Section 4
4-21
The alternative approach was to use the original algorithm and go to the expense of actually
nulling the Primary Current and core offsets. However, we have a linear detector, and so we
are actually measuring their effects and can account for them in the present algorithm.
The final thing to notice is that in the expression for resistance ratio, the nanovolt detector
readings only appear in the form:
nV(+t)(+)- nV +t)(-)
nV(-t)(-)- nV(-t)(+)
(
and so if a perfect nanovolt detector was replaced by one with a gain and offset error i.e.
nVA(nV)+B = nV
Then the offsets subtract and the gain errors divide leaving the same fraction!
Hence all we need is an uncalibrated linear detector that can be allowed to drift slowly
with time.
4.2.6. Graphical Representation of Turns Switching and nV readings
See Figure 4-8
Note that this describes the original (pseudo model 9975) measurement algorithm and gives
insight into turns and nanovolt detector readings.
The graph may look confusing but the main point is that the nanovolt detector readings take
"quanta" jumps for every turn switched in or out. If you plotted these nV readings for a
given turn you would find that they were all on a line and that the signal acquisition
algorithm described in section 4.2.4 guaranteed that T
n
gave the smallest absolute value of
the nanovolt detector reading and by switching turns ±2 turns about T
n
we could interpolate
the line to find out where it crossed the Turns axis (i.e. where the nanovolt detector would
read 0). By doing this for both current directions we could calculate a resistance ratio for
Rx/Rs that correctly accounts for thermals and works quite well for Rs 1 k. The problem
appears at Rs 10 k and is accounted for in the present measurement algorithm.
To Next Page
Loading...