Noise Figure – Test Method for High-Bias SOA

Preface
With the rapid development of the AI computing industry, optical transmission and optical computing applications impose new requirements on optical amplifiers, including miniaturization, low power consumption and broad operating bandwidth. This has directly driven the rapid adoption of SOAs (Semiconductor Optical Amplifiers).
SOAs are mainly classified into two categories by polarization characteristics:
Polarization-Insensitive SOA (PI-SOA), whose gain difference between TE and TM modes (also known as Polarization Dependent Gain, PDG) is typically less than 2 dB.
Single-bias / High-bias SOA (PM-SOA), featuring large PDG and high Polarization Extinction Ratio (PER).
Noise figure measurement for PI-SOAs generally follows the same scheme as conventional EDFAs, mostly using the interpolation method. In contrast, the polarization nulling method is more suitable for high-bias / polarization-maintaining SOAs. Combining national standards and internal practical experience, this paper fully describes the in-house noise figure measurement procedure for high-bias SOAs.
Referenced Standards
IEC 61290-3-1:2003 Optical Amplifiers – Part 3: Noise figure parameters
GB/T 16850.3-2021 Test methods for optical fibre amplifiers – Part 3: Test methods for noise parameters
The test is primarily implemented via the polarization nulling (PN) method specified in the above standards.
Test Environment
Polarization nulling test topology recommended by national standards
The polarization nulling test topology defined in the national standard is shown below:

Figure 1 Test Topology of Polarization Nulling Method for NF Measurement (Source: IEC 61290-3-1:2003 Figure 1)
Simplified Test Topology for High-Bias PM-SOA Adopted by Our Company
The output of our high-bias SOA chip is single-polarization TE light. TM-polarized light is significantly attenuated and suppressed by the waveguide. The butterfly package further integrates a polarization-dependent isolator on the output side. Its PER is typically greater than 20 dB, and TM-polarized ASE noise accounts for less than 1% of TE-polarized noise light, which can be nearly neglected.
Accordingly, the polarization nulling unit shown in Figure 1 can be omitted, and the simplified test topology in Figure 2 is adopted. Comparative experiments on our high-bias SOAs verify that the noise figure measured by the simplified optical path (Figure 2) is completely consistent with that obtained from the standard optical path (Figure 1).

Figure 2 Noise Figure Test Topology for Our High-Bias SOA
Notes in figure:
NLL/TLS: Either narrow-linewidth laser (NLL) or tunable laser source (TLS) can be selected as required. The light source shall feature spectral width <0.1 nm, side-mode suppression ratio (SMSR) >30 dB, and output optical power fluctuation <0.05 dB.
PC: Polarization Controller, used to adjust the polarization state of light; insertion loss shall be less than 0.2 dB.
Linear Polarizer: Extinction ratio >30 dB.
VOA: Variable Optical Attenuator; attenuation range >40 dB, stability better than 0.1 dB.
OSA: Optical Spectrum Analyzer; supports resolution (RES) ≤0.1 nm, stability better than 0.1 dB, wavelength accuracy better than 0.05 nm, and linearity better than 0.2 dB within its dynamic range.
SOA Driver: Supports bias current setting and TEC temperature control. Low electrical noise is required to minimize extra electrical noise introduction.
In addition, an OPM (Optical Power Meter) shall be equipped for optical power calibration of the OSA and auxiliary polarization tuning.
Noise Figure Calculation Formulas:
1.Noise figure calculation formula

2.Total Noise Formula

3. OSA Absolute Power Correction Factor (PCF)

4. Integrated Complete Noise Figure Formula

Test Procedures
Follow the operations specified in Clauses 6.3.2 and 6.3.3 of IEC 61290-3-1:2003.
Prepare the test setup
(1) Connect the light source module according to the test topology in Figure 2, switch on the light source, and set the optical wavelength to the target wavelength λₛ.
(2) Mount the SOA into the SOA driver. Set the maximum rated current limit, configure the SOA operating temperature (25 °C by default for standard tests) and operating current, then enable temperature control and the SOA sequentially.
(3) Set the OSA resolution (RES) to 0.1 nm and switch to the “optical amplifier noise analysis” mode.
Measure the OSA power correction factor (PCF) and the SOA input spectrum
(1) Set the wavelength of the light source module to the target wavelength λₛ. Connect the light source module directly to the optical power meter as shown in the figure below. Tune the VOA to set the source output optical power to the target value, and record the optical power Pₒₚₘ (dBm).

(2) Disconnect the optical power meter, connect the light source output to the OSA, and measure its output optical power \(P_\text{OSA}\) (dBm).

(3) Calculate PCF: \(\text{PCF}=P_\text{OPM}-P_\text{OSA}\). Navigate to the “Setup” page under the optical amplifier measurement function of the OSA, and enter PCF into the input/output optical power correction field. (This step enables the OSA to output NF automatically; this parameter setting is unnecessary if NF is calculated manually.)(4) “Save” the SOA input spectrum to Trace A and set Trace A to “Fixed”.
Adjust polarization to operate the SOA at maximum gain
(1) Make connections as shown in the figure below. Use the optical power meter (OPM) to monitor the SOA output optical power, and tune the polarization controller (PC) until the OPM reading reaches its maximum.

(2) Disconnect the OPM, connect the SOA output to the optical spectrum analyzer, and select to save the spectrum to Trace B.

Read the actual noise figure and apply standard correction
(1) The OSA directly calculates \(\text{NF}_\text{OSA}\). With the PCF correction configured earlier, the value provided by the OSA represents the true NF of the high-bias SOA; this value may be adopted for internal evaluation of chip performance.
(2) For factory outgoing delivery, customer acceptance and third-party testing, to avoid disputes over data and ensure unified and compliant test standards, the IEC standard with an additional 3 dB correction shall be followed. The final test result shall be corrected as NF(External) = NFOSA + 3 dB.

Test instance

Below is an example of noise figure measurement for the high-power polarization-maintaining SOA (JSA-BT525G35-PM).

Figure 3 A spectral example of SOA noise figure measurement
The yellow curve and cyan curve in the figure represent the input and output spectra, respectively. The parameters obtained from the figure are as follows:
• WL=1551.2263 nm
• Pin= -13.566 dBm
• PASE= -38.656 dBm
• Res= 0.069 nm (optical bandwidth B0 corresponding to the noise figure formula)
• G = 15.506 dB
• NF = 5.44 dB
It can be seen that the NF measured by the OSA is 5.44 dB, which is the actual noise figure of our high-bias SOA at an input power of -13.566 dBm and a bias current of 700 mA. For externally released reports, the NF is corrected to 8.44 dB to unify the test standards.
The following shows the NF test results of this revised SOA under different input optical powers and different driving currents (with a 3 dB correction already applied).

Figure 4 Relationship between the noise figure and input optical power of high-bias SOA