In the last article, we began our conversation about design tips for photodiode amplifiers.
Here, we’ll cover two concepts you may need to adjust in transimpedance amplifiers in photodiode circuits: leakage current and bandwidth.
If you’d like to learn more about photodiodes, don’t forget to check out my Introduction to Photodiodes series; the first article in the series covers light and pn junctions.
Minimizing Leakage Current
Photodiodes produce currents in the nanoamp and low-microamp range. With tiny currents like these, non-idealities that we often ignore can become noticeable and even problematic.
Op-Amp Input Bias Current
First, take a close look at the op-amp’s spec for input bias current. Ideally, zero current flows into or out of the input terminals, and all of the photocurrent flows through the TIA’s feedback resistor and contributes to the output voltage.
Unfortunately, a real-life op-amp requires some input bias current, and bias currents that would seem negligible in other applications may produce unacceptable errors in a photodiode system. With non-zero bias current, some of the photocurrent is diverted to the op-amp’s input stage, and if the photocurrent is in the low-nanoamp range, it wouldn’t take much current diversion to seriously alter the measurement reported by the amplifier.
Figure 1. This diagram demonstrates how some of a photodiode’s photocurrent is used as input bias current and therefore does not contribute to output voltage. In this configuration, the photodiode is reverse-biased by a positive voltage, and the diode’s orientation results in photocurrent that flows toward the output node.
In general, you will want an op-amp with a FET input stage. BJTs draw too much bias current. But even FET input stages have the usual protection diodes found in IC input circuitry; these diodes have leakage current, and this leakage current becomes much more significant as temperature increases. If you’re designing a photodiode amplifier for a high-temperature application, make sure you check the high-temperature specs!
Op-amps that are intended for TIA applications can achieve amazingly low input bias currents. For example, I did a quick search and found the LTC6268 from Analog Devices. At room temperature, its leakage current is only a few femtoamps. However, at 125°C, the spec is 4 picoamps (max)—an increase of three orders of magnitude!
Second, we need to remember that our PCB traces are not surrounded by materials that provide infinite resistance. If the connection to the photodiode is routed near traces or copper pours that create a significant potential difference, the DC leakage current through the PCB could be large enough to cause errors.
The photodiode’s input signal travels through a trace that leads to the op-amp’s inverting input terminal. The inverting input terminal is usually at or near ground, because the non-inverting input terminal is held at ground or a small offset voltage. Thus, the traces that are more likely to cause leakage-current problems are those with voltages that are not close to zero, such as positive or negative supply voltages. To maximize precision, leave as much space as you can (within reason) between these traces and the photodiode input trace.
Many photodiode applications do not require high-frequency response, and that makes life a bit easier, because designing an optimized photodiode circuit is tough even when speed is not a major concern. When you throw a requirement for wide bandwidth into the mix, the situation can become seriously challenging.
The circuit diagram presented in the previous article showed a normal capacitor (CF) included in the feedback path as a means of ensuring adequate stability:
Figure 2. Our example photodiode with transimpedance amplifier from our previous article
However, in high-speed photodiode applications, the optimal amount of feedback capacitance can be extremely small—much less than 1 pF, in some cases. This is especially true in high-gain applications, because the need for feedback capacitance decreases as feedback resistance increases.
Thus, wide-bandwidth photodiode TIAs may not need CF, either because the feedback pole is not located at a frequency that creates instability, or because the feedback path has so much parasitic capacitance that an intentionally installed capacitor is not needed.
Figure 3. The same circuit shown above, but without the amplifier
Taking this a step further, we see that the parasitic capacitance may actually be larger than the required compensation capacitance. In this case, the parasitic capacitance unnecessarily limits the TIA’s bandwidth, and the task of the designer is to reduce feedback capacitance so as to increase bandwidth.
In a tight layout with short traces, there’s not a lot that we can do to reduce the capacitance of the copper connections in the feedback path. We can, however, reduce the parasitic capacitance associated with the feedback resistor.
First, we can try to modify the resistor’s PCB footprint. In theory, capacitance can be reduced by decreasing the parallel-plate area of the resistor’s endcaps and by increasing the distance between the endcaps. Next, we can reduce endcap-to-endcap capacitance by running a ground trace between the pads in the resistor’s PCB footprint. You can read more about these techniques on pages 14 and 15 of the LTC6268/LTC6269 datasheet.
We’ve covered various interesting details related to TIA design, and I hope that you find this information helpful when you’re designing or analyzing a circuit that includes a photodiode amplifier. If you have any additional tips or tricks, feel free to share them in the comments section.