From: jl@glen--canyon.com
On Fri, 6 Feb 2026 05:15:17 -0700, Don Y
wrote:
>On 2/5/2026 9:22 AM, Don Y wrote:
>> On 2/5/2026 7:52 AM, Liz Tuddenham wrote:
>
>>> There is a wider problem: we have become 'travel junkies' and rely too
>>> much on transport. "I can get it cheaper from..." doesn't take into
>>> account the hidden costs such as liability to supply disruption, loss of
>>> control over the design and the fundamental need to teach underlying
>>> principles and train people in making things themselves.
>>
>> People are always looking for "more value" -- ignoring the costs
>> they incur in getting there.
>>
>> Look at how many product use Linux kernels -- yet likely have no one
>> on staff that understands or can maintain such a big chunk of code
>> ON WHICH THEIR PRODUCT DEPENDS.
>
>> Employers want "new hires" that are familiar with TODAY'S
>> technology. There is no concern for how well suited they
>> will be for TOMORROW'S! ("We'll just replace them!")
>
>We discussed this, tonight. But, instead of focusing on how much
>"cheaper" (in many senses!) products have become, we thought
>about how much cheaper *development* has become!
>
>No, per-employee costs haven't fallen. But, we require far fewer
>people and man-hours to achieve the same goals as previously.
>
>E.g., I don't draw with a pencil on a drafting board and
>reproduce those sheets with a Diazo machine.
I draw with pencil on D-size vellum, but now I take a picture instead
of using a blue-line machine.
I can draw way faster than I can do CAD entry. I don't have to create
library parts before I draw a new circuit.
It's good to get away from a screen and a mouse too.
Instead, it's all
>done on a glass screen and reproduced on whatever size hard
>copy I desire. Or, consulted on other "glass screens". How
>much easier it is to move a whole section of a schematic to
>accommodate a new component, etc. Bye-bye, electric eraser and
>erasing template!
I have an electric eraser and an electric pencil sharpener.
One can still (barely) buy good pencils and D-size vellum.
>
>I don't have to check the power dissipation of each component
>because the EDA tools can do that. Nor the setup and hold times
>of every signal path.
>
>I don't have a technician prototype my design but either
>simulate it or prototype it in foil. The "guy" who used to
>tape the layouts is gone. And, the time required to do so
>is reduced almost 10-fold with EDA tools.
>
>I can guarantee the final assembly will fit in the envelope
>defined by the ME/IE who designed the enclosure because I
>can view a model of the assembly before it is built.
>
>I can get a feel for how the design will react to temperature
>changes with modeling tools.
>
>I can have boards shot overseas and turned around in a couple
>of days -- instead of relying on in-house facilities or
>"local" service bureaus.
>
>I can have multiple copies created so others can evaluate
>specific aspects of the design without tying up "my" copy.
>
>I can use 4GLs to develop the codebase instead of dicking around
>with ASM. And, use code coverage tools and symbolic analysis
>to identify the "edge cases" and verify the logic.
>
>I can load the code into RAM (instead of burning ROMs) and
>probe it dynamically -- instead of having to purchase an
>expensive "development system" seat (per developer!).
We're using the RPi 400 as our dev system, with our own interposer
board for hardware testing. The Pi400 costs about $90, something like
that.
>
>Each of these things made possible by advances in technology
>(AI) that, in turn, was made possible by reductions in costs
>because of interacting and competing markets. Each coming at
>the "expense" of some previous "job position".
>
>Of course, we only have our own memories of each past project
>but we came to the concensus that things are probably 4 times
>faster than at the start of our careers. And, in adjusted
>dollars, probably a similar amount.
>
>But, a consequence of those economies is a narrower set of
>hiring constraints. You don't NEED a PCB guy. Or, as many
>technicians (esp in development -- how much hardware do
>you need to tinker with to justify those seats?). Or,
>*engineers* (this being a win on multiple levels as the
>cost and risk of communication is elided as the number
>of minds is reduced).
>
>The partitioning into "hardware" and "software" talent is
>largely gone; you can BUY hardware that has already been designed
>and debugged saving much of that development effort AND TIME.
>When was the last time you designed a power supply vs. resorting
>to a brick/wall-wart or drop-in-module?
>
>The real value-added comes in the modeling and software skills
>(software design is a modeling exercise) you can throw at a
>problem and the variety of domain experts at your disposal.
>Expecting "regular" engineers to make design decisions that
>are visible/exposed to the user is usually folly.
>
>All these advances are a consequence of people, parts and services
>becoming more affordable from "globalization" -- "travel/transport"
>in your terms. Imagine being forced to living without them or
>their consequences.
I was just talking over a beer with the founder of circuitly. It's
like Flux, namely a collaborative AI circuit designer. I'm skeptical.
The pcb layout and such might be automated, but I don't think AI can
invent things.
The folks doing this AI circuit design stuff seem to be software
people, not actual electronic designers.
Does AI really work for non-trivial pcb place+route?
John Larkin
Highland Tech Glen Canyon Design Center
Lunatic Fringe Electronics
--- SoupGate-Win32 v1.05
* Origin: you cannot sedate... all the things you hate (1:229/2)
|
