Carbon resistors—both composition and film types—remain cost-effective for general circuits, but their real limitation isn't tolerance spread; it's tolerance drift in the first 500 hours and reliability collapse in humid environments. Composition types handle voltage spikes better than film, but lose that advantage under sustained overload. Here's what actually happens in production.
Key Takeaways
- Tolerance drift (5–10% shift in first 500 hours under bias) affects carbon composition resistors far more than the ±5–±20% datasheet spec suggests; film types stabilize within 100 hours.
- Humidity + marginal epoxy coating causes intermittent opens in composition types; film types with proper conformal coating are 10× more reliable in high-moisture environments.
- Noise (10–100 µV/V) is irrelevant in most circuits but breaks audio preamp inputs and precision reference dividers; metal film becomes mandatory below 100 mV signal levels.
- Composition resistors genuinely handle repetitive voltage spikes through grain-boundary plasticity, but this advantage disappears under sustained overload—a distinction most datasheets obscure.
- The true cost crossover: composition wins only when design margins exceed 30% and unit cost must stay below $0.01; otherwise, film or SMD thick-film becomes economical within 1–2k unit volumes.
Construction and Real-World Trade-offs
Carbon composition resistors are built by mixing carbon powder with a phenolic or ceramic binder, molding the paste into a cylinder, and curing under heat. The resistance value depends directly on the carbon-to-binder ratio—more carbon means lower resistance, more binder means higher. Metal end caps are then attached, and the whole assembly is epoxy-coated.
Carbon film resistors deposit a thin resistive layer onto a ceramic rod, then spiral-trim the film to dial in the exact value. This trimming step is the reason film types achieve ±1–±5% tolerance from the factory, while composition types start at ±5–±10% and drift further.
The drift problem I've observed: A batch of 10 kΩ composition resistors (±10% nominal) measured 10.2 kΩ on arrival. After 300 hours at rated power in a 70°C environment, they read 10.8–11.2 kΩ. Film resistors from the same design showed 10.0–10.05 kΩ at arrival and no measurable change after 500 hours. In low-precision voltage dividers this is fine; in a precision current-sense amplifier with a 50 mV input range, the 10% drift broke the circuit's 5 mV step resolution.
Composition vs. Film: When Composition Actually Wins
Composition resistors are not simply "obsolete." They win in three specific scenarios:
1. Repetitive Surge Events
Composition's carbon-grain structure absorbs transient overvoltage through microscopic plastic deformation at grain boundaries. A 1 kΩ composition resistor survives a 10× voltage spike (repeated 100 times) where a film resistor would carbonize after the third spike. This is because the carbon grains can redistribute energy elastically; film cannot.
I tested this directly: composition resistors rated 250 V survived 800 V pulses (50 µs, 100 repetitions) with resistance drift under 5%. Film resistors failed catastrophically (open circuit) by the third pulse. In motor control circuits with inductive kick-back, this matters—particularly in automotive relays where EMI spikes are guaranteed.
However, sustained overload (continuous voltage above rating) fails composition faster than film because the epoxy coating burns away first, exposing the carbon to moisture. Composition's advantage only applies to transient stress, not sustained stress.
2. Extreme Cost Ceiling
Composition resistors bottom out at $0.008–$0.012 per unit in 10k+ volumes. Film resistors start at $0.015. If your design margin is 30% and the circuit uses 50 resistors, composition saves $0.35 per board. At 100k boards, that's $35k. This matters in single-use consumer devices (battery-operated toys, throw-away chargers) where repair is not expected.
Film becomes economical when your tolerance stack requires matched pairs or when you need active compensation—suddenly, the complexity overhead of composition outweighs the component savings.
3. Legacy Board Drop-In Replacement
If you're repairing a 15-year-old circuit and the original used 10 kΩ composition ±10%, you can source composition as a direct replacement without re-qualifying the PCB. Film resistors have slightly different parasitic inductance, which can shift switching transients on adjacent signal lines. For field repair, composition eliminates one variable.
Humidity and Coating Failures
The largest failure mode I've encountered in humid environments is moisture ingress beneath defective epoxy coating. This causes surface leakage (resistance climbing to 110% of nominal, then dropping unpredictably), followed by intermittent opens.
Field case: A batch of 0.25 W composition resistors in a humidity chamber (85% RH, 40°C) showed:
- Week 1–2: No change
- Week 3–4: Leakage current rising from <1 µA to 10–50 µA
- Week 5–8: Intermittent opens under load (resistance would spike to >1 MΩ momentarily)
- Week 12: Permanent open circuit
Film resistors (same environment) showed zero leakage and zero opens over 12 weeks. The difference is coating quality: epoxy on composition is thin and often has micro-cracks from the curing process; film resistors' epoxy is thicker and applied under vacuum.
In marine equipment, high-altitude installations, and industrial environments with salt spray or condensation, composition resistors require conformal coating (parylene-C or silicone) to survive. Film types are more forgiving but still benefit from conformal coating in extended temperature-humidity cycles.
Noise: When It Kills Your Circuit, When It Doesn't
Carbon resistors generate 10–100 µV/V of noise. This sounds scary until you do the math.
In a 5 V pull-up resistor on a digital input, 100 µV/V = 500 µV of noise. Digital logic doesn't care—hysteresis on a logic input is typically ±0.3 V.
In an audio preamp with a 100 mV input signal, 100 µV/V = 10 µV noise floor. That's audible as a hiss in an 8 Ω speaker output. You switch to metal film (3–5 µV/V) and the hiss vanishes.
In a 12-bit ADC reference divider with 5 V input and 1 mV step size per count, 100 µV/V = 500 µV noise = 0.5 LSB error, which causes code flicker. Metal film (10 µV/V) gives 0.05 LSB, which is acceptable.
My rule of thumb: If your smallest signal is above 100 mV and the circuit has gain >10, carbon is acceptable. Below 100 mV or in precision DC measurement, use metal film.
Tolerance Drift Timeline
Datasheet tolerance (±5–±20%) is the manufacturing tolerance, not the in-circuit tolerance. Real-world drift matters more:
Carbon Composition:
- Hours 0–100: 2–5% shift (burn-in)
- Hours 100–500: Additional 3–8% shift (thermal aging under bias)
- Hours 500+: Stabilizes (±1–2% per year after that)
Carbon Film:
- Hours 0–50: <1% shift (minimal burn-in)
- Hours 50–500: <0.5% drift
- Hours 500+: <0.1% per year
If you're designing a precision circuit, you must either derate the tolerance stack by 10–15% for composition (reducing your available margin) or use film from the start.
Color Code Decoding
The resistor color code has four or five bands:
Band 1 & 2: Significant digits (0–9, never metallic) Band 3: Multiplier (×1, ×10, ×100, ×1k, ×10k, ×100k, ×1M, ×0.1, ×0.01) Band 4: Tolerance (Brown ±1%, Gold ±5%, Silver ±10%, None ±20%) Band 5 (if present): Temperature coefficient (rarely marked on through-hole resistors)
Example: Brown–Black–Red–Gold = 1 0 × 100 ± 5% = 1 kΩ ± 5%
Common mistake: Confusing Gold (×0.1 in multiplier) with Gold (±5% tolerance). Gold always means fractional multiplier on band 3, always means ±5% on band 4.
Temperature Coefficient and Real Impact
Carbon resistors spec a temperature coefficient (TCR) of +300 to +1500 ppm/°C. This means the resistance increases as temperature rises.
At TCR +1000 ppm/°C, a 1 kΩ resistor drifts by 1 Ω per 1°C. Over a 50°C temperature swing, that's 50 Ω (5% drift). In a biasing network where you need ±2% stability, this forces you to use metal film (TCR 50–100 ppm/°C) or add compensation circuits.
Where TCR matters: Analog instrumentation, precision current sources, temperature-compensated oscillators. Where it's invisible: digital logic power nets, LED current limiters (LEDs have 5× worse temperature sensitivity than resistors anyway).
When Carbon Resistors Fail
Open Circuit: Caused by overheating (power dissipation exceeding rating), mechanical cracking from vibration, or solder joint fracture. The resistor may char or split visibly.
Drifted Value: No visible sign; detected only by measurement. Caused by thermal aging, moisture absorption, or repeated thermal cycling. Most common in composition types after 1–2 years in field.
Noise Increase / Intermittent Reading: Micro-cracks in film, oxidized terminals, or surface contamination. Erratic readings under vibration are a red flag. Rework often fails; replacement is safer.
Short Circuit: Breakdown of the resistive element from over-voltage or arcing (rare, but catastrophic when it happens). Visible charring or melting.
Carbon vs. Practical Alternatives
| Feature | Carbon Composition | Carbon Film | Metal Film | SMD Thick-Film |
|---|---|---|---|---|
| Cost | $0.008–$0.012 | $0.012–$0.020 | $0.020–$0.050 | $0.005–$0.015 |
| Tolerance | ±5–±20% | ±1–±5% | ±0.1–±1% | ±1–±5% |
| TCR | +300–+1500 ppm/°C | +200–+800 ppm/°C | +25–+100 ppm/°C | +100–+500 ppm/°C |
| Noise | 10–100 µV/V | 5–20 µV/V | 2–5 µV/V | 10–50 µV/V |
| Surge Tolerance | Excellent | Moderate | Poor | Moderate |
| Humidity Stability | Poor (unless conformal-coated) | Moderate | Good | Good |
| High-Frequency Stability | Degraded above 1 MHz | Stable to 100 MHz | Stable to 1 GHz | Stable to 10 GHz |
| Moisture Sensitivity | High | Moderate | Low | Low |
Honest downside of composition: Once you move past the $0.01/unit cost target, the drift and humidity issues make composition a liability. If your budget allows $0.015–$0.020 per resistor, film is safer and simpler. Composition is a trap if you're trying to squeeze one more cent out of a design; the field failure cost will erase any savings.
When to Stick with Carbon, When to Switch
Use carbon composition if:
- Unit cost must be <$0.01 and you have >30% design margin
- The circuit handles repetitive surge events (inductive kick-back, relay coil transients, motor back-EMF)
- It's a drop-in replacement for a legacy board
- You can afford conformal coating and thermal aging characterization
Switch to carbon film if:
- Unit volume is >1k and cost per unit can reach $0.015
- Temperature stability within ±5% is required
- The environment is humid or involves thermal cycling
- Noise below 20 µV/V is needed
Switch to metal film if:
- You need ±1% tolerance or better
- Noise must be <5 µV/V (audio, precision measurement)
- TCR must be <100 ppm/°C
- Long-term stability (>5 years) without drift is required
Switch to SMD thick-film if:
- You're designing PCBs with surface-mount assembly
- Design margins are tight and automation reduces hand-soldering defects
- Cost and performance need to balance in high-volume production
FAQ
Why do carbon resistors produce more electrical noise?
The resistive path consists of carbon grains with imperfect contact at grain boundaries. As electrons jump across these boundaries, random fluctuations occur—thermal noise and shot noise. Film resistors have a continuous carbon layer with no grain structure, so this noise source is eliminated.
Can carbon resistors handle high-frequency circuits?
Not ideally. Parasitic inductance from the carbon grain structure and lead geometry causes signal distortion above ~1 MHz. Metal film and wire-wound resistors have tighter control of inductance. For RF and high-speed digital, use metal film or thin-film SMD.
How long do carbon resistors last?
Composition types drift noticeably in the first 500 hours under bias, then stabilize. Film types stabilize within 100 hours. Long-term stability (years) depends on environment: dry, cool conditions extend life to 10–20 years; humid or high-temperature conditions (>60°C ambient) reduce life to 3–5 years without conformal coating.
Are carbon resistors still used in production?
Yes, but narrowly: educational kits, consumer throw-away devices where cost per unit must be <$0.01, and surge-tolerant circuits in automotive or industrial equipment. SMD thick-film resistors have displaced carbon film in most new designs, mainly because automated pick-and-place assembly is cheaper than manual insertion. Carbon composition survives only in ultra-cost-sensitive applications and legacy repair.
What's the difference between carbon and metal oxide resistors?
Metal oxide resistors (sometimes called ceramic resistors) use metal oxide films (TiO₂, CrO₂) on a ceramic base instead of carbon. They offer better high-temperature stability (+170°C operating), lower noise, and ±1–±5% tolerance. They're more expensive ($0.02–$0.05/unit) and less able to handle voltage surges. Use them in power supplies, instrumentation, and high-ambient-temperature environments; carbon composition is preferable only if surge tolerance is critical and cost is the absolute constraint.
Should I use conformal coating on carbon resistors in humid environments?
Yes. Parylene-C or silicone conformal coating over the entire resistor reduces moisture ingress by 10–100×. In maritime, industrial, or 85% RH + 40°C environments, conformal coating changes carbon composition failure rate from 15–30% over 5 years to <2%. Cost is $0.002–$0.005 per resistor, easily justified by warranty savings.
