top of page
news center(栏目banner).jpg

50G PON 1342nm 15dBm EML-SOA

A company mentioned at ECOC2022 that they made technical optimizations in 2020 to increase the output power of 1342nm EML (Electroabsorption Modulated Laser) for 50G PON.

1342 nm 15dBm EML-SOA
1342 nm 15dBm EML-SOA

ECOC2023, building upon last year's foundation, has been further optimized with the integration of SOA to reduce performance pressure on the EA region.

The integration of SOA is approached from three aspects:

  • How to increase output power.

  • How to reduce the impact of facet reflection.

  • How to reduce noise and improve signal (eye diagram) modulation quality.

SOA saturation power can be achieved by lowering the confinement factor.

EML-SOA performance and high power optimization
EML-SOA performance and high power optimization

The previous articles mentioned the down-tapered optical field design of manufacturers such as Lumentum, Densilight, and Xuchuang, which is the optical field design of the SOA in the equivalent DFB+SOA configuration within a CW light source.

To simplify these designs, they essentially involve adding a high refractive index InGaAsP waveguide region in the original n-InP region. This action shifts the optical waveguide down from the quantum well layers and amplifies it. There are several advantages to this approach. First, it enlarges the optical spot, reducing the confinement factor. This means that at the same optical power density, a greater optical power can be output. Secondly, the optical field is located in the n-type region, avoiding optical absorption losses caused by metal doping in the p-type region. Thirdly, it redistributes the divergence angle, enhancing fiber coupling efficiency.

DFB lasers are particularly sensitive to reflected light. Typically, a reflection level of -20dB, which is 1% of the reflected light, can cause significant parasitic resonance-induced noise and deterioration in RIN (Relative Intensity Noise).

The impact of reflections on DFB lasers includes reflections from the front facet and reflections at various nodes in the coupling optical path. For front facet coatings, it is usually necessary to control the reflection to 0.01%, which is -40dB.

If an SOA is integrated on the backend of the chip, the level of reflection at the front facet needs to be further reduced to less than 0.001%, which is equivalent to -50dB of reflection, and the transmittance should reach five nines (99.999%), with an anti-reflection coating achieving a transmittance of 99.999%.

There are two reasons for this requirement. First, reflected light passing through the SOA region will be amplified. Second, SOAs generate spontaneous radiation noise, some of which can enter the DFB in the reverse direction, necessitating a further reduction in front-facet reflections.

However, achieving 99.999% transmittance and 0.001% reflection is the design goal, which the current industry capabilities cannot meet. The approach is to use the commonly used industry standard of 99.99% transmittance and 0.01% reflection and then achieve an effective ultra-low reflection of 0.001% through internal optimization.

The method used involves angled light emission and an unconfined waveguide.

This approach, similar to Acacia's approach, shares a common purpose.

Thirdly, it involves balancing the gain of the SOA to reduce modal noise and improve signal quality.

Below are several eye diagrams, specifically for a 25Gbps signal, illustrating the eye diagram quality under three different combinations of DFB current and SOA current.

When both the DFB and SOA have either high or low currents, the eye diagram quality is generally good. However, a significant difference in current between the two can result in increased incident power, causing a noticeable gain mismatch in the SOA between P1 and P0, leading to a degradation of the eye diagram quality.

To avoid signal degradation caused by SOA gain mismatch, we did not choose the traditional -3dB saturation calibration region as the SOA's operating range. Instead, we opted to use the -2dB decrease region as the limit for its cutoff power, allowing the SOA to operate as much as possible in the linear range.

This calibration process involves first determining the laser output power without the SOA. Since the SOA is integrated at the front end of the EML, when the SOA is forward-biased, it acts as a semiconductor optical amplifier (SOA bias voltage). When the SOA is reverse-biased, it functions as a typical waveguide-type detector. By applying reverse bias voltage and measuring the detector current, we can directly test the output power of the EML because the EA is not provided with bias and modulation voltage, and the entire output power is solely from the DFB.

Then, by reversing the bias voltage of the SOA from reverse bias to forward bias and supplying current, we can calculate the gain of the SOA.

EML - SOA Performance chart
EML - SOA Performance chart

After fitting the linear graph, with an output power of 18.1dBm (65mW), LD current at 100mA, and SOA current at 100mA, the SOA's gain region is mostly within the linear range, allowing for the optimization of the eye diagram signal quality.

XINXIN GEM provides ceramic submount for the Electro-absorption Modulated Laser.

17 views0 comments


bottom of page