Ceramic vs Film Capacitors for Temperature Stable?

For temperature-stable capacitance, film capacitors (especially C0G/NP0 ceramic and polypropylene film) are the clear winners, while standard Class 2 ceramics like X7R and X5R drift significantly with temperature and DC bias. This article explains the differences, why temperature stability matters, and how to choose the right capacitor for your circuit.

Formula

The capacitance change due to temperature is modeled as:

C(T) = C₀ × (1 + α × ΔT)

Where:

• C(T) = capacitance at the new temperature (µF)
• C₀ = capacitance at 25°C reference (µF)
• α = temperature coefficient (ppm/°C, parts per million per degree Celsius)
• ΔT = temperature change from 25°C (°C)

For example, a 1 µF capacitor with α = +30 ppm/°C (typical C0G) heated by 50°C gives: C = 1 × (1 + 30e-6 × 50) = 1.0015 µF, a 0.15% increase. The same change in an X7R (α ≈ ±15% over range, not linear) would be roughly 15% – a hundred times larger.

Worked Example

Using the default values from the calculator: ΔT = 50°C, α = 30 ppm/°C, C₀ = 1 µF.

1. Convert α to a decimal: 30 ppm = 30 × 10⁻⁶ = 0.00003
2. Multiply by ΔT: 0.00003 × 50 = 0.0015
3. Add 1: 1 + 0.0015 = 1.0015
4. Multiply by C₀: 1.0015 × 1 = 1.0015 µF

The result is 1.0015 µF, a 0.15% change. The calculator rounds to 1.0015 µF and shows 0.15% drift.

Why Temperature Stability Matters

In timing circuits (RC oscillators, filters, ADC references), a capacitor that changes value with temperature shifts the frequency or timing. A 10% capacitance drift can turn a 1 kHz tone into 900 Hz, or cause a filter cutoff to wander outside its passband. In precision applications like audio crossovers, medical sensors, or aerospace telemetry, that drift is unacceptable.

Temperature stability is often conflated with accuracy. A capacitor can be accurate (±5% tolerance) but not stable (drifts ±15% over -55°C to +125°C). Stability is about change, not initial error.

Ceramic Capacitors: Class 1 vs Class 2

Ceramic capacitors split into two classes:

• Class 1 (C0G/NP0): Uses a non-ferroelectric dielectric (typically titanium dioxide). Temperature coefficient is ±30 ppm/°C, essentially flat. Capacitance drift is under 0.3% over the full military temperature range. Excellent stability, but low capacitance density – you rarely see C0G above 100 nF in a through-hole package.
• Class 2 (X7R, X5R, Y5V): Uses ferroelectric barium titanate. High permittivity allows microfarads in small packages, but the dielectric constant changes sharply with temperature. X7R is ±15% over -55°C to +125°C; X5R is ±15% over -55°C to +85°C; Y5V can be +22%/-82% over -30°C to +85°C.

Additionally, Class 2 ceramics suffer from DC bias effect – applying a DC voltage reduces capacitance, sometimes by 50% or more. This is separate from temperature drift but often compounds it. C0G is immune to DC bias.

Film Capacitors: The Temperature-Stable Alternative

Film capacitors use a plastic dielectric (polyester, polypropylene, polystyrene, PTFE). Polypropylene (PP) has a negative temperature coefficient around -100 to -200 ppm/°C, very linear and predictable. Polyester (PET) is higher, around +400 to +800 ppm/°C, but still far better than X7R. Polystyrene is near-zero but rarely used now due to temperature limits.

Film capacitors also have very low dielectric absorption, high insulation resistance, and self-healing properties (a small breakdown clears itself without shorting). They are the default choice for AC mains applications, audio signal paths, and precision timing where size is not a constraint.

Key trade-off: film capacitors are physically larger and more expensive than ceramics for the same capacitance and voltage. A 1 µF film capacitor might be the size of a thumbnail, while an X7R ceramic is a chip a few millimeters across.

How to Choose: Ceramic vs Film

Use this decision guide:

• Need < 100 nF and high stability? Choose C0G ceramic. It beats film in size and has comparable stability.
• Need > 1 µF and tight tolerance over temperature? Choose polypropylene film. It will hold value within a few percent over a wide range.
• Need high capacitance in small size and can tolerate ±15% drift? Choose X7R ceramic. Most decoupling and bulk storage applications fall here.
• Audio signal path? Film (polypropylene) is preferred for its low distortion and stability, though C0G is also excellent for small values.
• DC bias is a concern? Avoid Class 2 ceramics entirely. Use film or C0G.

Always check the datasheet’s temperature characteristic code (e.g., X7R, C0G) and the DC bias curve. The code tells you the temperature range and capacitance change, but not the voltage dependence – that requires a separate graph.

Common Problems and Misconceptions

Misconception: “All ceramics are unstable.” False – C0G is as stable as most films. The instability is specific to Class 2 dielectrics.

Misconception: “Film capacitors are always better.” Not true. Film capacitors have higher ESR at high frequencies and can be microphonic (vibration changes capacitance). For RF bypassing, C0G ceramic is often superior.

Problem: Capacitance drift with voltage. Class 2 ceramics lose capacitance as DC voltage increases. A 10 µF X5R at 6.3 V might measure only 4 µF at rated voltage. This is not a temperature issue but often surprises designers.

Problem: Aging. Class 2 ceramics lose capacitance over time (aging rate ~1% per decade hour). This is recoverable by heating above Curie point, but in practice it means the capacitor’s value slowly decreases. Film and C0G do not age.

Cost and Size Comparison

For a given capacitance and voltage, film capacitors are typically 2-5 times more expensive and 5-20 times larger than Class 2 ceramics. C0G ceramics are more expensive than X7R for the same value, but still cheaper than film for small values. If your design has strict board space, film may simply not fit. If your design has strict stability, the cost is justified.

For prototyping, it’s common to use X7R for initial testing and then switch to film or C0G for the final precision stages. This saves cost during development.

Frequently Asked Questions

Q: What is the temperature coefficient of X7R?
X7R is not specified with a linear ppm/°C. It is guaranteed to change no more than ±15% over -55°C to +125°C. The actual change is nonlinear, often peaking around 0°C and dropping at extremes.

Q: Can I use film capacitors for decoupling?
Yes, but they are usually too large and have higher inductance than multilayer ceramics. For high-frequency decoupling (100 MHz+), C0G ceramic is better. Film is fine for low-frequency decoupling (below 1 MHz).

Q: Why does my ceramic capacitor measure less than its labeled value?
Likely DC bias. Measure with a multimeter that applies a low test voltage (often 0.5 V). At higher DC voltages, capacitance drops. Also check aging – an old X7R may have lost 5-10% of its initial value.

Q: Is C0G the same as NP0?
Yes. C0G is the EIA code, NP0 is an older industry name. They are identical: temperature coefficient ±30 ppm/°C, no voltage dependence, no aging.

Q: What is the best capacitor for an RC oscillator?
For a stable frequency, use C0G ceramic or polypropylene film. The oscillator’s frequency depends directly on capacitance, so any drift changes the output. For high precision, consider a temperature-compensated crystal oscillator instead, which uses a quartz crystal, not a capacitor.

Q: Can I replace a film capacitor with a ceramic one?
Only if the circuit tolerates the temperature drift and DC bias. For audio coupling, film is preferred for low distortion. For snubbers or AC mains, film is required due to safety ratings – do not substitute ceramic unless it has the same X/Y safety certification.

Related

• Challenge an AI Hiring or Firing Decision at Work
• Awk Built-In Variables: Examples and How They Work
• AI Search Engines Need Trust—What If You Have None?

in t | Temperature change | °C | 50 | 0 | 200
in tc | Temp coefficient (ppm/°C) | ppm/°C | 30 | -1000 | 1000
in c0 | Capacitance at 25°C | µF | 1 | 0.000001 | 1000
out c | Capacitance at new temp | µF | c0 * (1 + tc * 1e-6 * t) | 4
out drift | Change in capacitance | % | (c - c0) / c0 * 100 | 2