The roadmap for the development of 1.6T optical modules was announced last year. It is expected that the application of 1.6T optical modules will be the trend in two or three years. What needs to be evaluated now is which solution to choose and what are its advantages and disadvantages.
Arista believes that 1.6T can be deployed in 2026, and the optical module adopts hot-swappable type, but in the future, it is possible to adopt denser packaging with super CPO to reduce power consumption.
1.6T adopts single-wave 200G optical technology, and the electrical signal may have a short-term gearbox for transition. In 2025, 1.6T DR8 and 1.6T-2FR4 are expected to be launched first
The speed of SerDes, on the one hand, 100G and 200G coexist, on the other hand, it will eventually transition to the 200G era, remove the gearbox, and have the advantages of low power consumption and low cost.
The 200G trend estimated by Arista, as shown in the red dotted line in the figure below, will tend to be saturated in 2027. This year’s report has made a slight change to this curve. 2027 and 2028 will still maintain a growth trend, which is another after the single-wave 100G technology. Large speed nodes.
To reduce the power consumption of optical modules, there are mainly four changes. The first is that the driver adopts a linear drive interface to reduce DSP/CDR power consumption, or cancel DSP/CDR. Choose a low-power modulator again, lower the drive voltage, and lower the insertion loss. Improve the electro-optic efficiency of the laser and improve the coupling efficiency. Reduce the connection length of the electrical interface, super densely package the electrical interface, and super CPO package the electrical interface.
In 800G optical modules, compared with the unit energy consumption of single-mode applications, there are two trends
One is that the energy consumption of thin-film lithium niobate is slightly lower than that of silicon photo-integration schemes.
One is to adopt a linear drive scheme without DSP, which can greatly reduce energy consumption.
By analogy, in the 1.6T optical module, the rate increases from 100G to 200G, followed by the development of the DSP node process from 5nm to 3nm, the estimated energy consumption of the 1.6T optical module is as follows.
In terms of the total power consumption of the switch, the 51.2T switch and the 102.4T switch can reduce the power consumption of the switch by 25% by canceling the DSP, adopting the linear drive scheme, and adopting the silicon photonics integration scheme.
For the modulator, the power consumption is proportional to the square of V, so reducing the half-wave voltage can effectively reduce the power consumption of the modulator. Reducing power consumption will naturally reduce the temperature of the optical module, thereby reducing the failure rate and improving reliability. The relationship between temperature and failure rate is exponential.
Regarding the electro-optic efficiency of the laser, high temperature will reduce the electro-optic efficiency of the laser, and the optical module does not want to use TEC to cool down. The working temperature of the laser inside the optical module is 75-85 ℃. Quantum dot technology can solve the contradiction between TEC and electro-optical efficiency. Quantum dot lasers can still maintain high temperature efficiency while being more reliable. The last point is to shorten the distance of the electrical connection and adopt a smaller size package. This is also very commonly used in the silicon photonics integrated packaging technology route.
To sum up, the 1.6T optical module has 40% expected space for reducing power consumption. Using linear drive and canceling DSP can reduce power consumption by 25%. With quantum dot lasers, power consumption can be reduced by approximately 5%. Ultra-dense co-packaging of switching chips and optical modules is used to shorten the distance of electrical signals, and it is estimated that power consumption can be reduced by 10%.
Commentaires