How can different sensor spectral-response units be related to quantum efficiency (QE)?
Asked 11/2/2024
4 views
2 answers
0
I'm compiling spectral-response data for monochrome camera sensors, but manufacturers present the Y-axis in different ways: relative response, quantum efficiency (%), spectral response (A/W), LSB/nJ/cm², and QE × fill factor (%). How do these measures relate to each other, and which can be converted to a QE(λ) curve?
Specifically, which are just normalized versions of the same thing, which can be converted using general physics, and which require extra sensor-specific data such as fill factor, conversion gain, ADC scaling, or pixel area?
Originally by LimeAndConconut. Source · Licensed CC BY-SA 4.0
LimeAndConconut
1y ago
2 Answers
1
Disclaimer: I am hobbyist who knows some physics and have been reading some technical literature from time to time.
- Quantum efficiency: linear scale, probability of photon causing electron to add to pixel well (capacitor) when it hits the pixel detector.
- Relative sensitivity: proportional to QE but normalized so that maximum = 1 (i.e. relative to peak sensitivity).
- Example C: the A/W is pixel well current per incident light power (Amperes per Watt) and is not fully proportional to QE because photon energy is proportional to it's wavelength. Ampere is unit of current (amount of electrons crossing a boundary) and is proportional to number of electrons.
- Example D: LSB is in the glossary there as "Least significant bit" and refers to ADC output, not pixel well. The graph is proportional to A/W graph. Iso-QE lines are for informative purpose and illustrate what I said about Ex. C.
- Example E: QE*FF is proportional to QE since FF = fill factor and is constant. Since sensor is not 100% covered by detectors' active surface QE might be measured for individual pixels or for sensor on average. QE*FF means explicitly that fill factor is included in QE.
So, to your questions:
- If your aim is to know the colour reproduction characteristics then all units can be converted according to what I've said, no additional information is needed.
- Relative efficiency cannot be converted to QE without knowing QE at least for one wavelength.
- Whether QE includes fill factor or not should be deduced from the document.
- LSB per power is fundamentally different from A/W because it very likely refers to ADC output, not pixel wells. But iso-QE lines are giving you the conversion ratio anyway (it seems to me that they correspond to some specific ISO sensitivity which is characteristic of ADC).
- Relative sensitivity is the largest common denominator of all those graphs. If you want to gather just the colour reproduction properties that's totally enough. So, in general, A/W must be corrected using wavelength and then it becomes fully proportional to QE and other units are proportional to either of those two.
Originally by Euri Pinhollow. Source · Licensed CC BY-SA 4.0
Euri Pinhollow
1y ago
0
Generated from our catalog & community — verify before relying on it.
Some of these curves are directly related, but not all are equivalent.
- QE (%): photon-to-electron conversion probability. This is the most direct measure.
- Relative response: usually just a normalized sensitivity curve, often proportional to QE, but scaled so the peak equals 1. You can recover shape, not absolute QE, unless a scale factor is known.
- Spectral response (A/W): convertible to QE using physics because responsivity depends on photon energy, which varies with wavelength. So A/W and QE are related, but not by a constant factor across wavelengths.
- QE × fill factor (%): this is not bare QE. To recover QE, you need the sensor’s fill factor.
- LSB/nJ/cm²: this includes the sensor’s optical response plus downstream electronics/ADC scaling. Converting it to QE requires sensor-specific information such as pixel area and electronic conversion factors, so it is not directly convertible from general physics alone.
In short:
- Same shape / normalized substitute: relative response ↔ QE (up to unknown scale)
- Directly physics-convertible: A/W ↔ QE
- Needs sensor-specific constants: QE × FF ↔ QE, and LSB/nJ/cm² ↔ QE
Recommended products
UniqueBot
AI1y ago
Your Answer
Related Questions
How is a camera sensor’s spectral sensitivity measured and expressed?
Can color accuracy be judged from a camera sensor's spectral sensitivity curves alone?
Why can RAW pixel values exceed the number of incident photons at high ISO?
How does luminance (Y) relate to the CIE 1931 chromaticity diagram and gamut limits?
Why can an external light meter work without knowing my camera sensor’s sensitivity?