Although advances in photonics and electronics are making it easier and cheaper to implement a light detection and ranging (LIDAR) system, the analog front-end receiver interface still presents many challenges for system designers, particularly when targeting modern high-speed solutions. Key among these challenges are issues arising from parameters such as dynamic range, noise, and stability. While there are many potential pitfalls in photodiode front-end circuit design, though, with proper knowledge and planning, it is often possible to find solutions for even the most challenging design requirements.
One of the most common LIDAR implementations uses a photodiode to receive the optical signal and then amplifies the photodiode’s output current through a transimpedance amplifier stage. Figure 1 shows an example block diagram of a photodiode-based LIDAR receiver that represents most of the circuits I’ll discuss here. While the theory of converting a photodiode current into a voltage is fairly simple, many challenges arise when designing the photodiode amplifier interface, particularly for unpredictable environments such as an automobile. In this article, we’ll discuss the mitigation of dynamic range, noise, and a few other issues that arise from photodiode front-end design, with a particular focus on the transimpedance amplifier stage.
Figure 1 This photodiode receiver block diagram represents most of the circuits used in LIDAR front ends.
For photodiode front ends, maximizing the dynamic range is often one of the largest issues for designers. Designs typically require as much gain as possible to resolve the smallest input signals, but designs also need to be able to function in the presence of inputs that could be several orders of magnitude larger. In LIDAR systems, this significant variation in signal amplitude results from the fact that optical signals in air reduce in amplitude by the square of the distance traveled. One of the best solutions to this problem is to use a logarithmic transimpedance amplifier, which is an amplifier where the gain changes logarithmically with input amplitude. Unfortunately, the bandwidth requirements of modern LIDAR applications often eliminate logarithmic amplifiers as a viable solution, requiring the use of linear amplifiers.
When facing dynamic range issues, designers may first try to design the circuit for their smallest expected input current, because this represents the furthest distance a LIDAR system can measure. However, it then becomes possible that the largest possible input current, such as from a zero distance reflection, will cause a very large output voltage that often saturates the system’s transimpedance amplifier. Some systems may be able to tolerate amplifier saturation, but the amplifier will take a significant amount of time to recover its output after saturation. As illustrated in the simplified example of Figure 2, a very large input current can saturate an amplifier output voltage and cause it to be “blind” to other input signals during its recovery time.
Figure 2 A saturated signal can “blind” a system to other inputs for a time.
There are several ways to alleviate issues from amplifier saturation, and each method has various benefits and drawbacks. These options include:
- Decreasing the system gain – This is often not possible, however, because it will limit the smallest measurable input signal.
- Using an amplifier with a higher output swing capability – This works in theory, but it is often not possible or practical to find an amplifier that can swing a large-enough signal at its output to cover a system’s whole dynamic range.
- Finding an amplifier that has a saturation recovery time fast enough for the system requirements – This solution can potentially work, as long as you include enough design margin to ensure that the system can always recover in time. Unfortunately, amplifier saturation recovery is often a function of input amplitude, time spent saturated and other factors. It is not easily characterized, and it is challenging to design safety margins properly for saturation recovery specifications.
- Using an amplifier with an output or input clamp that prevents saturation – This can be an external or built-in clamp that prevents the amplifier from saturating in the presence of a large input signal. This is one of the most common methods to avoid saturation issues; however, the clamp structures add additional parasitic elements to the circuit, which degrade potential performance. Clamping the output clamps also effectively reduces the gain, and the circuit loses its ability to transmit accurate amplitude information.
- Using a circuit with variable gain, either through the use of multiple amplifiers or a single amplifier with switchable gain – In theory, a variable gain circuit provides the best solution because it prevents saturation and still accurately transmits amplitude information. The circuit must be able to switch gains quickly enough to avoid saturating the higher gain stages, however, which is often a challenge in higher-speed systems.
A circuit that implements multiple methods to avoid saturation issues should be considered. The LMH32401 amplifier from Texas Instruments (TI), for example, uses both an input current clamp and variable gain modes to accommodate large input dynamic range requirements. The switchable gain allows the device to adjust the gain optimally for the input signal amplitude, while the input clamp ensures that the device does not saturate when the gain cannot be switched quickly enough.
Table 1 summarizes the benefits and trade-offs of the options listed earlier to resolve saturation issues in typical photodiode applications.
|Decrease gain||Output never saturates, no external components||Limits the smallest measurable input signal, increases noise|
|Use an amplifier with a larger output swing||Output never saturates, no external components, maintains gain||Requires a higher supply voltage, amplifier that meets circuit requirements may not exist|
|Use an amplifier with a faster recovery time||No external components, maintains gain||Hard to characterize saturation, will often be a higher-power amplifier|
|Use clamping||Allows the use of any amplifier, maintains gain for low input currents||Requires external components or a specialized device, reduces linear range|
|Use a variable gain circuit||Yields linear gain over the largest range||Requires a special circuit or amplifier, gain must switch fast enough to meet minimum recovery requirements|
Table 1 Benefits and tradeoffs of methods to solve dynamic range challenges
Like the challenges associated with dynamic range, noise in a photodiode also can limit the smallest recoverable input signal. For a photodiode and transimpedance amplifier circuit, noise sources include any noise from the photodiode itself, the amplifier’s input voltage and current noise, and the noise of the feedback resistor. The feedback resistor’s noise is directly related to the resistor’s value and cannot be changed without changing the circuit’s gain. The photodiode’s and amplifier’s intrinsic noise are set by the devices and can only be changed if you select new devices that have lower noise specifications.
The way that the noise sources are gained to the amplifier’s output will change with the circuit configuration. In general, a higher gain circuit will actually lead to less total input referred noise (more so for field-effect transistor [FET] input amplifiers than bipolar input amplifiers). Figure 3 shows a graph of input referred noise versus the feedback resistor for the OPA858 FET input amplifier and OPA855 bipolar input amplifier. For transimpedance circuits, the total input referred noise is typically expressed in root mean squared (RMS) amps and represents the total output noise divided by the circuit’s gain. This representation is helpful as a designer, because it makes it easy to see the smallest measurable root-mean-square current at the input.
Figure 3 Higher gain will result in less total input noise, especially in FET amplifiers, as this noise versus gain for the OPA858 and OPA855 with 10 MHz of bandwidth shows.
Of course, it is also possible to reduce the total noise from a circuit by lowering its bandwidth, which can be achieved by either filtering the circuit or choosing a lower-bandwidth amplifier. Filtering is not always practical for a design, though, especially in pulsed-based applications that require a wide bandwidth.
A potential design challenge related specifically to the nature of photodiodes is how the circuit must deal with “dark” current and current caused by ambient light. Dark current refers to how much leakage current the photodiode emits when there is no light source, and ambient light current is how much current the diode generates from non-signal light sources. It is possible to change both parameters by selecting different diodes. Dark current is affected by the type of diode, size, and biasing; selecting diodes with optical wavelengths further from the visible light spectrum and that have better optical filters will reduce ambient light interference. It is not always practical or achievable to eliminate all sources of unwanted photodiode current, however, so designers must either include margin for the expected value of these currents or use a specialized device such as TI’s LMH32401, which includes an ambient light and dark current cancellation circuit.
One of the challenges always presented by modern high-speed photodiode circuits is the risk of instability. The amplifier’s open-loop response, input capacitance, feedback resistor, and feedback capacitance will determine transimpedance circuit stability. There are many calculators and simulators available to help you check circuit stability, but these tools often do not account for printed circuit board (PCB) and component parasitics that affect the circuit. For slower circuits, parasitics are less of an issue because the values of parasitic inductances and capacitances are small enough that they do not cause an effect within the bandwidth of the circuit. But high-speed transimpedance circuits often use much smaller components (e.g., the feedback capacitor) which can be greatly altered by unwanted capacitances and inductances on the board or even on the discrete resistor and capacitor components.
To mitigate parasitics, it is best to select devices with packages designed to help overcome these challenges. Devices such as TI’s OPA855, OPA858 and OPA859, for instance, feature a feedback connection pin with an isolating no-connect pin to minimize the parasitic capacitance introduced across the feedback network. An even better solution is to purchase devices in die form, where the connections can be bonded directly to the device’s die, eliminating most board parasitics entirely. Die-form solutions do have the drawback of requiring significant infrastructure capabilities to properly mount and bond a bare die device, however. Finally, you could also look for a device with integrated gain, which will eliminate the parasitics from the feedback network on a PCB; you’ll only have to deal with parasitics from the input trace to the photodiode. Devices such as TI’s OPA857 and LMH32401 are examples of dedicated transimpedance amplifiers that feature integrated feedback networks. The tradeoff of these devices is that they allow for less flexibility in choosing gain values.
No matter what device type you choose for a design, it is always best to minimize trace lengths and follow good high-speed PCB layout techniques to minimize possible negative effects. Figure 4 shows example layout guidelines for the OPA855 high-speed operational amplifier that are applicable to any high-speed amplifier board layout.
Figure 4 These layout guidelines for the OPA855 are applicable to any high-speed amplifier boards.
Photodiode front-end circuits can be challenging to design, particularly when targeting modern high-speed solutions. There are many potential pitfalls in circuit design, but with proper knowledge and planning, it is often possible to find solutions for even the most challenging design requirements. In general, choose a device that most simplifies the circuit design, such as those with integrated features or protections. In cases where no specialized devices exist, a standard operational amplifier will often provide the best solution, as long as you take proper care to design the circuit. There are of course always more challenges that can arise from circuit design than those discussed in this article, but with proper planning – and by accounting for potential issues – you’ll be on the right path when developing photodiode front-end circuits.
Jacob Freet is the applications manager for Texas Instruments high-speed amplifiers products. He has worked with high-speed amplifiers for over seven years with a focus on high speed op-amps, fully differential amplifiers and transimpedance applications. Jacob received his BSEE from the University of Pittsburgh and worked in integrated circuit design and product definition before moving to his current role in applications.