My goal is to implement an analog-to-digital conversion. Therefor I used the LT6203 as an input buffer. The analog signal is directly applied to the OpAmp (IN+ and IN-) as shown in the attached schematic:
I supplied the LT6203 with +/-5V and have executed single-ended measurements with an input range of -5V to +5V, which were not satisfactory (nothing connected to BufferP and BufferN). While there is nothing to criticise for the results of negativ input voltages, in case of positive inputs, the deviation badly increases by approaching the upper rail. So I applied voltages from an external power supply unit to IN+ and referenced IN- to GND and there are deviations like this:
IN+ = 4.40V BufferP = 4.4019V BufferP-to-BufferN = 4.4011V
IN+ = 4.75V BufferP = 4.7522V BufferP-to-BufferN = 4.7514V
Is there a posibility to get more precise results (deviations beneath 1mV) or has anyone an idea what might be the reason of this inaccuracies?
I have also tested other amplifiers, like the LT1819 and the ADA4851, and repeated the same measurement, but the results were even more bad:
LT1819: IN+ = 2.0V deviation = 1.4mV
IN+ = 4.4V deviation = 19.5mV
ADA4851: IN+ = 3.5V deviation = -7.77mV
IN+ = 4.0V deviation = -396.0mV
Especially the ADA4851 delivers precise results for negative input voltges and for lower positive voltages (up to 3V). It would perform well, apart from the mentioned problems at higher voltages.
Am I doing something wrong or do I have overlooked something? I would appreciate if anyone has some tips, alternatives for more precise amplifiers or an idea what might be the source of error.