二維聚焦平面陣列雷射光偵測器開發

Abstract

本計畫執行製作一顆具有 2D 照像及3D 測距的功能CMOS 影像偵測器。 本計畫提出用接收調變雷射光源的差距/比例來計算3D 距離， 並已經用 16X16 CMOS 影像器的POST-LAYOUT-SIMULATION 驗證 (使用TSMC 0.35 um 2P4M 的CMOS 製 程) 。其中每一像素由光二極體， 一個自偏型反向放大器及兩個 S/H OTA amplifier 構成。 S/H OTA amplifier 則採用folded-cascoded type。兩個 S/H OTA amplifier 在 3D 測距扮演取樣接收雷射光源調變波正半週與負半週，訊號儲存在S/H 的 CAPACITOR， 因此 3D/2D 影像偵測器具有快照(SNAP-SHOT) 的功能。自偏型反 向放大器扮演放大光電流的角色， 來克服來自 SWITCH 的CHARGE INJECTION ERROR， CLOCK 的FEEDTHROUGH, 及S/H OTA amplifier 非理想效應 (gain, offset vltage) 。由於使用自偏型反向放大器， 因此需要兩個 RESET 控制訊號， 一個是光 二極體， 一個是自偏型反向放大器。 兩個 S/H OTA amplifier 取樣接收雷射光源調變波正半週與負半週之後， 則由後級的 DDS 做差距的動作。 然而在 POST-LAYOUT-SIMULATION 中發現一些非理想效應， 由於光電流非常微弱， 再經過 DDS 後的訊號只有MILLIVOLT。這微弱的訊號 ，如果再加上測試板上的雜訊 考量， 對 ADC 處理上具有相當的難度，因此在本計畫中預計將在最後一級加入放大 器, 並研發整合ADC 與CMOS 影像器的技術。 此外， 兩個 S/H OTA amplifier 的匹配問題， 將重新 layout 像素電路， 來增加兩個 S/H OTA amplifier 的匹配性。 本計畫也預計在下一版改善邊緣的像素偶合問題, 加入dummy cell 來改良boundary cell 的特性, 並找出偶合的源頭。

In this project, a CMOS imager is implemented with the capability to take 2D pictures and extract the 3D distance. We propose a difference-ratio method using a modulated laser source to extract the 3D distance of the target. Our difference-ratio method has been verified with the post-layout-simulation results of the implemented 16x16 CMOS imager in TSMC 0.35um 2P4M technology. The pixel we use includes a conventional 3T APS, a self-biased inverter, and two S/H amplifiers implemented in folded cascaded structures. By using the S/H amplifiers to hold the photo-signal, the snap-shot operation is achieved in our chip. By using the self-biased inverter to amplify the photo-signal, we therefore boost the photo-signal buried in the error sources coming from charge injection of the switch, the clock feed through, and the non-ideal effect of the S/H amplifiers (gain, offset). By the use of the self-biased inverter, we therefore need two reset controls, one for the photodiode, the other for the self-biased inverter. The reason to use two S/H amplifiers in a single pixel is that we use the exposure results from the positive and negative cycles of the pulsed lasers to do the difference-ratio method. The difference operation is readily done by the DDS operation. However some non-ideal post-simulation results are found in this version of our work. The weak photo-signal, though amplified, after the DDS is still on the order of 0.001 volt. This small voltage is difficult to be processed by the ADC if we take the noise from the test board. We therefore would like to add an amplification stage at the output stage and develop the technique to integrate the ADC on chip. Another problem is the match of the two S/H amplifiers in the pixel. We would like to re-layout this part to improve the match. The boundary cell also has bad coupling from somewhere, thus they have the worst simulated ratio compared with the actual ratio. We would add the dummy cells and find out where the coupling comes from.