Hi, in the datasheet of AD8475, it says that the voltage supply can be +-5V and +VS can be connected to 5V while -Vs can be connected to -5V or AGND. However, during simulation, it reports an error of singular matrix when connect -VS to -5V. Moreover, in CN-0385, CN-0345, CN-0269, all reference designs connect -Vs to AGND instead of -5V. So,
1, Is there any problem in connecting -Vs of AD8475 to -5V ?
2, I want to use ad8475 to drive AD4003, is it safe to use only 5VA and connect -VS to AGND without sacrificing any performances (i.e., CMR, THD, etc.)
Try ADA4940 or ADA4807. Also, please see Applications Information Section of the AD4003 datasheet.
Please note that simulation sometimes has computational problems that have nothing to do with actual part performance.
There are just two main concerns when setting the Vsupplies.
1. This part is specified only for total supply voltage difference up to 10 V. Beyond that some internal component may break down and make the part not function. Worse it may destroy the part and other circuitry connected to the part.
2. The output of these devices should not go much closer to any supply than about .4 to .5 V. You can tell that since the nonlinearity is specified with a 4 V swing when Vs+ is 5 V. and Vs- is 0 V. So if you have Vs+ = 5V. likely your performance will get better when you switch the -Vs to -5V. at the expense of doubling the power drain.
Concerning your final question you may sacrifice performance at Vs = 5 V. and 0 V. if you need to have Vout approach 0 V.
Even though the data sheet says the swing goes to -Vs + .05 V. there is no measure of quality, no criterion shown for this condition. The part may actually be railed in this condition generating much distortion.