39c3: In-house electronics manufacturing from scratch: How hard can it be? [video]
media.ccc.de259 points by fried-gluttony 4 days ago
259 points by fried-gluttony 4 days ago
This is painful. They got a used solder mask holder, a Lumen pick and place machine, a bunch of old Siemens feeders, and a small automatic reflow oven. All these tools needed major work. Everything with firmware needed firmware mods. Everything else needed assembly or major cleaning. Everything needed adjustment. They had to 3D print their own solder paste squeegee. They're six months in and still trying to produce one simple board.
I've been down this road of populating a surface mount board. There is a minimum size for a practical board-stuffing operation, and they are below it. They are using prototype techniques for 100 units or so, not techniques that scale.
Surface mount soldering requires applying hot air in a very controlled way, with the temperature ramping up, holding at the high temp for a few seconds, and then ramping down. On a small scale, you have a programmable oven which tries to do that. Those always have heat distribution problems. For production, you have a tunnel oven, with about six sections at different temperatures and a chain conveyor to take the boards through the tunnel. With the tunnel oven, you let the whole thing warm up and stabilize, and when all zones are at the right temperature, you can repeatably solder boards successfully.
They're using a hobbyist-grade pick and place machine. Slow, but cheap. Plus the software isn't ready for prime time. They looked at a used production machine. Runs Windows XP and wouldn't fit through the door. Rejected that.
They're about EUR 30,000 into this, not counting their own labor. This approach is not going to revive electronics in Europe.
Learning by mistake isn't painful; it is how you learn best. I keep iterating on that point to my children. But that isn't merely what these guys were doing. They were doing more, since they were documenting their (expensive!) learning process. Documenting your learning process and sharing it freely is allowing others to not make the very same mistakes, but to do better instead. It lowers the barrier of entry for competition, taunting competition who hopefully also share back. Like many talks on 39c3 (esp. the lightning talks), it is an invitation to collaboration.
Sharing the documentation is also an act of compassion, and very much in the spirit of FOSS & OSHW.
This talk was hands down my favourite talk (and not even in a subject I am familiar with!). These two guys shared a lot of info in little time, and were very humble. It was also a presentation which contains a political component (Europe's lack of independence, specifically hardware-wise), but it managed to avoid that discussion. Why, because it is assumed the attending public shares the same value. Instead, it maintains focus on the taking action part. I am not sure everyone here shares said value, but I do, and for whatever it is worth: USA is in a similar boat.
Hey! Talk speaker here, thank you so much for these words.
This is exactly what we wanted to convey: Let's act, our way isnt the best way, but it is the path we're on, and there is little we can do on our own to get to another path.
We don't want to build the european JLCPCB, we don't even know what our company will be in 20 years if it still exists.
What we want is to give knowledge and see more people get into the business of electronics. We also want to give meaningfull jobs to engineers and factory workers which will eventually join us.
We are not going to change the world, I would settle for selling 1 unit of 1 well made product to 1 customer. I would settle for giving one person a job that they love working with cool guys to make electronics. I would settle for the ability to pay my rent from this, from bringing value in the world.
Just curious, is it possible to get some basic production "kit" from China and go from there? I'd assume it's going to be new and cheaper and more "modular", and you only rely them on these basic items. I have never been into electronics manufacturing so not sure if it makes sense.
You are saying the same thing they said -- it doesn't scale. It's not how you build a large factory. They acknowledge this and pretty quickly move on to say that they are aiming for smaller and sustainable.
They even specifically call out why they chose not to use a conveyor based oven in the video.
Basically they believe they can be price reasonable at small scales, small batches. Build process knowledge and expertise over time, and then incrementally scale up after assessing bottlenecks.
I think the route of local sustainable, grow as needed or collaborate to expand capacity is pretty reasonable.
It's not possible to make a competitively priced product that way. What are you going to do, sell artisanal circuit boards on Etsy?
Here's a small US-based PCB board and assembly facility in the US, in Hesperia, California.[1] Looks like it might have 20 to 30 employees from the building picture. This is probably about as small as a viable business of this type gets. It doesn't have to be done in a huge plant like JLCPCB in Shenzhen.
Here's a company in India, Invariance, which makes low-cost semi-automatic machines to do exactly the same operations 39c3 is doing.[2] They have three machines - a solder paste spreader, a pick and place machine, and a mini tunnel reflow oven. They make all three machines. These machines intended for small companies who want to assemble their own boards in house. The solder paste spreader is just automated enough to do a consistent job, with pressure and timing controlled. The pick and place machine uses their own feeder design which runs off strips of component tape. The tunnel oven is small, only about a meter long.
That's close to a viable minimum production solution.
39c3 is the conference it is being presented at, not the presenter. 39th Chaos Communication Congress, the annual conference of the Chaos Computer Club.
This is a hacker messing around who is presenting to inspire others and get feedback. Many things presented at Xc3 are wildly impractical, potentially illegal, or not even technical at all and more on the side of activism and policy. Most are interesting and fun, which is the main goal.
That depends on your definition of "competitive", doesn't it?
Being 3x as expensive as China but 0.1x as expensive as current small-scale EU manufacturing can be extremely competitive. Plenty of people looking for <1000 unit runs would be willing to pay extra for a "made in EU" label.
COVID and the supply chain crisis made apparent just how over-reliant we are on Shenzhen and Taiwan for the most basic components. There are several hundred ICs on dozens of circuitboards in every car now that are dumber, slower, and less efficient than a 1990s calculator, that we have lost the ability to produce domestically. These are now bottlenecks to manufacturing in any disruption to world trade.
It doesn't need to be cutting edge, if you have a few board assembly shops and some fabs pumping out small chips in 20-30 year old process nodes it helps the resilience of the economy and geopolitical situation a great deal.
For that, there's a company named eurocircuits. Slightly more expensive than JLCPCB, but not 30x.
5pcs 2 layer default settings 80x100mm boards ($4 vs EUR 124) is pretty much 30x.
I double checked because these were not the prices I had in mind.
After looking through the options, I think that's because the designs I did quotations for had 0.2mm holes. This is standard for Eurocircuits, but high precision for JLCPCB.
Note that to get the price you quoted, you'll get lead in your PCB, and vias that are not plated, but plugged with conductive epoxy. Changing that gets you to $14 for 5 boards, which is still way cheaper than Eurocircuits.
I'll keep that in mind for the next PCB I design: keep holes bigger than 0.3mm if possible.
If the only service a CM brings to the table is cost reduction, than its a doomed business model.
Rule #23: Don't compete to be at the bottom, as you just might actually win.
Have a great day =3
Technically the differentiation isn't cost reduction its in shored manufacturer. That can become an important requirement as firms become more wary of IP theft and other issues.
Besides other than that what can you really differentiate in this field that isn't cost? Its a service that is standardized regardless of country. Maybe you can provide different "styles" of boards (ie. different specs) or improve the entire submit to production pipeline but thats about it.
The pipeline is a massive one - especially for PCBA. Being able to get an instant quote and order in 5 minutes is worth a lot. Nobody wants a 2-week-long back-and-forth with sales and engineering for a basic prototype run.
I don't know how much real innovation can happen there that the Chinese haven't done. In fact the Chinese haven't really innovated, that have optimized existing processes. Think about it: If you break it down: a "2-week-long back-and-forth with sales and engineering for a basic prototype run" is the same stuff happening in the background as the "get an instant quote and order in 5 minutes"...just much faster.
Thats not the innovation I was talking about: Let me give an example. When the Xbox was being developed, there was a push within Microsoft to create a standard CD‑based game console that was a more refined version of the existing competitors. The Xbox team pushed back and said: No, you don't enter a market and just copy the incumbents. You need to do something different and differentiated. So they innovated by adding a hard drive and standard networking to every console. The addition of these components as standard in each console opened up a new paradigm of gaming with Xbox Live and was a different vision compared to what the others were offering at the time. They eventually ended up moving the market in a fundamental way.
Now, going back to manufacturing, how do we translate that to PCB manufacturing? Again, I don’t know how you could possibly truly innovate the pipeline process more than it has already been done. Even just adding an AI‑handled pipeline process, it's just further optimizing what has already been done. There may or may not be an opportunity to truly move the market but if there is, then thats something the West can do to really compete.
China is a tough place to do business even for domestic small firms.
https://www.youtube.com/watch?v=FQrEDq8KPiU
>I don’t know how you could possibly truly innovate the pipeline process more than it has already been done
There are things China does well, things the US does well, and a lot of people building stuff for fun. Yet doing business in North America can be just as challenging for different "reasons". =3
Consumer hardware like the iPhone would not do well when the next model is 20% slower with a 35% retail price reduction.
>what can you really differentiate in this field that isn't cost?
Don't worry about it... =3
>Consumer hardware like the iPhone would not do well when the next model is 20% slower with a 35% retail price reduction.
Yes, American companies know this already: Apple managed to ship $1000 phone even the Chinese coveted for a long while and their innovation at the time was Face ID and the applications it enabled like Memoji.
Just improving spec sheets is a race to 0 profits.
>Don't worry about it... =3
There has got to be something that can move the market as relying on China is just a mirage anyway. Those demographic issues will catch up to everyone eventually. Im not knowledgeable enough to know what this magical innovation will be. (or else I'd pull my savings and get into the business :P )
>Just improving spec sheets is a race to 0 profits.
If and only if people chase the low-margin markets. Some people want the best value they can afford, and the ones that don't usually are not worth the sales effort.
Note, most current Apple products are just information-appliances tied to their media service businesses.
>Im not knowledgeable enough to know what this magical innovation will be
Whatever a unique imagination comes up with will probably be just as good or better. Don't worry about it... =3