Photon Shot Noise Calculation
Estimate photon rate, detected electron rate, photocurrent, shot noise current, and relative noise from optical power, wavelength, quantum efficiency, bandwidth, and integration time.
Enter source power before detection.
Common examples: 405, 532, 633, 850, 1064, 1550 nm.
Use detector QE in percent.
Noise bandwidth used for RMS current estimate.
Used to estimate total detected electrons and relative shot noise.
Shot Noise Trend Around Your Operating Point
Expert guide to photon shot noise calculation
Photon shot noise is the unavoidable fluctuation that appears because light arrives in discrete quanta rather than as a perfectly smooth flow. Even when a laser or lamp has stable average output, the actual number of photons detected during a finite interval varies statistically. In a quantum-limited optical measurement, those fluctuations become the floor below which signal extraction is impossible without increasing photon collection, extending integration time, or changing the system architecture. This matters in photodiode metrology, microscopy, spectroscopy, lidar, astronomy, fiber communications, quantum optics, and machine-vision sensors.
At the practical level, photon shot noise calculation answers a simple engineering question: given an optical power, wavelength, detector efficiency, and bandwidth, how much random fluctuation should I expect? Once you know that number, you can compare it with amplifier noise, dark current noise, thermal noise, and digitizer limits. If the shot noise is much larger than electronic noise, your instrument is operating close to the quantum limit. If electronic noise dominates, collecting more light will not help until the readout chain is improved.
The physics begins with the energy of a single photon:
E = hc / λ
where h is Planck’s constant, c is the speed of light, and λ is wavelength. Once photon energy is known, the photon arrival rate is simply optical power divided by photon energy. If your detector has quantum efficiency less than 100%, only a fraction of those photons generate charge carriers. The resulting electron stream produces a photocurrent, and because arrival events are discrete and random, the current contains shot noise.
Why shot noise follows a square-root law
For independent arrival events, the count statistics are well approximated by a Poisson distribution. In a Poisson process, if the mean detected count over an interval is N, the standard deviation is √N. That single result explains many familiar formulas:
- Photon count noise: σ = √N photons or electrons when quantum efficiency is included.
- Relative shot noise: σ/N = 1/√N.
- Photocurrent shot noise spectral density: √(2qI) A/√Hz.
- RMS shot noise in bandwidth B: √(2qIB) amperes.
This square-root dependence is why averaging helps. If you increase the number of detected photons by a factor of 4, the noise amplitude only doubles, but the signal also grows by a factor of 4. As a result, the signal-to-noise ratio improves by a factor of 2. In short, more photons almost always help, but the gains follow a square-root improvement rather than a linear one.
Step by step photon shot noise calculation
- Convert optical power into watts.
- Convert wavelength into meters.
- Calculate single-photon energy with E = hc/λ.
- Compute photon rate as P/E.
- Apply detector quantum efficiency to find detected electron rate.
- Multiply by elementary charge q to obtain average photocurrent.
- Use bandwidth to estimate RMS current noise: irms = √(2qIB).
- Use integration time to estimate total detected electrons N and relative shot noise 1/√N.
These equations are consistent with the fundamental constants maintained by the National Institute of Standards and Technology. For reference data on constants and radiometric foundations, see the NIST fundamental constants page. For detector and optical radiation references, you may also review material from NIST Optical Radiation Group and educational optics resources such as MIT optics lecture material.
Reference table: photon energy at common wavelengths
| Wavelength | Photon energy | Photon energy | Typical use case |
|---|---|---|---|
| 405 nm | 4.91 × 10-19 J | 3.06 eV | Blu-ray optics, fluorescence excitation |
| 532 nm | 3.73 × 10-19 J | 2.33 eV | DPSS green lasers, microscopy |
| 633 nm | 3.14 × 10-19 J | 1.96 eV | HeNe metrology, alignment systems |
| 850 nm | 2.34 × 10-19 J | 1.46 eV | VCSEL links, proximity sensing |
| 1064 nm | 1.87 × 10-19 J | 1.17 eV | Nd:YAG lasers, lidar |
| 1550 nm | 1.28 × 10-19 J | 0.80 eV | Fiber optics, telecom, eye-safer lidar |
The table highlights a subtle but important point: longer wavelengths carry less energy per photon. For a fixed optical power, that means a 1550 nm system delivers more photons per second than a 532 nm system. If detector efficiency is similar, the longer-wavelength system can produce a higher count rate and a lower relative shot noise. In practice, however, detector material response varies strongly with wavelength, so the actual advantage depends on quantum efficiency and excess noise factors.
Worked example
Suppose you detect 1 mW at 532 nm with a detector quantum efficiency of 85%. The photon energy is about 3.73 × 10-19 J. Dividing power by photon energy gives a photon rate of roughly 2.68 × 1015 photons/s. Applying 85% quantum efficiency yields about 2.28 × 1015 electrons/s. Multiplying by the elementary charge gives a photocurrent of approximately 0.365 mA. In a 1 MHz noise bandwidth, the RMS shot noise current is about 10.8 nA. If you integrate for 1 second, the total detected electron count is 2.28 × 1015, the standard deviation is about 4.78 × 107, and the relative shot noise is approximately 2.1 × 10-8, or 0.0000021%.
Comparison table: optical power versus photon statistics at 532 nm and 85% QE
| Optical power | Detected electron rate | Photocurrent | RMS shot noise at 1 MHz | Relative shot noise in 1 ms |
|---|---|---|---|---|
| 1 nW | 2.28 × 109 s-1 | 0.365 nA | 10.8 pA | 0.066% |
| 1 uW | 2.28 × 1012 s-1 | 0.365 uA | 0.342 nA | 0.0021% |
| 1 mW | 2.28 × 1015 s-1 | 0.365 mA | 10.8 nA | 0.000066% |
This comparison shows the defining trend of shot-noise-limited systems: absolute noise rises with signal, but relative noise falls. At 1 nW the absolute current noise is tiny, yet the relative uncertainty over short integration intervals is much larger than at 1 mW. Engineers often misinterpret this and assume that low absolute noise means high precision. In fact, precision depends on noise relative to the signal level and on the detection time available.
What the calculator includes and what it does not
This calculator focuses on the ideal primary shot noise mechanism associated with detected photoelectrons. It includes:
- Photon energy from wavelength.
- Incident photon rate from optical power.
- Detected electron rate from quantum efficiency.
- Average photocurrent.
- Current noise spectral density and RMS current noise in a specified bandwidth.
- Total detected electrons and relative shot noise over a chosen integration time.
It does not automatically include:
- Dark current shot noise.
- Thermal noise from resistors and transimpedance amplifiers.
- Avalanche photodiode excess noise factor.
- Laser relative intensity noise.
- Digitizer quantization limits.
- Optical coupling losses unless you already folded them into the optical power input.
How to use photon shot noise calculations in real design work
If you are designing a photodiode receiver, start by estimating expected optical power at the detector, not at the source. Include coupling loss, reflection, beam clipping, and filter transmission. Then choose the operating wavelength and detector quantum efficiency at that wavelength. Next, determine the effective electrical noise bandwidth of the readout chain. This is not always equal to the nominal sampling rate or analog cutoff frequency, so use the best available system model. Finally, choose an integration time or exposure time that matches your application. Imaging sensors, lock-in amplifiers, and pulse energy measurements all translate this parameter differently.
After calculation, compare the predicted shot noise with front-end electronics noise. A healthy quantum-limited design usually has electronic noise comfortably below the shot noise at the intended operating point. If not, there are several common remedies:
- Increase collected optical power within detector linearity limits.
- Increase integration time or reduce measurement bandwidth.
- Improve detector quantum efficiency by changing material or wavelength.
- Reduce optical losses using better coupling and anti-reflection control.
- Lower amplifier noise or optimize transimpedance gain.
Common mistakes in shot noise estimation
- Using incident photons instead of detected electrons: quantum efficiency matters.
- Mixing RMS and peak values: shot noise formulas usually give RMS values.
- Ignoring bandwidth definition: noise power scales with bandwidth.
- Confusing spectral density with integrated noise: A/√Hz is not the same as A RMS.
- Forgetting time-unit conversion: milliseconds and microseconds change count statistics dramatically.
- Assuming all detectors behave like ideal PIN photodiodes: APDs and PMTs may require extra factors.
Interpreting the chart below the calculator
The interactive chart visualizes how photocurrent and RMS shot noise vary as optical power moves around your chosen operating point. The relationship is not linear in every metric. Photocurrent scales linearly with optical power, while RMS shot noise scales with the square root of current and therefore grows more slowly. This is why increasing power improves signal-to-noise ratio: signal grows faster than the random fluctuation amplitude.
Final takeaway
Photon shot noise calculation is one of the most useful back-of-the-envelope tools in optics and electro-optics. It converts abstract quantum behavior into concrete engineering numbers such as photons per second, amperes, nanoamps RMS, and relative uncertainty. Once you know those values, you can judge whether your experiment is source-limited, detector-limited, or electronics-limited. For advanced work, extend the model with dark current, thermal noise, excess noise factors, and source intensity noise. But for first-order performance estimation, the shot noise framework remains the essential baseline.