Gaining the Whisper: How the Second-Stage Amplifier Turns Millivolts into Meaning |
Subtitle: A Deep Dive into the Gain Stage, Its Design, Its Trade-Offs, and Its Real-World Implementation - with Examples from Symbol, Zebra, Honeywell, Datalogic, Texas Instruments, and Analog Devices |
|
Opening Summary |
After the transimpedance amplifier (TIA) has converted the photodetector's tiny current into a voltage, and after the AC coupling network has removed the unwanted DC offset, the barcode signal is still a fragile thing. It is typically only a few millivolts in amplitude - a whisper that is easily lost in the noise. The next stage of the signal chain is the amplification stage, whose job is to raise this whisper to a robust level that can drive a comparator or an analog-to-digital converter. This is the gain stage, and it is the second hero of the barcode scanner. |
This article is dedicated to the gain stage - its design, its optimization, and its critical role in the overall system. We will explore the different topologies used for amplification, from the simple non-inverting amplifier to more sophisticated variable-gain and automatic-gain-control (AGC) circuits. We will examine the key trade-offs: gain versus noise, bandwidth versus stability, and cost versus performance. We will look at how major companies have implemented the gain stage in their products. We will see how Symbol (now Zebra) used a fixed-gain amplifier in the LS2208, and how they optimized the gain for the typical scanning conditions. We will explore Honeywell's use of a programmable-gain amplifier (PGA) in their imagers, allowing the scanner to adapt to different label contrasts. We will examine Datalogic's AGC circuit, which automatically adjusts the gain to keep the signal at a constant level, and we will look at how Texas Instruments and Analog Devices provide reference designs with carefully chosen gain stages. |
By the end of this journey, you will understand that the gain stage is not just a simple amplifier but a carefully engineered block that must balance competing demands. You will see how the choice of gain, the selection of the op-amp, and the design of the feedback network all contribute to the scanner's ability to read barcodes quickly, accurately, and reliably. |
|
Full Article |
Section 1: The Need for Gain - From Millivolts to Volts |
The output of the AC coupling network is a signal that swings around zero volts. Its amplitude is the peak-to-peak voltage of the barcode's black-to-white transitions. For a typical scanner, this amplitude is between 10 and 100 millivolts. This is too small to be used directly by a comparator (which needs a few hundred millivolts to switch reliably) or by an ADC (which needs a signal that spans most of its input range). The signal must be amplified by a factor of 50 to 100, bringing it up to 1 to 5 volts. |
The gain stage is the circuit that provides this amplification. It is usually a non-inverting amplifier based on an operational amplifier (op-amp). The gain is set by two resistors: a feedback resistor and a ground resistor. The gain is (1 + Rf/Rg), where Rf is the feedback resistor and Rg is the ground resistor. For a gain of 50, Rf/Rg = 49. For example, Rf = 100 kilohms and Rg = 2 kilohms gives a gain of 51. |
The gain stage is placed after the AC coupling network and before the comparator or the ADC. Its output is a larger version of the input signal, with the same waveform shape but with a much larger amplitude. |
|
Section 2: The Non-Inverting Amplifier - The Classic Topology |
The non-inverting amplifier is the most common topology for the gain stage. It has several advantages: high input impedance (which does not load the AC coupling network), low output impedance (which can drive the comparator or ADC), and a gain that is set by the ratio of two resistors, making it easy to adjust. |
The non-inverting amplifier uses an op-amp with the input signal applied to the non-inverting input. The feedback resistor (Rf) is connected from the output to the inverting input. The ground resistor (Rg) is connected from the inverting input to ground. The gain is 1 + Rf/Rg. |
The op-amp must be chosen carefully. It must have a low input bias current (to avoid loading the AC coupling network), a low noise (to avoid degrading the signal), and a sufficient bandwidth (to pass the barcode signal without distortion). The op-amp must also be stable at the gain and with the capacitive load of the comparator or ADC. |
|
Section 3: Symbol's LS2208 - A Fixed Gain of 68 |
Symbol's LS2208 uses a gain stage with a fixed gain of 68. The op-amp is the TLV2371 (the same op-amp used in the TIA). The feedback resistor (Rf) is 100 kilohms, and the ground resistor (Rg) is 1.5 kilohms. The gain is 1 + 100/1.5 = 67.7. |
The 68x gain is chosen to bring the typical TIA output (about 20 millivolts peak-to-peak) up to about 1.36 volts peak-to-peak. This is a suitable level for the comparator. The comparator's adaptive threshold is set to the midpoint of this signal. |
The LS2208's gain stage is a simple, robust design. The TLV2371 op-amp is cheap and readily available. The 100-kilohm and 1.5-kilohm resistors are standard values. The gain stage has been proven in millions of units. |
|
Section 4: The Gain-Bandwidth Trade-Off |
The gain stage's bandwidth is determined by the op-amp's gain-bandwidth product (GBW) and the gain itself. The closed-loop bandwidth is GBW / gain. For the TLV2371, the GBW is 3 MHz. With a gain of 68, the bandwidth is about 44 kHz. This is sufficient for a hand-held scanner (which needs about 100 kHz). The bandwidth is reduced by the gain, but 44 kHz is still acceptable. |
If a higher bandwidth is needed, a higher GBW op-amp can be used. For example, the AD8615 has a GBW of 10 MHz. With a gain of 68, the bandwidth is 147 kHz, which is more than enough. However, the AD8615 is more expensive than the TLV2371. |
The designer must balance the need for bandwidth against the cost. For most handheld scanners, a 3-MHz GBW op-amp is sufficient. For high-speed industrial scanners, a higher GBW is needed. |
|
Section 5: The Noise in the Gain Stage |
The gain stage amplifies not only the signal but also the noise. The noise sources are the op-amp's input voltage noise and the thermal noise of the resistors. The total noise at the output is the input noise multiplied by the gain, plus the noise contribution of the resistors. |
For the LS2208, the TLV2371 has an input voltage noise of 19 nV/(v)Hz. With a gain of 68, the output noise from the op-amp is 19 * 68 = 1.29 uV/(v)Hz. The thermal noise of the 100-kilohm feedback resistor is about 40 nV/(v)Hz. The thermal noise of the 1.5-kilohm ground resistor is about 5 nV/(v)Hz. The total noise is dominated by the op-amp's noise. The output noise is about 1.3 uV/(v)Hz. With a bandwidth of 44 kHz, the RMS noise is about 1.3 * sqrt(44,000) = 273 uV. This is small compared to the signal (1.36 volts), so the SNR is excellent. |
|
Section 6: Honeywell's Programmable Gain Amplifier (PGA) |
Honeywell's 1900 imager uses a programmable gain amplifier (PGA) in the sensor's analog front-end. The PGA is integrated into the CMOS image sensor chip. The gain can be set to one of several values, from 1x to 16x, in binary steps (1, 2, 4, 8, 16). The gain is controlled by the microcontroller via an I2C interface. |
The PGA is a switched-capacitor amplifier. The gain is determined by the ratio of the input capacitor to the feedback capacitor. The capacitors are switched in and out to change the gain. The PGA is a differential amplifier, preserving the differential nature of the sensor's output. |
The programmable gain allows the 1900 to adapt to different lighting conditions and label contrasts. For a high-contrast label, a low gain is used. For a low-contrast label, a high gain is used. The scanner automatically adjusts the gain during a pre-scan, ensuring that the signal is always at the optimal level for the ADC. |
|
Section 7: Datalogic's Automatic Gain Control (AGC) |
Datalogic's high-end scanners use an automatic gain control (AGC) circuit. The AGC continuously adjusts the gain to keep the signal amplitude constant. This is a feedback loop: the signal amplitude is measured, and the gain is adjusted to maintain a target amplitude. |
The AGC is implemented with a variable-gain amplifier (VGA) and a peak detector. The VGA is an amplifier whose gain is controlled by a voltage (the control voltage). The peak detector measures the peak amplitude of the VGA's output. The control voltage is generated by an integrator that compares the peak detector's output to a reference voltage. |
The AGC provides several benefits. It ensures that the signal amplitude is constant, regardless of the label's contrast or the scanning distance. This simplifies the design of the comparator and the decoder. It also extends the scanner's dynamic range. |
|
Section 8: The Variable-Gain Amplifier (VGA) - An Analog Multiplier |
The VGA is the heart of the AGC. The VGA can be implemented in several ways. One common method is to use an analog multiplier (such as the AD633) in the feedback path of an op-amp. The gain is controlled by the multiplier's control voltage. |
Another method is to use a transconductance amplifier (OTA) whose transconductance is controlled by a bias current. The bias current is controlled by the control voltage. This is a more integrated solution. |
Datalogic's AGC uses a VGA based on a Gilbert cell multiplier. The Gilbert cell is a classic analog multiplier circuit that is integrated into a single chip. The VGA has a gain range of 0 to 40 dB (1x to 100x). |
|
Section 9: The Peak Detector - Measuring the Signal Amplitude |
The peak detector is a circuit that measures the peak amplitude of the signal. It consists of a diode and a capacitor. The diode charges the capacitor to the peak voltage of the signal. The capacitor holds the peak voltage until the next cycle. |
The peak detector must have a fast response to track the signal's peaks, but it must also have a slow discharge to hold the peak value. The discharge time is set by a resistor in parallel with the capacitor. The time constant must be much longer than the period of the signal but much shorter than the changes in the signal amplitude. |
The peak detector's output is a DC voltage that represents the signal's peak amplitude. This voltage is used as the feedback signal for the AGC. |
|
Section 10: Zebra's DS3500 - A Two-Stage Gain |
Zebra's DS3500, the long-range scanner, uses a two-stage gain. The first stage is the TIA, which we have already discussed. The second stage is the gain stage, which provides an additional gain of 50. The total gain (TIA + gain stage) is about 5,000,000 (100 kilohms * 50). This is a high gain, needed for the very weak signals from long-range scanning. |
The gain stage uses a non-inverting amplifier with a gain of 50. The op-amp is the AD8615, which has a low noise and a high GBW. The feedback resistor is 100 kilohms, and the ground resistor is 2 kilohms. The gain is 1 + 100/2 = 51. |
The two-stage gain provides high gain without compromising the stability. The TIA provides the initial gain and bandwidth; the gain stage provides the additional gain. |
|
Section 11: The Gain Stage and the Comparator's Input Range |
The comparator has a limited input range. It can only accept signals that are within its power supply range (e.g., 0 to 5 volts). The gain stage's output must be within this range. If the output exceeds the range, the signal is clipped, and the barcode edges are distorted. |
To avoid clipping, the gain is set so that the maximum output voltage is just below the comparator's supply voltage. For a 5-volt supply, the maximum output is about 4.5 volts. The gain is set so that the maximum signal (the white level) is at 4.5 volts, and the minimum signal (the black level) is at a few hundred millivolts. |
The LS2208's gain of 68 gives a maximum output of about 1.36 volts (for a typical signal). This is well within the 5-volt supply. The gain could be increased, but then the signal would be closer to the supply rails, and the risk of clipping would be higher. |
|
Section 12: The Gain Stage and the ADC's Input Range |
In an imager, the gain stage's output is fed to an ADC. The ADC has a fixed input range (e.g., 0 to 3.3 volts). The gain is set so that the signal's peak-to-peak swing spans most of this range, maximizing the ADC's resolution. |
The PGA in Honeywell's 1900 adjusts the gain so that the signal's peak-to-peak swing is about 2.5 volts. The ADC's input range is 3.3 volts, so the signal uses about 75% of the range. This is a good compromise: it provides enough headroom for the signal to vary without clipping, and it uses enough of the ADC's range to provide good resolution. |
|
Section 13: The Gain Stage and the Signal-to-Noise Ratio |
The gain stage amplifies the signal and the noise equally. The SNR at the output is the same as the SNR at the input, assuming the gain stage itself adds negligible noise. The SNR is determined by the TIA and the photodetector. |
The gain stage does not improve the SNR. It only increases the signal's amplitude. The SNR is set by the earlier stages. This is why the TIA and the photodetector are so critical. |
If the gain stage is noisy, it can degrade the SNR. This is why a low-noise op-amp is used for the gain stage. The TLV2371 and the AD8615 are both low-noise op-amps. |
|
Section 14: The Gain Stage and the Power Supply |
The gain stage's power supply must be clean. Any noise on the supply is added to the signal. The gain stage is usually powered by a linear regulator (LDO) after the main switching regulator. The LDO provides a clean, low-noise supply. |
The gain stage's power supply rejection ratio (PSRR) is important. The PSRR is the op-amp's ability to reject supply noise. The TLV2371 has a PSRR of 80 dB at 100 kHz. This is excellent. The AD8615 has a PSRR of 100 dB. |
|
Section 15: The Gain Stage and the Temperature |
The gain stage's gain is determined by the ratio of two resistors. The resistors have a temperature coefficient (TC). If the TC of the two resistors is different, the gain will change with temperature. To minimize the temperature drift, the resistors are chosen to have the same TC (e.g., both are 50 ppm/C metal film resistors). |
The op-amp's gain also drifts with temperature, but the gain of the non-inverting amplifier is determined by the resistor ratio, not the op-amp's gain. The op-amp's drift is negligible. |
|
Section 16: The Gain Stage in Texas Instruments' TIDA-00857 |
Texas Instruments' TIDA-00857 reference design uses a gain stage with a gain of 68. The op-amp is the TLV272 (a dual version of the TLV2371). The feedback resistor is 100 kilohms, and the ground resistor is 1.5 kilohms. The gain is 1 + 100/1.5 = 67.7. |
The TIDA-00857's gain stage is virtually identical to the LS2208's. This is a testament to the robustness and simplicity of the design. The gain stage is a standard, proven circuit. |
|
Section 17: The Gain Stage in Analog Devices' Reference Design |
Analog Devices' reference design uses a gain stage with a gain of 100. The op-amp is the AD8615. The feedback resistor is 100 kilohms, and the ground resistor is 1 kilohm. The gain is 1 + 100/1 = 101. |
The higher gain is used because the reference design is intended for a wider range of applications. The AD8615's low noise allows the high gain without degrading the SNR. The reference design also includes a baseline restorer. |
|
Section 18: The Gain Stage and the Capacitive Load |
The gain stage may need to drive a capacitive load, such as the input of an ADC. A capacitive load can cause the op-amp to oscillate. To prevent oscillation, a small resistor (e.g., 100 ohms) is often placed in series with the output. The resistor isolates the op-amp from the capacitive load. |
The resistor does not affect the gain, but it does create a voltage drop if the load draws current. The voltage drop is small (a few millivolts) and can be neglected. |
|
Section 19: The Gain Stage and the DC Offset |
The gain stage amplifies any residual DC offset that is present at its input. The residual offset comes from the AC coupling network, which may not completely remove the DC component. The amplified DC offset can cause the comparator to trigger incorrectly. |
To remove the residual DC offset, the gain stage is often AC-coupled as well. A second AC coupling capacitor is placed at the output of the gain stage. This capacitor blocks the amplified DC offset. The second AC coupling is common in laser scanners. |
The LS2208 does not have a second AC coupling. The gain stage's output goes directly to the comparator. The comparator's adaptive threshold tracks the average signal level, so the residual DC offset is not a problem. |
|
Section 20: The Gain Stage and the Barcode's Dynamic Range |
The barcode's dynamic range is the ratio of the maximum to the minimum signal amplitude. A high-contrast label has a large dynamic range; a low-contrast label has a small dynamic range. The gain stage must be able to handle the full dynamic range without clipping or losing the signal. |
The gain stage's linearity is important. The amplifier must be linear over the entire dynamic range. The TLV2371 and the AD8615 are both linear over a wide range. |
The AGC, used by Datalogic, automatically adjusts the gain so that the signal always uses the full dynamic range of the comparator or ADC. This is the most flexible solution. |
|
Section 21: The Gain Stage and the Scanning Speed |
The scanning speed affects the signal's frequency. A faster scan produces a higher frequency. The gain stage must have a bandwidth that is sufficient for the highest frequency. The bandwidth is determined by the op-amp's GBW and the gain. |
The LS2208's gain stage has a bandwidth of 44 kHz, which is sufficient for the typical hand-scanning speed. For a faster scan, a higher GBW op-amp (e.g., the AD8615) with a lower gain would be needed. |
|
Section 22: The Gain Stage and the Comparator's Hysteresis |
The comparator's hysteresis is a voltage offset that prevents oscillation. The hysteresis is typically set to 50-100 millivolts. The gain stage's output must be larger than the hysteresis. The gain is set so that the signal's amplitude is at least 10 times the hysteresis. |
For the LS2208, the signal amplitude is 1.36 volts, and the hysteresis is 50 millivolts. The ratio is 27, which is more than enough. |
|
Section 23: The Gain Stage and the ADC's Resolution |
The ADC's resolution is the number of bits it has. A 10-bit ADC has 1024 levels. The gain stage's output must use as many of these levels as possible to maximize the resolution. The gain is set so that the signal's peak-to-peak swing spans most of the ADC's range. |
In Honeywell's 1900, the PGA adjusts the gain so that the signal uses about 75% of the ADC's range. This provides good resolution while leaving some headroom. |
|
Section 24: The Gain Stage and the Microcontroller's Input |
In some scanners, the gain stage's output goes directly to a microcontroller's analog input. The microcontroller's ADC is used to digitize the signal. The gain must be set to match the microcontroller's ADC input range. |
The microcontroller's ADC input range is usually the same as the supply voltage (e.g., 3.3 or 5 volts). The gain is set so that the signal's peak-to-peak swing is close to the full range. |
Section 25: The Gain Stage and the Test Points |
The gain stage often has test points (e.g., a test pad on the PCB). The test points allow the technician to measure the signal with an oscilloscope. The measurements are used to verify the gain and to troubleshoot the scanner. |
The test points are usually accessible through a small hole in the scanner's housing. The technician can insert a probe and measure the signal. |
|
Section 26: The Gain Stage and the ESD Protection |
The gain stage's input is protected against ESD. The protection is usually a pair of clamping diodes, similar to the protection on the TIA. The diodes shunt the ESD current to the supply rails. |
The protection diodes must have a low leakage current. A high leakage current would create a DC offset at the gain stage's input. The offset would be amplified and could cause problems. |
Section 27: The Gain Stage and the PCB Layout |
The PCB layout of the gain stage is important. The feedback path must be kept short to minimize parasitic capacitance. The input and output traces must be routed away from high-current digital traces. A ground plane is used to provide a low-impedance return path. |
The layout of the gain stage is less critical than the layout of the TIA, but it is still important. A poor layout can cause oscillation or noise pickup. |
|
Section 28: The Gain Stage and the Power-Up Sequencing |
The gain stage's power-up sequence is important. The op-amp must be powered up before the signal is applied. If the signal is applied before the op-amp is powered, the op-amp's internal protection diodes can be damaged. |
The power-up sequence is controlled by the microcontroller. The microcontroller enables the regulator that powers the gain stage, and it waits for the regulator to settle before enabling the TIA. |
Section 29: The Gain Stage and the Output Impedance |
The gain stage has a low output impedance (typically less than 100 ohms). This is important for driving the comparator or the ADC. A low output impedance ensures that the signal is not loaded by the input impedance of the next stage. |
The low output impedance is provided by the op-amp. The op-amp's output is a low-impedance source. |
|
Section 30: The Gain Stage and the Input Impedance |
The gain stage has a high input impedance (typically greater than 1 megaohm). This is important for not loading the AC coupling network. The high input impedance is provided by the op-amp's non-inverting input. |
The AC coupling network's output impedance is the resistor (e.g., 10 kilohms). The gain stage's input impedance is much higher, so the loading is negligible. |
Section 31: The Gain Stage and the Slew Rate |
The slew rate is the op-amp's maximum rate of change of the output voltage. It is measured in volts per microsecond. The slew rate must be high enough to handle the fastest signal changes. |
For the LS2208, the signal changes at a rate of about 1.36 volts per millisecond (for a 1-kHz signal). The slew rate is 1.36 V/ms = 1.36 mV/us. The TLV2371 has a slew rate of 2 V/us, which is much higher than needed. |
|
Section 32: The Gain Stage and the Settling Time |
The settling time is the time it takes for the amplifier's output to settle to within a small percentage of its final value after a step change. The settling time is determined by the op-amp's bandwidth and slew rate. |
For the LS2208, the settling time is about 5-10 times the time constant of the amplifier. The time constant is determined by the gain and the GBW. The settling time is about 10 microseconds, which is much shorter than the barcode's pulse width (hundreds of microseconds). |
Section 33: The Gain Stage and the Overload Recovery |
The gain stage can be overloaded if the input signal is too large. The output hits the supply rail. When the input signal returns to normal, the gain stage must recover from the saturation. The recovery time can be much longer than the settling time. |
To reduce the overload recovery time, the op-amp may include a clamp that limits the output voltage. The TLV2371 does not have a clamp, but the overload recovery time is still short (about 100 microseconds). |
|
Section 34: The Gain Stage and the Multi-Channel System |
Some scanners have multiple photodetectors and multiple gain stages (e.g., a linear imager with a row of photodiodes). The gain stages must be matched - their gains must be very similar. The matching is achieved by using integrated circuits with precise resistor ratios. |
The matching is important for the scanner's accuracy. If the gain stages are not matched, the signal from each photodiode will be amplified differently, causing errors in the barcode decoding. |
Section 35: The Gain Stage and the Diagnostic Tests |
The gain stage can be tested by injecting a known test signal into its input and measuring the output. The test signal is generated by a digital-to-analog converter (DAC) in the microcontroller. The measured output is compared to the expected output. If the output is out of range, the scanner signals an error. |
The diagnostic test can detect a faulty op-amp, a broken feedback resistor, or a shorted ground resistor. |
|
Section 36: The Gain Stage - A Summary of Best Practices |
Based on our exploration, let us summarize the best practices for designing the gain stage in a barcode scanner: |
1. Determine the Required Gain: The gain must be high enough to bring the signal to a level that is usable by the comparator or ADC, but not so high that the signal clips. The gain is typically 50 to 100. |
2. Choose the Op-Amp: The op-amp must have a low input bias current, low noise, sufficient bandwidth (GBW/gain > signal frequency), and a high slew rate. The TLV2371 is a good choice for most applications; the AD8615 is a better choice for high-performance designs. |
3. Set the Gain with Two Resistors: The non-inverting amplifier topology uses a feedback resistor (Rf) and a ground resistor (Rg). The gain is 1 + Rf/Rg. Choose Rf and Rg values that are practical and that have low temperature coefficients. |
4. Consider a Programmable Gain: For scanners that need to adapt to different conditions, a programmable gain amplifier (PGA) or an automatic gain control (AGC) is beneficial. Honeywell and Datalogic use these techniques. |
5. Protect Against ESD: Use clamping diodes to protect the gain stage's input. |
6. Layout the PCB Carefully: Keep the feedback path short, use a ground plane, and route the sensitive traces away from the digital traces. |
7. Test the Gain Stage: Verify the gain, the bandwidth, the noise, and the settling time. |
|
Final Summary |
The gain stage is the second hero of the barcode scanner. It takes the fragile, millivolt-level signal from the AC coupling network and amplifies it to a robust, volt-level signal that can drive the comparator or the ADC. The gain stage is a critical component that must be designed carefully to balance gain, noise, bandwidth, and stability. |
We have seen how major companies have implemented the gain stage in their products. Symbol's LS2208 uses a simple, fixed-gain non-inverting amplifier with a gain of 68. Honeywell's 1900 imager uses a programmable gain amplifier (PGA) that adapts to different label contrasts. Datalogic's high-end scanners use an automatic gain control (AGC) that continuously adjusts the gain to maintain a constant signal amplitude. Zebra's DS3500 uses a two-stage gain to achieve a very high total gain for long-range scanning. Texas Instruments and Analog Devices provide reference designs with carefully chosen gain stages. |
The gain stage is a testament to the power of analog design. It is a simple circuit - an op-amp and two resistors - but its design requires a deep understanding of the system's requirements. The gain stage is the bridge between the fragile signal of the photodetector and the robust digital world of the comparator and the microcontroller. It is the second hero that ensures that the barcode's message is not lost in the noise. |
cs@easiersoft.com