I appreciate your feedback but it concerns me. It sounds like a lot of trial and error with components that cost money. One concern I have with that is it's dealing with electricity. What's the odds of getting components fried or yourself electrocuted just doing trial and error learning from tutorials and datasheets? And when you've done a bunch do you even really understand what you're doing when given fresh components?
"If you need to know about design, read application notes from manufacturers. There are thousands and thousands. "
I'll try to remember that.
"You will almost always be constrained by more practical issues like enclosure sizes or where you have to put connectors."
I'll definitely believe that. Thanks for the tips and different perspective.
If you make electronics, the odds of frying a component or electrocuting yourself is 100%, regardless of how educated or talented you are. The following happen to me on a regular basis, whether I'm working on high speed digital/RF or making something simple like an arduino shield, even though I learned as an apprentice under a brilliant EE:
* PCB with poorly aligned copper layers shorting the second the prototype is plugged in, usually destroying at least one chip. Lots of smoke and burning FR4
* Solder in $100+ high power transistors the wrong way. BOOM
* Using a counterfeit capacitor from shady vendor that either shorts internally or just plain explodes. Happens a lot when I need a really large capacitor and have to get it on short notice
* Forget to use a little extra flux and tin 'whiskers' form between freshly soldered pins that short them the second the device is powered up. This is so common that NASA has a whole website dedicated to the topic [1]
* Use wrong temperature profile or make the pin layout a thousandth of an inch too small or large and bam, two solder balls on a BGA flow together, requiring hours for reflow and reballing if you're lucky, and a new $2,000 FPGA if you're not.
Any nontrivial circuit is going to be impractical to simulate (and impossible to describe analytically as a whole) for all but the most well funded projects so I'd say 90+% of EE is trial and error, even for the most experienced designers. There's many rules of thumb and you develop an intuition for a wide variety of situations just like you do in programming, but it's just a fundamentally different field with different constraints.
Wow. That's some crazy stuff. So, the scary, trial-and-error is unavoidable then. Thanks for the feedback. Btw, I just recommended a few books here based on feedback from other EE's...
Any thoughts on them? Particularly, a combo of something like Malvino and Circuit Designer's companion to get a good head start on analog and PCB's respectively. Or do you have other references that kick ass in teaching practice more than theory? Gotta build up links for new people to accelerate hands-on part of their learning just like others did for programming.
Note: Art of Electronics is usually in my list but that link was for digital learner. Not sure if it's needed there.
What sort of circuits do you want to design? RF? Audio? Analogue is a big field! It's like saying "I want to write programs, which books do you recommend?"
I'm gathering information to help everyone out then organizing, cataloging, and sharing it. You could say it falls into some basic categories:
1. Enough knowledge to get designs working on a FPGA plus integrate that with other chips on a PCB. OSS HW with minimal analog.
2. Enough knowledge to design basic analog circuits for control and stuff. Alternatively, to design digital cell libraries as there's almost nothing available for academic toolbuilders.
3. The serious, mixed-signal shit that lets me do some parts in digital and some parts in analog where it handles it better. I've seen analog coprocessors with 100x performance at 1/8th power on ODE's and stuff. It also seems like certain signal processing or media codec tasks would be crazy fast/efficient in analog. I know high-end ASIC's make extensive use of such techniques. What tidbits I see in blog comments and papers can only be described as black magic without a more thorough resource. :)
4. RF books outside of ARRL that's been recommended to me. Need a lot of people experimenting with this stuff to reinvent things like TEMPEST that are classified. They need some good resources to get head-start.
So, those are some basic categories where I'm looking for both accessible, foundational material and cookbooks with heuristics. Being able to combine COTS components like MCU's and FPGA's on custom PCB's is major help to hobbyists. Being able to make the cells and basic, analog components required in about any ASIC in conjunction with tools like Qflow OSS Synthesis could get custom stuff going quicker. More thorough stuff for mixed-signal for its advantages plus to explore analog and digital interactions in digital systems that can screw either up. And RF for reasons stated.
Whatever you have. Drop it here or email it to me in my profile address. I'll keep circulating that along with others tips and resources whenever people ask.
I've never had electric shocks before, but frying components sure. Don't work with mains voltage directly if it worries you. You can power most hobby projects from a USB port or a wall wart. Designing power supplies is one area where you might want to read up on things like trace clearances, but again, look at YouTube for PSU teardowns (bigclivedotcom has plenty).
As always it's mostly human error. I have never (literally) fried a component from overheating, it still amazes me how hardy modern ICs are. I've also never paid too much attention to ESD protection, though if my job depended on it then I would. What has happened is shorts, often. Even the GPU guy routed his board with GND and VCC back to front, it happens to the best of us. Most chips are at least partially tolerant to silly things like overvoltage, so even if you accidentally short some GPIOs on a micro, the protection circuitry might save you. Simple advice is to put a low-current polyfuse on every prototype you make. It's saved me so many times when I've shorted power supplies by accident.
The more complicated your circuit, the more likely it will be that you mess up. Don't try and solder a 150-pin BGA on your first board. Build some breakout boards for sensors, build your own microcontroller dev board (ARM if you want a challenge) or pick a project from the internet.
Odds of messing up you first board in some subtle way? Unless it's a very simple board, > 80%? Components are cheap though, roll with it! Plus you probably won't brick all the components if something goes wrong, the magic smoke will usually only come out of one.
When I make mistakes, usually they're footprint (e.g. wrong pinout) or construction errors (shorts between pads, etc). If the schematic is incorrect then that's another issue, but most often it's things like not reading a datasheet properly and forgetting to connect a pin, tying an positive-enable pin to GND rather than VCC, etc. In principle layout engineers assume that the circuit diagram is gospel, so the blame doesn't always lie with them. Of course if you're the designer and the layout engineer...
I'm now at a stage where I can get a board back from the fab and it'll usually work :)
Just to add some: There's a lot of specialization in EE like any other field and you can't really dive into everything. I find power electronics to be more of a black art than RF or high speed digital so I try to never design my own. I've seen people spend years working on a single design for a solar inverter making dozens if not hundreds of iterations, testing with all types of switching circuits and chips, comparing the behavior of one vendor's capacitor to another, understanding different transistor behavior, and so on. Thankfully, nowadays you can go to TI and use their automated schematic generator for your power supplies. They have many reference designs and tons of documentation on PCB layout in a variety of situations. Debugging them, however, is a whole different story. If you can afford it just buy a module and never think of it again. Polyfuses are a godsend, especially if you're designing a high current PCB like a motor controller. It really doesn't take much to melt a 1.5 oz 10 mil trace.
How expensive trial and error is depends on your experience. When I'm done with using a board or have some old electronics to throw away, whether I bricked it or its obsolete, I always throw it in a pile. When I have time I just go through and desolder each nontrivial part because it helps build an intuition for how each type of solder will respond to heat and flux, how wick looks as it absorbs solder and how to move it to get all the solder without overheating the chip, and so on. You need to develop that muscle memory like a surgeon would because once you're good enough you can do crazy things like snake a tiny wire under a BGA chip under an xray to fix flaws in the design or reflow. In more expensive designs I'll regularly take small gauge wire and solder it all over to fix the design as well as cut traces or lift copper layers after stripping the FR4.
Hand assembling electronics is largely more craft than engineering.
"If you need to know about design, read application notes from manufacturers. There are thousands and thousands. "
I'll try to remember that.
"You will almost always be constrained by more practical issues like enclosure sizes or where you have to put connectors."
I'll definitely believe that. Thanks for the tips and different perspective.