Mixed signal ICs lend flexibility to EDFA control applications

April 1, 2002

by Niall Lyne

Automatic power-control architecture can stabilize optical-fiber amplifiers through use of optical power detectors and mixed-signal ICs, despite variations in temperature and input power.

Continuing innovations have been made toward stabilizing the output of an optical transmission signal by use of erbium-doped fiber amplifiers (EDFAs). The main element within an EDFA that controls the gain and the stability of the amplifier is the method of stabilizing the output of the pump laser diode. This element requires high-performance automatic closed-loop controls made up of mixed-signal electronic components to control the pump laser, which in turn ultimately controls the output of the EDFA.

The main circuit of a basic single-pump EDFA consists of the erbium-doped fiber in the center, two optical isolators, and two couplers outside these isolators (see Fig. 1). Inside the isolators, there are two wavelength-division multiplexers. The wavelength of the optical transmission signal is near 1550 nm and the optical excitation signal is either 980 or 1480 nm. The erbium-doped fiber disperses an excited optical signal, and the second optical isolator on the output blocks the optical signal reflected from an output coupler. The amplified optical signal travels to the output coupler via the second optical isolator. The input and output couplers divide the received optical signal at a predetermined ratio of x:y, where x = 1% to 5%, y = 99% to 95%. An output photodiode senses the intensity of the output "x," and the output "y" is loaded on an outer optical fiber.

The optical excitation signal is controlled by the pump laser-diode current control loop (this loop controls the pump laser-diode current driver). A digital signal processor (DSP) or microprocessor generates the control signal to stabilize the final output of the optical fiber amplifier or the optical excitation signal of the pump laser diode. The pump laser diode provides energy to the EDFA, while a photodiode detector in the pump module senses the output power of the pump laser. External to the module there is also a pump laser-diode current driver that supplies the current to drive the pump laser. The first digital-to-analog converter (DAC) converts the digital control signal received from the DSP/microprocessor to a current or voltage (depending on the architecture) that is applied to the pump laser-diode current driver. This driver supplies the current to the pump laser that generates an excitation signal from the pump of 1550 ±60 nm.

There are five main control loops within a basic single-pump EDFA. In one optical-monitoring feedback loop, the input-signal photodiode detector detects an optical signal divided at a predetermined ratio by the input coupler—for example, 5%:95%—and converts the optical signal to an electrical current. The measured current could range from 100-pA to 10 mA (160 dB electrical, 80 dB optical dynamic range) in very high-end applications. A logarithmic amplifier or transimpedance amplifier (TIA) detects and converts this current into a voltage directly proportional to the intensity of the input optical signal. The first analog-to-digital converter (ADC #1) then converts this voltage to an equivalent digital signal and applies it to the DSP/microprocessor.

In a second optical-monitoring feedback loop similar to the input feedback loop, the output photodiode detects an optical signal from the output coupler and converts it to a current, again somewhere in the range of 100 pA to 10 mA. A TIA or log amplifier again gains this current and converts it into a voltage. This voltage is converted (via ADC #4) to an equivalent digital value and applied to the DSP/microprocessor. In a third feedback loop, the photodiode detector associated with the pump detects the optical excitation-signal output from the pump laser (approx 20 dB optical dynamic range) and converts it to a current. The log amplifier or TIA and ADC #2 in turn convert the equivalent current to a voltage, then to a digital signal and onto the DSP/microprocessor.

A fourth feedback loop is associated with the pump laser-diode bias-current sensing circuit. The pump laser diode bias-current sensor (a TIA) senses the bias current of the pump laser diode, and ADC #3 converts it to digital and feeds it to the DSP/microprocessor. A fifth feedback loop, associated with the pump laser thermoelectric cooler (TEC) controller, sets and stabilizes the temperature of the TEC. A voltage is applied to the controller (via DAC #2) corresponding to a target-temperature set point. The appropriate current is then applied to the TEC to either pump heat to or away from the pump laser, the temperature of which is being regulated. The temperature of the pump laser is measured by a thermistor and is fed back to the controller to correct the loop and settle the TEC to the appropriate final temperature. In some cases, a temperture monitor is put in place, ADC #5, which is used to feed back the temperature value to the DSP/microprocessor.

The DSP/microprocessor compares the digital signal from the pump laser-diode bias-current sensor to a bias-curent limit set by a user. Depending on the result, the DSP/microprocessor controls the optical excitation-signal output of the pump laser, increasing or decreasing the power based on the EDFA input and output feedback-control circuits. A comparison of the input power versus the output power of the EDFA is then compared with an intially set reference EDFA-output value, to stabilize the final output optical signal. Simultaneously, the DSP/microprocessor adjusts the output of the pump based on the digital signal received from the third feedback-control circuit (the reference pump laser-diode output), also preset by a user. These control loops and multiple others simultaneously interact. Additional feedback-control circuits may be used—more common in systems with two or more pump lasers.

The performance of high-end EDFAs is quickly approaching its limit because of the physical constraints imposed by high-speed and high-resolution mixed-signal components in closed-loop controllers. The need for high resolution in DWDM architectures requires optical and analog feedback signals, which set this limit.

The typical EDFA-control signal chain requires a processor core and a generic set of peripheral function blocks to interface between the digital processor and the analog and optical signals. In a high-performance (fast settling time <10 μs) EDFA control system, a powerful DSP core with the set of peripherals completes the signal chain. Control loops for EDFAs in long-haul DWDM systems typically have loop bandwidths on the order of 400 kHz to 1-MHz. These loop bandwidths, coupled with the 14- to 16-bit resolutions of ADCs and DACs, are required by the high-end EDFAs. A 16-bit general-purpose DSP can handle this performance.

Typical DSP performance requirements are 80 to 160 MIPS with 80 to 160 kbytes of on-chip RAM depending on the application. These devices are well-suited for EDFA control applications due to their high-speed signal-processing capabilities. There are many reasons for moving toward a completely digital control system with DSPs at the core. The primary reason is that a digital system offers the most flexible control-system architecture.

Although DSPs have the computing power to control high-bandwidth current loops, they require additional hardware to implement some of the EDFA-control peripheral functions, unlike some conventional microcontrollers. These functions have been implemented using standard components such as ADCs, gate arrays, or application-specific ICs. Depending on the application and the feedback resolution required, however, this may not be cost-effective.

Responding to the challenge, silicon providers are pushing the limit of mixed-signal integration of high-performance analog-to-digital converters and digital-to-analog converters on the same substrate as a DSP or microcontroller. This push has led to a family of microconverter products that integrate successive approximation register or sigma-delta ADCs with industry standard 8051 microcontrollers for closed-loop-control applications, similar to that required by an EDFA.
In a lower-cost, lower-performance (settling time <5 ms) EDFA control system, a microconverter combines an 8051 microprocessor core, multichannel 12-bit ADCs, 12-bit DACs, and other integrated peripherals with an external TEC controller and some external power-monitoring amplifiers to complete the signal chain (see Fig. 2). This is sufficient for single-channel EDFAs in shorter-haul applications.

In optical power monitoring, traditional linear techniques using TIAs provide a linear transfer function that requires multipoint calibration to achieve acceptable accuracy. Optical coupling uncertainty along with variations in photodiode responsivity may cause the transfer function to slide to the left or right (see Fig. 3). The traverse slide of the transfer function can generate gross errors if left uncalibrated. As a result, it is necessary to generate a calibrated lookup table to accurately estimate an optical power level based upon a sampled digital value.

The dynamic range of a classical TIA used to amplify the photocurrent of a fiberoptic photodiode is limited. For an output voltage swing of 5 V, the TIA offers less than 30 dB of optical dynamic range per selected feedback resistor. The linear nature of this optical-to-electrical conversion requires a higher-resolution 16-bit ADC following the TIA to maintain precise measurements over a wide dynamic range.

By comparison, the same uncertainties in the transfer function of a log amplifier results in a fixed intercept offset that can easily be removed with a single-point calibration when using a log conversion. For the same 5-V output swing, 100-pA intercept, and slope of 500 mV/decade, the log amplifier provides 80 dB of optical dynamic range with a constant volt-per-decibel relationship. The advantage of increased dynamic range and constant decibel slope makes the log amplifier front end a superior method of conditioning the signal prior to the ADC.

With the log amplifier approach, a wide dynamic range can be compressed and simultaneously transformed to a decibel representation. This facilitates the use of a lower-resolution 12-bit ADC to that required for the TIA. One case in which the TIA is better than a log amplifier is when faster loop response is needed in an EDFA control loop. This requires a higher-bandwidth amplifier, typically greater than 1 MHz in the sub-nA measurement range or 10 MHz in the higher mA range.

Control applications for EDFAs, including test equipment for optical components, require an ADC with 12- to 16-bit resolution. This is because of the dynamic range of electrical current equivalent to the light strength (10 pA to 10 mA) measured. Furthermore, because control loops must acquire data quickly and act on it, high-resolution ADCs have to be very fast, up to 1 Ms/s. Control loops by their nature cannot tolerate latency, so the ADC has to instantly provide a digital representation of the sampled analog signals. The capability to multiplex incident signals is desirable, whereas a pipeline-delay architecture is not.

TEC CONTROLLERSComplete TEC controllers require a precision input-amplifier stage to accurately measure the difference between the target temperature and the actual temperature of the TEC. They also require a compensation amplifier to optimize the temperature step response of the TEC, a high-output current stage to provide the required current to the TEC, and phase adjustment on a TEC controller for multipump EDFAs.

To maintain accurate wavelength and power from the laser diode, the difference voltage between the input and the error amplifier must be as accurate as possible. For this reason, self-correction auto-zero amplifiers are used in the input stage of controllers, resulting in final temperature accuracy within ±0.01°C in typical applications. Because of the high output currents involved, a TEC controller should operate with high efficiency to minimize the heat generated from power dissipation (see Fig. 4). In addition, an effective controller should operate down to +3.3 V and indicate when the target temperature is reached. Finally, phase adjustment on a TEC controller is advantageous for multipump EDFAs because this allows two or more TEC controllers to operate from the same clock frequency and not have all outputs switch simultaneously, which could create an excessive power-supply ripple.

The automatic power-control architectures discussed can stabilize EDFA output power by using optical power detectors and mixed-signal ICs. These components provide a stable final output of the optical fiber amplifier, despite possible inconsistencies caused by variations in temperature and in the power of an input optical signal.

Niall Lyne is a field applications engineering manager with Analog Devices, 3550 North First Street, San Jose, CA 95134. He can be reached at [email protected].