OPA2381: Ability of Voltage Output Swing (Low)

Shinichi,

Please see my comments below:

(1) Could I think 50mV is three sigma? 

A short answer is NO - output voltage swing does not follow a Normal Gaussian distribution.  Since 50mV Vout swing above negative rail is a condition of AOL test (see below), 100% of all units must pass this test - this assures a minimum 110dB linearity.  

(2) Is there temperature spec?（For example, Ta vs. Low output voltage）

Since the spec is shown in BOLDFACE, it applies over -40C to 125C temp range.

(3) Is there a distribution data like a Figure of OFFSET VOLTAGE PRODUCTION DISTRIBUTION? 

Any unit that does not swing within 50mV of the negative rail would fail AOL test and thus would be discarded but we do NOT test how much closer to negative rail the output may get.  The actual swing to the rail is a function of output transistor sizing, output current, and P+, N+  doping levels.  Since the transistor size is fixed and output current the same under AOL test, the only variation in swing comes from the variation in the fab processing doping levels.  This is the reason why output swing may vary from 30mV to 50mV for wafers coming from different wafer lots BUT the output swing for the units coming from the same lot are almost exactly the same - for that reason any histogram of Vout swing would be a single column for each wafer lot.  Under no circumstances, however, should you ever find a unit that does not get within 50mV of negative rail with RL=10k connected to Vs/2 since 100% of all the units are being tested in the final test for this condition.