Operating Principles and Limitations of Exposure Meters

Hello. Last lesson compared cameras that preview the taking lens’s image with cameras that use a separate viewing path. An exposure meter poses a related question: which light does it measure, and how faithfully does that measurement predict the light reaching the recording surface?

This lesson follows the signal from scene light to an exposure recommendation. We will compare three detector technologies—selenium, cadmium sulfide (CdS), and silicon—then examine how through-the-lens (TTL) metering incorporates a detector into the viewing path you have just studied. The key engineering distinction is between detecting light accurately and deciding how a scene ought to be exposed.

---

What an exposure meter can—and cannot—observe

Recall that, to first order, exposure at the image plane scales with scene radiance, shutter time $t$, lens transmission $T$, and the inverse square of f-number $N$:

$$H\propto \frac{LTt}{N^2}.$$

Film speed tells the meter how much exposure the chosen material requires. The detector supplies an estimate of the light available; a calibrated mechanical or electronic system combines that estimate with film speed and camera settings. A needle position or recommended shutter time is therefore an interpretation of a detector signal, not a direct measurement of the film’s eventual density.

Most camera-mounted meters measure light arriving from the scene. For an approximately diffuse surface, its radiance is proportional to both its illumination and its reflectance:

$$L_{\lambda }\approx \frac{{\rho }_{\lambda }{E}_{\lambda }}{\pi },$$

where $E_{\lambda }$ is incident spectral irradiance and ${\rho }_{\lambda }$ is spectral reflectance. A camera cannot generally infer those two factors separately from the reflected light alone. A bright reading might indicate strong illumination, a pale subject, or both.

This matters even with a perfect detector. Suppose two uniformly lit surfaces fill the meter’s sensing area, with reflectances $0.8$ and $0.2$. The first produces roughly four times the reflected-light signal. If the meter treats either surface as a reference midtone, it recommends two stops less exposure for the pale surface—even if you intend to preserve its pale appearance. Dark subjects create the converse problem. Exposure compensation or a deliberately chosen spot reading addresses this interpretive limitation; replacing the detector material does not.

An incident meter instead measures illumination at the subject, typically through a diffuser. It avoids dependence on the subject’s reflectance, but requires access to an appropriate measurement position and knowledge of which illumination matters. Keep that distinction in mind: reflected versus incident describes the measurement geometry; selenium versus CdS versus silicon describes how light becomes an electrical signal. Neither distinction is synonymous with TTL.

---

Selenium and CdS: two different electrical effects

Read the selenium and CdS subsections of Exclusive Architecture’s Photoelectric Exposure Meters for the historical transition from a self-powered meter to a compact, battery-powered one. Treat its broad claims about spectral matching as descriptions of particular implementations, not universal properties of each material.

A selenium photovoltaic cell generates an electrical output when light is absorbed. In the construction described in the reading, selenium lies between a supporting contact and a thin upper contact. Absorbed photons create mobile charge carriers; separation at a rectifying interface allows current to flow through an external circuit. More incident light generally produces more photocurrent. A sufficiently large cell can drive a sensitive meter movement without a battery.

That battery-free operation shaped the camera. A front-mounted selenium cell could measure light without taking any from the viewfinder, but its area had to be large enough to deliver a usable signal, especially in dim conditions. Such a cell measures the light falling on its own window. Unless additional optics constrain its field appropriately, that is not necessarily the field, angular weighting, or light transmission of the taking lens. Ageing of the cell and its contacts can also shift calibration. “Self-powered” is thus an electrical advantage, not a guarantee of sensitivity or accuracy.

CdS is ordinarily used as a photoconductor, not as the selenium cell’s direct photovoltaic substitute. Absorption increases the population of mobile carriers, raising conductivity and lowering resistance. A battery drives current through a circuit containing the CdS element; the resulting voltage or current changes with illumination. The cell can be small, making it easier to place within a camera’s optical system, and it can give useful readings at light levels where a needle driven directly by a practical selenium cell struggles.

The cost is that conductivity is not solely a function of present illumination. Carriers can become trapped and released gradually. After a bright scene, some CdS cells therefore respond slowly when pointed into darkness—the “night-blindness” described in the reading. Battery condition and circuit calibration also matter. Unlike an ideal instantaneous detector, a CdS meter may briefly report something closer to the recent history of the light than to the light now in front of it.

Neither material simply “sees what film sees.” A meter’s spectral response includes the detector, its window, any filters, and its optics; film response depends on the particular emulsion. A close match under one illuminant does not ensure a close match under another.

---

Silicon: fast detection, with the circuit still doing essential work

Hamamatsu Photonics’ Technical Note: Si Photodiodes provides a detector-level account. Read the indicated portions rather than the full device catalogue. In particular, distinguish the photodiode’s light-generated current from the powered electronics a camera may use to amplify and interpret it.

In a silicon photodiode, a photon with sufficient energy can excite an electron across the band gap, leaving an electron–hole pair. The field in the junction’s depletion region separates carriers; carriers generated nearby can also diffuse into that region and be collected. Over an appropriate operating range, collected photocurrent is approximately proportional to the rate of absorbed photons.

That does not mean a silicon photodiode fundamentally requires a battery to generate a signal. Like a selenium photovoltaic cell, it can produce a light-dependent output without an externally applied bias. In a camera meter, however, power is normally useful for amplifying a small current, displaying or processing the result, and controlling exposure. A circuit may hold the diode near short circuit for a proportional current measurement, or apply reverse bias to meet particular response or linearity requirements. Hamamatsu notes the trade-off: reverse bias can improve some operating characteristics while increasing dark current and noise.

Silicon also makes detector limitations easier to express quantitatively. At room temperature its band gap corresponds to a long-wavelength cutoff of about $1100\mathrm{nm}$, beyond the visible range. An unfiltered silicon meter can therefore respond to near-infrared light that a particular film or intended visible-light calibration weights differently. At low light, the limit comes not merely from “insufficient sensitivity” but from photocurrent competing with dark-current, resistive, and amplifier noise; at high light, the readout can saturate.

Silicon’s rapid response and compact size made it attractive for changing scenes and more elaborate metering patterns. Its photocurrent remains a spectrally weighted measurement, however. Filters and calibration can bring that weighting closer to the intended photographic response, at the cost of rejecting some light. Likewise, a low-noise diode alone cannot overcome noise or saturation in its readout circuit.

---

TTL: measure light after the taking lens

The previous lesson’s SLR geometry provides a route to a better-matched measurement. With the mirror down, the taking lens forms a preview image on the focusing screen. A small sensor can sample light associated with that image. The meter then includes much of what an external, front-facing cell misses: the mounted lens’s field, its transmission, attached filters, and—in suitable arrangements—changes such as lens extension.

Watch the physical demonstration in Technology Connections’ 1970’s Camera Tech: How they showed you what settings to use. It shows one historical solution, not the optical layout of every TTL meter.

Another design gathers light travelling from the focusing screen through the pentaprism. The following image traces that arrangement. Its cyan and white rays begin at the same point on the focusing screen: they illustrate different directions in which light leaves that point, not a beam splitter dividing one ray.

The historical examples and this prism arrangement are described in Exclusive Architecture’s “Through-The-Lens Metering.” Read them as examples of where a meter can intercept preview light, paying particular attention to the loss of light between the focusing screen and the meter.

TTL does not mean the meter sits at the film plane. In these SLR arrangements it usually reads before exposure, with the mirror down; the recording surface receives light later, after the mirror moves. A focusing screen scatters light, and the meter collects only some of it. That loss is acceptable if its relationship to image-plane exposure is known and stable enough to calibrate.

Aperture timing is another important qualification. Some cameras meter with the lens stopped down to the intended taking aperture: the light reduction is present in the reading itself, though the preview becomes dimmer. Others meter wide open for a bright finder and account for the selected taking aperture through lens–body coupling or electronic information. In the latter case, a correct detector reading can still yield a wrong exposure if the camera has incorrect information about the lens or aperture. TTL incorporates the taking lens’s optical path, but it does not remove the need for mechanical and electronic calibration.

---

From a reading to a decision

Detector technology, optical placement, and scene interpretation form three distinct error budgets:

Question	Representative limitation
Did the detector respond to the light?	Selenium output too weak in low light; CdS slow after a brightness change; silicon signal limited by noise or readout saturation
Was it the relevant light?	A front cell sees a different field; a TTL meter samples only part of the preview light; spectral weighting differs from the recording material
Was it interpreted appropriately?	A reflective reading cannot tell a pale subject in weak light from a darker subject in stronger light

A meter may also weight different parts of the frame differently. A central sensing region, an average over the frame, or multiple metering zones each makes a different choice about which radiances should govern the recommendation. More zones supply more scene information, but cannot uniquely determine the photographer’s intended subject tone. A bright window behind a subject remains a decision problem, not just a detector problem.

This distinction explains the historical progression without suggesting that each new cell made previous exposure errors disappear. Selenium enabled practical battery-free photoelectric meters. CdS offered small, sensitive elements that were readily incorporated behind the lens, while introducing response-time and battery dependencies. Silicon supported fast, compact, approximately proportional photocurrent measurements and increasingly capable readout electronics. TTL improved the relationship between what the meter sampled and what the taking lens supplied. None of these developments made reflected light an unambiguous instruction about the desired photograph.

---

Conclusion

Selenium photovoltaic cells generate a light-dependent output that can drive a meter without a battery; CdS cells use light-dependent resistance in a powered circuit; silicon photodiodes collect junction-generated carriers and are typically paired with powered readout electronics. Their practical limits differ in signal strength, response history, spectral weighting, noise, and calibration.

TTL metering changes where light is sampled: after the taking lens, often from the SLR preview path. It accounts for more of the actual optical system than an external cell, while still depending on the preview-to-recording-path calibration and on a choice about how scene brightness should be rendered.

The next lesson turns from measuring image brightness to measuring image sharpness: how contrast and phase differences provide autofocus signals, and how a camera uses those signals to move a lens.