FEDEVEL
Platform forum

Andrew Green , 01-11-2025, 05:04 PM

Need your advice on how to break into IC design at the age of 34.

I got my B.S. in CSE from UNR 11 years ago but I feel like I’ve learned very little. Got an internship that I botched with my poor programming skills and little clue how C++ compilers work. I heard horror stories about crunch in video game industry. Since then I spent years in sales and material handling. Now I want to earn a living designing IC's - controllers, memory chips, etc. The best response I got on Quora and Reddit was that I have to zero in on PCBs or IC's, not both as they require different skillsets. I think starting with IC's makes more sense.

My completed projects:
- Ben Eater's CPU board, VGA board, EEPROM programmers
- a solderless board for plotting pixels on a 16x2 character LCD
- a flashlight with a hand-wired prototyping board and a battery

Now I want to design and prototype the keyboard encoder, the USB controller, and the PCB circuits (fill me in if I am missing something) for a 3x2 keypad that supports USB-A 3.0 - one circuit at a time, though I'm not exactly sure where to start.

Plan:
- find work with a PCB manufacturer (I had a shot with TTM Technologies!),
- apply for admission into a bunch of colleges in the Bay Area + San Jose State University,
- get career counseling,
- find the right program,
- learn its curriculum,
- take some online classes to prepare,
- nail the curriculum

Not employed by a sponsor, so apprenticeships are out. I would continue on to my M.S. if I retained any interest in much higher-level software and my university credits weren't so old. However, I feel like my heart is much closer to C, assembly, ISA CPU architectures, and development of BIOS routines. I'm OK at Bash, Python, C, and x86 assembly and yet to learn Verilog and CAD and sim tools. I am taking courses on Udemy. Already got a LinkedIn profile and two GitHub repos.

Lastly, I got an interview with a Panasonic rep but now my gut tells me it's a waste of time.

Thank you.

QDrives , 01-11-2025, 07:40 PM

Robert made (multiple?) videos with Matt Venn on IC design:
https://www.youtube.com/watch?v=YounUJvIW04
https://www.youtube.com/watch?v=ILZ6fDHZ_eo

IC design is a much slower process, as in between design and seeing the product work. You will also do a tiny part of a bigger whole.
From what I heard, universities are not the best places to learn electronics. A lot of (outdated) theory and little practical. Here in the Netherlands, the students need to do 'projects' in order to design a board. That project is a 'break' of one year.

As for PCB or IC - Yes, there is a difference in skillset and, yes, it is (almost) either or. But more importantly, which one would you prefer to do?

Robert Feranec , 01-12-2025, 12:53 PM

I would also add this video - it can give you an idea how to start: https://youtu.be/caXwuuXSB-A

lookfwd , 03-16-2025, 07:23 PM

I graduated 1st on one of the best MScs in Microelectronics (Southampton, UK) back in 2005. Then I worked for ARM for 4 years till 2009. I would like to gently express that your view of the industry might be a little bit romanticized, and the YouTube videos further romanticize the industry. This is an industry with EXTREME specialization. Here's a high-level list:

* Digital IC Design
* Analog/Mixed-Signal/RF IC Design
* Verification and Validation
* EDA Tool Development
* Packaging
* Process Technology
* Various architect and Product

In a company you will likely be something as specialized as "debug interface engineer" or "core GPU verification engineer".

The closest you can get to a solo person doing IC design is to first write some Verilog, which might take one unit of time (might be a day or a year). Then you likely spend 3 units of time verifying this design which means writing testbenches and loading on an FPGA and ensuring it does what you want. Then you will work with some people that will synthesize your design for a technology and then other companies will fabricate the chip and you will have a test chip in your hands. It's likely that there will be a couple of test chips before you get it right, but by using hardware fuses, you can have something somewhat working faster. If you want something more complete than a simple digital IC e.g. with Analog or RF parts you will have to collaborate with IP vendors that will help you get that right. Bottom line, we're likely talking millions of dollars and quite some time (* there are exceptions). The most important thing you need to be to succeed, is a business person. You will need to discover a significant need that will make your project financially viable.

Bottom line it's nothing like software. There are lots of very specialized jobs that you have to explicitly choose over and target with a multi-year career path. I think the #1 skill in the industry is Verilog, so definitely try that to see if you like it.

Denis , 03-16-2025, 09:00 PM

I have found this video some time ago.
https://www.youtube.com/watch?v=DdF_nzMW_i8&ab_channel=bitluni
The author designed some hardware on verilog and it was manufactured together with others designs.
The production is very expensive. It can cost millions of USD to produce first chip.
To reduce a cost companies can collaborate and order their IC together.

lookfwd , 03-17-2025, 01:48 AM

There are some tiny academia and hobbyist scenes. But industry is winner takes it all. Right now, Nvidia makes money, AMD doesn't, Intel hopes for a miracle for survival. Imagine the little boutique Verilog shop. The last thing that was a glimpse of opportunity for the boutique shop was crypto mining hardware that had a relatively straightforward ROI. But creating a better processor? If it hasn't been done yet, there's a reason. Even Risk-V didn't bring any big change in the scene.

Beyond that side of the business, on a personal level, you have to be a pretty clever person to do hardware. Almost everything you do is distrubuted-systems level work i.e. the hardest part of databases. The testing part is insane. That's because you really can't get things wrong, since you don't just push to git and auto-deploy to production. The hardware bug is very expensive. How does the industry deal with that? Comparatively very little innovation i.e. if something works, you double the width of the bus or something and you call it the "next generation" of the family. What I'm trying to say, is that you will highly likely find much better use for your talents (including 5x higher salary) doing something other than designing ICs.

I loved this industry. I felt it let me down. But I might have been unlucky. In any case, Verilog is the first and most important step

Denis , 03-17-2025, 08:23 AM

I guess I was more lucky with University. We studied VHDL/Verilog for one year.
On the first year we used Modelsim and opensource simulators.
Another year was spent on testing/verification and FPGA.
It was enough for me to pass interview and be recruited as verification engineer.
University gave a lot of theoretical information. Can not say that it was outdated.