Comment by kazinator | Hacker News Reader

Comment by kazinator | Hacker News Reader

Parent Full thread

I self-learned enough electronics a few years ago to get to 16.5K karma on the Electronics StackExchange.

I was motivated by doing some audio projects. Projects have real requirements, and so they force you to iterate on the design until you hit all your requirements: power supply logistics, signal purity, enclosure, ...

Get a good textbook like Horowitz The Art of Electronics.

Learn how to use a CAD-based circuit simulator program like LTSpice. Build the circuits you read about in the simulator, and run them: apply signals, and look at how the voltages behave at various nodes in the circuit, as a function of time.

Read schematics.

Read schematics for equipment that you know. If you're into vintage audio, that is not hard to come by.

Recently I was looking at the schematics for a "Furman PQ-3" parametric equalizer (Google for it). I blinked twice and did a "double take" and then immediately recognized that its filter bank consists of "state variable filters": https://en.wikipedia.org/wiki/State_variable_filter

Bam! Didn't even know what that was some four, five years ago.

Here is one copy of the schem: https://www.gearslutz.com/board/attachments/so-much-gear-so-...

Check out the power supply: the output of the transformer goes to a dual-voltage regulator. That feeds the chips. The unregulated voltage is also tapped and that is used for an emitter-follower output-stage on the upper left.

This is completely pointless. The op-amp IC's have such stages inside them too; why do they get regulated power and this one doesn't? On a dual supply, op-amp chips don't really need regulation.

If I built a clone of the device, I'd completely leave out this discrete component output stage; it is pointless. You're not going to drive speakers with this thing, but relatively high-impedance inputs (the next device in the chain, possibly a power amp).

So you can see what I'm doing here; critically looking at (the electronic aspect of) a complete product. Doing that requires some learning, but it also produces learning bit by bit.

You ask questions: why is that stage here? Why did they include this component? What is this transistor/resistor/diode doing here? Is there a pattern to this, and where have I seen it before? Is it really the same pattern and is it justified in this context? And so on.

I took a physics class using the Art of Electronics in my senior year of undergrad (my major was nuclear eng. and my lab partner was another nuclear eng. friend).

This class ranked right up there with my nuclear engineering labs in the following sense: 1) using an oscilloscope in a lab setting takes a lot of patience & hard work (similar to radiation detectors) 2) I wasn't prepared for how "fuzzy" (sorry, I know that is not the right word) electronic components behave when examined in a lab setting. I was used to resisters and capacitors, and in previous labs they behaved fairly well. This class showed me how complex it all is, and "Art" is not a bad word to describe it at all.

I learned a lot and strongly 2nd the Horowitz recommendation if you want to really get down into the nitty gritty. Maybe it isn't the first book you pick up depending on your background, I dont' know. AND, I hope oscilloscopes and their user manuals have gotten a lot more friendly in the intervening years since 1991 :-)

> "Art" is not a bad word to describe it at all.

I consider analog electronics to be more akin to "dark magic" than anything else...

Analog wizards say that about RF engineers.

I spent quite a few years as an wide-band RF engineer and I still think it's black magic. If you're not being snobby, you can learn a lot from an experienced bench technician. If you want to get into RF electronics, you need specific features in your spectrum analyzer along with a very good return loss bridge at your target impedance.

I'm another fan of Horowitz and Hill's _The Art of Electronics_ - it's my go-to reference after 35 years in the business.

Been doing RF/Microwave design for 24 years. It’s not black magic, though that’s what led me into it.

As a "dark magician" I think that about digital electronics.

And it's really not that hard.

Back in '77, op-amps weren't as good as they are now in terms of PSRR and output power. That push-pull output stage is quaint but given that they wrapped it in the feedback loop of the last op-amp, it may have been a necessity at the time.

For sure. The kind of thing that should get picked on is cheaping out and putting an analog section directly on a rippling rail.

Also, the discrete output section helps power dissipation in the regulator. Especially if the output gets shorted.

... And I was curious, so I just looked at the datasheet for RC4558. Maximum supply voltage is +/- 18V. So you'd probably have to bring that unregulated "20V" rail down to 12-14V actual to stay within that.

> The kind of thing that should get picked on is cheaping out and putting an analog section directly on a rippling rail.

Ah, but that section here is a feedback-stabilized amplifier stage; and the power rail is dual-voltage. The ripple in the positive rail swings opposite to the negative one.

Lots of power amplifiers use no voltage regulation for the rails. The amplifier is a kind of voltage regulator already anyway; if you add a regulator, you're basically adding another amplifier to the amplifier.

(Hey, I heard you like amplifiers, so I put an amplifier in your amplifier, ...)

I get the point about the noise being balanced, I just personally wouldn't default to trusting it.

But most of my experience is analog conditioning for microcontrollers - more sensitive, higher frequency noise, and admittedly single rail with today's op amp technology. An extra component doesn't necessarily add complexity like in software, but can actually simplify one's mental model of the circuit - nice stable supply rails.

If you look at that original circuit, how much did the regulator cost compared to say the transformer? Eliminating the regulator would have required changing the transformer, at least.

I'm just trying to get across that it's a bit of a red herring to prioritize cost-optimizing the circuit if you're setting out to build a one-off clone. Even hobbyist debugging time is worth more than a few ten cent parts.

Every time i've tried to use some combination of a Spice program and something like EagleCad I get really lost quickly. It's clearly software written for people who already know what they are doing.

Do you have any good resources on learning such conceptually?

I use CircuitLab, which is very helpful from a testing standpoint. As a coder, one of the things that really stresses me about electronics engineering is how untestable everything else, and CircuitLab gives me the ability to mock up simple unit-test-like circuits and see what the expected values should be, including when a power source fluctuates.

It costs a little, if that's a concern.

https://circuitlab.com

(CircuitLab dev here) Thanks for the link. I've also been writing an online electronics textbook with simulations built in https://www.circuitlab.com/textbook/ which should be relevant.

Post a new topic about it as you make updates man! People here would eat that up.

I have a circuit simulator program called EveryCircuit on my ipad (it looks like it's available on Chrome) and it's very simple to use and develop intuitions about basic components.

You really don't need to use EagleCad to lay out your own PCBs for a very long time into the hobby. Perf board and jumper wires are sufficient for prototyping for a long time, most SMD/ball only chips/components have prototyping boards from sparkfun or adafruit or whatever.

I took a few stabs at eagle before trying KiCad which I found MUCH easier to learn. The “Getting to Blinky” series on youtube was what i used but there are probably other good tutorials too.

+1 for KiCAD - it's somewhat hotkey-based, but once you get the hang of the key commands it is very easy and intuitive.

Honestly, after a few designs, making a circuit board feels as easy and fun as playing a puzzle game.

Hackaday has a good series of blog posts for familiarizing yourself with the basic commands:

https://hackaday.com/2016/11/17/creating-a-pcb-in-everything...

Also check out http://librepcb.org/. Probably just a few months left until the first early release. Watch the FOSDEM talk recording for more information.

I'd recommend not linking to that series of posts. The author is a hothead and generally disliked by the entire electronics community.

Actually, there are some rumors about him that are quite unsavory, and I'm waiting until the cetacean equivalent of the #metoo movement to make an appearance on twitter to see the fallout from that.

Fair enough I guess, but I would also hate to deny people the same sort of easy path that allowed me to get interested in the subject and access a complicated topic.

Do you have any alternatives to recommend?

That comment is from the series' author, just doing a bit of his signature trolling for which he is so beloved on Hackaday.

I know he (Brian) has expressed that those articles generated underwhelming metrics for the amount of work they take, but they really are a useful resource for people looking to get an overview of their options. And I'm sure the Whalebait fiasco will blow over soon enough.

Hilarious.

What's he doing with whales?

What's he doing with whales?

Falstadt (iirc) is a great web based one that is pretty easy to use. It's not as accurate as ltspice though but it's good for simple stuff.

http://www.falstad.com/circuit/

It used to be a Java applet, now rewritten in JavaScript. Sadly, the Java version vastly outperforms the JS one, at least on my machine, but Java applets are annoying to run nowadays

The JS one is plenty performant on my machine. I love that simulator. Helps me intuitively understand the circuit when I can see the voltage and current on each connection.

I do love that you can double click on a wire and check the box to show the current/voltage there.

Try my electron wrapper:

https://github.com/Immortalin/Simulacra

yes. Kicad and two video series made by Chris Gammell...

https://contextualelectronics.com/courses/shine-on-you-crazy...

https://contextualelectronics.com/courses/getting-to-blinky/

I'd recommend doing them quick, as Kicad is on the verge of a new major release(5.0) and these series might become outdated.

Some fun schematics to "read" are Fender's digital microcontroller adjustable tube amps; basically the modern incarnations of integration of tube technology with solid-state and modern microcontrollers. Last I knew (when I did web dev for them 5-6 years ago), you could download them from their site.

You can find similar circuits from the 1950s-60s Popular Science back-issues on Google Books; that's the time period when hobbyists were transitioning from tube-based stuff to more solid-state and transistor stuff (transistors came down in price enough, plus they were more reliable) - so occasionally, you can find an article on some project combining both technologies.

Microcontroller adjustable tube amps are just descendants/knock-offs of the ADA MP-1: what I use. :)

One reason it can be so frustrating (to me) to practice the “read a bunch of schematics” approach is that, unlike code that tends to be filled with inherent textual clues like filenames, function names, and variable names, electronic schematics tend to be very cryptic. Single letter names, only a subtle visual grouping of components to show functional units, no hint of the rationale behind component value choices. If this equalizer were digital, the filter would be a function called StateVariableFilter(), and you wouldn’t have had to intuit that from looking at it.

Basically, whenever I look at a schematic, I think “why do analog engineers like to work in the equivalent of assembly language?”

I do more digital than analog electronics, and based on Verilog and VHDL, those folks seem to be working with “stone knives and bear skins” too! At least they get to have real names for things.

Not being at all a professional electronic engineer, I’m sure this a misguided reaction, but I’m not sure exactly how.

I got a hand me down IBM PC jr in 1988, when I was in elementary school. 5.25" floppy, no hard disk, RAM measured in Kilobytes.

It came with a spiral bound manual that taught GW-Basic. I didn't learn a damn thing at the time, I just slowly typed the lines of code into the PC. I stuck at it long enough that I eventually drew a star on the monitor, as "Twinkle Twinkle Little Star" beeped at me its 8 bit glory

In that singular moment, I hadn't "learned" anything, but I knew then, I needed to take apart every single piece of electronics I could get my hands on. I never "learned" anything about schematics or what all of these pieces of metal do, yet I remained endlessly fascinated.

30 years later, I do embedded development. There's not a day that goes by when I "learn" anything. But the sheer joy of my continued failures, along with the rare, occasional success, has made me a very happy person, who backed into somehow figuring out how to read schematics, prototype a proof of concept, layout PCBs, order the parts from digikey, order boards from Dirty PCBs, solder them on to the PCB, program Assembly, C, Python, JS.

But when it comes to the folks that can do devops, thats just plain magic.

I don't think there is a "fast path" to electronics. I didn't learn it overnight or in a year or even two years, I didn't learn it in a course or lecture, and neither did anybody I know who is a decent circuit designer.

Second the art of electronics. That said, I recommend getting the lab course over the book. It's much better for hands on learning.

Reply on news.ycombinator.com