there are two issues:

1) unknown resistor value: you can easily work that into the program, assuming that the resistor value is stable.

let's say that you apply Vs to the divider network / adc and the adc reading is ADCs.

1. #define Vs   10000ul //Vs is 10000mv
   
2. #define ADCs 456 //associated adc reading is 456
   
3. 
   
4. #define ADC2V(adc)  ((adc)*Vs / ADCs) //convert adc reading to voltage reading, in mv

复制代码this macro assumes 1) linearity and 2) no zero offset

2) to correct for non-linearity (and offset), you need a lot more work and the likely solution is a table.

rclong 发表于 2012-4-16 19:10
线性测量参考方程Y=aX+b 通过两个点(X1,Y1),(X2,Y2)求解实际电路的a和b
可以通过两者计算增益误差和偏移误 ...

我也是这样
我的做法是均匀取10点，分别用AD7705和34401A测量，然后直线拟合得到y=ax+b的a、b。
AD7705很准，就是太慢了，后来用了LTC1865，也还不错

关注下。哈哈。

本帖最后由 hhxb 于 2012-5-1 21:01 编辑

imfv 发表于 2012-4-16 20:49
如果能够用程序来实现阻值的误差，岂不是更加好，电阻我准备买0.1%的电阻！ ...

我去，软件能全校准的话，估计安捷伦就可以破产了

梅特勒是采用的5点校正法，安捷伦同样存在软件校准，当然这需要有精密基准