Author Topic: The End of Electrical Engineering Design?  (Read 5265 times)

0 Members and 1 Guest are viewing this topic.

Offline MuhScopeBrokeTopic starter

  • Contributor
  • Posts: 35
  • Country: us
The End of Electrical Engineering Design?
« on: July 12, 2022, 02:13:46 pm »
https://techcrunch-com.cdn.ampproject.org/c/s/techcrunch.com/2022/07/06/celus-which-uses-ai-to-automate-circuit-board-design-raises-25-6m/amp/

Curious to hear opinions. Obviously the technology isn't there yet but perhaps one day all board design will be automated  :-BROKE
 

Online ejeffrey

  • Super Contributor
  • ***
  • Posts: 4034
  • Country: us
Re: The End of Electrical Engineering Design?
« Reply #1 on: July 12, 2022, 04:29:53 pm »
More and more PCB design is "dropping one or a handful of giant chips on a board and then carefully designing the PDN and thermals.  More and more of circuit design is going to chip design or some form of "software" (including FPGA configuration).  It's sort of the opposite of the old days when you would build a complex circuit and then drop a 7805 regulator on it for power.

Given that it's not surprising that this is more amenable to automation all the time but I don't think they are going to put board designers out of a job.  Assuming it actually works (a huge assumption) it's going to let designers turn out more boards faster and therefore lower cost.  Lowering cost almost always increases demand, very often more than the decreased cost.
 
The following users thanked this post: Alex Eisenhut

Offline free_electron

  • Super Contributor
  • ***
  • Posts: 8550
  • Country: us
    • SiliconValleyGarage
Re: The End of Electrical Engineering Design?
« Reply #2 on: July 12, 2022, 04:47:36 pm »
meh... another fad based on clown (cloud) computing.

all the systems i have seen are banked around a bunch of pre-chewed circuits that are clicked together and the interconnects are automatically routed.
Grab a 5 volt supply , an arduino, a spi to wifi module, a serial port (db9 and max232) . Connect wifi to spi 1 , connect serial port to uart 1 , apply power to all modules. Place the blocks ( which invariable yield a huge circuit board as the parts inside the blocks are pre-placed and not modifiable and the block shape is immutable). Click route and export Gerber.

This nothing more than assembling kits based off demoboards. This is NOT board design. as for intelligence ? most 8 year olds can do better

Watch this as they show absolute crap schematics built using their 'blocks' (featherwing here, other hobby crap there)  and then 'route' (it's more like somebody regurgitating a plate of squid pasta and then throwing a firecracker in it and take a picture of what stuck to the wall)


« Last Edit: July 12, 2022, 04:52:16 pm by free_electron »
Professional Electron Wrangler.
Any comments, or points of view expressed, are my own and not endorsed , induced or compensated by my employer(s).
 
The following users thanked this post: Zucca, pardo-bsso, tooki, RJSV

Online CatalinaWOW

  • Super Contributor
  • ***
  • Posts: 5571
  • Country: us
Re: The End of Electrical Engineering Design?
« Reply #3 on: July 12, 2022, 04:58:04 pm »
I don't see this as a short term threat.  Doing a good board design involves juggling a great many things.  Power and signal integrity.  Placement of large components in three dimensions.  Thermal control.  And many more. 

I see two approaches to doing this automatically.  The first, most obvious and most difficult is to get real intelligence in the AI so that the program will get similar results to a good board designer.  There has been effort in this direction for three decades that I know about, and progress has been limited at best.  Not a short term threat.

The second is more like the chess programs.  Just try every possible combination and evaluate the results.  Eventually good enough designs will pop out.  But it seems to me that the resources to do this will not be free, and it will likely turn out to be cheaper to hire a good designer.  That isn't to say that the bottom quartile of designers and possibly even the bottom two won't be threatened.  But no one will feel threated because we all know that no one is in those quartiles.  Just ask them.
 
The following users thanked this post: tooki

Online Siwastaja

  • Super Contributor
  • ***
  • Posts: 9333
  • Country: fi
Re: The End of Electrical Engineering Design?
« Reply #4 on: July 12, 2022, 05:20:45 pm »
Bullshit as usual. Nothing to see here.
 
The following users thanked this post: Zucca

Online mawyatt

  • Super Contributor
  • ***
  • Posts: 4119
  • Country: us
Re: The End of Electrical Engineering Design?
« Reply #5 on: July 12, 2022, 05:24:46 pm »
For those in the IC design world this isn't new, Digital IC designs have been ported from one process into another process for some time now. Even "dropping in" complex digital IP has been around for awhile. However, the software to do this is quite expensive and the folks involved are highly skilled and specialized.

About 5 years ago we attended a DARPA Conference where Cal Berkeley presented a sponsored project where they were porting analog IC designs from one process to another, almost completely "hands off". The preliminary results were quite impressive at the time, and they were working on the thermal aspects affecting analog performance as well and even looking to the ultra-precision type analog circuits where not only thermal but process gradients and local proximity (localized doping gradients) come into play.

Of course the use of such complex transformation algorithms and software requires extremely knowledgable folks in all aspects of IC design, simulation, layout and fabrication, so likely displacing some IC designers and augmenting others.

Very sophisticated and involved effort by Berkeley, and we expect to "see" these efforts begin to filter into high-end analogish IC design in the near future, although at the very high end!!!

Best,
Curiosity killed the cat, also depleted my wallet!
~Wyatt Labs by Mike~
 

Online Benta

  • Super Contributor
  • ***
  • Posts: 6420
  • Country: de
Re: The End of Electrical Engineering Design?
« Reply #6 on: July 12, 2022, 05:43:36 pm »
Idiotic spin.
PCB design today is 5% of design effort and has very little to do with engineering. It's more in the Draftsman trade.
 

Online mawyatt

  • Super Contributor
  • ***
  • Posts: 4119
  • Country: us
Re: The End of Electrical Engineering Design?
« Reply #7 on: July 12, 2022, 06:06:53 pm »
Idiotic spin.
PCB design today is 5% of design effort and has very little to do with engineering. It's more in the Draftsman trade.

Good point!!

Long ago EEs had to design individual gates with transistors/resistors, then they evolved to individual gates on a chip, then simple logic functions on chip, then complex logic functions on a chip, then FPGAs and synthesis and so on. Funny thing is there are more EE around than ever, and we are continuously pushing the more mediocre design tasks onto the computer or AI, whatever you want to call it!!

So think EEs will be around for a long time, they just will be handling more complex tasks as time marches on!!

Best,
Curiosity killed the cat, also depleted my wallet!
~Wyatt Labs by Mike~
 

Offline jonpaul

  • Super Contributor
  • ***
  • Posts: 3656
  • Country: fr
  • Analog, magnetics, Power, HV, Audio, Cinema
    • IEEE Spectrum
Re: The End of Electrical Engineering Design?
« Reply #8 on: July 12, 2022, 06:20:20 pm »
Just another vaporware startup....

Try their AI on some multiKW Power electronics? High voltage? Data Acquisition? Measuring instruments?

j
An Internet Dinosaur...
 

Offline SiliconWizard

  • Super Contributor
  • ***
  • Posts: 15797
  • Country: fr
Re: The End of Electrical Engineering Design?
« Reply #9 on: July 12, 2022, 06:22:45 pm »
Uh. Guys. There WILL be millions of startups founded with some crap AI at their core. We know we can't escape this, as some user told us a couple days ago. :-DD
 

Offline free_electron

  • Super Contributor
  • ***
  • Posts: 8550
  • Country: us
    • SiliconValleyGarage
Re: The End of Electrical Engineering Design?
« Reply #10 on: July 12, 2022, 06:39:18 pm »
Idiotic spin.
PCB design today is 5% of design effort and has very little to do with engineering. It's more in the Draftsman trade.
You've probably never seen a real pcb.
Try a modern motherboard. or a high power inverter with embedded transformers. Maybe something with all kinds of controlled impedance, signal integrity, looptime compensation , 8 layers, embeddded heat medallions, embedded wires , stacked / dogboned laservias / vippo / elic ? something that has 0.5mm BGA's and 1000 ampere running through it and 800 volts busbars. Then you will understand how much engineering goes in the board. Without that engineering your nice schematic is SNOT. if the layout is not done properly , at switch-on time you'll end up with a smoldering crater where once was your lab bench.
The time where draftsmen made single layer hand taped stuff for through hole parts is over.
Professional Electron Wrangler.
Any comments, or points of view expressed, are my own and not endorsed , induced or compensated by my employer(s).
 
The following users thanked this post: daqq, tooki, newbrain, Jacon

Offline SiliconWizard

  • Super Contributor
  • ***
  • Posts: 15797
  • Country: fr
Re: The End of Electrical Engineering Design?
« Reply #11 on: July 12, 2022, 06:43:58 pm »
Yeah, I don't necessarily agree with the fact PCBs are very little effort in a given project. It all depends on the project. Some can be very tricky, but they are probably a niche, overall.
Doesn't mean that all this AI frenzy crap is going to help anything and that those startups are not just scams.
And well yeah, this thread's title is at best misleading. ;D
 

Offline ConKbot

  • Super Contributor
  • ***
  • Posts: 1400
Re: The End of Electrical Engineering Design?
« Reply #12 on: July 12, 2022, 07:35:49 pm »
I'll stop rolling my eyes into the back of my head when someone makes an autorouter that doesn't take longer to setup to produce good results than is does to just route a board manually.  Currently I can take a fully routed chip, drag it 3mm to the side, and when Altium tries to autoroute to re-complete what I moved, a lot of times it manages to confuse itself and block itself off from completing.   ::)

Walk before you run.
 
The following users thanked this post: tooki, eugene

Offline bd139

  • Super Contributor
  • ***
  • Posts: 23099
  • Country: gb
Re: The End of Electrical Engineering Design?
« Reply #13 on: July 12, 2022, 09:40:02 pm »
How AI works

1. Someone finds a niche where no one has solved any problems with it.
2. Rinse investors for money
3. Create hype
4. Deliver something which does about as good as a job as a one armed brain damaged monkey.
5. Claim next generation AI will solve all the problems.
6. Go to (2)

Celus is at stage 3 iteration 1.

Currently dealing with an ML system which "learns" infrastructure trends and then tells you when something unpredictable is about to or is happening. 50% of the time it goes off when there aren't any problems. The other 50% of the time it doesn't go off when there are problems. If that can't reason about simple trends then I doubt it can reason about PCB design which is far far far far more complicated :-//
« Last Edit: July 12, 2022, 09:42:40 pm by bd139 »
 
The following users thanked this post: free_electron, tom66, Zucca, MT

Offline free_electron

  • Super Contributor
  • ***
  • Posts: 8550
  • Country: us
    • SiliconValleyGarage
Re: The End of Electrical Engineering Design?
« Reply #14 on: July 12, 2022, 09:54:31 pm »
now the iot buzz has flunked they found a new cow to milk : AI.
Professional Electron Wrangler.
Any comments, or points of view expressed, are my own and not endorsed , induced or compensated by my employer(s).
 
The following users thanked this post: bd139

Offline tszaboo

  • Super Contributor
  • ***
  • Posts: 8218
  • Country: nl
  • Current job: ATEX product design
Re: The End of Electrical Engineering Design?
« Reply #15 on: July 12, 2022, 10:01:17 pm »
So AI designs stuff? Much wow.
I can pick up the phone, call a supplier, talk for 5 minutes, and make a deal with them for a part, which makes the design 2% cheaper, which pays for my yearly salary and makes the AI unemployed.
Almost everyone can design functional electronics, that's not a valuable skill.
 
The following users thanked this post: bd139

Offline tooki

  • Super Contributor
  • ***
  • Posts: 13157
  • Country: ch
Re: The End of Electrical Engineering Design?
« Reply #16 on: July 12, 2022, 10:58:39 pm »
I'll stop rolling my eyes into the back of my head when someone makes an autorouter that doesn't take longer to setup to produce good results than is does to just route a board manually.  Currently I can take a fully routed chip, drag it 3mm to the side, and when Altium tries to autoroute to re-complete what I moved, a lot of times it manages to confuse itself and block itself off from completing.   ::)

Walk before you run.
Right?? In Altium, if I want to move a component and let it re-route, I frequently just do it in many tiny steps, otherwise it completely loses the plot, even if it could have just pushed without any actual re-routing!
 

Offline MT

  • Super Contributor
  • ***
  • Posts: 1687
  • Country: aq
Re: The End of Electrical Engineering Design?
« Reply #17 on: July 13, 2022, 12:25:05 am »
I'll stop rolling my eyes into the back of my head when someone makes an autorouter that doesn't take longer to setup to produce good results than is does to just route a board manually.  Currently I can take a fully routed chip, drag it 3mm to the side, and when Altium tries to autoroute to re-complete what I moved, a lot of times it manages to confuse itself and block itself off from completing.   ::)
Walk before you run.
Something Protel99se also used to do back in the day. Back in the day we used to buy very expensive autoroters from companies who specialized in such things.
The AI hype is a penis extender for Schwab and Kurzweil and some eugenics bankster oligarchs for their Great Reset program where you own nothing and you are happy.
« Last Edit: July 13, 2022, 12:28:56 am by MT »
 

Online Benta

  • Super Contributor
  • ***
  • Posts: 6420
  • Country: de
Re: The End of Electrical Engineering Design?
« Reply #18 on: July 13, 2022, 12:58:18 am »
You've probably never seen a real pcb.
Try a modern motherboard. or a high power inverter with embedded transformers. Maybe something with all kinds of controlled impedance, signal integrity, looptime compensation , 8 layers, embeddded heat medallions, embedded wires , stacked / dogboned laservias / vippo / elic ? something that has 0.5mm BGA's and 1000 ampere running through it and 800 volts busbars. Then you will understand how much engineering goes in the board. Without that engineering your nice schematic is SNOT. if the layout is not done properly , at switch-on time you'll end up with a smoldering crater where once was your lab bench.
The time where draftsmen made single layer hand taped stuff for through hole parts is over.
I seem to have stepped on your toes. Why so aggressive?
And don't worry, I've seen plenty of PCBs. And designed them too...tedious job. In 95% of the cases it's draftmanship.
 

Offline free_electron

  • Super Contributor
  • ***
  • Posts: 8550
  • Country: us
    • SiliconValleyGarage
Re: The End of Electrical Engineering Design?
« Reply #19 on: July 13, 2022, 11:22:28 am »
I seem to have stepped on your toes. Why so aggressive?
And don't worry, I've seen plenty of PCBs. And designed them too...tedious job. In 95% of the cases it's draftmanship.
Because this is what i do : PCB design. It was the same thing when i was doing IC design. polygon-pushers ,p-cell placers , line drawers ... derogatory names for extremely complex work that the "electrical engineers" couldn't even do.

The same is in the PCB world "Draftsmen". Like we're sitting behind a large sheet of paper with a ruler and a set of rotring pens... drawing "lines" without understanding.

That stuff was true 50 years ago. Modern boards require highly complex analysis. Field solvers, thermal solvers, building a layerstack that works for all the impedance criteria , velocity is different for inner/outer, signal integrity , power delivery network analysis, signal integrity.
Many times we have to tell the "engineers" that their schematics suck. Proper termination , what to be placed where (spatially). proper routing , testpoint usage. half of the designs are not manufacturable. Place a part the "engineers" selected and it can't be routed: the voltage standoff needs 10mm clearance between two pins and they picked a ... 5mm pitch part.

Then there is the mechanical engineering involved. Boards are part of an assembly. Boards with embedded components. mating stuff with heatsinks , pressfit .
Design for manufacturing. You may have the most wonderful schematic design but if it can't be built ...(at an affordable price)
Optimisation for cost : reducing layers, faster assembly , fewer processing steps.

A board in a smartphone is a highly complex design done by highly specialized engineers.

What is the most often heard comment about the PCB design ?
Why does it take so long ? it's only lines from point a to b. we gave you the netlist.
How come these draftsmen need so much time to do that ? They have all these powerful computers and it still takes weeks. yeah, cause there are so many criteria it grinds even the most powerful computer to a halt trying to run the DRC with 15000 rules when moving a resistor.

engineers.. lol. They can design billion gate processors but if the fuse in their desklamp breaks they need to go ask the lab tech to replace it. And never, ever let them around a soldering iron, you will run out of burn cream in your first-aid kit before you can blink. I've had to apply the burn cream to those "engineers". Grabbed the iron by the hot end, and then complained to his manager "why do i have to do this lab work ? that's for the technicians". The boss agreed. It's too complicated and dangerous for the engineers to do, they are better off doing simple things behind their keyboard.

And that billion gate design ? big whoop.. They wrote 5 lines of verilog to glue some ip blocks together and they had a netlist. Work done. No, the synthesizer did it. Talk about menial work. So simple a computer program can translate it. You don't even need an AI. it's algoritmic. But trying to get that things to run timing wise, parasitic wise ? layout wise ? that's where the real engineering starts. It all looks good on paper and in theory. Designed with ideal components. Translating 5 lines of code to a netlist can be done mby machine, easily. placing and routing that netlist ? not so easy ...

And software writers ? lol. 5 lines of python (or whatever language du jour) and their 5 GHz 128 gigabyte 64 core dual xeon grinds to a halt due to sheer overhead. 32 kilobyte and 4MHz got us to land on the moon.
Software for PCB design ? even bigger lol. 50 years they have been yakking about how "soon" the "draftsman" would be replaced by software , soon .. as soon as those damn hardware jockeys can build us  will have faster computers ... now excuse us while we do another round of heated debate on vi vs emacs , gnome vs kde , bsd vs linux vs mac vs windows, we're waiting for the machine to reboot after our latest build crashed it , again... must be all the comic radiation in this room.

Meanwhile the computer scientist go "coders ? you mean the keyboard pounding code monkeys right ?" ahh , those will be replaced by our "artificial intelligence" . when ? soon ... real soon , as soon as when the engineers can make a quantum processors and the code monkeys can implement our AI. But it seems everybody is waiting for the draftsmen to complete the damn layout.

Somewhere on a factory floor, there is a line operator going "effing ididots, the whole bunch of em, this part doesn't fit, and that one doesn't solder properly, and this goddamn AOI software keeps seeing ghosts".
Meanwhile a supply chain engineer goes "damn it, they specced another part that is unobtainium , obsolete , or nowhere in stock and without substitutes".

And the customers keep screaming : i need a better phone , with a better camera , and faster video call , so i can post clearer pictures on my facetwitgramterestblogtube so my followers can send me more likes and i will be the most followed person in the world ! People used to be paranoid and think they were being followed. They now actively seek followers !

So, yeah, it rubs the wrong way when someone talks down about something that isn't their field.
Professional Electron Wrangler.
Any comments, or points of view expressed, are my own and not endorsed , induced or compensated by my employer(s).
 
The following users thanked this post: thm_w, pcprogrammer

Online CatalinaWOW

  • Super Contributor
  • ***
  • Posts: 5571
  • Country: us
Re: The End of Electrical Engineering Design?
« Reply #20 on: July 13, 2022, 01:14:34 pm »
I seem to have stepped on your toes. Why so aggressive?
And don't worry, I've seen plenty of PCBs. And designed them too...tedious job. In 95% of the cases it's draftmanship.

Because this is what i do : PCB design. It was the same thing when i was doing IC design. polygon-pushers ,p-cell placers , line drawers ... derogatory names for extremely complex work that the "electrical engineers" couldn't even do.

The same is in the PCB world "Draftsmen". Like we're sitting behind a large sheet of paper with a ruler and a set of rotring pens... drawing "lines" without understanding.

That stuff was true 50 years ago. Modern boards require highly complex analysis. Field solvers, thermal solvers, building a layerstack that works for all the impedance criteria , velocity is different for inner/outer, signal integrity , power delivery network analysis, signal integrity.
Many times we have to tell the "engineers" that their schematics suck. Proper termination , what to be placed where (spatially). proper routing , testpoint usage. half of the designs are not manufacturable. Place a part the "engineers" selected and it can't be routed: the voltage standoff needs 10mm clearance between two pins and they picked a ... 5mm pitch part.

Then there is the mechanical engineering involved. Boards are part of an assembly. Boards with embedded components. mating stuff with heatsinks , pressfit .
Design for manufacturing. You may have the most wonderful schematic design but if it can't be built ...(at an affordable price)
Optimisation for cost : reducing layers, faster assembly , fewer processing steps.

A board in a smartphone is a highly complex design done by highly specialized engineers.

What is the most often heard comment about the PCB design ?
Why does it take so long ? it's only lines from point a to b. we gave you the netlist.
How come these draftsmen need so much time to do that ? They have all these powerful computers and it still takes weeks. yeah, cause there are so many criteria it grinds even the most powerful computer to a halt trying to run the DRC with 15000 rules when moving a resistor.

engineers.. lol. They can design billion gate processors but if the fuse in their desklamp breaks they need to go ask the lab tech to replace it. And never, ever let them around a soldering iron, you will run out of burn cream in your first-aid kit before you can blink. I've had to apply the burn cream to those "engineers". Grabbed the iron by the hot end, and then complained to his manager "why do i have to do this lab work ? that's for the technicians". The boss agreed. It's too complicated and dangerous for the engineers to do, they are better off doing simple things behind their keyboard.

And that billion gate design ? big whoop.. They wrote 5 lines of verilog to glue some ip blocks together and they had a netlist. Work done. No, the synthesizer did it. Talk about menial work. So simple a computer program can translate it. You don't even need an AI. it's algoritmic. But trying to get that things to run timing wise, parasitic wise ? layout wise ? that's where the real engineering starts. It all looks good on paper and in theory. Designed with ideal components. Translating 5 lines of code to a netlist can be done mby machine, easily. placing and routing that netlist ? not so easy ...

And software writers ? lol. 5 lines of python (or whatever language du jour) and their 5 GHz 128 gigabyte 64 core dual xeon grinds to a halt due to sheer overhead. 32 kilobyte and 4MHz got us to land on the moon.
Software for PCB design ? even bigger lol. 50 years they have been yakking about how "soon" the "draftsman" would be replaced by software , soon .. as soon as those damn hardware jockeys can build us  will have faster computers ... now excuse us while we do another round of heated debate on vi vs emacs , gnome vs kde , bsd vs linux vs mac vs windows, we're waiting for the machine to reboot after our latest build crashed it , again... must be all the comic radiation in this room.

Meanwhile the computer scientist go "coders ? you mean the keyboard pounding code monkeys right ?" ahh , those will be replaced by our "artificial intelligence" . when ? soon ... real soon , as soon as when the engineers can make a quantum processors and the code monkeys can implement our AI. But it seems everybody is waiting for the draftsmen to complete the damn layout.

Somewhere on a factory floor, there is a line operator going "effing ididots, the whole bunch of em, this part doesn't fit, and that one doesn't solder properly, and this goddamn AOI software keeps seeing ghosts".
Meanwhile a supply chain engineer goes "damn it, they specced another part that is unobtainium , obsolete , or nowhere in stock and without substitutes".

And the customers keep screaming : i need a better phone , with a better camera , and faster video call , so i can post clearer pictures on my facetwitgramterestblogtube so my followers can send me more likes and i will be the most followed person in the world ! People used to be paranoid and think they were being followed. They now actively seek followers !

So, yeah, it rubs the wrong way when someone talks down about something that isn't their field.


I only disagree with you on one point.  Regardless of your job title or training, you are what I call an engineer.  So your polemics against engineers vs the board designers are weird.  Now if we change the discussion to describing what good engineers do, I am totally on board.  There are plenty of not so good engineers.  Almost all are not so good early in their careers.  Unfortunately quite a few don't improve much over time.
 

Online coppice

  • Super Contributor
  • ***
  • Posts: 10034
  • Country: gb
Re: The End of Electrical Engineering Design?
« Reply #21 on: July 13, 2022, 01:19:10 pm »
Idiotic spin.
PCB design today is 5% of design effort and has very little to do with engineering. It's more in the Draftsman trade.
You've probably never seen a real pcb.
Try a modern motherboard. or a high power inverter with embedded transformers. Maybe something with all kinds of controlled impedance, signal integrity, looptime compensation , 8 layers, embeddded heat medallions, embedded wires , stacked / dogboned laservias / vippo / elic ? something that has 0.5mm BGA's and 1000 ampere running through it and 800 volts busbars. Then you will understand how much engineering goes in the board. Without that engineering your nice schematic is SNOT. if the layout is not done properly , at switch-on time you'll end up with a smoldering crater where once was your lab bench.
The time where draftsmen made single layer hand taped stuff for through hole parts is over.
So, why are only 5% of the engineers doing PCB layout in a typical design centre?
 

Offline free_electron

  • Super Contributor
  • ***
  • Posts: 8550
  • Country: us
    • SiliconValleyGarage
Re: The End of Electrical Engineering Design?
« Reply #22 on: July 13, 2022, 04:02:13 pm »
I only disagree with you on one point.  Regardless of your job title or training, you are what I call an engineer.  So your polemics against engineers vs the board designers are weird.  Now if we change the discussion to describing what good engineers do, I am totally on board.  There are plenty of not so good engineers.  Almost all are not so good early in their careers.  Unfortunately quite a few don't improve much over time.
Because it is all about the "attitude" ! The endless being looked down on because "you only draw lines, push polygons". In some schools this is spoon fed that they are better than board designers, lab techs , factory workers, reworks technicians, test technicians and anything else that is involved in getting a product to market. Just because they have a fancy paper on their wall. Sure, i won't deny they have done heavy studies with lots of maths. But does that make them good at actual work ?
I've seen it so many times in 30+ years in the industry. the theoretical scientists laugh at the experimental scientists , they all laugh at the engineers , who in turn laugh at the layouters, who laugh at the factory assemblers...

"it's 95% line drawing (draftsmen work)" ... yeah , no it ain't... and draftsmen produce far better drawings than 99% of the engineers. You should see some of the schematics that "engineering" produces.

it gets old after 30 years...
Professional Electron Wrangler.
Any comments, or points of view expressed, are my own and not endorsed , induced or compensated by my employer(s).
 

Online Siwastaja

  • Super Contributor
  • ***
  • Posts: 9333
  • Country: fi
Re: The End of Electrical Engineering Design?
« Reply #23 on: July 13, 2022, 04:34:38 pm »
In my opinion, board design is an integral part of the design itself. It all interconnects starting from component selection ending to mechanical design.

If the target is to make good product, the team should be able to make changes all the way to component purchase order based on mechanical constraints from the enclosure.

For one-man show, this is all trivial, but for a team, tight communication is needed, where design changes can be requested in very quick cycle, from any design stage. Top-down flow "component choice" -> "schematic capture" -> "PCB layout" <- "mechanical design", with one way communications, will result in a product with inferior specifications, longer development time, and higher cost.

Being physically present in the office and discussing design over cup of coffee works best. One-man show can also work well, but that of course requires people with both wide and deep competence, and will put a limit on the total complexity of the project.
« Last Edit: July 13, 2022, 04:36:30 pm by Siwastaja »
 
The following users thanked this post: free_electron, bd139

Online Benta

  • Super Contributor
  • ***
  • Posts: 6420
  • Country: de
Re: The End of Electrical Engineering Design?
« Reply #24 on: July 13, 2022, 05:19:29 pm »
Just to be precise on my point about "idiotic spin": that adressed the original article which made it sound like AI designing PCBs would make engineers obsolete... as if electronic design is only about laying out a PCB.
Total rubbish.
I'm not denigrating PCB design, but presenting it as "this is all the engineering needed" is patently ridiculous.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf