Working on a camera of our own using a specific IR CMOS sensor connected to an AD9826, we’re experiencing strange “noise” issues.
Our sensor sends out 3 pixels at a time which are sampled with AD9826 using the RGB inputs in 3 channels SHA mode.
The datasheet recommendation of having CDSCLK2 rising edge occur after ADCCLK falling edge is met (as outline in capture TEK00003.BMP)
The 3 pixels are available at sensor output about 180ns before CDSCLK2 falling edge (so we’re sure the signal is settled). What we observe, is that at CDSCLK2 rising edge then falling edge, a transient is observed on the 3 analog signal lines, about 350mV to 400mV peak peak (see TEK00002.BMP and TEK00007.BMP). To verify the origin of the transient, we disconnected (physically) the sensor outputs from ADC inputs, and we can observe the transient amplitude is reduced to less than 150mV peak peak (see TEK00004.BMP and TEK00006.BMP).
At last, I tried to remove physically the ADC part from the PCB to verify what its contribution to the remaining transient would be, and the result is about similar to the result observed with analog outputs disconnected from ADC inputs.
Anybody can give me leads or pointers on how to reduce the transient.
Thanks in advance.