buying stuff from mouser, shipping fees are crazy high

[marcn]

Try it
As you mentioned 10uf in ceramic is hard to come by and if you find it its gonna be pretty big. That's reason enough to go electrolytic.
In case you're worried about 'sound' these sit in your psu path and all they do is store a charge so that the supply to your IC's or transistors is constant.

[mlatu]

Ah, so it's their size why you would avoid them, I haven't had my hand on many capacitors so I don't know how big they're gonna be.

In the picture they all look the same to me XD

https://i.ebayimg.com/images/g/y90AAOSw ... s-l300.jpg

[cygmu]

Huh, as you probably can imagine I didn't see that tiny plus sign.

I'm still curious about what would happen if I used an unpolarized cap there.

You can always replace a polarized capacitor with an unpolarized one. The polarization is a restriction on use rather than being advantageous in any way.

The issue with such a replacement is that you will inevitably be using a different type of capacitor (film or ceramic in place of electrolytic or tantalum, say) and different caps have different characteristics.

It is not at all true that ceramic caps are "close to ideal". Regular ceramics, with dielectrics characterised as things like X7R, have some decidedly non-ideal properties. The main one is that their capacitance varies with the voltage across them. In the audio path, this leads to distortion: as the signal (and therefore voltage) varies, the capacitance varies so the way in which the signal is transmitted varies.

For power supply decoupling, where the voltage is steady, it means that you may not be getting the capacitance you expect: a 10uF ceramic with 12V across it could appear as a 4uF capacitor, for example.

The exception to this is C0G ceramics, whose properties are really excellent in almost every respect. They are very common in small values, and essentially unavailable in larger values.

The big caps at the power input of a module should probably be electrolytic, not because of the polarisation but just because they are among the best caps for the purpose at the value you need. The smaller decoupling caps near ICs can be regular ceramic, e.g. 100nF, because even if they behave as 10nF caps when voltage is applied, they will still do the job. For caps in the audio path, C0G or film are good.

Well, that's my understanding anyway.

[mlatu]

Hm, allright, I was refering to some chart comparing various capacitor types' impedences at increasing frequencies.

http://www.iequalscdvdt.com/esr.html


I have to admit I didn't read the text, yet. I probably should.



You said:

For power supply decoupling, where the voltage is steady, it means that you may not be getting the capacitance you expect: a 10uF ceramic with 12V across it could appear as a 4uF capacitor, for example.

Is that example cap rated for around double the voltage?

[cygmu]

For power supply decoupling, where the voltage is steady, it means that you may not be getting the capacitance you expect: a 10uF ceramic with 12V across it could appear as a 4uF capacitor, for example.

Is that example cap rated for around double the voltage?

Yes, the surprising finding in this report
https://www.maximintegrated.com/en/app- ... vp/id/5527
was that the derating of capacitance with bias voltage does not depend as much as you might expect on the voltage rating of the cap. The package size seemed to matter more -- smaller package caps behaved worse. This might mean that through hole parts are among the better ones, because they will tend to be large, though that might not be universal.

[mlatu]

This is ridiculous.

Like selling a tank that decreases in size if you increase the diameter of the hose you want to fill it with... I know it's not a good analogy, and I guess there are usecases where it doesn't matter. Or perhaps where this is even beneficial.

I guess I'll try both, film and ceramic and try to see where I can get away with using the cheaper ceramic ones.

[cygmu]

You will almost certainly get away with using any kind of capacitor in the decoupling role anyway. I imagine there is no precise way to calculate what is needed in a module without knowing what the rest of the system is, so people use 10uF or 22uF as a rough “good enough” attempt to smooth out power transients.

This thread has made me a bit nostalgic for the days when my Mouser orders were below the free shipping threshold and I had to combine them or wait or find other sources. Doesn’t seem to happen now...