TLC271: TLC271 drift

Chen,

Thanks for your question.  The problem you are having is due to the large offset voltage on the TLC271 multiplied by the gain.  The maximum input offset voltage for TLC271 is Vos = 10mV.  This offset will be multiplied by the amplifier gain which is 160 V/V (160kΩ/(1kΩ).  The maximum output due to Vos is 10mV x 160 = 1.6V.  Note that when T+ and T- are shorted the gain is set by R44 and R40.  Op Amp Offset Voltage and Bias Current Limitations and the Op amp Precision Labs section on input offset voltage cover this topic in detail.  You may want to consider zero drift amplifiers (lower cost options) as these devices have very low offset.  Alternatively, choosing any general purpose device with offset of about 1mV will reduce your error by a factor of 10.

I hope this helps.  Best regards, Art

Chen,

It means that for the measured population, there are two devices that are outside the graph boundaries.  These devices are considered to be "outliers" because they are not part of the normal distribution.  Note that for TLV271 the Vos temperature drift specification is given as a typical specification only.  Technically this means that it is possible that it is possible to get large values for drift.  The typical value has a statistical meaning(± one standard deviation of 68% of the population).  If the typical is given as ±2uV/C, 68% of the devices will have a drift in the range of ±2uV/C.  Without a maximum specification you can estimate that the majority of the population is ±3 standard deviations (99.7% of population).  So the tails of the distribution reach to ±6uV/C.  However, without a maximum it is possible that there will be some non-gaussian outliers.  If there is a maximum limit that means the devices are tested over temperature and the outliers would be discarded.  Thus, if temperature drift is critical you need to choose a higher precision device (higher cost also) that has a maximum limit for drift.

Best regards, Art