Good Gravy, These Ceramic Discs Are Garbage

[Ginsly]

Although I haven't used them in a build yet I have an assortment of ceramic discs from a few different sources, and I was shocked to see how low they were when tested. For instance, several 100nf caps tested at around half of the stated value, and it didn't end there. Used a couple different testers.

It kinda made me re-evaluate my capacitors and why I use certain types in certain situations. For the most part, I've stuck to this:

pF values - C0G/NP0 MLCC (TDK blue guys)
nF values (inc 1uf at times) - polyester box film (mostly Kemet white 100v - R28 series)
uF values - electrolytic (mostly Nichicon/Rubycon various voltage)

All of the above seem to have good tolerances.

*I also have non-C0G/NP0 yellow MLCCs that I use for primarily for breadboarding
*Additionally, I have mylar greenies and some reds (metallized polyester or polypropylene?) that I've used in builds, but can get large.

Some thoughts that popped up:

- Although many use them without issue, I clearly have some really bad ceramic discs - maybe they're generally known for bad tolerances, though..? I'm tossing them, but is there any reason to replace them with better discs rather than simply focusing on MLCC, box, and mylar for non-polarized?

- To that end I rarely use the yellow MLCCs (likely X7R?) caps in builds even though they're nice and small - if box film fits, it seems to be the better choice with better tolerance. Is that a good baseline approach when in the audio path? It sounds like I could easily use X7R MLCCs without issue for any capacitor that goes to ground or is used for DC filtering - seems like this may be a system some builders use. That being said, I'm largely building dirt pedals (for now), and I could likely use any type of MLCCs without issue (as long as they're within an acceptable tolerance).

- I have some tantalum caps, but honestly don't know where or why I would use them. I've seen some people here use them in place of ALL electros, and some people never use them. There are some builds that use both - I have a Klon board that features a bunch of electros and one single tantalum - wondering why that might be.

- I doubt silver mica is necessary for any of the builds I'm doing, nor can I drop that kinda dough on caps!

This is all a longwinded way of saying that what I'm doing at the moment works, but I'm still a little unclear about a few things. I've pored over several posts, but would love some additional opinions!

 

[PunchySunshine]

Yes, yes they are garbage in my experience.

 

[Brett]

I don't recall ever using a ceramic disc capacitor in an audio application. The cost of MLCC and poly film are plenty affordable enough to justify using a better quality material.

I have some tantalum caps, but honestly don't know where or why I would use them.

Tantalum can be useful in situations where leakage (common with general aluminum electrolytic caps) is a concern. Some folks are against using them based on principal (because sourcing the raw materials needed to make them isn't always ethical).

I doubt silver mica is necessary for any of the builds I'm doing

Necessary? Probably not.
Expensive? Absolutely.
Useful? Possibly.

Silver Mica is widely considered superior to ceramic and remains a popular choice for sub nF caps in hifi applications. I personally like to use Silver Mica caps in the negative feedback path for op amp clipping stages. To each their own. Most values can be had through AES at a reasonable price if you decide to try them.

 

[Ginsly]

I don't recall ever using a ceramic disc capacitor in an audio application. The cost of MLCC and poly film are plenty affordable enough to justify using a better quality material.

Yeah, I can't find a good reason to use the discs. Space aside, is there ever a reason/situation to use an (nF) MLCC over poly film though? I've read that MLCCs are better for DC filtering than film caps, although it might have just been a "use the cheaper option since it doesn't matter" kinda thing.

Tantalum can be useful in situations where leakage (common with general aluminum electrolytic caps) is a concern. Some folks are against using them based on principal (because sourcing the raw materials needed to make them isn't always ethical).

Gotcha. I'm not sure I'll know when electrolytic leakage is a concern, but that's because I'm still very much learning. My guess is that it doesn't matter as much in the trashy fuzz I've been making, but may become an issue in more sensitive circuits. Thanks for clueing me in to that!

Silver Mica is widely considered superior to ceramic and remains a popular choice for sub nF caps in hifi applications. I personally like to use Silver Mica caps in the negative feedback path for op amp clipping stages. To each their own. Most values can be had through AES at a reasonable price if you decide to try them.

Ah, ok - so it's mainly useful for pF values. I had gathered that C0G/NP0 MLCCs were a solid, reliable choice for sub-nF caps, but I'm sure there's a reason people use silver mica. I have some, but only a couple values - not particularly useful! Do you keep a certain roster of pF values on hand? Are there a handful of SM values that are generally good to have on hand, or is it very much a situational thing?

 

[HamishR]

I only use silver mica in amplifiers where they fit more easily than in pedals. For amps where the voltages are generally rather higher than in a pedal I have become a bit careful about which SM I use. Since watching a fair few Psionic Audio vids I have started avoiding the SM caps which are black and have "SM" marked on them. Unfortunately they are the cheaper ones (of course)! Mr Psionic says that in his experience they often fail. So I am using the CDE caps I can get from Antique Electronics. They're +/- 5% and start at $2.35.

But also because of Mr Psionic I am using Vishay MKT1813 630V caps for coupling caps etc in amps instead of Sozo now and that saves me a fortune!

 

[lowpitch]

Yeah, I can't find a good reason to use the discs. Space aside, is there ever a reason/situation to use an (nF) MLCC over poly film though? I've read that MLCCs are better for DC filtering than film caps, although it might have just been a "use the cheaper option since it doesn't matter" kinda thing.

A lot of op amp datasheets seem to specifically call for MLCCs, usually 100nF, for decoupling the power pins. As to why, I don't know, but if the manufacturer recommends it I'll stick to it. Sorry for the non-answer-answer

 

[mybud]

You might find this useful: http://www.modular.fonik.de/files/capacitors_101.pdf

 

[Ginsly]

I only use silver mica in amplifiers where they fit more easily than in pedals.

Yep they do seem a little large... I'm nowhere near dabbling with amps, but I can see that they involve a whole different set of considerations.

A lot of op amp datasheets seem to specifically call for MLCCs, usually 100nF, for decoupling the power pins. As to why, I don't know, but if the manufacturer recommends it I'll stick to it. Sorry for the non-answer-answer

Good to know, thanks - do you know of an example of this off the top of your head? I'll certainly keep this in mind.

You might find this useful: http://www.modular.fonik.de/files/capacitors_101.pdf

That is VERY straightforward and informative, much appreciated! I actually have some random caps, and this might help me identify what in the heck they are exactly:

The pdf is helping me understand why these red film caps are SO much smaller than the greenies, even at higher values. Based on the article, it seems that both are polyester mylar caps, but greenies are "film and foil" whereas the red ones are "metallized film" - is that right?

 

[lowpitch]

Good to know, thanks - do you know of an example of this off the top of your head? I'll certainly keep this in mind.

You'll see this in almost any TI chip datasheet. I like and trust those because they're some of the most comprehensive datasheets of all the manufacturers. Some examples:
OPA2134 https://www.ti.com/lit/ds/symlink/opa2134.pdf "Bypass the power supply pins with 10nF ceramic capacitors or larger to minimize power supply noise."
RC4558 https://www.ti.com/lit/ds/symlink/rc4558.pdf "Connect low-ESR, 0.1μF ceramic bypass capacitors between each supply pin and ground, placed as close to the device as possible. A single bypass capacitor from V+ to ground is applicable for single-supply applications."
TL07xx https://www.ti.com/lit/ds/symlink/tl072.pdf "Connect low-ESR, 0.1-μF ceramic bypass capacitors between each supply pin and ground, placed as close to the device as possible. A single bypass capacitor from VCC+ to ground is applicable for single-supply applications."
OPA167x https://www.ti.com/lit/ds/symlink/opa1678.pdf "Connect low-ESR, 0.1-μF ceramic bypass capacitors between each supply pin and ground, placed as close to the device as possible. A single bypass capacitor from V+ to ground is applicable for single-supply applications."

 

[Ginsly]

You'll see this in almost any TI chip datasheet.

Huh! I’ll have to peruse some PedalPCB layouts and see if MLCCs are indicated in those positions…

 

[Brett]

You'll see this in almost any TI chip datasheet.

There may very well be a performance advantage to use ceramic over poly film for decoupling, but I don't specifically recall reading about it.

I'd assume that TI is suggesting the use of ceramic capacitors in power decoupling positions for the following reasons:

1. Ceramic is more cost-effective than polyester film and will work adequately for power decoupling.
2. If going through-hole, MLCCs are physically smaller than polyester box film caps allowing more optimal placement in certain cases.
3. New designs using TI's parts are likely surface-mount. I'd assume that TI understands that and (to some extent) writes their datasheet copy with that in mind. By way of example, look at many of the modern datasheet layout recommendations. Most of them are SMD. Ceramic capacitors are widely available, cost-effective, and highly miniaturized in SMD format. Polyester/silver mica/polystyrene? Not so much.

Electrolytic capacitors aren't the best candidate for the job for op amp-adjacent decoupling. You'd be hard-pressed to find them in such low capacitance ratings (100nF), and due to their size, they'd likely prevent optimal placement in close proximity to the op amp pins where decoupling is recommended. Op amp decoupling at the pins isn't "always" absolutely necessary, though it is usually recommended and is often considered good design practice.

 

[lowpitch]

There may very well be a performance advantage to use ceramic over film for decoupling, but I don't specifically recall reading about it.

I'd assume that TI is suggesting the use of ceramic capacitors in power decoupling positions for the following reasons:

1. Ceramic is more cost-effective than polyester film and will work adequately for power decoupling.
2. If going through-hole, MLCCs are physically smaller than polyester box film caps allowing more optimal placement in certain cases.
3. New designs using TI's parts are likely surface-mount. I'd assume that TI understands that and (to some extent) writes their datasheet copy with that in mind. By way of example, look at many of the modern datasheet layout recommendations. Most of them are SMD. Ceramic capacitors are widely available, cost-effective, and highly miniaturized in SMD format. Polyester/silver mica/polystyrene? Not so much.

Electrolytic capacitors aren't the best candidate for the job for op amp-adjacent decoupling. You'd be hard-pressed to find them in such low capacitance ratings (100nF), and due to their size, they'd likely prevent optimal placement in close proximity to the op amp pins where decoupling is recommended. Op amp decoupling at the pins isn't "always" absolutely necessary, though it is usually recommended and is often considered good design practice.

Yes, there are probably a lot of caveats. Not least of which: "Is it really necessary for building guitar pedals?" And probably no, you can throw in a 100nF film cap to decouple the power pins and it'll be fine because film caps also have quite low ESR compared to electrolytics. Maybe omit it altogether, still the same, because we're just not operating at the same frequency (literally) as modern high-tech equipment, where there may actually be a difference in performance. I think everybody kind of has to draw their own lines here. BUT this also was just one example of how ceramic caps aren't completely useless, even in guitar pedals.

 

[owlexifry]

Huh! I’ll have to peruse some PedalPCB layouts and see if MLCCs are indicated in those positions…

aionfx does quite a bit

 

[jwin615]

Yes, there are probably a lot of caveats. Not least of which: "Is it really necessary for building guitar pedals?" And probably no, you can throw in a 100nF film cap to decouple the power pins and it'll be fine because film caps also have quite low ESR compared to electrolytics. Maybe omit it altogether, still the same, because we're just not operating at the same frequency (literally) as modern high-tech equipment, where there may actually be a difference in performance. I think everybody kind of has to draw their own lines here. BUT this also was just one example of how ceramic caps aren't completely useless, even in guitar pedals.

It's more necessary on higher frequency as they can oscillate at much higher(I e. More) frequencies.
Old low frequency jellybeans just won't be as susceptible.
There's also the case of ringing the decoupling caps. If you use say a 100u electrolytic at power input and a 100n electrolytic at the opamp, the large cap will impose an inductance on the small cap. Lower ESR caps can create a higher Q LC circuit and create a resonance in the smaller cap,.which is next to the IC... Bad news.
Using a ceramic instead of avoids this ringing from potentially happening. Using non-low ESR electrolytics (regular power caps...) as the main filter caps--not using low ESR in both positions-- can also help.
I think this ringing can occur with bipolar supplies that are decoupled with 2 caps between + and G/G and -, but I may be remembering that incorrectly.
At least that's my understanding. I had a circuit a year or so ago(can't remember what atm) that I had used a solid aluminum cap as the main filter cap. It would oscillate and shutdown(the opamp), even with no input. When I changed that cap(after days of WTF) for a cheap chin-co lytic cap, all was well. Those Al caps had minute ESR, sub-ohm iirc.

 

[Ginsly]

Still curious as to what everybody's "cap rules" are (outside of "no fedoras")... This is kinda where I'm at:

- There are better options than ceramics discs, not worth saving a few cents to use the few that test ok.

- I think sticking with C0G/NP0 MLCCs for all pF values sounds wise. Not messing with discs, and it still seems like those yellow MLCCs of unknown dielectric might present some issues (microphonic behavior, etc?), depending on where in the circuit they're used. Silver mica look like they have the best specs, but they're a bit large and a little pricey.

- Outside of ICs (since there's already an ongoing discussion about that), are there certain nF positions where any of you ONLY use MLCC? I can see using these mystery yellow MLCCs (X7R?) in positions that connect to Ground, and as Miller Caps, for instance (though that would be pF) - I'm not sure how much the dielectric would really affect those situations.

- Tantalum sounds like something I'd use only once in a while when a build doc specifically calls for it. Electros may have leakage and a life span, but tantalum sounds a little too finicky to be worthwhile.