Three generations of photoelectric cell have powered hand-held and in-camera exposure meters, and each carries a distinct electrical character that shows up directly in the negative. A meter is only as honest as its cell, and the differences between selenium, cadmium-sulfide, and silicon are not merely historical curiosities: they determine whether a reading can be trusted in dim light, under tungsten, or after the cell has spent an hour in bright sun.

Selenium: Self-Powered but Slow to See in the Dark

The selenium cell is photovoltaic. Light striking the selenium layer generates a small current directly, with no battery required, which is why meters built around it remain serviceable in fully mechanical bodies decades after manufacture. The trade-off is sensitivity. The output current is proportional to incident light, but at low levels that current becomes too small to deflect a galvanometer reliably. Selenium meters typically run out of usable range somewhere around domestic interior lighting and cannot register candlelight, moonlight, or starlight at all. The selenium itself is also slowly consumed in service, so cells drift low over many years. Spectrally, selenium responds across most of the visible band with a peak in the green, reasonably close to the human eye, which is part of why these meters earned a reputation for pleasant, forgiving readings in daylight.

Cadmium Sulfide: Sensitive, Eye-Matched, but with a Memory

Cadmium-sulfide (CdS) cells are photoresistors: their resistance falls as illumination rises, so they require a battery to drive the measuring circuit. The reward is far greater low-light sensitivity than selenium, which is why CdS displaced selenium in the 1960s for available-light and through-the-lens metering. Their peak response sits roughly between 520 and 650 nanometers, in the green-to-red region, closely tracking the photopic sensitivity of the eye. CdS carries two well-documented faults. First, a pronounced memory effect: resistance depends on the cell’s recent illumination history, drifting after a change in light level and taking minutes, sometimes far longer, to settle, so a reading taken immediately after moving from bright to dim conditions reads high. Second, sluggish recovery makes CdS slow to respond at the lowest light levels precisely where it is most needed.

Silicon: Fast, Linear, and Hungry for Infrared

The silicon photodiode is also photovoltaic but generates far less voltage than selenium, so it depends on an amplifier and a battery. In exchange it offers a near-instantaneous response, no measurable memory effect, and excellent linearity over a very wide range, which makes silicon-blue-cell meters suited to flash measurement and rapid changing-light work. Its weakness is spectral. Bare silicon is sensitive from roughly 400 to about 1100 nanometers, peaking deep in the near-infrared around 800 to 900 nanometers, well outside what panchromatic film records as luminance. Left uncorrected it over-reads under tungsten and other infrared-rich sources. The fix is an integral color-correction filter that suppresses the infrared and reshapes the response toward the photopic curve; a cell so corrected is usually marketed as a silicon-blue or SPD cell, and the quality of that filter largely determines a silicon meter’s color accuracy.

Choosing by Failure Mode

The practical distinctions reduce to a few failure modes. Selenium is battery-free and benign in daylight but blind in the dark and prone to age-related drift. CdS reaches into dim light but must be allowed to settle and treated with suspicion just after a large brightness change. Silicon is fast, linear, and stable, yet only as accurate as the filter taming its infrared appetite. Knowing which cell sits behind the needle explains most of the discrepancies between two meters pointed at the same scene.