LiDAR SoC Process Choice: 7nm vs. 28nm? Power and Heat Are Key

Been looking at domestic LiDAR solutions lately. Everyone is hyping their point cloud density and detection range. But from a chip design perspective, if you pick the wrong underlying SoC process node, all subsequent performance is built on sand. I saw one company using 28nm for MEMS drivers and data processing, and they hit a power wall that locked them at a 60-degree temperature rise, processing less than 200k points per frame. So, the trade-off between area and performance at this node needs to be handled very carefully. If you don't move to advanced nodes, you can't increase the parallelism of TDC and histogram processing, meaning you simply can't catch distant small targets. But moving to 7nm brings challenges with high tape-out costs and thermal design. To my brothers doing system integration: what junction temperatures are you measuring on your LiDAR main control chips? Let's compare notes.