The CD Optical Pickup

Salvage a pickup from a dead CD player and identify the flexible cable's pinout from the part number, or use a discrete four-quadrant photodiode with a bare laser diode. Build one transimpedance amplifier PER SEGMENT: this is the circuit from the avalanche-photodiode blueprint, repeated six times, on +9 V and -9 V from two 9 V batteries. Do NOT tie photodiodes together to sum their currents. It saves parts and adds all their capacitances at the summing node, which destroys the RF bandwidth. Sum and difference are taken AFTER the transimpedance stages, always. Two different jobs from one detector. The RF signal must pass the disc's shortest pits: 3T pits at the 4.3218 MHz channel rate make a 720 kHz fundamental, so the RF path needs well over a megahertz, and its four stages and the summer run on OPA1612s at a low transimpedance (10k). A TL072 would not do for the summer: four inputs give it a noise gain of 5, which would pull its bandwidth down to about 600 kHz. The focus and tracking signals need only tens of kilohertz, so the side-spot stages run on TL072s at high transimpedance (1M) and are much quieter. Checked in ngspice with behavioural op-amp models and 5 pF per segment (it varies by pickup): about 12 MHz per fast stage and 8.4 MHz through the RF summer; about 76 kHz for the slow stages; and with deliberately unequal test currents, RF = -(A+B+C+D), FE = (A+C) - (B+D) and TE = E - F came out exactly. RV_BAL trims the weight of A+C against B+D, both ways, to null FE with the spot centred. Drive the laser diode from the constant-current driver of the laser-diode blueprint, at the datasheet current and no more. These diodes are cheap to kill.