There's misleading stuff on the internet about the options with injection moulding tools (terminology: the moulds are called tools) and this article doesn't help clear that up much.
For a normal hard plastic like ABS, you basically have two options: a "soft" tool or a "hard" tool. These are both usually made of steel, even aluminium can be too soft at temperature and pressure for good parts.
The difference is that a soft tool is a fairly normal steel that can be milled on a standard 3D CNC machine. This makes it maybe $10k and a few weeks to make, given a suitable design. You can make probably 1000 parts with a tool like this (maybe more depending on the part and the tolerance needs).
A hard tool is made from a very hard "tool steel". There are some different alloys in use but there all extremely difficult to machine - the main method f is Electro-discharge machining. This is a rather involved process, usually involving milling heads that can then spark erode metal to he same shape (look up EDM if you haven't seen it, it's cool). This whole process is more like $300k and takes 3-6 months for a 1-2 action tool. It's likely to be the biggest capex line item for setting up mass manufacture.
For prototyping and low volume manufacture, I strongly recommend Model Solutions in South Korea (https://www.model-solution.com/en/m/main). For us they used to turn around injection moulded parts in quantities of 1k+ in 2 weeks (including postage), at very reasonable prices (used to because I haven't needed them for a couple of years).
Your maths is wildly off in my experience. We had 4 tools for custom parts making ABS plastic parts made in 2013. These were considered hard tools and produced over 100,000 parts each across 5 years and cost us £30k made in the UK in about 4 weeks after material was available. The parts were each around the size of an adult fist and the tools were multi impression. They weren’t large enough to have any hot runners (I think I’m remembering that term correctly) but they did have movers in them.
If we had wanted to make the parts out of PVC we would have had to use stainless steels tools from memory and that would have been more expensive.
I'm curious about your pricing ranges - they seem to be vastly different to what I've experienced over the years (in both hard+soft molds for ABS injection).
Our 2-piece "high-run" tool, built out of tool-steel cranked out its 1M iteration a few months ago before we retired it. The tool cost 6k USD to produce.
These prices also included the model prep (IE, taking an STP of the final product, inserting the ribs, internal structures, closures, etc required to make a real product) and tool design (Modeling the tool itself gates, etc etc).
Perhaps it's the complexity involved, as injection-molding can scale from small parts to giant parts, but our pricing experience seems vastly different.
I've never encountered a high volume mould tool that was less than $50k. In addition to being two piece is it also a low complexity part, e.g. just a few straight faces with pretty lose tolerances? I'm still amazed you got it that cheaply - where did have it made?
I should have stated above that I am neither an injection moulding expert, nor even a mechanical engineer. My experience is as tech lead for years of projects needing injection moulded parts, so I've worked through options with mechies, recommended things for purchase etc. Would be very to happy to learn of cheaper ways of getting stuff made!
I've ordered many high volume tools that were way less than 50k, more like 5. Things in this world have vastly changed in the last 2 decades. Places like Xcentric Mold, Xometry, Fictiv, Baro, etc. all offer very good quality tooling production and molding, very quickly, and the line between soft and hard tooling is much more blurry now. A lot of the advice others are posting here re: injection molding is true for 20 years ago, but doesn't really reflect the current state of the market.
Our manufacturing (tool design, machining and production) is based in Guangdong, China. We have strong ties (15+ years working relationship) with the factory, which may have some pricing impact.
The tolerances on the high-run part are non-critical (but still need to match the other part for a seamless ultrasonic weld). It's got curved surfaces, and the surface texture was EDM'd on.
Likewise, not at all an expert - but have about 10 injection-moulded products under my belt; I just got back from China pricing out two new projects tooling projects. My email is in my profile if you ever want to reach out!
Would it be possible to design say 1000 of the soft tools and have the machines auto-replace the "tips", as if it were an ink cartridge that you replace but in an automated way, with an auto-feeding mechanism after their useful life?
No, the tools are big chunks of metal that are carefully machined to very tight tolerances and polished or textured very carefully. You can use something called a MUD, which is basically a carrier block that you put inserts into which create the mold, this is what most of the quick turn shops do
Surely that $10k figure for 'soft' tooling includes the iteration and engineering time required to get the mould right? I can't imagine that it'd cost $10k to produce another set of tooling if it can be done on a standard CNC machine once you have the CAD/CAM files created and tested.
There's kind of two parts to the design: design a part that is suitable for injection moulding, and then designing the tool itself - where are the inlets, dividing points etc. If it's multi-action, how do the action mechanisms work. Similarly making the tool isn't just "CNC a block of metal and then you're done", you've got to make and assemble the mechanisms and there's things like acid washing the mould surface to control what the resulting plastic surface will look like. It adds up.
Thanks for the extra insight! Assuming a fairly simple part, surely most of the expense is still in the initial design and iteration, though? Even if there's more to it than re-running the CAM job, I'd imagine producing a second set of tooling would be much cheaper than the first?
Alternately, would this be the point where you'd get the same tooling made up but this time in hardened steel? Or is it common to jump straight for the high-end tooling material?
That depends or raw material costs, tool supplier availability and a ton of other factors. One of which is the initial tool design. Also, maybe a new tool design is needed. So, the answer is: it depends.
I'm just an armchair YouTube-viewing machinist, but I'm pretty sure you don't need EDM for tool steel, carbide or a not less hard steel will do won't it? Or you could start with untreated steel of similar grade/purity/whatever and then harden it after machining?
Hundreds of thousands of cycles usually. Bear in mind that your tool is probably multi-cavity, i.e. making multiple copies of your part in each cycle so you might get a up to few million from the tool.
John puts out a bunch of good information and I don't want to fail to acknowledge that while sharing the following. A few years ago, when he was launching Hardware Academy, he advertised somewhere (probably on EEVBlog) for people to join at a reduced price to help get it off the ground. I signed up, only to receive the following a few days later.
> Thank you so much for applying to be a Founding Member at the Hardware Academy.
> There were only 20 slots available for the first batch of Founding Members, yet within just 24 hours I had 130 people apply.
> Although you were not selected as one of the first Founding Members, you will still be given a great discount and early access to the Academy. You will only pay $35/month or $309/year.
As far as I recall, there was no disclosure of a limit on the number (only on time). They say people don't remember what you say, but they never forget how they made you feel. I felt like this was some combination of deception to pump up his email list and/or bait and switch on the offering.
Sorry that happened, but this really seems like a hanlon's razor situation - i.e. never attribute to malice that which is adequately explained by stupidity. It's easy to mess up messaging, or even omit a limit you thought you had put in there. How you handle it after the fact matters, and you're perfectly entitled to be annoyed by the mistake and the mishandling after - but rounding it up to scammy behaviour doesn't really track for me. This is all just my opinion - take it as you will. :)
Hardware products have a missing-middle problem when it comes to enclosures and packaging. For expensive, custom, one-off products, the costs are manageable and expectations around customization (printed logos, die-cut packaging inserts) are lower. For large scale productions (+10k) injection molded plastic or custom metal cases are relatively easy and cheap to produce, and the tooling cost of cardboard dies is easily spread over all those units. But for something you plan to sell 50-1000 of, minimum orders and tooling costs are a pretty large concern. As an example: packaging companies have minimum orders no lower than 1k-2k, and the die will cost you $500-$1000.
A few tricks I've picked up for dealing with being mid-scale:
1. Rubber stamps are cheap and a low-environmental-impact way to brand packaging. Stickers are cheap too, but less low-impact. If you live in a major city, there's probably a stamp maker near you with a 1 or 2 day turnaround.
2. Packaging inserts can be laser cut out of cardboard in small quantities, if you can find a hack space near you with a shared laser. You can do cardboard design using the sheet-metal tooling in fusion 360, but do the prototyping with scrap cardboard and an exacto knife - it's way faster than starting digitally.
3. Cardboard sourcing is harder than you'd think - you run into minimum order issues just like with buying custom boxes.
They are so very much "it depends" that there's not much point. You have a million possible choices to make all of which will move the number around.
Very roughly for feasibility I would say the sale price should be four to five times the oneoff BOM cost, plus allocate $50k upfront for NRE including certifications. Experienced hardware developers can do better than that, but if you've not done one of these before you need to budget for errors.
I've done some of this on a small scale, and the number of potential surprises is just really high. Things like your aluminum frame being subject to US anti-dumping regulations. Or the supplier sending you power supplies with fake UL marks, and now the whole shipment gets seized...not just the power supplies, but the other stuff in the container too. Or your mold is off by just enough to make the already bought-and-paid for aluminum end pieces not fit. PCB oopsies that can't be jumpered around, etc. Or, one of your suppliers likes your idea and copies it and undercuts you on the finished products. Or that terrific, low-cost/high-quality supplier just disappears overnight, and you can't find a suitable substitute.
Hard to translate into numbers, but generally plan on some amount of throw-away work, get some experienced advice if you can, and grow slowly until you understand the risks.
It seems like you've been burned too many times with these issues. I'd be curious to know what you have built / brought to market and what you think about this rule of the retail price being 3 ton 4x the BOM.
Small/medium sized signage related stuff. For me, since I was low volume, a BOM rule of thumb wasn't super useful, as things like shipping from the manufacturer, breakage, reliability, etc, were just as big a driver and varied by product. BOM mattered more for small/light things, but I did variety in small quantities, so I don't have any useful rules of thumb.
One thing notable that I didn't predict well was competitors that would sprout up selling for way too low. They wouldn't last long, eventually disappearing, but hard to compete with that.
Depending on use case, you can dodge the enclosure part by using premade enclosures that come with a machining service to make cutouts for ports etc. Eg Hammond, or Polycase. OOM cheaper at small to moderate runs than injection molding; ie $100 set up fees vice $10k+.
This is notable because the enclosure costs for injection molding can dominate costs otherwise. Eg, getting a PCB manufacturered and assembled from places like JLC is cost effective at any volume. There's no comparable option for injection molding.
This is interesting to me, and the article implicitly touches on it; the parts that seem most important: PCB design and firmware writing, other than the time costs, are dwarfed once you introduce injection molded enclosures. Or if you need certification for an intentional radiator.
https://crowselectromusic.com/products/ovum/ is definitely an aluminum PCB - I don't know about any of the others. Plenty of eurorack manufacturers use aluminum PCBs as panels - but keep in mind the printing options are super limited at small scale (e.g. 5 colors for the background soldermask, 2 colors for foreground silkscreen). Using an FR4 PCB has the added benefit that if you leave off soldermask in an area the panel is slightly transparent so you can see LEDs thru it (e.g. http://www.makenoisemusic.com/modules/richter-wogglebug)
You can reduce upfront costs for plastic injection mold making to zero by 3d printing parts at scale, but does anyone know a scalable low-cost way to polish/sand these pieces so they are nicely finished?
You’d have to come up with some way of doing that with machine tools, and in the process of spending all that time, money and difficulty you’d work out why people tend not to do that at any decent volumes! Excepting of course where there are very specific reasons to, such as some of the metal printing done in jet and rocket engines where the geometry would be impossible to do in single parts with casting and machining. But again that’s not exactly low cost, volume manufacture.
Does anybody do this at scale though? Also vapor smoothing can mess with your tolerances (I.e., small edges will unevenly “pull back”, making split enclosures no longer mate properly), so I’m not sure it’s a great idea for enclosure parts.
the right concentration of acetone vapor will melt them slightly, giving them a polished look. Also, resin printers don't have that problem and some can print in color.
My wife does product lifecycle management and development, and it’s been fascinating watching just how hard it is to actually make and sell something physical. Her company’s products aren’t even electrical, but require complex textile work that has its own set of challenges.
Concept to sales in stores can be a multi-year cycle. There are late night meetings with factories in China. Patent issues that can require workarounds that seem absurd to those of us in software. Iteration time of product concepts that require a physical prototype can take months of work between a designer and the manufacturer.
And then there is one of the most important final steps of having the product end up in stores across the world: shipping!
Pre-COVID a shipping container ran between $6-9k, during COVID it ballooned up to $36k. Margins suddenly got waaaay smaller and that had a trickle down effect where even retailers were then pushing for higher margin to compensate —- squeezing her company on both ends.
After you have a detailed cost estimate, triple it to account for inevitable cost overruns due to various "unanticipated" reasons, which always materialize, eg: tooling error, parts out of stock, incorrect tolerance, Chinese new year, or manufacturer just doesn't deliver.
Quadruple that number again to estimate your retail price.
It may be tempting to develop your own physical product, but if you can build it in software and run it on an existing platform (e.g. iOS), then that is much preferable considering the waste of natural resources.
This is a narrow view of physical products. I encourage you to look around where you're sitting now. Observe the electronics devices, and assess how many should be written as software for your phone etc.
Mine is almost none. Oven, microwave, dishwasher, wristwatch, drones, 3d printer, soldering iron, wrist gyro with LEDs, oscilloscope, thermostat, router, smart light bulbs, game controller, headphones, microphone, power brick for laptop etc.
Why eat physical food? Just buy some banana NFTs and eat those instead. Much less waste of natural resources. Same with housing. Just live in your Sims home.
Some are. I have a… I don’t know, 25 year old calculator that’s still trucking along. My stud finder is at least 10 years old and works great. The light programming and battery charging circuits in my headlamp are great, and as long as the lamp is working, I’ll presumably be content with them.
There are plenty of basic circuits like this which don’t really need to adapt or improve over time. I like these a lot. What concerns me are those which rapidly go obsolete or lose utility.
I’m also not so sure things need to be useful forever. Many older technologies are totally obsolete and therefor not practically useful anymore. The key thing is, if this is no longer useful, how can we recycle it? How can we remove it from the world without it being a burden?
That’s probably the hardest part.
Edit: just realized a calculator is arguably a perfect candidate for being an app. I suppose my point was that it is still working well, though. A stud finder or an oven as someone else suggested is a much better example.
I suppose it depends on the product thermostats? watches? ovens?. The lifecycle of an unmaintained iOS app is about 18 months. So that's what we are comparing.
Most of the apps I have purchased in the past no longer honor it and have switched to a subscription because it doesn't pay for yearly maintenance.
For a normal hard plastic like ABS, you basically have two options: a "soft" tool or a "hard" tool. These are both usually made of steel, even aluminium can be too soft at temperature and pressure for good parts.
The difference is that a soft tool is a fairly normal steel that can be milled on a standard 3D CNC machine. This makes it maybe $10k and a few weeks to make, given a suitable design. You can make probably 1000 parts with a tool like this (maybe more depending on the part and the tolerance needs).
A hard tool is made from a very hard "tool steel". There are some different alloys in use but there all extremely difficult to machine - the main method f is Electro-discharge machining. This is a rather involved process, usually involving milling heads that can then spark erode metal to he same shape (look up EDM if you haven't seen it, it's cool). This whole process is more like $300k and takes 3-6 months for a 1-2 action tool. It's likely to be the biggest capex line item for setting up mass manufacture.
For prototyping and low volume manufacture, I strongly recommend Model Solutions in South Korea (https://www.model-solution.com/en/m/main). For us they used to turn around injection moulded parts in quantities of 1k+ in 2 weeks (including postage), at very reasonable prices (used to because I haven't needed them for a couple of years).