article in Fast Company:
As Isaacson relates in Steve Jobs, most parts cast in molds have an angle that is slightly greater than 90 degrees, because the extra degrees make it much easier to get the parts out of the mold. That’s the kind of compromise neither Esslinger or Jobs was willing to make for the NeXT, arguing it would ruin the “purity and perfection” of the NeXT cube. So the sides had to be produced separately, using molds that cost $650,000, at a specialty machine shop in Chicago.
Just because the mold was really expensive doesn't mean that the marginal cost of each unit created by it is more than a unit produced by a cheap mold. In many cases, the marginal cost is significantly less as you are able to use less raw material thanks to variable thicknesses, etc. There's a break-even point where the expensive mold becomes cheaper/better and that's another area where experienced manufacturing consultants help out.
The reason is that you want the tool to release the piece and not get bound up. Shape and cooling also factor in, but for simple box type shapes with no large cutouts, this is usually predictable.
Here's a story [1] of a company that found that the Taiwanese plastic company that made its custom parts stopped responding after 15 years. They got another one of their Taiwanese suppliers to go to the plastic company to see what was going on--and found that the factory was now a hotel and the sales office was a now a strip mall.
[1] https://www.stlpr.org/arts/2013-07-24/on-chess-the-case-of-t...
Also if we’re talking about China and not Taiwan, even if you do have people on the ground, how likely are you to win any kind of contract dispute?
Same with with US based PCB shops. They all want to do high dollar ITAR restricted boards for govt contractors.
There is a middle ground. US sub shops that oversee captive Chinese operations with their own crew and keep you in the loop. This worked well for my personal projects and wasn't too horrendous. About 16k for a simple design that I could've had done with our work team for about 8k. But I never had to fly over, play WeChat tag in Mandarin, and I got PowerPoint DFM updates regularly.
1) They are unlikely to work with you at all unless you anticipate high volume orders (thousands at least, they prefer 100's of thousands).
2) Price: Get a quote from a Chinese supplier, double it and add a zero.
3) Lead Time: Get a lead time quote from a Chinese Supplier. Add shipping + customs slack. Double it and add another 30% for the excuses that are going to come.
4) Quality: Surface finish, dimensional accuracy, materials and process will be poor and no one will care. Refunds will be extremely difficult to get with Western suppliers.
5) Efficiency: The ordering process is usually very efficient with Asian companies. Western suppliers will drag you through a fully manual quoting process that involves days, weeks or months of bullshit back and forth while they wait for a slot in their broken production process to then rush you into commitments for.
5) Labor: Many "Western Suppliers" sub everything out to a random, lowest bidder in China anyway, even after all the bullshit.
6) There are now some 3rd party service aggregators (e.g., Protolabs and Xometry) that are attempting to collect rent on Chinese labor. They charge Western prices, offer Western lead times and Western levels of support, then sub the work out to random shops in China with poor communication, no consistency, bad incentives and unacceptable results.
7) There are some truly incredible, honest, hard-working Western Suppliers doing truly incredible work. If you've found one, do not take them for granted (e.g., Digikey, Craft Cloud, Adafruit, Sparkfun, Make Augusta, and more).
And of course, YMMV
It's interesting that China usually has the reputation of producing low quality crap, but somehow they get the large scale manufacturing right.
Were it not for those supply chain constraints, these companies would almost all disappear overnight.
That's a terrifying proposition, as we're in no position to compete with China.
Yeah, China earned a bad rap in many ways, but there are lots of companies there that have figured out how to do excellent business with westerners. See JLCPCB and PCBWay as two examples.
https://www.youtube.com/watch?v=RMjtmsr3CqA
Interesting history of a process that's everywhere and invisible at the same time.
It seems everyone reinvents the plastic-rectangle-with-grooves. Is there some obvious reason for this?
And is there some canonical place you can get some proven high-quality boxes in bulk?
A friend and I made some sensor boards and used off the shelf waterproof enclosures - but I wonder if there is some go-to standard enclosures that people go for
No to mention, if you want a unique industrial design for your product to stand out then you are going to need custom tooling.
I do this for my personal projects (volume < 5) and the results are nice and functional. You can also cut holes for buttons and whatnot and you get something well tailored for your needs.
Laser cutting can be quite big and make a large volume box, if that's what you need. And transparent plexi is nice (if it fits your needs).
For small production runs, definitely. There are standard aluminum extrusions for boxes with PC boards. You slide the PC board in, and provide custom flat end plates with holes for connectors and controls. Here's one of mine.[1] The end plates were cut on a laser cutter. Here's a supplier in China.[2] For small boxes, the aluminum extrusion alone should cost a few dollars.
Somewhere above a few thousand, custom injection molding becomes cheaper. Amusingly, these are better boxes than plastic injection molding, but don't look like consumer products.
[1] https://github.com/John-Nagle/ttyloopdriver/raw/master/board...
[2] https://www.alibaba.com/product-detail/CHANGHE-electronics-a...
And even further down the road: make sure to account downtime in your production schedule for mold maintenance.
I see others commenting on registration, certification and approval matters and I’ll only say: do not underestimate the time and cost of that for a global product. This landscape changes all of the time. One minute a country will accept evidence of an FCC filing; the next minute they require in country testing with a local authority.
But they cannot compete with the purple network in the USA right now.
I've suggested this before but if they somehow partnered with WeatherSTEM that could change things in a hurry. Even if they had to provide discounted units. WeatherSTEM appeared to have their own open database so I think the costs of long-term data storage would be avoided once integrated.
AirGradient's outdoor sensor costs $155, Purple's is $229, is 50% higher price "far too expensive"? From your comment, I expected Purple's to cost several times more.
Can you provide ball-park costs - how much was your mold, and how much did you pay for 5,000 units to be made?
Even if just a rough estimate that would be really helpful, thanks.
Can you do XYZ impossible feature? yes! Can you do it for cheap? Yes! Etc,etc
The guy at the front of the business will BS you hard to get you in.
Later their DFM guy will make subtle, or not subtle changes to your CAD files because they realized they have to actually make it.
Or, they sub it out and take a cut off the top.
I have seen designs completely reverse engineered and recreated wholesale from a step file to fix numerous unpredictable features.
The client did not realize this until the tooling was already produced.
The YOLO factor cannot be understated. It can both work for and against you.
The other thing you have to watch out for in China is using low grade steel for molds. That might be a perfectly appropriate move... so long as it's what you are expecting (and paying for). Not so great if you paid for the best and got pot metal!
Both problems exist everywhere but are particularly important to watch for in China. The best shops will never do either, but it's still better to cut off these entire classes of problem by being proactive.
Also, each production run is unique and it's own beast. Never assume past success will guarantee future success.
- Need to be incredibly explicit. Call out tolerances, how you need it to be created, expectations, everything you can think of. They'll do the work to what you specify, but if you don't specify something, there will be issues there.
- They are as precise as you ask them to be.
- Be careful of prototypes to full orders. We got a lot of parts where the prototypes were perfect, but the when they mass produced them, the method was different, causing an issue with the part.
- You need to be the knowledgeable party of your design. From our interactions, they are implementing your design, not necessarily helping you make it work. If you don't have a lot of plastics knowledge, there will be a lot of issues and retooling to get it to work, so I would go with a manufacturer that is more collaborative.
I've been in the market for a few more CO2/air quality monitors around the house, but I'm unhappy with the very proprietary nature of the uHoo and it's incredible expense.
https://www.airgradient.com/indoor/#:~:text=The%20AirGradien....
[1] https://www.airgradient.com/indoor/
[2] https://www.airgradient.com/integrations/home-assistant-esp-...
One escape hatch is custom machined and printed ABS or metal enclosures, like from Polycase and Hammond. It is limiting and forces you to design around the enclosure, but is much cheaper for small/medium runs than injection molding. Ie a few hundred $USD upfront depending on turnaround time, enclosure size and material etc, then not much more than the plain enclosure per-unit.
If you are injection-molding, hopefully you can use a 3D printer to prototype, but you may still end up screwing up the first (or more) mold due to issues highlighted in this article.
I did work for a US contract moldmaker. The amount of knowledge about the performance of thermoplastics and metals was really impressive. They used some interesting off-the-shelf software to do simulations of mold performance, too. It was a fun gig. I appreciated the tactile, tangible technology that went into their operation.
- What is the plastic used for the part (ABS, PC)? - What was the material for core/cavity (tool steel)? - What was the lowest tolerance on the part? - Were US/EU manufacturers considered at all?
- DIN 2738 tool steel
- I don't know the tolerances. My engineer probably knows
- Not considered
For one of our products, we used an off the shelf plastic enclosure. Unfortunately, the process of getting our equipment certified with an off the shelf enclosure was much more difficult than expected. Finding an enclosure manufacturer who would support us through certification was challenging.. the sticking point was trying to get manufacturers to provide test samples of the gasket materials being used.
Out of curiosity: what else have you considered as material?
How did you find your mold maker and consultant?
> It is super important that you review in depth this mold 3D with your own expert as sometimes mold makers try to do shortcuts that save them money but could create problems later on.
Any examples?
They could have powder 3D printed this stuff for a very long time given the NRE on that complex of a mold.
If they needed to control costs, they could have negotiated with a case supplier: https://www.takachi-enclosure.com/
Only after they were shipping enough product to make it worthwhile (100K+ units), should they even think about making a mold.
Many of these things you would expect not to leave the initial design phase that way.
It's kind of cool that someone can basically walk up to manufacturing and still get it done, though. I never would have considered that.
At those kinds of quantities, I would just fire up the 3D printer, and leave it printing in the corner of the office for a few weeks (with an auto-ejector).
It would end up costing ~10 cents/part, assuming you already have a 3d printer for prototyping and already have someone who knows how to load a new reel of filament into it every morning and evening.
Thats hard to beat for anything under 10k units...
Makes me think about early production FRP sailboats and stories about Company B purchasing hull mold from Company A when they're starting out.
HN hug of death maybe, now that it got upranked?