Seven steps to successfully achieve ultra-low light signal conversion

In solid-state detectors for light detection, photodiodes are still the basic choice (Figure 1). Photodiodes are widely used in optical communications and medical diagnosis. Other applications include color measurement, information processing, bar codes, camera exposure control, electron beam edge detection, facsimile, laser collimation, aircraft landing assistance, and missile guidance.

In solid-state detectors for light detection, photodiodes are still the basic choice (Figure 1). Photodiodes are widely used in optical communications and medical diagnosis. Other applications include color measurement, information processing, bar codes, camera exposure control, electron beam edge detection, facsimile, laser collimation, aircraft landing assistance, and missile guidance.

Seven steps to successfully achieve ultra-low light signal conversion
Figure 1. Photodiode equivalent circuit

The light energy is transmitted to one of the sensors to generate current, which is further processed by the high-precision preamplifier. Analog-to-digital conversion and digital signal processing form the rest of the signal chain. The process of selecting a sensor and designing an analog front end can be reduced to seven steps:

1. Describe the signal to be measured and the design goal.
2. Select the appropriate sensor and describe its electrical output.
3. Determine the maximum gain that can be obtained.
4. Determine the optimal amplifier for the preamplifier stage.
5. Design a complete sensor and preamplifier gain module.
6. Run the simulation.
7. Build the hardware and perform verification.

Step 1: Signals and goals

According to the equivalent circuit in Figure 1, the output current is calculated as:

Seven steps to successfully achieve ultra-low light signal conversion

To convert light into electrical signals for further processing, you need to understand the AC and DC characteristics of the light source, the signal amplitude of the light source, the expected measurement resolution, and the power supply available in the system. Understanding the signal amplitude characteristics and noise level provides a basis for how to select the sensor, the necessary gain in the gain module, and what input voltage range and noise level may be required when selecting an analog-to-digital converter (ADC).

Assume that at room temperature, a light source emits 1 kHz light pulses with a brightness of 50 pW to 250 nW (0.006 lux). This is a very low amount of light and requires very precise signal conditioning and signal processing chains. The goal is to capture and process this signal with 16-bit resolution and precision. Achieving this resolution means that the measurement accuracy needs to reach 3.8 pW.

In addition, it is assumed that +12 V and C12 V power supplies can be used in the system. On this basis, the designer can calculate the signal-to-noise ratio (SNR) and design the circuit.

Step 2: Sensor selection

During the design process, photodiodes for photovoltaic mode or photoconductive mode are often optimized. Responsivity refers to the ratio of the detector output to the detector input, and is a key parameter of the photodiode. The unit is A/W or V/W.

Large-area indium gallium arsenide (InGaAs) photodiodes are used in instrument measurement and sensing applications. Compared with high-speed photodiodes used in high-speed analog and digital communication systems, instrument measurement and sensing applications, large-area indium gallium arsenide (InGaAs) is less than 600 Better response performance from nm to 800 nm.

When there is no light, apply a voltage to the short-circuited photodiode to get the current IS. Illuminating the diode will produce a counter current I proportional to the light intensitylight. The total current Isc is:

Seven steps to successfully achieve ultra-low light signal conversion

In Equation 2, the second and third terms limit the linearity of Isc, but they can be ignored in a broad sense. In fact, the relationship between Isc and the incident light level is basically linear, which can be approximated as:

Seven steps to successfully achieve ultra-low light signal conversion

To detect a small amount of light, the designer must specify a large-area photodiode whose minimum expected emitted light multiplied by the responsivity is greater than the dark current of the photodiode. This will produce a signal higher than the noise floor of the photodiode sensor. For silicon photodiodes with light wavelengths exceeding 1100 nm, the responsivity is usually less than 0.7 A/W. In this example, select Hamamatsu S1336 in Table 1.

The expected current of the photodiode can be calculated from the expected optical power of the light source, the formula is:

Seven steps to successfully achieve ultra-low light signal conversion

If the light source consumes energy on the effective area of ​​the selected photodiode, only formula 4 is required. For 16-bit conversion, it is necessary to resolve to half of the least significant bit (LSB) or 0.95 pA.

The effective area of ​​the Hamamatsu photodiode is 5.7 mm2, And use a circular array. It is also necessary to use an optical fiber between the sensor and the light source. The cross-sectional area of ​​the fiber optic cable may be larger than our photodiode. Usually, the optical power measurement unit in this case is W/cm2. The area of ​​the photodiode in cm2Indicates that the result is 57 x 10C3 cm2. For the unit of measurement W/cm2For the same 25 pA light source output current, the necessary power is:

Seven steps to successfully achieve ultra-low light signal conversion

The noise characteristics of silicon photodiodes determine the lower limit of light detection. The photodiode equivalent circuit shown in Figure 1 captures three noise sources:

Seven steps to successfully achieve ultra-low light signal conversion

The noise of the diode is the thermal noise generated by the shunt resistor of the diode.

Step 3: Gain block calculation

The preamplifier extracts the small signal generated by the sensor in a high background noise environment. There are two types of photoconductor preamplifiers: voltage mode and transconductance (Figure 2).

Pre-amplification is the first step to extract the small signal generated by the sensor from the background noise. There are three possible configurations to interface the photodiode with the transconductance amplifier. The transconductance amplifier configuration shown in Figure 2c uses “zero bias” to achieve precise linear detection of the photodiode. In this configuration, the photodiode output has a short circuit, according to Equation 3 (Isc = Ilight), there is basically no “dark” current. The ideal relationship (gain) between the output voltage and the input current of the I/V (transconductance) converter can be expressed as:

The ideal relationship (gain) between the output voltage and the input current of the I/V (transconductance) converter can be expressed as:

Seven steps to successfully achieve ultra-low light signal conversion

The value of the feedback resistor used defines the gain (sometimes called sensitivity) of the converter. If the current-to-voltage gain is extremely high, the generated Rf is as large as the allowable value of other restrictions. The designer should choose a resistance that is large enough to allow the minimum output current of the sensor to be sufficient for measurement, and the maximum current will not saturate the amplifier.

If the resistance value increases, this resistance will also exhibit obvious thermal DC voltage drift, which is the same as the situation reflected by the temperature coefficient of the amplifier’s input current. In order to compensate for this error, the same resistor is usually connected in series with the non-inverting input of the amplifier, and most of its noise is eliminated through capacitive bypass. To achieve the maximum signal-to-noise ratio, multiple gains must be avoided.

As the resistance continues to increase, its tolerance and temperature ratings drop significantly. For example, it is relatively simple to find a 1 k resistor with a tolerance of 0.01%, but it is difficult and costly to find a 10 M resistor with the same tolerance.

A series of low-value resistors can be used to form a larger resistance value, a low-value resistor can be used in conjunction with a multi-stage gain, or a “T-type network” circuit can be used to solve this problem. Unfortunately, these advantages need to be balanced. Using large feedback resistors may make mistakes and may cause instability. These issues will be resolved later.

In addition, this design example uses a very high resistance resistor: Rf = 80 M. This should convert the lowest and highest photodiode currents into output voltages that are easier to measure. The formula is:

Seven steps to successfully achieve ultra-low light signal conversion

Seven steps to successfully achieve ultra-low light signal conversion
Figure 2. Photodiode and transconductance amplifier interface

Step 4: Determine the optimal amplifier for the preamplifier stage

When the photodiode is exposed to light and the circuit of Figure 2c is used, the current will flow to the inverting node of the op amp, as shown in Figure 3. If load (RL) Is 0 and VOUT = 0 V, the photodiode will theoretically be short-circuited. In fact, neither of these two situations will occur. RL is equal to Rf/Aopen_loop_Gain And VOUTAnd VOUT It is the virtual ground applied by the amplifier feedback configuration.

Seven steps to successfully achieve ultra-low light signal conversion
Figure 3. Photodiode in short-circuit condition

If a voltage is applied to a short-circuited photodiode in the absence of light, a so-called “dark current” is generated. Therefore, the amplifier must have a large open-loop gain, and the designer must create the best “virtual ground”. This means that the error between the amplifier inputs is extremely small. RphotoDeviations from 0 V will cause error currents due to non-ideal amplifier conditions. These sources of error are obvious:

Seven steps to successfully achieve ultra-low light signal conversion

This requires an amplifier that introduces the least error. In other words, the amplifier selected by the designer has its feedback resistance configured as Rf = 80 M, the output error must not exceed 2 mV. It must also be ensured that the rise and fall times of the amplifier are less than the rise and fall times of the excitation laser diode source.

Several other amplifier parameters that do not appear in Equation 9 but can improve the design accuracy are:

• Low offset voltage temperature drift
• Low input bias current temperature drift
• High input impedance
• Low input capacitance
• Low input current noise density
• Wide bandwidth

The price, package size, and power consumption must also be considered when selecting the appropriate amplifier.

As mentioned above, the relationship between the output voltage and input current of the I/V converter is actually the gain of the converter. The formula is:

Seven steps to successfully achieve ultra-low light signal conversion

As shown in these formulas, VOThe formula has an error term, which must be reduced as much as possible. For example, if you choose aOFor very large amplifiers, the term aοβ increases, while 1/aοβ decreases. This makes the error term smaller.

In this example, we choose the precision operational amplifier AD8627 with a working voltage of ±12 V, which has extremely low noise, low bias current and wide bandwidth. Check the data sheet of AD8627, you can get the following characteristics and specifications: IB = 1 pA, ft = 5 MHz, en = 16 nV/√Hz at f = 1 kHz Ccom =3.8 pF, Cdiff = 4.1 pF. IC manufacturers provide online search and selection tools to select components according to user requirements. Table 2 lists several amplifiers suitable for photoelectric sensing.

Table 2. High-voltage FET input amplifiers suitable for photodiode applications

Product number

VOS (µV)

IB(pA)

UG bandwidth (MHz)

Noise (nv/rtHz)

Encapsulation

AD8610/AD8620

100

10

25

6

MSOP

ADA4610-2

400

25

9.3

7.3

MSOP

AD8625/AD8626/AD8627

750

1

5

16

SC-70

AD8641/AD8642/AD8643

750

1

3.5

27

SC-70

Step 5: Gain module

When the photodiode is connected to the amplifier configured in Figure 2c, the amplifier often oscillates. As mentioned earlier, the large resistance in the amplifier feedback can cause abnormal behavior and cause oscillations. The designer must ensure that the appropriate amplifier is selected and that the amplifier works stably when used in combination with a sensor.

The circuit’s response or bandwidth, peaking or overshoot, noise or signal-to-noise ratio (SNR) and other performances will become very complex, nonlinear, and mainly depend on the difference between the active and passive components in the converter circuit. affect each other. Alternative circuit models can be used to arrive at a more realistic analysis (Figure 4). Figure 4 considers all the non-ideal conditions of this solution, allowing designers to implement modeling and use pole/zero analysis to avoid future problems.

Seven steps to successfully achieve ultra-low light signal conversion
Figure 4. Alternative circuit model can be used to analyze the circuit shown in Figure 2c

The interaction between the large feedback resistance and the input capacitance will introduce the zero to the pole/zero stability analysis. If CphotoLarge enough, at the crossover frequency where the open-loop transconductance gain intersects the noise gain function (Figure 5), the closed-loop phase shift will be close to −180.

Seven steps to successfully achieve ultra-low light signal conversion
Figure 5. Stability analysis illustrates the interaction between the large feedback resistance and the input capacitance

In order to maintain the 45-phase margin and stability, a small capacitor and Rfin parallel. The value of this capacitance is related to the input capacitance at the input of the amplifier. The input capacitance of the amplifier is small, then CfThe value is small. Use a smaller C in the circuitfAt that time, more bandwidth of the amplifier will be used for existing applications.

Bandwidth and sensitivity are directly related, by choosing RfTo weigh the use. For example, the feedback capacitor (Cf = 2 pF) in parallel with Rf, Cphoto = 20 pF and Rf= 80 M photodiode will have a maximum bandwidth of 1 kHz.

In addition, if a 10 kHz bandwidth is required, the designer may have to choose the maximum Rf = 8 M, CfThe capacitance can still be 2 pF. This concept can be used to design programmable bandwidths to handle different input signals, such as 1 kHz and 10 kHz.
The designer must analyze the bandwidth and noise to confirm that the selected amplifier is just right for the design. In this way, we can understand why it is important to choose the AD8627 with low input capacitance and bandwidth.

Seven steps to successfully achieve ultra-low light signal conversion

As you can see, choose CphotoA large-area photodiode with a higher value has a much smaller fx (that is, a lower bandwidth). One possible remedy is to choose an amplifier with a very wide bandwidth (ft). However, this can cause other problems, such as more noise.

In this case, the AD8627 amplifier must have extremely low voltage noise in order to obtain lower total noise in large-area photodiode transconductance amplifier applications. This is necessary because the noise gain of the transconductance circuit rises sharply with frequency (noise gain = 1 + Zf/Xc), it affects voltage noise and resistance noise, but does not affect current noise. As shown in Equation 12, the total noise above 0.01 Hz when the I/V converter and photodiode are combined is calculated:

Seven steps to successfully achieve ultra-low light signal conversion

Assuming AD8627 (IB = 1 pA, ft = 5 MHz, en = 16 nV/√Hz at f = 1 kHz Ccom = 3.8 pF, Cdiff = 4.1 pF) and Hamamatsu photodiode (Rphoto = 2 GΩ, Cphoto = 20 pF) together. In addition, Rf = 80 MΩ, Cfeedback = Crf + Cf = 2 pF. According to the above information, the input capacitance is Cin = Cphoto + Ccom + Cdiff = 20 pF + 3.8 pF + 4.1 pF = 28 pF.

The noise gain of the I/V circuit is mainly restricted by en, And inAnd iRThe noise gain of is consistent with the signal gain (Figure 6). The main noise is fxNearby enoeAnd fpNearby thermal noise enoR

Seven steps to successfully achieve ultra-low light signal conversion
Figure 6. According to Equation 12, the noise gain of the transconductance circuit is mainly restricted by en, And inAnd iRThe noise gain is consistent with the signal gain

Seven steps to successfully achieve ultra-low light signal conversion

Because this design uses the JFET input amplifier, the current noise is almost negligible.

Step 6: Simulation

The photodiode manufacturer does not provide the SPICE model of the product, but the SPICE model of the amplifier can be downloaded from the Analog Devices website. Designers can also download the free version of SPICE simulation software MultiSim provided by National Instruments from the ADI website.

The software environment provides LabVIEW for photodiodes®The transconductance model allows for customization based on the specific photodiode used in the design example (Figure 7a). The simulation must be run before any board is built. Due to the introduction of zeros in the noise gain path (Figure 7b), instability may occur.

The National Instruments MultiSim user interface provided for the ADI transconductance amplifier example model has the characteristics of Hamamatsu photodiodes and can be further analyzed (Figure 7a). MultiSim simulation illustrates the instability caused by the introduction of zeros in the noise gain path (Figure 7b). Changing the capacitance on the feedback resistor will affect the available bandwidth (Figure 7c).

As mentioned above, a 2 pF capacitor must be placed on the feedback resistor to introduce a pole, thereby canceling this zero. 2 pF feedback capacitance is the theoretical value. The influence of different values ​​on the available bandwidth of the designed circuit can be analyzed (Figure 7c). The circuit bandwidth can also be verified by monitoring the output, and its −3 dB bandwidth is 1 kHz.

Seven steps to successfully achieve ultra-low light signal conversion
Figure 7. Noise circuit analysis of photodiode circuit

Step 7: Hardware Verification

In addition to using a very clean power supply, the ultra-low light detection circuit also needs to adopt best-practice rules and practices to reduce the noise of all noise sources, including environmental electromagnetic noise and all leakage source noise. Low-voltage applications can use batteries, but need to use RC or LC filters to set the power supply bypass.

Other key factors for success include circuit boards made of materials with high insulation resistance. In order to avoid leakage into the measurement circuit, a guard ring or Teflon support must be used to wirelessly connect the photodiode pin to the input terminal of the operational amplifier. The same is true for feedback resistors and capacitors.

Shielding cables and circuits using metal shielding boxes is a good measure to prevent electromagnetic interference (EMI). In more demanding occasions, designers can use optical fibers between the light source and the photodiode.

Analog Devices has designed an example circuit in the range of 25 pA to 125 nA. Any signal outside this range will saturate the amplifier and affect overall performance. If a wider range is needed, a low-leakage switch can be connected in series with the feedback resistor. Then other feedback circuits can be used according to different sensitivities.

Reference circuit

Jung, Walter G. High impedance sensor. Operational amplifier application manual. Part 4-4, Analog Devices, 2006.
Technical guide MT-059. Compensate the effect of input capacitance on voltage feedback and current feedback operational amplifiers used in current-to-voltage converters. Analog Devices, 2009.
S1336 data sheet. Silicon photodiode. Hamamatsu Photonics, Bridgewater, New Jersey, 1991.

author

Seven steps to successfully achieve ultra-low light signal conversion
Reza Moghimi

Reza Moghimi is an applications engineering manager in the Precision Signal Conditioning Division of Analog Devices (San Jose, California, USA). He received a bachelor’s degree in electrical engineering from San Jose State University in 1984 and a master’s degree in business administration (MBA) in 1990.

The Links:   SKM200GB124D SK45GAL063

Related Posts

Leave a Reply

Your email address will not be published. Required fields are marked *