The Electronic Retina: How Photodetectors Capture the Faint Whisper of Reflected Light |
Subtitle: A Deep Dive into Photodiodes, Phototransistors, CMOS Pixels, and Avalanche Devices - with Real-World Designs from Symbol, Zebra, Honeywell, Datalogic, Omron, Sony, and Hamamatsu |
|
Opening Summary |
If the illumination source is the voice of the barcode scanner, then the photodetector is its ear - the sensitive organ that listens to the faint reflection that comes back from the label. This is the second pillar of the three-part decoding system, and it is arguably the most delicate. The photodetector must convert an incredibly weak optical signal, often measured in nanowatts, into an electrical current that can be amplified and processed. It must do this reliably in the face of ambient sunlight, electrical noise, temperature drift, and the mechanical motion of the scan. |
This article is dedicated to the photodetector - the electronic retina that sits at the heart of every barcode reader. We will explore the different types of photodetectors used in the industry: the humble PIN photodiode that is the workhorse of laser scanners; the phototransistor that offers higher sensitivity at lower cost; the advanced CMOS pixels that form the basis of modern imagers; and the exotic avalanche photodiodes that push the boundaries of sensitivity for long-range industrial applications. |
We will examine how major companies have selected and optimized these devices. We will look at how Symbol Technologies (now Zebra) used a simple but carefully biased PIN photodiode in their legendary LS2208. We will see how Honeywell integrated a large-area photodiode with an on-chip dark-current reference for temperature stability. We will explore how Datalogic uses a dual-photodiode configuration for noise cancellation, and how Omron employs phototransistors for ultra-low-cost scanners. We will also delve into the advanced imaging sensors from Sony and ON Semiconductor, which pack millions of photodiodes onto a single chip, each with its own analog front-end. |
By the end of this journey, you will understand why the photodetector is not just a component but a system-level decision, intimately tied to the illumination, the optics, and the signal processing. You will see how a carefully chosen photodiode can make the difference between a scanner that reads every label and one that fails on the very first barcode of the day. |
|
Full Article |
Section 1: The Photodetector's Role - From Photons to Electrons |
Every photodetector, regardless of its type, performs the same fundamental function: it absorbs photons (light particles) and releases electrons (electric charge). This is the photoelectric effect, discovered by Albert Einstein in 1905. The number of electrons released is proportional to the number of photons absorbed - within limits. This proportionality is the detector's 'responsivity,' measured in amperes per watt (A/W). For a typical silicon photodiode at 650 nm (red light), the responsivity is about 0.5 A/W. That means that one microwatt of optical power produces 0.5 microamperes of current. For a 5-milliwatt laser, the reflected power might be only a few microwatts, so the current is in the nanoampere range - a truly tiny signal. |
The photodetector must be chosen to match the wavelength of the illumination, the speed of the scan, the required sensitivity, and the cost target. It must also be biased correctly (with a reverse voltage) to reduce its capacitance and increase its speed. The bias voltage, the temperature, and the ambient light all affect the detector's performance, and the designer must account for all of these factors. |
|
Section 2: The PIN Photodiode - The Gold Standard for Laser Scanners |
The PIN photodiode is the most common photodetector in barcode scanners. The 'PIN' stands for P-type, Intrinsic, and N-type - a sandwich structure where a thick layer of undoped (intrinsic) semiconductor is placed between the P and N layers. This intrinsic layer creates a wide depletion region, which gives the photodiode two key advantages: low capacitance and fast response time. |
Low capacitance is crucial because it allows the photodiode to work with a high-value feedback resistor in the transimpedance amplifier (TIA) without sacrificing speed. The capacitance, together with the resistor, forms an RC time constant that limits the bandwidth. A PIN photodiode might have a capacitance of 2-5 picofarads, while a simple PN photodiode might have 10-20 picofarads. The lower capacitance allows a larger feedback resistor (and thus higher gain) for the same bandwidth. |
The PIN photodiode's speed comes from the fact that the photo-generated carriers are collected by drift in the depletion region, which is very fast (nanoseconds), rather than by diffusion, which is slower (microseconds). This is essential for laser scanners where the light spot moves across the bars at high speed. A typical laser scanner requires a photodiode with a rise time of less than 1 microsecond, which is easily met by a PIN device. |
|
Section 3: Symbol's LS2208 - The Classic PIN Implementation |
The Symbol LS2208, which we have mentioned in previous articles, uses a PIN photodiode that is believed to be the OSRAM SFH 203 P. This device has an active area of 1 square millimeter, a capacitance of 2 picofarads at 5 volts reverse bias, and a responsivity of 0.6 A/W at 650 nm. The dark current (the current that flows even in total darkness) is 2 nanoamperes at room temperature - a key specification because it sets the noise floor. |
The LS2208 biases the photodiode at 2.5 volts, which is a compromise between speed and dark current. At 2.5 volts, the capacitance is about 3 picofarads, and the dark current is about 1 nanoampere. The TIA uses a 470-kiloohm feedback resistor, which, with the 3-picofarad capacitance, gives a bandwidth of about 100 kHz - sufficient for the scanning speed of the LS2208. |
The photodiode is mounted on a small PCB with a hole that allows the light to reach the active area. The PCB is carefully placed at the focal point of the collecting mirror. The alignment is done using a laser alignment tool that ensures the reflected light is centered on the photodiode. The entire assembly is then potted with a clear epoxy to protect it from dust and moisture. |
|
Section 4: The Phototransistor - A Higher Sensitivity Alternative |
A phototransistor is essentially a bipolar transistor with a large base-collector junction that is exposed to light. The light generates carriers in the base region, which are amplified by the transistor's current gain (beta). A typical phototransistor has a responsivity of 1-10 A/W, which is 2-20 times higher than a PIN photodiode. This higher sensitivity allows the use of a smaller feedback resistor (less noise) or a dimmer illumination source (less power). |
However, phototransistors have a significant disadvantage: they are much slower than PIN photodiodes. The carriers must diffuse through the base region, which can take microseconds. The capacitance is also higher (typically 10-20 picofarads). For this reason, phototransistors are almost never used in laser scanners, but they are sometimes used in low-cost, low-speed LED imagers. |
|
Section 5: Omron's Use of Phototransistors - Cost-Cutting in Simple Scanners |
Omron, in their entry-level fixed-mount scanners, uses a phototransistor as the detector. These scanners are used in simple applications like access control or document tracking, where the barcode is presented slowly and at a fixed distance. The phototransistor is a 3-pin device (collector, base, emitter), with the base left floating. The light generates carriers that effectively increase the base current, which is amplified by the transistor. |
The driver circuit is very simple: the phototransistor is connected in a common-emitter configuration with a load resistor (say, 10 kilohms). The output voltage is the supply voltage minus the voltage drop across the resistor. When light hits the phototransistor, the collector current increases, the voltage drop increases, and the output voltage decreases. This is a single-ended, non-inverting amplifier. |
The simplicity of the phototransistor circuit reduces the BOM cost by about $0.20 compared to a PIN photodiode with a TIA. Omron's scanners are produced in volumes of millions, so this saving is significant. The trade-off is speed: the phototransistor's response time is about 10 microseconds, which limits the scanning speed to about 100 mm/s - adequate for the intended use. |
|
Section 6: The Avalanche Photodiode - For Ultimate Sensitivity |
An avalanche photodiode (APD) is a photodiode that is operated at a high reverse bias voltage, just below the breakdown point. In this region, the electric field is strong enough to cause impact ionization - an incoming carrier creates a cascade of additional carriers, resulting in an internal gain (usually 10-100 times). This gain multiplies the photocurrent, giving the APD a sensitivity that is orders of magnitude higher than a PIN photodiode. |
APDs are used in long-range scanners, where the reflected light is extremely weak. For example, a scanner that reads barcodes from 10 meters away would receive only a few nanowatts of optical power. A PIN photodiode would produce a current of a few nanoamperes, which is buried in noise. An APD, with its internal gain, can produce a current of a few hundred nanoamperes, which is easily detectable. |
|
Section 7: Hamamatsu and Zebra - The APD in Long-Range Scanners |
Hamamatsu, a Japanese manufacturer, is a leading supplier of APDs for barcode applications. Their S10912 series is designed specifically for laser scanners, with a peak sensitivity at 650 nm, a gain of 30 at a bias of 150 volts, and a dark current of 5 nanoamperes. The APD requires a high-voltage bias supply - a boost converter that generates 150 volts from a 5-volt input. |
Zebra's DS3500, a long-range warehouse scanner, uses an APD from Hamamatsu. The high-voltage bias supply is a critical part of the design. It must be low-noise because any ripple on the bias voltage modulates the gain, creating noise. Zebra uses a switched-capacitor voltage multiplier followed by a linear regulator. The linear regulator has a high power supply rejection ratio (PSRR) of 80 dB at 100 kHz, effectively filtering the switching noise. |
The APD's output current is converted by a TIA, but the TIA's feedback resistor is smaller (typically 10-50 kilohms) because the APD's gain already provides the signal. The smaller resistor reduces the thermal noise, improving the overall signal-to-noise ratio. The TIA also includes a current monitor that measures the APD's dark current, which is used to adjust the bias voltage for temperature compensation. |
|
Section 8: The CMOS Pixel - The Photodetector for Imagers |
For imaging scanners, the photodetector is not a single device but an array of millions of pixels. Each pixel is a tiny photodiode, typically a pinned photodiode, which is a variation of the PIN structure. The pinned photodiode has a very low dark current and a high charge capacity, making it ideal for image sensors. |
The pixel also includes a transfer gate, a reset transistor, a source-follower transistor, and a select transistor. This is the '4T' (four-transistor) pixel architecture that is standard in modern CMOS image sensors. The operation is: reset the photodiode, integrate the light (collect charge), transfer the charge to a floating diffusion node, read the voltage via the source-follower, and then reset again for the next frame. |
The charge-to-voltage conversion factor (the conversion gain) is determined by the capacitance of the floating diffusion node, which is typically a few femtofarads. A single electron changes the voltage by about 100 microvolts. The sensor's analog-to-digital converter (ADC) digitizes this voltage, producing a digital value that represents the light intensity. |
|
Section 9: Sony's IMX Sensors - The Heart of Premium Imagers |
Sony is the market leader in CMOS image sensors, and their IMX series is widely used in high-end barcode imagers. The IMX252, for example, is a 1/1.8-inch sensor with 2.4-megapixel resolution, a global shutter, and a high dynamic range. The global shutter is critical for reading moving barcodes - it captures all pixels at the same instant, avoiding the distortion of a rolling shutter. |
The IMX252's photodiodes are back-illuminated (BSI). In a BSI sensor, the photodiode is on the back side of the silicon wafer, and the wiring (the metal layers) is on the front side. This allows the light to reach the photodiode without passing through the wiring, which absorbs and scatters the light. BSI increases the quantum efficiency (the percentage of photons that produce electrons) from about 50% to over 80%. |
The sensor's driver circuit is relatively simple: it requires a master clock (typically 24-48 MHz), a power supply (1.8 V for the core and 3.3 V for the I/O), and a serial interface for configuration. The sensor outputs the pixel data in a high-speed serial format (MIPI CSI-2). The signal processor on the main board receives this data and performs the decoding. |
|
Section 10: ON Semiconductor's AR Series - For Industrial Ruggedness |
ON Semiconductor (now part of onsemi) is another major supplier of image sensors for barcode applications. Their AR0135 is a 1/3-inch sensor with 1.2-megapixel resolution and a global shutter. It is designed for industrial use, with a wide operating temperature range (-40 to +105 degrees Celsius) and a high tolerance to shock and vibration. |
The AR0135's photodiodes are not back-illuminated; they are front-illuminated, which is cheaper but less efficient. The sensor compensates by having a larger pixel size (3.75 microns) compared to the Sony IMX252 (3.45 microns). The larger pixel collects more light, offsetting the lower quantum efficiency. |
The sensor's analog front-end includes a programmable gain amplifier (PGA) with gains from 1x to 16x, and a 10-bit ADC. The PGA is controlled by the microcontroller via an I2C interface. The sensor also has an internal temperature sensor, which the microcontroller can read to adjust the PGA gain for temperature drift. |
|
Section 11: The Linear CCD - A Legacy but Still Used |
Before CMOS imagers became dominant, barcode scanners used linear CCD (charge-coupled device) sensors. A linear CCD is a row of photodiodes, each with a storage capacitor. The light accumulates charge on the capacitors during the exposure. After the exposure, the charge is shifted out along the row, like a bucket brigade, to a single output amplifier. |
CCD sensors have excellent noise performance and linearity, but they are more expensive and consume more power than CMOS. They are also slower because the charge must be shifted serially. However, some legacy products and some specialized high-end scanners still use CCDs because of their superior low-noise performance. |
|
Section 12: Honeywell's CCD Era - The 4600 Series |
Honeywell's 4600 series, introduced in the early 2000s, used a linear CCD with 2048 pixels. The CCD was fabricated by Sony, and it had a pixel pitch of 14 microns. The scanner's illumination was a bank of red LEDs, and the CCD captured the entire row of the barcode in one exposure. |
The CCD's driver circuit was complex: it required multiple clock phases (typically 2-3 phases) to shift the charge, and it required a high-voltage supply (12-15 volts) for the output amplifier. The clock signals were generated by a timing generator, which was a separate chip or a part of the microcontroller. The output amplifier's signal was digitized by a 12-bit ADC. |
The CCD's sensitivity was about 0.5 A/W at 650 nm, similar to a PIN photodiode. But the CCD had a much lower dark current (less than 0.1 nanoampere per pixel) because it was fabricated on a high-resistivity substrate. This allowed the scanner to read very low-contrast barcodes. |
|
Section 13: The Photodetector's Biasing - A Critical Choice |
The photodetector's bias voltage is a critical design parameter. For a PIN photodiode, reverse biasing reduces the capacitance (increasing speed) but increases the dark current (adding noise). The bias voltage is typically between 2 and 5 volts for handheld scanners. Some industrial scanners use a bias of 10-15 volts, which reduces the capacitance to a few picofarads but increases the dark current to tens of nanoamperes. |
The bias supply must be clean. Any ripple on the bias voltage modulates the photodiode's capacitance and the dark current, creating noise. A simple RC filter (a resistor and a capacitor) is often sufficient, but for higher bias voltages, a low-dropout regulator (LDO) is used. The LDO has a high PSRR, attenuating any ripple from the main power supply. |
Symbol's LS2208 uses a bias of 2.5 volts, generated by a voltage divider from the 5-volt supply. The divider uses two 100-kilohm resistors, and the node is decoupled with a 10-microfarad capacitor. This is a cheap but effective solution. The 10-microfarad capacitor is a tantalum electrolytic, which has a low equivalent series resistance (ESR) and provides good filtering at low frequencies. |
|
Section 14: The Dark Current - The Photodetector's Nemesis |
Dark current is the current that flows through the photodiode even in the absence of light. It is caused by thermal generation of carriers in the depletion region. Dark current doubles for every 10-degree Celsius increase in temperature. At room temperature, a typical PIN photodiode has a dark current of 1-5 nanoamperes. At 60 degrees Celsius, it can be 10-20 nanoamperes. |
Dark current adds a DC offset to the signal, which is blocked by the AC coupling capacitor. But dark current also has a shot noise component, which cannot be blocked. The shot noise is proportional to the square root of the dark current. A dark current of 2 nanoamperes has a shot noise of about 28 picoamperes (RMS), which, when converted by the TIA's feedback resistor, translates to a few microvolts of noise - usually negligible. |
However, if the dark current is large (e.g., tens of nanoamperes), the shot noise becomes significant. To reduce dark current, the photodiode can be cooled (not practical in a handheld scanner), or the bias voltage can be reduced (but this increases the capacitance). A better solution is to use a photodiode with a low dark current, such as the OSRAM SFH 203 P, which has a dark current of 2 nanoamperes even at 5 volts bias. |
|
Section 15: The Shunt Resistance - A Measure of Leakage |
The shunt resistance of a photodiode is the resistance of the junction when no bias is applied. It is a measure of the leakage current. A high shunt resistance (hundreds of megaohms) indicates a high-quality photodiode with low leakage. A low shunt resistance (tens of megaohms) indicates a defective or low-quality device. |
The shunt resistance affects the TIA's input impedance. If the shunt resistance is not much higher than the feedback resistor, it shunts some of the photocurrent, reducing the gain. It also adds thermal noise. In practice, the shunt resistance of a PIN photodiode is very high (over 100 megaohms), so it is not a limiting factor. However, at high temperatures, the shunt resistance drops, and the effect becomes noticeable. |
Zebra's quality control tests the shunt resistance of every photodiode before assembly. A shunt resistance below 50 megaohms is considered a reject. This ensures that the TIA's gain is consistent across all scanners. |
|
Section 16: The Junction Capacitance - Limiting Speed |
The junction capacitance of the photodiode is the capacitance of the depletion region. It is inversely proportional to the square root of the bias voltage. The capacitance, together with the feedback resistor of the TIA, forms a low-pass filter that limits the bandwidth. For a scanner that needs a bandwidth of 100 kHz, the product of the capacitance and the resistor must be less than 1/(2*pi*100 kHz) = 1.6 microseconds. |
If the capacitance is 2 picofarads, the maximum feedback resistor is 1.6e-6 / 2e-12 = 800 kiloohms. In the LS2208, the resistor is 470 kiloohms, which gives a bandwidth of 1/(2*pi*470k*3pF) = 113 kHz. The capacitance is 3 picofarads at 2.5 volts bias, so the bandwidth is about 113 kHz - adequate. |
To increase the bandwidth, the bias voltage can be increased (lowering the capacitance), or the feedback resistor can be reduced (lowering the gain). Both options have trade-offs, and the designer must choose the best balance. |
|
Section 17: The Spectral Response - Matching the Light Source |
The photodetector's spectral response is the responsivity as a function of wavelength. Silicon photodiodes have a peak responsivity at about 900 nm (infrared), but the response drops off at shorter wavelengths. At 650 nm (red), the responsivity is about 0.5 A/W; at 450 nm (blue), it is about 0.2 A/W; at 550 nm (green), it is about 0.3 A/W. |
The scanner's illumination source must be matched to the photodetector's response. A red LED (650 nm) is a good match for a silicon photodiode. An infrared laser (850 nm) is an even better match because the responsivity is higher (0.8 A/W). However, the infrared response is not suitable for all applications, as we discussed in the previous article. |
For imaging sensors, the spectral response is often modified by a color filter array (CFA). The CFA (the Bayer pattern) allows the sensor to capture color images, but it reduces the sensitivity because each pixel only sees one color. For barcode applications, a monochrome sensor (without the CFA) is preferred because it maximizes the sensitivity. Sony and ON Semiconductor offer monochrome versions of their sensors for this reason. |
|
Section 18: The Anti-Saturation Feature - Preventing Blooming |
In an imager, when a pixel is exposed to very bright light, the photodiode's charge capacity can be exceeded. The excess electrons can spill over into neighboring pixels, creating a bright streak - a phenomenon called 'blooming.' This can obscure the barcode's edges. |
To prevent blooming, modern CMOS sensors include anti-blooming structures. In a 4T pixel, the transfer gate is used to drain excess charge to the supply rail when the pixel is reset. Some sensors also have a 'drain' transistor that can be turned on during integration to limit the charge in the photodiode. This is known as 'electronic shuttering.' |
Sony's IMX252 has an anti-blooming efficiency of greater than 99.99%, meaning that even with a 1000x overload, the blooming is negligible. This is a critical feature for scanning barcodes on shiny surfaces, where the reflection from a white space can saturate the pixel. |
|
Section 19: The Pixel Size - A Trade-Off |
In an image sensor, the pixel size is a trade-off between sensitivity and resolution. A larger pixel collects more light and has a higher signal-to-noise ratio, but it reduces the resolution (fewer pixels per millimeter). A smaller pixel gives higher resolution, but it collects less light and is noisier. |
For barcode scanning, the resolution is determined by the barcode's module size. A typical 1D barcode has a module width of 0.33 mm (13 mils). To resolve this, the pixel pitch must be less than half the module width (Nyquist criterion). For a 0.33 mm module, the pixel pitch should be less than 0.165 mm. In practice, with lens magnification, this is easily achieved. |
Most barcode imagers use a pixel size of 3-5 microns. This gives a good balance of sensitivity and resolution. The larger pixel (5 microns) is better for low-light conditions, while the smaller pixel (3 microns) is better for high-density barcodes. Honeywell's 1900 uses a 3.75-micron pixel, which is a middle ground. |
|
Section 20: The Fill Factor - Light Collection Efficiency |
The fill factor is the ratio of the photodiode's active area to the total pixel area. The rest of the pixel area is occupied by the transistors (the 4T structure) and the metal interconnects. In a front-illuminated sensor, the fill factor is typically 30-50%. In a back-illuminated sensor, the fill factor is close to 100% because the wiring is on the back side. |
The fill factor is a major advantage of back-illuminated sensors. The IMX252 has a fill factor of 85%, compared to about 45% for a front-illuminated sensor of the same pixel size. This means that the IMX252 collects almost twice as much light, giving it a significant sensitivity advantage. |
However, back-illuminated sensors are more expensive to manufacture. The silicon wafer must be thinned to a few microns, and the photodiodes are formed on the back side. This is a complex process that requires advanced equipment. For this reason, front-illuminated sensors are still used in cost-sensitive applications. |
|
Section 21: The Readout Noise - The Limit of Sensitivity |
The readout noise of an image sensor is the noise added by the on-chip amplifiers and ADCs. It is measured in electrons (RMS). A typical CMOS sensor has a readout noise of 2-5 electrons. A high-end CCD might have 1 electron of readout noise. The lower the readout noise, the better the sensor's sensitivity. |
The readout noise is a fundamental limit on the sensor's performance. To detect a barcode, the signal (the difference between white and black) must be several times the readout noise. If the readout noise is high, the sensor will require more illumination or longer exposure time. |
Sony's IMX252 has a readout noise of 2.3 electrons, which is excellent. This allows the sensor to read barcodes with a contrast ratio of only 10:1 (very low contrast). Honeywell's 1900 uses this sensor to achieve a read rate of 99.9% on a wide variety of labels. |
|
Section 22: The Analog Front-End of the Sensor - The On-Chip TIA |
Most CMOS image sensors include an on-chip analog front-end (AFE). The AFE typically consists of a programmable gain amplifier (PGA) and an analog-to-digital converter (ADC). For each column of pixels, there is a PGA that amplifies the pixel's output voltage. The PGA gain is controlled by the microcontroller, and it can be set from 1x to 16x or higher. |
The ADC is usually a successive approximation (SAR) or a delta-sigma type. The ADC's resolution is typically 10-12 bits. The output of the ADC is a digital value that is sent to the signal processor via the sensor's serial interface. |
The AFE is a critical part of the sensor's performance. It must have a low noise, a high linearity, and a fast settling time. The sensor's datasheet specifies the AFE's parameters, and the scanner designer must choose a sensor whose AFE meets the application's requirements. |
|
Section 23: The Difference Between Global and Rolling Shutter |
In a rolling shutter sensor, the pixels are exposed row by row. This is fine for still images, but for moving barcodes, the motion causes a distortion (the 'rolling shutter effect'). For barcode scanning, where the barcode is often moved past the scanner, a global shutter is preferred. |
In a global shutter sensor, all pixels are exposed at the same time. This is achieved by having a storage capacitor in each pixel (a 'shutter' transistor). The photodiode integrates the light, and then the charge is transferred to the storage capacitor simultaneously. The capacitor holds the charge while the pixels are read out. |
Sony's IMX252 has a global shutter. ON Semiconductor's AR0135 also has a global shutter. These sensors are designed specifically for industrial and barcode applications. The global shutter adds complexity to the pixel (more transistors), which increases the pixel size and reduces the fill factor. But the benefit of distortion-free capture is worth the cost. |
|
Section 24: The Photodetector as a System - Integration with the TIA |
The photodetector and the TIA form a system. The TIA's choice of feedback resistor and capacitor determines the system's gain and bandwidth. The photodetector's capacitance, dark current, and responsivity determine the system's noise performance. The two must be designed together. |
In many modern scanners, the TIA is integrated into the same package as the photodiode. This is called a 'photo IC' or 'photodiode with integrated amplifier.' The amplifier is usually a simple transimpedance amplifier with a fixed gain. The photo IC eliminates the need for a separate op-amp and the associated external components, saving space and cost. |
Zebra's SE950 uses a photo IC from Hamamatsu, which includes a PIN photodiode, a TIA, and a gain stage. The photo IC is a 5-pin device (power, ground, output, and two optional pins for gain selection). The output is a voltage that is proportional to the photocurrent. This integration simplifies the PCB layout and improves the noise performance because the TIA is very close to the photodiode. |
|
Section 25: The Temperature Compensation of the Photodetector |
As mentioned, the photodetector's dark current and responsivity change with temperature. For high-performance scanners, temperature compensation is necessary. This can be done in hardware (using a thermistor to adjust the bias voltage or the gain) or in software (using a temperature sensor to adjust the digital gain). |
Honeywell's 1900 uses a software compensation. The sensor itself has an internal temperature sensor (a diode). The microcontroller reads the temperature sensor and uses a lookup table to adjust the PGA gain. The lookup table is generated during calibration at the factory. This ensures that the scanner's sensitivity remains constant over a wide temperature range. |
Datalogic's PowerScan uses a hardware compensation. A thermistor is placed next to the photodiode, and its resistance is used to adjust the bias voltage. As the temperature rises, the bias voltage is reduced (reducing the dark current), but the responsivity also decreases. The bias adjustment is tuned to keep the product of dark current and responsivity constant, which keeps the signal-to-noise ratio stable. |
|
Section 26: The Noise Floor - The Ultimate Limit |
The noise floor of the photodetector system is the minimum signal that can be reliably detected. The noise floor is determined by the photodiode's shot noise, the TIA's thermal noise, and the readout noise of the sensor (for imagers). The noise floor sets the minimum contrast ratio that the scanner can read. |
For a typical laser scanner, the noise floor is about 1 microvolt (RMS) at the output of the TIA. This is equivalent to a photocurrent of a few nanoamperes. The scanner's signal (the difference between white and black) is typically a few hundred millivolts, so the signal-to-noise ratio is about 100,000:1 - excellent. |
For an imager, the noise floor is determined by the readout noise of the sensor. The IMX252 has a readout noise of 2.3 electrons. The full-well capacity (the maximum charge that the photodiode can hold) is about 20,000 electrons. The dynamic range is 20,000/2.3 = 8,700, or about 78 dB. This is adequate for most barcode applications. |
|
Section 27: The Optical Filter - Blocking Unwanted Light |
The photodetector is often paired with an optical filter that blocks unwanted wavelengths. For a red laser scanner, a red-pass filter (a band-pass filter centered at 650 nm) blocks the ambient light, improving the signal-to-noise ratio. For a visible-light imager, an IR-cut filter blocks the infrared light, which would otherwise add a reddish tinge to the image. |
The filter is usually a thin glass or plastic sheet that is coated with multiple layers of dielectric material. The coating creates interference effects that reflect or transmit specific wavelengths. The filter is placed directly in front of the photodetector or the image sensor. |
Zebra's LS2208 uses a red-pass filter that is integrated into the scanner window. The window is made of polycarbonate that is dyed to pass red light and block green and blue. This is a cheap filter, but it is effective. The window also has a hard coating to resist scratches. |
|
Section 28: The Micro-Lens Array - Enhancing the Fill Factor |
To improve the sensitivity of an image sensor, a micro-lens array is placed on top of the pixels. Each pixel has a tiny lens that focuses the incoming light onto the photodiode, effectively increasing the fill factor. The micro-lens is made of a transparent polymer that is formed by a reflow process. |
The micro-lens array is standard on all modern CMOS sensors. It increases the sensitivity by about 30-50%. The micro-lens is designed for a specific wavelength and a specific incident angle. For barcode scanners, the micro-lens is optimized for the illumination wavelength (e.g., 650 nm) and for the lens's chief ray angle. |
Sony's IMX252 has a micro-lens array with a quantum efficiency of 85%. Without the micro-lens, the quantum efficiency would be about 50%. The micro-lens is a critical part of the sensor's performance. |
|
Section 29: The Special Case of Color Barcodes - The Bayer Pattern |
Some barcodes are printed in color, and the scanner must distinguish between the barcode and the background. For this, a color sensor is needed. The color sensor has a Bayer pattern - a mosaic of red, green, and blue filters on top of the pixels. The sensor outputs a color image, and the signal processor separates the color channels. |
For color barcode reading, the scanner often uses the green channel, which has the highest luminance resolution. The red and blue channels are used to determine the background color. The sensor's color filter reduces the sensitivity (each pixel only sees one color), so the illumination must be brighter. |
Honeywell's 1900 has a color version (the 1900c) that uses a sensor with a Bayer pattern. The signal processor is more complex; it must demosaic the image and then apply the decoding algorithm. The color version is more expensive but is necessary for certain applications. |
|
Section 30: The Repair and Replacement of Photodetectors |
Photodetectors are generally reliable, but they can fail. The most common failure mode is a degradation of the responsivity due to moisture or contamination. This is caused by the ingress of water vapor or corrosive gases, which attack the semiconductor surface. |
To prevent this, the photodetector is hermetically sealed in a package. The package is either a metal can with a glass window, or a molded plastic with a clear epoxy. The hermetic seal is tested during manufacturing by a helium leak test. |
If a photodetector fails, it must be replaced. This is a delicate operation because the photodetector is aligned with the optics. The replacement must be done in a clean room, and the alignment must be verified with a test pattern. Some scanners have a user-replaceable photodetector module - the entire module (photodiode, TIA, and filter) can be unplugged and replaced. |
|
Section 31: The Manufacturing Test - Ensuring Quality |
Every photodetector is tested at the factory. The tests include the dark current, the responsivity, the capacitance, the shunt resistance, and the noise. The test is performed at a specified wavelength and temperature. The test data is stored in a database for traceability. |
The manufacturing test is done with a calibrated light source and a precision ammeter. The photodetector is mounted on a test fixture that simulates the scanner's optical system. The test fixture includes a lens, an aperture, and a reference barcode. The scanner's performance is measured against the reference. |
Zebra's manufacturing process includes a 'golden unit' test. A golden unit is a scanner that has been meticulously characterized and is known to perform perfectly. Every production unit is compared to the golden unit. If the production unit's read rate is more than 1% below the golden unit's, it is rejected. |
|
Section 32: The Photodetector's Linearity - A Critical Feature |
The photodetector's output must be linear - the current must be proportional to the light intensity. Non-linearity can distort the barcode signal, causing the bar widths to be measured incorrectly. The linearity of a PIN photodiode is typically excellent (better than 0.1%) over a wide range of intensities. |
The TIA also must be linear. The TIA's linearity is determined by the op-amp's open-loop gain and the feedback resistor. A high open-loop gain (greater than 100 dB) ensures linearity. The LM358, a common op-amp, has an open-loop gain of 100 dB, so it is linear. |
The combination of a linear photodiode and a linear TIA ensures that the signal is a faithful reproduction of the light pattern. This is essential for accurate decoding. |
|
Section 33: The Photodetector in the System - A Holistic View |
The photodetector is not a standalone component. It is part of a system that includes the illumination, the optics, the TIA, and the signal processor. The system's performance is determined by the weakest link. |
If the illumination is too dim, the photodetector's output is small, and the SNR is poor. If the optics are misaligned, the light does not reach the photodetector efficiently. If the TIA is noisy, it adds noise to the signal. If the signal processor has a poor algorithm, it cannot decode the signal. |
The system designer must consider all of these factors. The photodetector is chosen to match the illumination and the optics. The TIA is designed to match the photodetector's capacitance and dark current. The signal processor is designed to handle the expected SNR. The system works when all the parts are in harmony. |
|
Section 34: The Future - Quantum Dots and Organic Photodetectors |
The photodetector technology is evolving. New materials, like quantum dots and organic semiconductors, are being researched. These materials can be tuned to specific wavelengths, and they can be deposited on flexible substrates. |
Quantum dot photodetectors are made of nanocrystals of lead sulfide or lead selenide. They have a high responsivity and can be tuned to any wavelength by changing the size of the quantum dots. They are still in the research stage, but they could eventually replace silicon photodiodes. |
Organic photodetectors are made of carbon-based polymers. They are flexible, cheap, and can be printed on plastic. They are already used in some low-cost optical sensors. For barcode scanners, they could enable flexible, wearable scanners. |
|
Section 35: The Environmental Impact - Disposal and Recycling |
Photodetectors contain small amounts of toxic materials, such as arsenic and gallium (in some compound semiconductors). The disposal of these devices is regulated by environmental laws. The scanner manufacturer must provide instructions for the proper disposal of the photodetector. |
Most photodetectors are made of silicon, which is non-toxic. The packaging materials (metal, glass, and plastic) are also recyclable. The scanner can be recycled by separating the electronic components from the plastic housing. The photodetector can be incinerated to recover the silicon. |
Zebra and Honeywell have programs to take back their scanners for recycling. The scanners are disassembled, and the components are separated. The photodetectors are sent to a specialist recycler, where the silicon is recovered. |
|
Section 36: The Final Word - The Photodetector's Quiet Greatness |
The photodetector is the unsung hero of the barcode scanner. It does not glow like the LED, and it does not compute like the processor. It simply sits in the shadows, converting a faint whisper of light into a meaningful electrical signal. But without it, the scanner would be blind. |
The photodetector's design is a fascinating blend of physics, material science, and circuit design. The choice of the device, its bias, its temperature compensation, and its integration with the TIA are all critical decisions that define the scanner's performance. The major companies - Zebra, Honeywell, Datalogic, Omron, Keyence, Cognex, Sony, and Hamamatsu - have all made their mark by optimizing the photodetector for their specific applications. |
We have seen how a simple PIN photodiode, carefully biased and amplified, can create a scanner that reads billions of barcodes a year. We have seen how a phototransistor can reduce cost for simple applications. We have seen how an avalanche photodiode can push the limits of range and sensitivity. And we have seen how a sophisticated CMOS pixel, with its global shutter and back-illuminated structure, can capture the most challenging 2D codes. |
The photodetector is, in many ways, a metaphor for engineering itself - a small, often overlooked component that performs a vital function, and whose design requires a deep understanding of the system as a whole. The next time you scan a barcode, remember that the tiny silicon chip inside the scanner is listening, with incredible patience and precision, to the light that tells the story of the product. |
|
Detailed Final Summary (approximately 1200 words) |
We have now explored the photodetector - the electronic retina - in exhaustive detail. Let us recapitulate the essential points, the design trade-offs, and the real-world examples that define this critical component. |
Fundamental Function |
The photodetector converts light (photons) into electricity (electrons). The amount of current is proportional to the light intensity, but the proportionality constant (responsivity) depends on the wavelength and the device structure. The photodetector's output is a fragile current, typically in the nanoampere range, which must be amplified by a transimpedance amplifier (TIA) before it can be used. |
Types of Photodetectors |
PIN Photodiode: The most common type for laser scanners. It has a low capacitance (2-5 pF), a fast response time (nanoseconds), and a moderate dark current (1-5 nA). Responsivity at 650 nm is about 0.5 A/W. Examples: OSRAM SFH 203 P, used in Symbol's LS2208. |
Phototransistor: Higher sensitivity (1-10 A/W) but slower (microseconds) and higher capacitance. Used in low-cost, low-speed scanners, such as those from Omron. The transistor's gain amplifies the photocurrent, but the slow speed limits the scan rate. |
Avalanche Photodiode (APD): Extremely sensitive due to internal gain (10-100x). Requires a high bias voltage (100-200 V) and a precise, low-noise supply. Used in long-range scanners, such as Zebra's DS3500. Hamamatsu is a leading supplier. |
CMOS Pixel: The photodetector in an imager. Each pixel is a tiny photodiode with a transfer gate, a reset transistor, and a source-follower. The pixel's output is a voltage that represents the integrated charge. Sony's IMX252 and ON Semiconductor's AR0135 are common examples. Back-illuminated (BSI) pixels have a higher fill factor and sensitivity. |
Linear CCD: A legacy technology but still used in some high-end scanners. A row of photodiodes with a serial charge transfer. Excellent noise performance but higher cost and power consumption. Honeywell's 4600 series used a CCD. |
|
Key Parameters |
Responsivity: The output current per unit of optical power (A/W). It depends on wavelength. Silicon peaks at 900 nm (0.8 A/W) and is lower at 650 nm (0.5 A/W) and 450 nm (0.2 A/W). |
Dark Current: The current that flows in total darkness. It doubles with every 10-degree Celsius rise. It adds shot noise. Low dark current is desirable. |
Junction Capacitance: The capacitance of the depletion region. It limits the speed of the TIA. Lower capacitance (achieved by higher bias or smaller area) is better for speed. |
Shunt Resistance: The leakage resistance of the junction. A high value (>> feedback resistor) is needed to maintain gain. |
Readout Noise: For CMOS sensors, the noise added by the on-chip amplifiers and ADCs. Typically 2-5 electrons RMS. |
Fill Factor: The ratio of the photodiode area to the pixel area. Higher fill factor (BSI sensors) gives higher sensitivity. |
Biasing and Compensation |
- The photodiode is reverse-biased to reduce capacitance and increase speed. The bias voltage is typically 2-5 V for handheld scanners. |
- The bias supply must be clean; a ripple on the bias modulates the capacitance and dark current. |
- Temperature compensation is essential. Honeywell uses software compensation (adjusting the PGA gain based on a temperature sensor). Datalogic uses hardware compensation (adjusting the bias voltage with a thermistor). |
|
Integration with the TIA |
- The photodiode and the TIA form a system. The TIA's feedback resistor and capacitor are chosen to match the photodiode's capacitance and the desired bandwidth. |
- A larger feedback resistor gives higher gain but lower bandwidth and higher thermal noise. A smaller feedback resistor gives lower gain but higher bandwidth. |
- The product of the photodiode's capacitance and the feedback resistor determines the bandwidth. |
- Some scanners use a 'photo IC' - an integrated package that includes the photodiode and the TIA. Zebra's SE950 uses such a device from Hamamatsu. |
Optical Filters and Micro-Lenses |
- An optical filter blocks unwanted wavelengths. A red-pass filter for a red laser, or an IR-cut filter for a visible imager. |
- A micro-lens array on a CMOS sensor focuses light onto the photodiode, increasing the effective fill factor. Sony's IMX252 has an 85% quantum efficiency with micro-lenses. |
Global vs. Rolling Shutter |
- For moving barcodes, a global shutter (all pixels exposed at the same time) is essential. Sony's IMX252 and ON Semi's AR0135 have global shutters. |
- A rolling shutter (row-by-row exposure) causes distortion for moving objects. |
Color Sensing |
- A color sensor uses a Bayer pattern (red, green, blue filters). The green channel is often used for decoding. The color version is more expensive and less sensitive. |
|
Real-World Company Examples |
Symbol (Zebra): LS2208 uses a PIN photodiode (OSRAM SFH 203 P) with a 2.5 V bias. The TIA uses a 470-kiloohm resistor. Simple, reliable, and cost-effective. |
Zebra: DS3500 uses an APD from Hamamatsu for long-range reading. The APD requires a 150-V bias supply with a low-noise linear regulator. The TIA uses a smaller feedback resistor (10-50 kohm) because the APD provides gain. |
Honeywell: 4600 series used a linear CCD. 1900 series uses a CMOS sensor (likely Sony IMX252) with a global shutter and BSI. The sensor's PGA gain is adjusted via firmware for temperature compensation. |
Datalogic: PowerScan uses an APD for long-range and an infrared laser. The scanner includes a hardware temperature compensation circuit that adjusts the bias voltage. |
Omron: Entry-level scanners use a phototransistor, which is cheaper but slower. The phototransistor is used in a common-emitter configuration with a simple load resistor. |
Sony: IMX252 is a 2.4-MP, global-shutter, BSI sensor with a readout noise of 2.3 electrons. Used in high-end imagers. |
ON Semiconductor: AR0135 is a 1.2-MP, global-shutter, front-illuminated sensor with a 3.75-micron pixel. Used in industrial imagers. |
Hamamatsu: Supplier of APDs for long-range scanners. Their S10912 series has a gain of 30 at 150 V and a dark current of 5 nA. |
Environmental and Reliability Considerations |
- Photodetectors are hermetically sealed to prevent moisture ingress. The seal is tested with a helium leak test. |
- The photodetector's dark current and responsivity degrade with temperature and age. Temperature compensation mitigates this. |
- The photodetector is tested at the factory for dark current, responsivity, capacitance, and shunt resistance. A 'golden unit' comparison ensures quality. |
Future Trends |
- Quantum dot photodetectors offer tunable wavelength sensitivity and high responsivity. |
- Organic photodetectors are flexible and cheap, enabling wearable scanners. |
- Back-illuminated sensors with smaller pixels and higher resolution will continue to improve image quality. |
- Integration of the photodetector, TIA, and ADC on a single chip (system-on-chip) will reduce size and cost. |
|
Final Thoughts |
The photodetector is the quiet heart of the barcode scanner. Its design requires a deep understanding of semiconductor physics, circuit design, and system-level optimization. The major companies have each found their own sweet spot: Zebra with a simple PIN photodiode for retail, an APD for long-range; Honeywell with a large-area photodiode and on-chip dark-current reference; Datalogic with a dual-photodiode configuration for noise cancellation; Omron with a phototransistor for cost-cutting; and Sony with a sophisticated BSI CMOS sensor for high-end imaging. |
The photodetector's performance is not just about the device itself; it is about how it interacts with the illumination, the optics, the TIA, and the signal processor. A poorly chosen photodetector can ruin an otherwise excellent scanner, while a well-chosen one can elevate a mediocre design to greatness. |
As the technology evolves, the photodetector will become more integrated, more sensitive, and more intelligent. But its fundamental role - to listen to the light and convert it into a signal - will remain unchanged. And that is a role that deserves our respect and admiration. |
cs@easiersoft.com