OPA992: Suggestion about provide bias for TIA

Hi Dewei,

of course, you can do so, allthough you only save two resistors per each TIA? But in any case you would need to isolate the output of OPA992 from the capacitive loads presented by the decoupling caps at the +inputs of TIAs. An advantage of this method is that all TIAs see exactly the same bias voltage. A disadvantage is that the very sensitive and vulnerable +input of TIA does not see a short wiring directly and closest to the OPAmp any longer but a largely pulled apart wiring being extended over 16 channels. This makes the bias line very sensitive to EMI and a simple decoupling cap may not be enough to gurantee proper operation. You would need to add a resistor at the +input of each TIA again to form a low pass filter in combination with the decoupling cap at the +input.

But there's still another issue. If you want highest precision, then you should run the photodetectors with zero bias. For a high precision photodetector all the specifications are usually given at zero bias. Going away from zero bias will introduce additional errors. Because of that I would connect the +input of each TIA directly at the OPAmp to signal ground (which is the best way to minimize EMI anyway) and would power the TIAs (and eventually the following OPAmps) with a small negative supply voltage of some hundreds of mV. The LM7705 may be useful.

Kai