LiDAR: The War Between 905 and 1550

However, a 905 LiDAR manufacturer pointed out that the 1550 has the following two disadvantages when operating at higher pulse power and repetition frequency:

The price of significantly increasing the power per point is that the 1550 cannot achieve the same number of point clouds per second as the 905, which in turn affects the resolution.

In order to increase the FOV and point cloud number, single laser solutions basically use two-dimensional rotating mirrors, which makes its structure much more complicated than multi-laser solutions (usually using one-dimensional rotating mirrors). Correspondingly, in order to maintain performance and stability, the required R&D cost is also higher.

However, in response to the question raised by some 905 laser radar manufacturers that a single 1550 laser has "too heavy a workload (several or dozens of times the workload of a 905 laser), which will affect its life, reliability and robustness ", Bao Junwei responded that the workload required by the current laser radar is not "too heavy" for a 1550nm fiber laser, even if the frequency is increased by more than ten times. As long as the design process is well done, this load is not a problem at all.

5. Can 1550 be made into a VCSEL chip solution?

It has been repeatedly mentioned above that the 1550 is too large, which results in a low number of lasers. So, can the 1550 be made into a VCSEL chip (using multiple lasers) to achieve the same high level of integration as the 905?

Here’s the answer first: No.

Hesai said:

"Not everything can be turned into chips if we want to. A necessary condition is that the non-chip system is already a mature and stable system that has been verified. However, the 1550nm system is not mature enough today. Its reliability and performance have not yet met the requirements. Therefore, it is unlikely to be turned into chips."

Semiconductor laser chips are divided into two types: EEL (edge-emitting laser chip) and VCSEL (surface-emitting laser chip) according to the different resonant cavity manufacturing processes. EEL is to form a resonant cavity by coating optical films on both sides of the chip, and emit lasers parallel to the substrate surface, while VCSEL is to coat optical films on the upper and lower sides of the chip, which can emit lasers perpendicular to the chip surface. VCSEL has the advantages of low threshold current, stable single-wavelength operation, high-frequency modulation, easy two-dimensional integration, no cavity surface threshold damage, and low manufacturing cost, but the output power and electro-optical efficiency are lower than those of edge EEL.

Hu Panpan said that there are indeed some research institutions and manufacturers working on 1550 VCSELs, and samples have already come out, but their power is relatively small.

However, most respondents believe that 1550 is unlikely to be made into VCSEL.

A researcher from a 1550 LiDAR manufacturer said: "1550 uses fiber lasers, which require a gain fiber amplification process. Its output power is much higher than that of VCSEL - VCSEL is just a laser diode , and most of them are directly coupled to spatial light in array form without an amplification process."

However, according to Kuang Guohua, an expert in the field of optical communications and operator of the self-media "Optical Communications Women", fiber lasers have high power and large size. If they are made into VCSEL surface emission, there will be a limiting hole in the electronic circuit, which will lead to reliability risks.

In addition, Kuang Guohua also gave a detailed explanation of this topic in his article "Why 1310/1550 VCSELs are rarely seen" written in 2019. The author summarizes his core points as follows:

The resonant cavity of VCSEL relies on the DBR reflector, which is composed of two materials with high and low refractive index (the material of the reflector used in VCSEL is the same - GaAs and AlAs, and it can adjust the reflection of different wavelengths by simply adjusting the thickness).

To achieve a relatively high total reflectivity, if the difference in refractive index between the two materials is large, only a few layers are needed; if the difference in refractive index between the two materials is small, multiple layers are needed.

Excerpted from "Women of Optical Communications"

It is relatively easy to find two materials with a large refractive index difference in materials with a luminous wavelength of 850-905 (AlGaAs with different ratios), while the refractive index differences between several materials in the 1550 band of indium phosphide are very small (3.21, 3.37, 3.35). This means that the demand for materials for VCSELs made with 1550 will be much greater than that of 905, which leads to extremely high costs.

Image from "Women of Optical Communications"

In addition, the thermal conductivity of several 1550 materials (InGaAs or AGaInAs) is much lower than that of 905 materials. If they are made into VCSELs, they are likely to "burn as soon as the power is turned on."

How to understand this? If you choose 1550, its luminous material cannot be gallium arsenide series. It must use indium phosphide series, indium gallium arsenide or aluminum gallium indium arsenide. However, the refractive index difference of indium gallium arsenide is very small, which is not suitable for DBR. If you want to make it, because the refractive index difference is relatively small, its single-layer reflectivity is relatively low. It needs to stack a lot of layers to achieve a reflectivity close to 100%. In other words, if indium gallium arsenide material is used to match the gain layer of 1550 as a reflective layer, then it is difficult to get heat.

According to Dang Na, Onna also tried the 1550 VCSEL solution, but found that it "didn't work", so it gave up.

However, according to Hu Xiaobo, chairman of LeiShen Intelligent, there is "no point" in making 1550 into VCSEL, because "the beam quality of 1550's fiber laser is very good. We want to use 1550's optical laser as the light source, and we mainly focus on the beam quality of the fiber laser. If it is made into VCSEL, it will be no different from the 905 semiconductor laser."

6. Improving the detection end: 1550 is more difficult than 905

As mentioned before, one reason why the 1550 has a high transmission power is that the detection end has a relatively low sensitivity. In order to achieve a relatively long detection distance, the transmitter naturally needs a higher power than the 905. This means that one way to reduce the power consumption of the 1550 is to increase the sensitivity of the detection end. Is this idea feasible?

Since the silicon-based detector of 905 cannot absorb 1550nm light, it is necessary to pair the detector made of InGaAs material with the 1550nm InGaAs laser. Bao Junwei and Dang Na both said that relying on the development of the optical communication industry, the 1550nm detection end is already very mature. Dang Na believes that the next focus is to strive to expand its temperature range and improve its reliability so that it can better meet the needs of vehicle-mounted scenarios.

However, at this stage, even if we put aside the yield and cost factors, in terms of detection efficiency, response speed, and operating temperature range, silicon detectors for 905nm have obvious advantages over 1550.

Image source: https://www.ledinside.cn/qiye/20211222-51334.html

Hesai believes that although the power of the 905 laser is ultimately limited by human eye safety, a longer detection distance can be achieved by improving the efficiency of the detector. Currently, the mainstream 905 has begun to replace the APD with SPAD or SPiM with higher photosensitivity. In the future, the improvement of the 905 detector will mainly be to continuously improve the detection efficiency of single photons.

SPAD or SPiM are both "single photon detectors". The concept of "single photon detector" is relative to photodiodes (such as avalanche photodiodes APD) and CMOS  image sensors, which have a detection limit of tens to hundreds of photons. Compared with the latter, the main advantage of single photon detectors is: higher photosensitivity. 

Before explaining the detection efficiency, let's first explain the working principle of single-photon detectors. SPAD and SPiM are essentially photodiodes ( PDs). PDs will enter avalanche mode (also known as "Geiger mode") by continuously increasing the reverse voltage. In avalanche mode, a photon incident on the detector can generate a gain of 10 to the sixth power. Therefore, the effect of single-photon detection can be achieved.

The detection efficiency of a single-photon detector is a probabilistic indicator. That is to say, when I receive a photon, there is a probability that it will trigger an avalanche, and there is also a probability that there will be no reaction. For example, if the detection efficiency is 10%, it means that when 100 photons come in, there is a probability that 10 of them can trigger an avalanche.

Detection efficiency = quantum efficiency * photon avalanche probability * filling factor.

For silicon detectors, the quantum efficiency is basically over 90%; the photon avalanche probability is related to the production process; and the filling factor of Hamamatsu's detectors is over 70%.

According to Wang Lizidong, assistant to the chairman of Fushi Technology, "To improve detection efficiency, one is to improve the probability of capturing photons through SPAD device design and the other is to improve the material mixing and process level. However, these two are not immediate, because as the efficiency is improved, the noise will also be amplified; so most of the time, efficiency and noise should be considered comprehensively, and the higher the efficiency, the better. For example, a large Japanese manufacturer suppresses noise very well, so even if the detection efficiency is a little lower, it will not affect anything."

At present, the detection efficiency of single-photon detectors is about 10%, that is, every 10 photons received will trigger an avalanche. According to Hesai, this value will gradually increase to 20% and 30% in the future.