Open Circuits: The Inner Beauty of Electronic Components
pluralistic.net
pluralistic.net
It covers everything from railroad tracks, antenna towers, highway overpasses, power lines, grain elevators, oil refineries, steel mills, and more… Lots of great pictures and detailed explanations.
And I also bought "Open Circuits" and find it to be both beautiful and illuminating.
Typeset in the future https://typesetinthefuture.com/
The Rudiments of Wisdom by Tim Hunkin (cartoons not photographs)
The Thrilling Adventures of Lovelace and Babbage by Sydney Padua (cartoons with epic footnotes and endnotes)
https://nostarch.com/download/OpenCircuits_Chapter1.pdf
$30 for the ebook. Free on the high seas for whose who don't have $30.
Ta-MnO₂ capacitors do have that one, minor, tiny, insignificant problem of starting on fire sometimes. This is not actually much of a problem anymore because people have learned how to build these things well (they're 100% screened at manufacturing), specify them (derate to 50-75% of nameplate voltage, modified by max temp), and protect them (surge/inrush current is mostly what actually ignites them, so let's not do that if we want them to live). The easiest way to use them well is to keep them inside designs (i.e., on internal signals and rails, don't let things off the board dump current through one) and they'll be fine.
They're great, and every designer should use them when appropriate!
Or are most sourcing issues solved by now?
[1]: https://www.kyocera-avx.com/docs/corporate/Responsible-Miner...
[2]: https://www.kemet.com/content/dam/kemet/lightning/documents/...
One can frequently see an 6.3V 22uF to 100uF SMD tantalum capacitor being used for the bulk capacitance right after the DC-DC converter. Closest MLCC you can buy is 25V 22uF X7R, which costs slightly more and provides less capacitance (even when comparing 22uF nominal) with further loss due to aging. I have yet to see 100uF MLCC.
Not all designs strictly require them, though. When for example dropping 5V to 3.3V using an LDO, having all components decoupled well enough locally, you might get by without bulk capacitance. Some boards are large enough for electrolytic caps that can be combined with ceramics to get you close.
Correct! This, plus their CV product, is one of the reasons they get used.
> So you don't have to use caps rated for like 10x your working voltage to get anywhere near the rated capacitance when decoupling.
You still need to derate them appropriately, to 50-70%. They have a rail voltage and category voltage in addition to their nominal voltage. It can get kind of complex, but at least it's well documented in the manufacturer literature. (These days.)
Ceramic DC bias derating is not related to voltage derating. It's really most correlated with package size. (Which used to track voltage rating pretty well, but doesn't these days.) There are also more and less expensive dielectric formulations, even within X5R or X7R classes, and guess what? The more expensive ones tend to lose less capacitance with increasing voltage. Go figure. Using the manufacturer tools is critical here.
> I have yet to see 100uF MLCC.
They're all over the place. You can even get 330uF 1210s from Murata. And probably bigger from someone else. These guys are popular among certain classes of design (mostly size or height constrained). Not my favorite choices though... because as SimSurfing will show you, they're sure not good for that 330uF: you get about at 240uF at 1.8V or 130uF at 3.3V. Which is still pretty great for 1210! But tantalum (AVX TAJ series, your boring basic tantalums) will give you 150-220uF in nearly the same space (1411/B case) depending on how hard you feel like pushing them. (Ceramics can beat tantalum's CV product for the small cases as there's some fixed overhead to just getting the tantalum pellet hooked up.) But then if you use MLCCs you have to deal with the underdamped PDN you just created somehow. (Which isn't hard, unless you forget about it.)
> Not all designs strictly require them, though. When for example dropping 5V to 3.3V using an LDO, having all components decoupled well enough locally, you might get by without bulk capacitance.
Please don't do this. (Unless you're trying to keep me employed.) Yes, you can engineer the bulk caps out of your design, and that's a viable strategy. But leave the footprints in your first spin so you can make it work, then Muntz them out after you know everything's OK. You will save everyone a lot of grief this way.
> Some boards are large enough for electrolytic caps that can be combined with ceramics to get you close.
This is actually one of the best overall strategies if the form factor allows it. Ceramics in parallel with tantalums or aluminum electrolytics is a fantastic combination and gives you the benefits of everything and drawbacks of... well, the electrolytics are still big.
Old RIFA film capacitors are known to fail, explode, and cause collateral damage to nearby components. I’ve seen it happen in several vintage Apple ][ and Mac power supplies.
1 - Take out all of the tubes, except the high voltage rectifier, if possible (some systems wire them in series, in that case, this won't work) Be sure you have good and correct documentation about tube placement, you wouldn't want to make it worse.
2 - Measure the voltage across the resistor in series with said capacitors. You should read zero volts across them... even with 250 volts B+. If there is a voltage, you either have a resistive path to ground you didn't see, or a bad capacitor, leaking to ground. (There will be zero current through a good capacitor, thus zero volts across any resistor in series with it)
3 - Replace all the tubes
This saved us hours of trouble once I figured this trick out.
If that was a real problem, people would die a lot more often than they do, just from touching random things in the world!
No, the capacitor organic electrolytics are things which are not really very bad (mostly glycol-y things and organic acid-y things, I think?) but which are best washed off your hands. That's all. There aren't any really nasty things in them, fortunately (and I have read enough RoHS/REACH/Prop65 crap to be confident of that, sadly).
You saying that I agree doesn't mean that I actually agree. Also, putting that silly legalese at the top of the page is pretty annoying, for what turns out (after a dozen paragraphs) to be a chatty book review.