EE Grad - far removed - need insight on latest technologies

[chris_n]

Hello all.  I'm very happy that I've found this forum.  I look forward to contributing as best I can.

I'll try to be as brief and concise as possible.  I am an EE grad from 2003.  I am very far removed from hardware design, I actually never delved into that career path.  I was a systems engineer for a short stint and then I went into operations management.  I still work for an engineering heavy organization however.

What are some of the must haves skills for a new hardware design guy trying to dip his toes back into the industry?  What are some of the newer technologies out there?  I know AI is a standard answer but anything else?

Thank you in advance.
Chris

[tggzzz]

Hello all.  I'm very happy that I've found this forum.  I look forward to contributing as best I can.

I'll try to be as brief and concise as possible.  I am an EE grad from 2003.  I am very far removed from hardware design, I actually never delved into that career path.  I was a systems engineer for a short stint and then I went into operations management.  I still work for an engineering heavy organization however.

What are some of the must haves skills for a new hardware design guy trying to dip his toes back into the industry?  What are some of the newer technologies out there?  I know AI is a standard answer but anything else?

Thank you in advance.
Chris

AI is a red herring in that context. Modern AI (cf good old fashioned AI) is statistical processing of the training data. The problems with such AI are nothing to do with (digital) hardware.

Apart from that, your question is too vague: there are too many types of electronic circuits. Start with analogue, digital, RF, DSP - and progress from there. Design starts by partitioning the application requirements into analogue/digital/RF/DSP domains - and that implies knowing those domains' relative strengths and weaknesses.

If by "hardware design" you mean digital hardware, then start by realising that digital circuits are actually analogue circuits that interpret voltages/currents as digital signals. The outputs of modern jellybean logic gates contain significant energy >1GHz; hence RF/microwave design techniques need to be employed. Ensuring correct interpretation requires good "signal integrity", a whole discipline in its own right.

[Psi]

Yep, we need more info about what sort of EE job you want to get back into

[JustMeHere]

Get your favorite micro and build a custom board for it.  Don't worry about getting too fancy.  For example, built a board that plugs into the back of a LED panel and drives it.   I used an ESP32.  In another project I made a board that interfaced with a RFID scanner.  It used an Atmega32U2.  I've done others too.   Don't be afraid to do 4 layer boards, and SMD.  They are actually quite a bit easier than 2 layer trough hole.

[T3sl4co1l]

The average EE is probably closely integrated with embedded systems.  Knowing at least one platform and how it works from top to bottom, even if it's Arduino (because once you know it, the Arduino layer doesn't matter, it's just an ATMEGA or STM32 or whatever with code on it), can do real work.  Many projects have wireless involved, WiFi, Bluetooth, etc., IoT generally.

If you're not working with or supporting an MCU or other computer system, then it's probably power electronics -- power always and ever will need to be transformed, switched, controlled, alternated.  Or even if so -- many such controllers benefit from an MCU, even just a simple one to make the logic programmable, as opposed to rev-ing the circuit to rewire some gates.  Or even just save the cost of more than a couple gate ICs.

There are the odd jobs dealing with analog applications, whether it's as simple as an amplifier, a proprietary sensor, or as complicated as RF something or other; but these are pretty minor cases for a beginner.

Analog I wouldn't say is more difficult than software, I would say software is geometrically exponentially combinatorially more complex, but it is a different box of tools and thought processes.  In particular the reciprocity of node voltages and current flows makes for more considerations (think if a software function's output depended on what function's receiving it), plus managing the nonlinearity and potential chaotic dynamics of real devices (most commonly, transistors' exponential transfer function, and their nonlinear capacitance, as well as some inductor and capacitor types with significant nonlinearity).

RF in particular, if you want to get into it, may be worth continuing education to refresh or pick up the theory to work with it, and to get some practice on current tools -- field and circuit simulators, and network solvers, are quite important on the design side, as are VNAs and test jigs in the lab.

Tim

[chris_n]

Thank you for the reply guys.  I appreciate it.

I knew my question was vague while I was typing it but I figure I'd give it a shot.

I currently work in the RF/DSP/VHDL world of electrical engineering.  There is software design mixed in as well.

I've always had a curiosity with small handheld LCD GUI design.  Nothing too complex.  What would you all suggest would be the best starting path for that?

Thank you
Chris

[Psi]

There's plenty of ESP32 modules with build-in 3.5" 480x320 color LCD touchscreen that you can mess around with.

I got the ESP32 touchdown, but there are others.

https://esp32touchdown.com/

Find something very popular with lots of online help/tutorials

[abeyer]

I've always had a curiosity with small handheld LCD GUI design.

Given you've focused on that: I'd argue that your problem is 80% design, 19% software, and only 1% hardware related. There is a limited pool of micros that are capable of the io and throughput to drive a reasonably responsive lcd interface... but once you've committed to that as a minimum spec there's not a huge difference between the options. If you have very specialized interfacing requirements beyond the ui, then maybe that is something to guide your choice, but otherwise I'd tend to say just pick a reference platform and build on that.

[T3sl4co1l]

What do you mean by "limited"? Most anything Cortex M0 or equivalent power will do.  Small LCDs don't even need that, AVR or PIC will suffice.

Like, I did one project with a 160 x 128 display that's 10MHz SPI, so you can't go very fast anyway, and drivers written in C are fine even on AVR.  That's mostly going to be precooked graphics, basic blitting, no framebuffer of course, and maybe (or even preferably?) palettized graphics to save on space.  With a parallel interface via GPIOs, access can be awkward, and ASM helps on more meager platforms; most ARM chips are either fast enough that this isn't a problem, or have a bus interface that might be shared with external SRAM or Flash as well.  I once played around with a 800x480 display with a parallel bus controller, and fill rate is comparable to, say, putting a low-end SVGA on an XT PC:

https://imgur.com/gallery/display-initialization-photo-w1FjqoX

PCs certainly weren't useless at the time (say mid 80s), despite the slow e.g. EGA/VGA graphics.  Of course, this improved rapidly into the 90s, by which point full-screen gaming, with more than just screen scroll gimmicks, became practical.  (The analogy is pretty strong; AVR is a bit faster than an 8086 overall, I think, but the limited memory capacity, and most access requiring GPIOs (unless you splurge for a chip with bus interface) limits it for general-purpose use, not to mention the Flash program memory.)  But you could still do some pretty fancy interfaces this way, if you are careful with how much you have to redraw at a time.

In any case, needless to say, most any MCU in the 100s of MHz can do very well at display graphics, even with software routines for drawing shapes (antialiasing included), frame buffer (if external memory provided), implementing a canvas, etc.

Tim

[EEVblog]

What are some of the must haves skills for a new hardware design guy trying to dip his toes back into the industry?  What are some of the newer technologies out there?  I know AI is a standard answer but anything else?

As other have said, it really depends on what sort of industry you want to get into, and what aspect of design.
Electronics design is so broad, it's pretty hard to make generic recommendations, as there are literally dozens of specialities we could rattle off.
Like, do you want to be a "jack of all trades" electronics designer, or a specialist in some aspect?

Generally speaking it would pay to be up the date on the latest embedded platforms and how they intergrate with the the web and other services as one example.

[EEVblog]

I knew my question was vague while I was typing it but I figure I'd give it a shot.
I currently work in the RF/DSP/VHDL world of electrical engineering.  There is software design mixed in as well.
I've always had a curiosity with small handheld LCD GUI design.  Nothing too complex.  What would you all suggest would be the best starting path for that?

Depends on whether you want to do "chip level" design, or do higher level stuff like use existing modules which make it pretty easy. e.g https://4dsystems.com.au/

As always, the end product market needs drives the design descisions. e.g. if you are designing a small volume or one-thing for a client, then you wouldn't waste your development time cost-optimising the design (choosing the right chip and LCD, writting the graphics routines etc), you'd usually just want to get the job done with a higher level off-the-shelf module like above and program at the higher level.

[tggzzz]

Like, do you want to be a "jack of all trades" electronics designer, or a specialist in some aspect?

That is always a good "touchstone" question, since your answer will shape your future.

There is no right/wrong answer; both choices are equally valid. It is just that they have different risks and rewards.