如何保持功率准确度

如果您查看E4419B的功率计数据表，则声明的绝对精度为+/- 0.02dB。
规范还规定，如果超过传感器的某个功率水平，则需要添加线性偏差。
根据我所做的许多读数，似乎如果你保持低于那个功率水平，那么你可以假设传感器的线性偏差可以忽略不计。
也许这是一个错误的结论。
当我使用安捷伦的不确定度计算器时，我对如何解释结果感到有些困惑。
在我考虑了传感器选择后，如果我保持在最佳线性度范围内，我最终会看到不确定性水平如下：ULU：+ 0.075dB LLU：-0.076dB似乎永远无法击中+ /
如果不确定性大得多，则为-0.02dB规格。
我在这里想念的是什么？



     以下为原文

  If you look at the E4419B's power meter data sheet there is a declared absolute accuracy of +/-0.02dB.  The spec also states that you need to add in the linearity deviation if you exceed a certain power level to the sensor.   Based on many readings I have done, it seems that if you stay below that power level then you can assume that sensor linearity deviation is fairly negligible.  Perhaps this is a wrong conclusion.  When I use Agilent's uncertainty calculator I am somewhat confused in how to interpret the results.  After I factor in the sensor choice, and provided I stay within the best linearity range, I end up seeing uncertainty levels as follows:

ULU: +0.075dB
LLU: -0.076dB

It seems that one would never be able to hit the +/-0.02dB spec if the uncertainty is so much greater.  What am I missing here?