Ceramic cap in schematic, actual/derated value or ideal?

[jnz]

My schematic calls for a 1uF cap.

It’s a 3.3V VDD line.

I could use a 0402 1uF 50V part and get damn near 1uF. Or a 0805 6.3V part. Or a 0603 3.3uF 10V part.  The actual capacitance after DC bias let’s pretend are all the same.

A. Does it matter in other ways?

B. Did my schematic mean “use some typical 1uF part” or “get 1uF how ever you can get it”?

[bob91343]

Very few capacitors would be stressed excessively on a 3 Volt line so use whatever is cheap and has good high frequency characteristics.  The voltage rating is pretty much immaterial.

The exception to that would be an electrolytic capacitor, which is not likely under consideration here.

[jnz]

Bob,

This is about DC bias derating. Not necessarily about stress.

Although the exact same question applies to 12V.

I want the smallest part I can fit in this application. A 1210 is going to run pretty darn close to 10uF for example, but the exact same specs in 0805 I will only get an effective 4uF for example.

TDK has an excellent tool in their site for figuring this out. https://product.tdk.com/en/search/capacitor/ceramic/mlcc/list#ref=characteristic&1shape%5B%5D=2+Terminal&1shape%5B%5D=Low+ESL+Reverse+Geometry&lw%5B%5D=3.2mm+x+1.6mm+%5BEIA+1206%5D&lw%5B%5D=3.2mm+x+2.5mm+%5BEIA+1210%5D&1rv%5Bcomp%5D=%3D&1rv%5Bdc%5D%5B%5D=50&1rv%5Bdc%5D%5B%5D=100&1a_dcbias%5Bt%5D=14&1a_dcbias%5Bl%5D=1&1a_dcbiasc_f%5Bt%5D=4&1a_dcbiasc_t%5Bt%5D=8&1a_dcbiasc_t%5Bl%5D=1.00E-06&_l=20&_p=2&_c=2el_dcbias_meas-2el_dcbias_meas&_d=1&_106=1 

So the question is, does the schematic specify an approximate part you want, or does it specify the value you need? The latter makes more sense.

[bdunham7]

B. Did my schematic mean “use some typical 1uF part” or “get 1uF how ever you can get it”?

If you have a schematic without a BOM and the schematic doesn't specify a part number or something more specific that just "1 uF", then you have to decide for yourself.  Typically a schematic (or BOM) will simply indicate the nominal value of the part, not the derated value at the voltage present in the circuit.  If you are putting an MLCC in place of something else, then you do need to think about the voltage derate.  If this is just a decoupling or filter cap, the value isn't really all that critical, so your best guess is likely fine.  However, if the circuit was not designed for an MLCC, using one might actually cause problems, so don't assume MLCC is the only way to go for this.  If you are looking for 'more', you can get tantalums at 10uF @ 10V in size 0603.

[Whales]

Yes.  If it's right next to a regulator: check the regulator's datasheet to see if there is a minimum capacitance value for stability. 

[floobydust]

A real design take in account component tolerances and temperature. An X5R can be -15% and worse with aging and temperature, and for this reason I never put in the minimum value part. We don't know what "my schematic calls for" means and what about 2.2uF?

[Siwastaja]

You need to understand what went into designing that circuit, there is no way around it. Schematic itself is just a small part of design. Component selection is another equally important part, and so is PCB layout.

True schematics of real designs have part numbers for everything (sometimes generic values for parts that can be widely replaced). Not usually visually in the schematic itself, but in a separate BOM listing.

If the schematic just says 1uF and nothing else, it's not a complete design. You can't assume anything, you need to understand the circuit to finalize the design (i.e., do component selection and layout).

Of course we can guess that you are probably OK with any type of 1uF MLCC, maybe excluding the worst offenders like Y5V dielectric in tiny packages. But this is just guessing.

Maybe show us the schematic?

For example, bypass caps next to digital ICs are chosen by ESL and ESR, not capacitance, so it doesn't matter if it's Y5V 0.22uF or X7R 2.2uF (both have near-zero ESR as MLCCs, and ESL is defined directly by package size), as long as it's decently small package (say, 0805 or smaller) and placed close to the pins. On the other hand, output capacitors for linear regulators are finicky - depending on the regulator IC - so you need to read the regulator datasheet carefully to choose right amount of capacitance and ESR.

[tszaboo]

You don't need to de-rate ceramic caps.
Also the voltage rating of the cap has nothing to do with the loss of capacitance at a DC bias. A 50V rated 0603 will not be any better than a 6.3V rated 0603 from the same MF/series.
Size does, bigger will have less DC loss.

[mon2]

https://www.maximintegrated.com/en/design/technical-documents/tutorials/5/5527.html

[Siwastaja]

You don't need to de-rate ceramic caps.
Also the voltage rating of the cap has nothing to do with the loss of capacitance at a DC bias. A 50V rated 0603 will not be any better than a 6.3V rated 0603 from the same MF/series.
Size does, bigger will have less DC loss.

This, always look at the actual curves (if available), but failing that and having to assume, it's best to base the assumption on the claimed volumetric energy density, i.e., how big the capacitance seems compared to the actual package size. If it seems "too good to be true", then it will suffer more under DC bias.

A 4.7uF in 0603 is highly unlikely to deliver over 30% under 5V DC bias, but 4.7uF in 1210 very likely does at least 50-60% under the same 5V DC bias. This is regardless their voltage ratings, temperature ratings, tolerances, or dielectric classes. Y5V is known as a "bad guy", but X7R can be colossally bad, too: the red flag is the small package and large capacitance.

It is even possible that a manufacturer sells the exact same part with different voltage ratings.

But always try to find the actual data for the exact product in question.

[strawberry]

could test capacitor performance with AWG and various DC offset levels
too small capacitance may upset powersupply feedback loop