Millions of photovoltaic detectors are used to sense and measure radiant energy in every type of industrial, medical, consumer, and scientific instrumentation application. By properly selecting materials and structure, solid-state photovoltaics can be made to respond to wavelengths ranging from the far-infrared through the visible spectrum and into the ultraviolet.
Moreover, when compared as a class to their chief competition, photoconductive detectors, photovoltaics generally display superior dc accuracy (zerooffset and linearity) and flicker-noise characteristics. Photovoltaics, therefore, tend to excel in precision photometric applications. This versatility suits photovoltaics to such diverse jobs as chemical spectral analysis, colorimetry, non-contact thermometry, flamedetection, and non-invasive blood-gas monitoring.
The basic signal conditioning circuit for photovoltaics is the current-to-voltage or transimpedance amplifier (Fig. 1). The chief shortcoming of this classic circuit is the inability of one value of feedback resistor to adequately accommodate the four, five, or more decades of dynamic range displayed by many detectors. Even when RF is made adjustable, the finite resolution of both mechanical and digitally controlled feedback resistances fails to fully resolve the dynamic-range problem.
The circuit in Figure 2 combines two Xicor X9258T digitally controlled potentiometers with an AD822 low-noise dual op amp. This ultimately creates a flexible, digitally calibrated, wide-dynamic-range transimpedance amplifier topology that can be used with virtually any photovoltaic detector technology. The amplifier output is given by:
VO = IS(1M) (1 + P1)/(256 − P1)
where P1 is the 8-bit (0 to 255) digital value written to DCP1.
Of particular interest is the pseudo-logarithmic behavior of the circuit’s transimpedance pin as a function of P1. The transimpedance pin of this amplifier varies over the range of 1/256M to 256M as P1 varies from 0 to 255. Meanwhile, the gain-factor resolution never gets worse than 10% per P1 increment over a 400:1 (52 dB or nearly 9-bit) range of 50k to 20M. Transimpedance settings covering an even wider span are accessible—4k to 256M corresponding to full-scale IS values from 1 mA to 16 nA—albeit with reduced resolution. A linear pin adjustment could never do such a good job of achieving both a multidecade gain adjustment range and good adjustment resolution throughout the range.
Another useful feature of the Figure 2 circuit is that the gain is independent of both the P1 element and the wiper resistances. The trick of using the pot wiper as an input terminal effectively moves the element temperature coefficient and wiper contact resistance inside the feedback loop of A1. Thus, they’re removed as gain-error terms, which improves the time and temperature stability of the gain setting.
DCP2 is used to null the amplifier zero-point. It works by varying the voltage at the noninverting input of A2 over ±2-mV range with a resolution of 16 µV.