… Since starting MacroFab 7 years ago, we have held a strong belief that manufacturing should work like a cloud resource – software-driven, transparent, and easily able to scale from low to high volume production. Over the last few years, more of this vision has become a reality. …
I agree that it should work as a cloud resource, and for simple designs or very low volume designs I think it’s very possible to use MacroFab or similar vendors and have it really be that way. But for sufficiently complex designs or high volume designs, especially with high speed circuits or RF, I don’t think there’s any way to really make that a reality today.
In my experience, even changing a PCB vendor for a complex design and nothing else, where the new vendor swears they’re using the same stackup and materials, can result in a design which was very robust becoming a scrap nightmare. Qualifying new vendors takes a LOT of time and effort.
But also, being able to go on-site to the assembly and test vendor is super helpful. With a hands-off approach with a cloud vendor you really can’t do this. But if you have somewhat local manufacturing, then when something goes wrong you really can go visit and debug it on the line with the vendor. This can be super helpful for improving yield and for improving design-for-manufacture/test to reduce prices.
One of my customers just got back some MCU boards back from SEEED, and a R,C pair were misplaced (each was rotated 90 deg with one connection on the other part’s pad). If the manufacturer was doing functional testing, this would have been caught before it left the factory.
I like SoftIron’s approach :
It seems here they are offering a high quality/secure product because of the high level of integration between design and manufacturing. Likely not the lowest cost product, but neither are Apple products.
It seems for really great products, there has to be a deep level of collaboration instead of a throw-it-over the wall model.
The automation and tooling companies like Macrofab and SEEED are investing in seems to make sense – no matter what scale. It increases transparency and eliminates a ton of overhead/gate-keeping – even if it is all internal to a company(s). To have a company culture that embraces tools/automation is difficult as someone usually feels threatened that automation will replace what they do, instead of looking forward to having more time to do other higher value activities.
I suspect “cloud” manufacturing will have a lot of parallels with cloud software. It is a good fit for low volume and perhaps even mid-size ventures. It enables you to get started at very low cost before you can afford your own infrastructure. But once you scale up, or want more custom solutions, then companies start building out their own infrastructure.
Just learned about JLCPCB – they seem to be a big player in this space, and seem connected with LCSC and EasyEDA:
As I look through these sites, it is apparent someone has a vision for how this should all work, and is very focused on IT/automation.