notesonly.in

One notebook for every subject — open it anywhere.

Log in

Photodetectors: PIN and APD: Complete Beginner's Guide

EngineeringTry the interactive 3D model →

Photodetectors are the receivers of optical communication, converting light back into current at the fiber's end. The two dominant types are the PIN photodiode and the APD (avalanche photodiode). The PIN is simple, cheap and linear; the APD adds internal gain for far weaker signals — at the cost of high voltage, extra noise and temperature sensitivity.

The choice sets the sensitivity floor of every optical link. PINs suit short, strong-signal links; long-haul receivers need APD gain to lift tiny photon streams above the noise. Responsivity, quantum efficiency and avalanche multiplication let you calculate receiver performance instead of guessing.

Key Concepts

  • PIN structure: a wide intrinsic layer between P and N. Reverse bias depletes it; photons create electron-hole pairs swept out as photocurrent — one pair per photon, no gain.
  • Responsivity: R = ηq/hν A/W (η = quantum efficiency). Typical: 0.5 at 850 nm, 0.9 at 1310 nm, ~1.0 at 1550 nm.
  • Quantum efficiency: fraction of photons creating collected carriers; good PINs reach 0.7–0.9.
  • APD gain (M): high reverse bias (tens–hundreds of volts) triggers impact ionization — avalanche multiplication, M ≈ 10–100.
  • Excess noise F(M): avalanches are random, so gain multiplies noise too — gain helps only until multiplied shot noise dominates thermal noise.
  • Dark current: leakage with no light, roughly doubling every 8–10 °C, setting the noise floor.

Formulas

R = ηq/(hν) = η × λ(µm)/1.24 A/W. At 1550 nm with η = 0.8: R = 0.8 × 1.55/1.24 ≈ 1.0 A/W. Photocurrent Ip = R × Popt; for an APD, Ip = M × R × Popt. SNR improves only while multiplied shot noise stays below amplifier thermal noise — optimum M is ~10–30 for silicon, lower for InGaAs. Bandwidth trades against responsivity: thinner depletion layers are faster but absorb less — the classic responsivity-bandwidth tradeoff.

Worked Example

A 1550 nm receiver gets Popt = 10 µW (−20 dBm); detector η = 0.8, so R = 1.0 A/W. Step 1: PIN gives Ip = 10 µA. Step 2: APD with M = 50 gives 500 µA — 50× the signal. Step 3: noise check. Shot noise power scales ~M²×F(M); with F ≈ 5, noise rises ~250× while signal power rises 2500×, so SNR still improves ~10× — a real win. Step 4: cost. The APD needs ~40–60 V bias with temperature compensation versus 5 V for the PIN. Verdict: the APD earns its complexity only where thermal noise dominates — the long-haul case.

Common Misconceptions

  • "APD is always better." Gain multiplies noise too. For strong signals a PIN matches the SNR at 5 V bias, lower cost, no temperature compensation.
  • "Responsivity above 1 A/W is impossible." Quantum efficiency caps a 1550 nm PIN at ~1.25 A/W — but APD effective responsivity M×R reaches 10–50 A/W via gain.
  • "Photodiodes respond instantly." Transit time and junction capacitance limit bandwidth; large-area PINs may manage only MHz while small high-speed ones exceed 40 GHz.

Real-World Applications

  • Telecom: PINs for short reach; InGaAs APDs for long-haul 10–100 Gbps receivers.
  • LIDAR: APDs and Geiger-mode SPADs detect single-photon returns.
  • OTDR: APD instruments locate fiber faults from faint backscatter.
  • Everyday: PINs in remote-control receivers, smoke detectors, pulse oximeters, light meters.

Photodetectors close the optical loop: the PIN gives honest one-electron-per-photon conversion at low voltage; the APD trades high bias and noise for gain that rescues faint signals. Compute responsivity from wavelength, apply gain where justified, and receiver sensitivity stops being a mystery.

Related blogs