I use your company's typical circuit AD8251+AD8475+AD7982 applied to multi-channel acquisition.
1.The test attachment figure. when the AD8251 input +5V, gain1, AD8475 input +5V, the AD8475 gain is connected to the IN0.4X pin, the AD7982 output reading is 0xCCBD (correctly so); When the AD8251 input +5V,gain2, AD8475 input +10V, in the AD7982 output readings 0x19979 (or so correct). correct linear. question: When the AD8475 input +5V readings 0xCCBD, sometimes change 0x2CCBD. the AD8475 input +10V readings 0x19979,sometimes change 0x39979.Symbol sometimes is "1". I do not know why?
2.Applications, the actual input is a positive signal, i.e., +5V or +10V. So can shield the sign bit.Use only the readings.But the conversion precision is only 17 bit.How positive signal can be used in 18 bit?
3. I used AD7982 ADR435 reference voltage is 5V output.The AD8475 is an attenuation of 0.4 times.+10V input does not correspond to the full value 0x1FFFF.If I use the 4.096V output of the ADR434,able corresponding to the full-scale value 0x1FFFF?
4.Attachment figure, the AD7982 digital interface signal through ADuM5401 and ADuM3100 isolated to the FPGA.AD7982 to achieve the sampling rate of 1MHz the FPGA provided SCLK rate 60MHz above.ADuM5401 only 25M rate.There to meet more than 60MHZ(3.3V) isolation chip?