NEWS

MLCC Dielectrics Explained: C0G, X5R, X7R and Why Your Capacitor Loses Capacitance

Short answer: the three-character code on a ceramic capacitor tells you its temperature range and how much capacitance it is allowed to lose across that range. It tells you nothing about what happens when you apply voltage — and for Class II parts, applied DC voltage is usually the larger effect.

How to read the EIA code

For Class II dielectrics such as X7R and X5R, each of the three characters means something specific.

Character What it sets Values
1st (letter) Minimum temperature X = −55 °C, Y = −30 °C, Z = +10 °C
2nd (digit) Maximum temperature 4 = +65, 5 = +85, 6 = +105, 7 = +125, 8 = +150 °C
3rd (letter) Max capacitance change R = ±15%, S = ±22%, U = +22/−56%, V = +22/−82%

So X7R reads as: −55 °C to +125 °C, capacitance staying within ±15% across that span. X5R is the same tolerance but only to +85 °C. Y5V runs −30 °C to +85 °C and is permitted to lose 82% of its nominal value at the extremes.

The common dielectrics side by side

Dielectric Class Temp range Cap change Typical use
C0G / NP0 I −55 to +125 °C ±30 ppm/°C Timing, filters, RF — anywhere the value must hold
X5R II −55 to +85 °C ±15% Decoupling in consumer and handheld
X7R II −55 to +125 °C ±15% General industrial and automotive decoupling
X6S II −55 to +105 °C ±22% High density where some drift is acceptable
X7S II −55 to +125 °C ±22% High capacitance in a small case
Y5V II −30 to +85 °C +22 / −82% Legacy and cost-driven bulk decoupling only

Class I and Class II are different components

This is the distinction that matters most, and the part number does not spell it out.

Class I (C0G/NP0) uses a paraelectric ceramic. Its capacitance is stable, predictable and essentially linear. It does not lose value under DC bias, it does not age, and it is not piezoelectric. The trade-off is capacitance density: a C0G part of a given value is far larger and more expensive than a Class II equivalent.

Class II (X5R, X7R, X6S, X7S, Y5V) uses a ferroelectric barium-titanate ceramic. That gets you far more capacitance in the same case size, at the cost of four behaviours designers routinely forget.

The four Class II behaviours that cause field problems

1. DC bias — usually the biggest effect

Apply DC voltage across a Class II capacitor and its effective capacitance falls. The loss grows as applied voltage approaches the rated voltage, and it is worse in smaller case sizes and in parts with high capacitance for their size. For small-case, high-density X5R and X7R parts operated near their rating, the loss can exceed half the nameplate value.

Nothing on the label warns you about this. The only way to know is the manufacturer DC bias curve for that exact part number, case size and voltage rating. Two parts with identical markings in different case sizes can behave very differently.

The practical defence is to derate. Choosing a voltage rating two or three times the applied DC voltage, or stepping up one case size, recovers most of the loss.

2. Ageing

Class II ceramics lose capacitance logarithmically over time after firing, as the crystal structure relaxes. Manufacturers specify this as a percentage loss per decade of hours. Heating the part above its Curie point resets the clock — which is why a reflow cycle restores capacitance, and why measurements taken immediately after assembly differ from measurements taken weeks later.

3. Temperature

The dielectric code caps the change but does not make it linear. An X7R sits near its nominal value around room temperature and drifts toward the limits at the extremes. If a design depends on the exact value across temperature, Class II is the wrong choice regardless of how tight the tolerance code looks.

4. Microphonics and piezoelectric noise

Ferroelectric ceramics are piezoelectric. Mechanical vibration generates a voltage, and an applied AC voltage makes the part physically deform — which is the source of audible singing-capacitor noise in switch-mode supplies. Class I parts do not do this.

Choosing quickly

  • The value must be accurate and stable (timing, oscillators, filters, sample-and-hold, RF matching) → C0G / NP0, no exceptions.
  • Bulk decoupling, value not critical, ambient below 85 °CX5R.
  • Decoupling in industrial or automotive, up to 125 °CX7R.
  • Maximum capacitance in a small case, ±22% acceptableX6S or X7S.
  • Y5V → avoid in new designs. An 82% permitted loss is rarely a real engineering choice.

Frequently asked questions

Can I substitute X5R for X7R?

Only if the maximum operating temperature never exceeds 85 °C. The tolerance band is the same (±15%); the temperature ceiling is not. X7R to X5R is a downgrade; X5R to X7R is generally safe on temperature but may differ in DC bias behaviour and price.

Why does my 10 µF capacitor measure 4 µF in circuit?

Almost always DC bias, sometimes compounded by ageing. Check the DC bias curve at your actual operating voltage. A meter reading taken with no bias applied shows close to nominal, which is why bench measurement and in-circuit behaviour disagree.

Is C0G worth the extra cost?

Where accuracy matters, yes — it is not really a cost decision but a functional one. Where you simply need bulk capacitance on a rail, Class II gives far more capacitance per unit cost and per unit volume.

Does a tighter tolerance code fix DC bias loss?

No. Tolerance describes manufacturing spread at zero bias and reference conditions. A ±5% X7R loses capacitance under DC bias exactly like a ±20% one.


Venatech Limited is an authorised distributor of Walsin passive components. If you are selecting a dielectric or need a cross-reference for an existing part, talk to us — we will check the DC bias data with you before you commit the design.

Share This, Choose Your Platform!