Hacker News new | past | comments | ask | show | jobs | submit login
How to Make Everything Ourselves: Open Modular Hardware (lowtechmagazine.com)
319 points by dvfjsdhgfv on June 28, 2018 | hide | past | favorite | 76 comments

Nobody actually builds stuff that way, except for prototypes. But people do build things from T-slotted aluminum extrusions.[1]

Their heating appliance designs are not too good. The one that heats a water vessel has liquids above a non-sealed box with electrical wiring. The toaster uses parts from another toaster, plus CNC milled plates and 3D printed parts. More expensive than a standard toaster, and looks iffy on electrical safety.

[1] https://www.walmart.com/ip/80-20-1010-72-T-Slotted-Extrusion...

Walmart? Would not have guessed they sold 80/20. Here's a more typical site that thousands of engineers will use today to purchase extrusion for all kinds of industrial projects:


Or, if you're a makerspace in the market for larger quantities and want to standardize, find a local vendor of extrusion (80/20, Misumi, Futura, MB Kit, etc) that will have the oddball fittings you may need on hand locally.

It's true that 80/20 will not sell you their dies, but they are happy to give away free CAD libraries of all their parts. There are subtle differences between different manufacturers, but everyone makes parts that are compatible with it and you probably shouldn't depend on the tiny features that won't match. It looks like OpenGrid saw a proliferation of too many 'standards' that were not open-source, and thought "We'll make a new standard that will be better than the others because it's open", and ended up adding a worse (less strong, less adjustable) standard product with limited adoption.

On the electrical boxes, they seem to be proud of their homemade enclosures. What they probably don't realize is that you can buy an IP67, hinged, IEC-rated enclosure with continuously welded seams from Saginaw or Hoffman with built-in grounding lugs, backplane, and latches for $80.

> Walmart? Would not have guessed they sold 80/20.

they don't, their website is a marketplace just like amazon's, these days.

Walmart is a bit more picky about what sellers they accept, from what I've heard.

The article mentions 80/20:

> There exist modular construction systems that operate according to the same principles, like the T-profiles made by 80/20 inc. However, in the modular construction systems that we have introduced above, everyone is allowed to design and produce parts, as long as these parts are compatible with the basic set of rules.

Which also includes a link to http://www.8020.net/ .

For projects in which 80/20 would be overkill, there are luckily smaller options available, such as 15mm (openbeam, makerbeamxl), 10mm (makerbeam), and 5mm (nanobeam). I imagine the 15mm ones are the most popular because you can use standard M3 fasteners with it easily.

the whole metal fitting (no clue how to call it correctly in english, this was my best guess) is standardized just like the article suggests.

millions of manufacturers over the world, all doing their own designs, but you can screw most cabinet door hinges with the same holes in the wood just fine.

or vesa mounts. etc. lots of industries already saw how good it is.

now, others profit from the lie that this is impossible. like sony using its monopoly on some markets to try to push silly things like minidisc, memorystick, etc. or Dyson making vacuum attachments that aren't interchangeable with even their own equivalent models. Which make them filty rich since a fool is born every minute, unfortunately.

> The one that heats a water vessel has liquids above a non-sealed box with electrical wiring.

Which seems to be a standard feature on all 4 apartments/houses heating appliances I am familiar with, sourced from three different brands.

Friends don't let friends use 80/20.

I've used exclusively https://us.misumi-ec.com/ for all my extrusion needs. They're one of the best companies I've ever worked with. Every order is a "custom order" - you tell them what you want including cutting, tapping, drilling.. You name it. Tolerances for a 20mmx20mm is 1mm cut. 15x15 tolerance is down to .5mm !

Their prices are also very reasonable, and shipping is good. I've also had a weird question about tolerance of a right-angle bracket, and their system imported the data wrong. Told that to the tier 1 support, and they went and found an engineer. Yeah, 5 minutes from calling to talking to the engineer responsible for that line. (Damn!)

The only caveat is they can't easily calculate shipping the first time a package shows up in the area. But after 1 delivery to a zip code, they can accurately estimate it.

My printer in action: https://www.youtube.com/watch?v=7EG-x7TBVng

The scale of the machine: https://imgur.com/a/GJWSf4t

Discovering T-slot was a revelation for me, lego for "grown-ups" basically. There's a good reason why the majority of decent 3D printers are built from it.

I would guess that the majority of 3D printers by shipped units are made of bent sheet metal.


- ed

I should qualify this, lest it come across as mere snark - from my meanderings on the internet and a couple of visits to 3D printing shows, the machines that seemed capable/large/demi-industrial were made from extrusions and t-slots.

Certainly, where it comes to rigid forms necessary for (none- heavy industrial) CNC and such, they are heavily used.

eg. Germany's BigRep http://c93fea60bb98e121740fc38ff31162a8.s3.amazonaws.com/wp-...

Ah. I was thinking something like the Prusa, which is $1000, and it’s derivatives which are $200-500. The BigRep is apparently around $40k. I think the cheaper printers ship in higher volumes.

As far as “decent”, I do a lot with my sub $1k machines. The expensive ones seem to work better, but the price is so high I don’t consider them. And I come from a background of CNC machining where I’m familiar with $100k machines. Now I’m really happy with my $500 printer.

Actually I was thinking machines like the Creality3D ones (like the Ender 3S, I just bought one recently) which while far from perfect perform miracles at their (<$200) price point, in large part due to having a stiff frame with few resonances due to being built from commodity T-slot components. Even the Prusa i3 mk3 now uses T-slots for the base, though in their case it seems to be more of a DFM improvement. You're right though that sheet metal is very prevalent in the mid-range 3D printers and is an excellent choice for machines built in volume.

This reminds me of the MIT course "How to make (Almost) Anything"


... as well as the follow on course "How to make something that makes (almost) anything".


See also Marcin Jakubowsky's work in this vein.


I am compelled to say it is spelled Jakubowski.


Not that I know of, it's a relatively common Polish surname.

Oh, those guys are still at it?

They have a good idea with the engine-driven hydraulic power pack. Except that you can buy those cheaply enough that making them isn't worth the trouble.[1][2] All they're really making is the frame; the engine, pump, hoses, radiator, and controls are all purchased.

(Those guys seem to have a thing about not painting metal, so that it Looks Hand Made. Bare steel rusts. Rapidly. Paint is cheap.)

[1] https://www.ebay.com/itm/STANLEY-GT13-HYDRAULIC-POWER-PACK-S...

[2] https://www.alibaba.com/product-detail/Mini-order-Hydraulic-...

you can buy those cheaply enough that making them isn't worth the trouble

This is something that bugs me probably far more than it should. I'm really sick of seeing half-assed implementations of stuff that you can buy cheaply off the shelf. Don't people do research before starting to design things anymore?

It's pretty rare that you can build something for less than the equivalent off the shelf item. Now, if that item is missing a feature you need, that's a different story.

The short video on the front page is really interesting:


One thing that Lego does very well despite (or perhaps because of) being a closed system is ensuring the consistent quality of all its bricks. In the years I played with Legos as a child I never encountered two bricks that didn't fit together perfectly. I wonder if an open standard would be able to achieve the same consistency.

Like you said, the standard is dead easy it's paying for the QA

I can't find it now, but I remember reading about the QA process that they use. If I remember correctly, each mold is uniquely identified, so if you do find a bad part, the mold is taken out of circulation.

A Lego brick is a remarkably sophisticated piece of technology. Achieving that level of precision involves clever engineering and a very expensive piece of tooling.


How long ago was that? Lego from the 80's sometimes got stuck together. I even remember two (2-brick) sets of bricks that stayed together to this day.

You could tell the sticky pieces by the teethmarks from pulling them apart :D

By that standard, nearly all of my pieces must have been sticky.

(And yes, I did grow up in the '80s).

Have you tried these:

They work very, very well.

The other thing about Lego is that it's a bit trickier to clone than you'd think. Most of the other 'compatible with other major systems' brands don't come apart as well as Lego does.

It seems part of the trick of Lego is to make things that are appropriately easy to take apart, as well as put together.

I used this one:


I wonder about the design change, with this plus-like protrusion at the top. Is it to enforce correct direction of operation when using the separator?

It is to push out cross axles that are part of Lego Technics.

See for example set 10074-1.

I guess that "plus-like protrusion" is designed to fit "tubes" on the high bricks, primary use being removing low bricks from high bricks. I guess one could use two separators side-by-side to firmly hold 4x2 brick, IIRC 2x2 had one "tube" and 4x2 had two "tubes".

Once the lego patents expired all the lego clones had perfect compatibility with lego.

Any of the clones I’ve tried with my kids have barely had compatibility with themselves - at least not to the standard LEGO sets. They don’t grip as well, they fall apart, even their colors can vary. Stay away.

It is a matter of quality. LEGO uses more expensive plastic, and scraps their molds at the first sign of wear. The competitors use cheaper plastics and keep using their molds until they are significantly out of tolerance. This is how they are able to be much cheaper.

There isn't anything magic about what lego does, other than they do it.

I really love the aesthetic of all the demo pieces for OpenStructures (mentioned in the article). Unfortunately it seems it didn't really catch on, and is not an active project. Also, I've found that many standard metric parts that could fit in the system are hard to get or expensive here in the US, alas.

This is the curse of Home Depot, which doesn't stock much, if any, metric. I was recently in a Home Depot where an Asian woman was trying to get some fasteners to repair some piece of metric restaurant equipment, and wasn't aware how backwards the US is on metric standardization.

The US needs to go hard metric. That means metric fasteners and components. The rest of the world does not have inch tools. This hurts exports. The 95% of the population that can't use any tool beyond a screwdriver can still talk of inches and miles, but physical stuff needs to go metric. Autos, aircraft, and DoD switched decades ago. It's mainly building components that are still inch.

In the US it probably makes more sense to go with an inch-based standard, since it's easier to find materials in those sizes. I've been making up my own 'standard' for rack-mount electronics with an eye towards making it work nicely with as many other defacto standards as possible (19" racks, standard lumber sizes, pegboard, unistrut, erector sets): http://www.nuke24.net/docs/2018/TOGRack.html

You can just buy the closest match and cut off a bit of the end. In the end it's much better for everyone to standardize on the superior units

Indeed. Which is why I standardized on inches, which match the hardware at Home Depot and whose rulers make it easy to divide by powers of 2. :P

Aw shucks. Downvoted for disagreeing with the HN groupthink.

First rule of HN groupthink is don't acknowledge HN groupthink. Just downvote anyone who's different.

I get it, but the execution is going in the wrong direction.

I want good documentation for everything in current consumer products. Good documentation makes things modular, not pre-drilling a bunch of holes in steel beam. If every printer came with a good BOM and links to datasheets, it would be much easier to reuse those parts, both for consumers and other manufacturers.

Such standardization makes it easier to design and cheaper to manufacture, all without artificially setting a standard. I hope with new "right to repair" legislation, we will start to see a bit more light shed by OEMs as to what parts their products use and datasheets for them.

Oh heh, my project (Bitbeam) is listed there. I use it as the foundational building block for the robots I make at my company.

I googled for 'bitbeam' and found a blog but no description of what it is.

Try clicking the "About" link on the blog? https://bitbeam.org/about/

Basically, Bitbeam is 3D printable, CNC millable, or lasercutable parts for making things like small robots. The geometry of the parts are compatible with Lego Technic beams.

Some of the links to other sites on that page no longer work.

Nice project! I particularly like your factory, that looks like a fun setup that you have.

Thanks. Yeah, I'm years overdue in updating the Bitbeam site. But I do use Bitbeam all the time in my work.

Would have. This is what it looks like in Firefox: https://imgur.com/a/yA5dhm2

Modularity in every industry is 3-10 times more expensive than monolithic commercial products, and this is no exception. I can buy a metal bed frame for $40 at Walmart, or buy ~10 $40 pieces to make my own. It's the price you pay for having flexible parts.

The synthesizer world knows this well, where a $3000 modular synthesizer is required to build a patch for a $300 portable synthesizer, although you gain flexibility with the modular version.

Pretty sure that’s an economies of scale issue. If the demand for modular was as large as the demand for complete synths, they’d be dirt cheap - but they are essential different products. Though we’ll see how that plays out if behringer ever makes good on their threat to get into modular...

I think it has more to do with baseline product overhead. The cost of bringing a single electronic product to market is $50-100/unit + variable amount based on the scale. The solution to make the product cheaper is to offer monolithic functionality in a single unit, with the same baseline overhead. The same is true for assembled "monolithic" bed frames vs. modular aluminum beams, just with different numbers.

> The cost of bringing a single electronic product to market is $50-100/unit. The solution to make the product cheaper is to offer monolithic functionality in a single unit

Another solution is to reduce the baseline overhead.

Can't. People require a particular amount of support, your marketing costs scale with product sold, and international shipping costs can't get any cheaper. The baseline is not much less for Behringer as it is any reputable modular manufacturer. What they do instead is reduce the parts, manufacturing, and R&D/unit cost, but they're still restricted from going below $50-100/unit even if they could magically clone modules.

The point of this article is "making everything ourselves." What you're describing only applies if the traditional manufacturer/sales channel/consumer paradigm is maintained. I'm proposing to think further outside the box than that.

For an example of what is possible: you can buy Arduino clone hardware on eBay direct from China much cheaper than anything from authorised suppliers.

Any successful "making things ourselves" community (consider Pure Data as an example) is self-supporting and doesn't expect support from manufacturers/creators. Similarly, such a community spreads awareness without marketing investment.

You have to buy the aluminum bars, screws, brackets from somewhere. It will always be cheaper for a manufacturer to buy them in bulk at a price much cheaper than you could get them and sell it to you for cheaper than the sum of the parts (at your price).

> The cost of bringing a single electronic product to market is $50-100/unit + variable amount based on the scale.

that doesn’t seem right... there is an abundance of electronics products priced under $50... are they loosing money on those?

Sorry, I should have specified. Electronics the size, functionality, and build quality of each synthesizer module.

I'm liking the idea but I just can't see how we can get anywhere close to the current size and aesthetics of the current stuff we have. Being modular and standard leaves you much less ability to optimistic to the most compact and cheap design. Desktops have this kind of modular design and they are huge and mostly full of empty space. I think it would be very hard to design a laptop using standard modules without everyone just building the exact same laptop.

The issue with Desktops is that motherboard layouts suck. 2' x 2' square with video cards nearly a foot long. Not much room to kill area unless you want fill the random gaps around your video card.

The real place to make things more custom is when you can do small and stick the computer in places it would not normally go. Raspberry pi or a micro atx in the glove box of your car kind of thing.

I like the idea of GridBeam for furniture. Break a chair leg, just throw a new one on. Build a crib, convert it to a toddler beb then to twin. Later reformat it into the first chair for the child's new apartment.

I used Gridbeam to make the fume hood for my 3D printers: https://twitter.com/hugs/status/824090649540562946

This stuff is cool, but it is also ridiculous over designed and over thought out. Anybody with a welder, angle grinder, plasma cutter, can do all of these things in seconds, from scrap on the side of the road. I guess what this is proposing is having a bunch of pre-fabricated metaL with holes cut. So, 95% of your holes won't be used -- which is a massive overhead for nothing. I can drive in any neighborhood on a Saturday, find a broken threadmil, and cut it up with my plasma cutter to whatever size I need and weld it. If I make a mistake, I can grind the weld off and re-do it. Also I would never trust that bed frame that looks like a deathwish.

I see your point, there are a lot of tools required for that approach though and those tools are expensive and will see - at best - part time use making each of those welds and cuts a relatively expensive affair.

If you're doing production work then sure, you will need that kind of optimization to stay in business.

But if instead you are looking to produce one-offs with relatively low skill and a limited amount of money to spend on tooling then I totally see the point in making things from re-usable and easy to size components.

And then there is the garbage disposal element: almost none of this is waste, even at the end of life of the original object. The same goes for Lego, all it is is parts, and parts have a life of their own (in the decades!) that is totally different from the life of the objects made from them (minutes to several years).

So the over-design and the over-thinking go towards doing that work once so that the users of the system won't need a tools budget in the 10's of thousands of dollars (welder, plasmacutter, grinder, machine shop to go with it) and still build functional equipment that is not readily available.

As for how sturdy the bed frame looks: that's mostly a function of the design of the frame, not of the parts that went into it (Ikea sells kids bedframes that look quite close to what is shown in the article), you could do a much better job if you wanted to with the same parts.

I also note a lot of 'cheating' where the system doesn't work anymore (for instance: the stays to stabilize the bed and the legs under the sled).

Most if these systems looks similar to meccano, so anyone w/ access to a drill and measure could actually make stronger parts.

And the strength concern is all the holes weakening the very spindly looking frames. I'm not bored enough to do the stress calcs, but it doesn't look sturdy enough that I'd trust a kid on that bed.

I buy the cheapest tools I think if I remember right, here's what I got

170 AMP harbor freight welder -- $125 160 AMP Arc welder, amazon $199 CUT-50 Plasma $225 Pizza style air compressor, on sale HF $49 this weekend Angle grinders - I have a bunch $39-59

Anyway I need these tools because I am constantly breaking my farm equipment. Also, instead of diddling around with things I just tack weld things, like my plow blade broke off I couldn't find the bolt so I just welded it in a few spots.

The CUT-50 and the cheap AC/DC arc welder are the best tools, but I use the MIG welder as my first go-to

I "modified" the arc welder to make small metal molds in a carbon crucible. I use aluminum, but it will melt steel. I can lathe/mill that to what I need. I also have the cheapest HF mini lathe and a nicer mill. So yeah all this stuff cost me $2000-$3000 but I use it. I also have a completely different CNC setup.

They made a kit called the Elektor set in the 1940s-50s, my dad gave me, it is the same concept as this post.

That sounds like a very useful skill to have. I’d love to learn, but like so many people I live in small apartment - so this kind of garage workshop stuff doesn’t work well for me. I do occasional small woodworking projects on my balcony, but it requires a lot of forethought and planning... e.g. mostly hand tools, and i can’t just buy an extra sheet of plywood in case I need it.

First thing I do whenever I move is change my Dryer to a NEMA welder plug and outlet. You can sort of use your dryer as a work bench. They make this mini arc welder that is really small for $140, I've been trying to justify owning two arc welder to just try it. But they all use the same NEMA plug, you just need to switch your dryer over to it and you're good.

Said skills and tools are both less than common and scale worse. Scale also makes things cheaper counterintuitively to hand crafting. If you have say a drill press that can easily drill every hole at once it makes for better logistics to just drill every hole and stock them than keeping a ton of variants or worse drilling on demand.

I can certainly understand that mental blindspot however given I can't help but think of enterprise software as drastically overpriced for stuff I could slap together myself.

I think one of the points is to be able to build things without the need of expensive/extensive tooling.

There is a middle ground. I make quite a bit of stuff out of square and angled aluminum stock using only a few basic tools - drill, file, hack saw, propane torch, and aluminum brazing rods.

I would use open modular hardware if it were available everywhere like plain metal stock is, but it's not so I've learned to improvise.

I think about this all the time, i'm not sure how far it could go but it'd nice to witness

Could be a nice approach on Mars

I am sure they have thought about it already.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact