6 ms·

> I think big things are coming for this world and by and large they are not ready for it. With regard to PCB design, I'm curious if can you add more color to

by varenc 28d ago

> I think big things are coming for this world and by and large they are not ready for it. With regard to PCB design, I'm curious if can you add more color to how the world isn't ready for this in particular? I don't think I quite grasp all the ramifications of AI becoming PCB design experts.

tomhow 28d ago
I can speak from experience here (I’ve been around small-scale electronics design/manufacturing for decades via my father’s business and was hands-on with it on an ag-IoT project for about 10 years). The limiting factor in electronics/PCB development and manufacturing isn’t engineering expertise (there are plenty of sufficiently competent electronics engineers in the world, especially in Asia), nor parts/materials or assembly costs. It’s the endlessly laborious grunt work of testing a design and making minor improvements until it is finally ready for volume production. That process can easily take a year or two even for a fairly simple design. Even a really skilled/experienced engineer can make a simple mistake in a schematic or PCB layout, which only gets picked up weeks later when prototypes are assembled and tested, or even later when a batch has been made and shipped to customers. So, each design-assemble-test cycle takes weeks/months, and each testing/diagnosing/repairing session takes hours or days. And it mostly has to be undertaken by degree-qualified engineers or experienced technicians. And that’s mostly before you’ve written all the firmware, which is incredibly laborious in its own right. The outcome is that only very large companies with big sales volumes can afford to do it (or niche players able to sell at huge margins, most commonly in defense or medical industries). Outside of those sectors, it’s now just not really viable to be a small-medium business selling moderate volumes of units anymore, the way it was in the 80s-90s. I’ve seen it up close in agricultural IoT. There are no big/dominant players (customer needs are too heterogeneous and distribution is so hard). There are many small vendors, but they find it hard to make products that work well and that customers love, especially at a price point most growers will go for. So, no vendors are big commercial winners and not many customers are very happy - even though there’s a great need for technology in this sector. And it’s really not because the vendors and their engineers are incompetent; it’s just really hard. With AI to take care of so much of the grunt work involved with circuit/schematic design, PCB layout, pre-manufacturing verification, testing, and firmware dev, it’s realistic to me that you could cut down on R&D costs by 90%, make it much faster to get products onto the market, and make it possible for so many more high-quality niche products to be commercially viable.
- rjh29 28d ago
  I don't understand your argument. PCB design is not that hard you can learn how to do it in a month or so. JLCPCB can print (or even assemble) PCBs for next to nothing. Irrespective of AI it seems easier than ever to produce prototypes right now.
  - tomhow 28d ago
    Yes, you can learn PCB design “in a month”, design your board and get a batch of prototypes made up by JLCPCB. All of that takes 2-3 months absolute minimum. Then you have to test the boards. You need a lot of costly equipment and expertise to test the board and fix any problems with it. That can take days to weeks of work. Then ordering your next batch of prototypes carries a 2-4 week turnaround time including shipping. Repeat this for any design error or just any improvement you want to make once you’ve tried working with it for a bit. Then you have to write the firmware. That’s weeks or months of work. (And much of the embedded firmware code I’ve seen come out of small business engineering departments or consulting shops is terrible, because they don’t have the time/budget or organizational structure that delivers clean, readable, well-structured, well abstracted C code.) All this is before you get an MVP into the hands of your customers. Then you start getting customer feedback and have to start making more improvements. “Never been easier” it may well be to get a hobby project prototypes. But for a product with any amount of serious functionality, it’s at least months and likely 1-2 years to get anything into volume production. But the potential for saving huge amounts of time particularly on design/layout, pre-manufacturing error detection, and production of well-structured and debugged firmware code are huge.
    - rjh29 28d ago
      You lost me when you referred to a multimeter as "costly equipment"
      - tomhow 28d ago
        I don’t know why you need to be snarky. I’ve been living this for over a decade with engineers with decades of experience who have done some of the most advanced electronics work imaginable (satellites, military drones). To test and rework a surface-mount PCB, you may need: - logic analyzer or oscilloscope; - microscope; - high precision soldering equipment/ability; - solder paste and oven; - cutting equipment to cut through 4-6 PCB layers and connect tracks with hair-width wire; - Years/decades of education and experience to understand what to look for and what to do when you find a problem.

by varenc 28d ago

tomhow 28d ago

I can speak from experience here (I’ve been around small-scale electronics design/manufacturing for decades via my father’s business and was hands-on with it on an ag-IoT project for about 10 years). The limiting factor in electronics/PCB development and manufacturing isn’t engineering expertise (there are plenty of sufficiently competent electronics engineers in the world, especially in Asia), nor parts/materials or assembly costs. It’s the endlessly laborious grunt work of testing a design and making minor improvements until it is finally ready for volume production. That process can easily take a year or two even for a fairly simple design. Even a really skilled/experienced engineer can make a simple mistake in a schematic or PCB layout, which only gets picked up weeks later when prototypes are assembled and tested, or even later when a batch has been made and shipped to customers. So, each design-assemble-test cycle takes weeks/months, and each testing/diagnosing/repairing session takes hours or days. And it mostly has to be undertaken by degree-qualified engineers or experienced technicians. And that’s mostly before you’ve written all the firmware, which is incredibly laborious in its own right. The outcome is that only very large companies with big sales volumes can afford to do it (or niche players able to sell at huge margins, most commonly in defense or medical industries). Outside of those sectors, it’s now just not really viable to be a small-medium business selling moderate volumes of units anymore, the way it was in the 80s-90s. I’ve seen it up close in agricultural IoT. There are no big/dominant players (customer needs are too heterogeneous and distribution is so hard). There are many small vendors, but they find it hard to make products that work well and that customers love, especially at a price point most growers will go for. So, no vendors are big commercial winners and not many customers are very happy - even though there’s a great need for technology in this sector. And it’s really not because the vendors and their engineers are incompetent; it’s just really hard. With AI to take care of so much of the grunt work involved with circuit/schematic design, PCB layout, pre-manufacturing verification, testing, and firmware dev, it’s realistic to me that you could cut down on R&D costs by 90%, make it much faster to get products onto the market, and make it possible for so many more high-quality niche products to be commercially viable.

rjh29 28d ago
I don't understand your argument. PCB design is not that hard you can learn how to do it in a month or so. JLCPCB can print (or even assemble) PCBs for next to nothing. Irrespective of AI it seems easier than ever to produce prototypes right now.
- tomhow 28d ago
  Yes, you can learn PCB design “in a month”, design your board and get a batch of prototypes made up by JLCPCB. All of that takes 2-3 months absolute minimum. Then you have to test the boards. You need a lot of costly equipment and expertise to test the board and fix any problems with it. That can take days to weeks of work. Then ordering your next batch of prototypes carries a 2-4 week turnaround time including shipping. Repeat this for any design error or just any improvement you want to make once you’ve tried working with it for a bit. Then you have to write the firmware. That’s weeks or months of work. (And much of the embedded firmware code I’ve seen come out of small business engineering departments or consulting shops is terrible, because they don’t have the time/budget or organizational structure that delivers clean, readable, well-structured, well abstracted C code.) All this is before you get an MVP into the hands of your customers. Then you start getting customer feedback and have to start making more improvements. “Never been easier” it may well be to get a hobby project prototypes. But for a product with any amount of serious functionality, it’s at least months and likely 1-2 years to get anything into volume production. But the potential for saving huge amounts of time particularly on design/layout, pre-manufacturing error detection, and production of well-structured and debugged firmware code are huge.
  - rjh29 28d ago
    You lost me when you referred to a multimeter as "costly equipment"
    - tomhow 28d ago
      I don’t know why you need to be snarky. I’ve been living this for over a decade with engineers with decades of experience who have done some of the most advanced electronics work imaginable (satellites, military drones). To test and rework a surface-mount PCB, you may need: - logic analyzer or oscilloscope; - microscope; - high precision soldering equipment/ability; - solder paste and oven; - cutting equipment to cut through 4-6 PCB layers and connect tracks with hair-width wire; - Years/decades of education and experience to understand what to look for and what to do when you find a problem.