Campbell Scientific CS505 Fuel Moisture Sensor - Manual (Page 14)

Appendix A. Explanation of Experimental Data
A-2
The full scale range is broken into four sub ranges: 0-10, 10-20, 20-30, and
30-50 percent. The computed rms errors are shown in Table 1. For each range
band an error band was determined in which more than 90 percent of all
measurements fall (also shown in Table A-1 and Figure A-1).
TABLE A-1. Accuracies at Measurement Ranges
Range 90% of All
Measurements
rms Error
0 to 10% ±2% ±1%
10 to 20% ±3% ±1.5%
20 to 30% ±5% ±2.2%
30 to 50% ±6% ±3%
*At least ninety percent of all experimental data fall within
these error bands at the noted range divisions.
The above accuracy is a static accuracy derived at slow changing conditions
with experimental data. In rapidly changing conditions (e.g., in the beginning
of a rain event) the rate of response for each wood dowel differs. After
sufficient time to equilibrate, the dowels will come to the same water content.
This difference in rate of response is inherent in the traditional weighing fuel
stick racks. The effects are lessened by the additional sorting of the dowels
performed at the factory. Readings above 50% are caused by surface water on
the probe or other external factors. Readings are mathematically limited to
70%.
The sensor error is the sum of three influences:
1. Error due to the variability and non-reproducibility of the response
characteristics of the wood: density, sorption characteristics, resin
content, etc.
2. Error due to the aging of the wood (cracking, discoloration, mass loss as
resin evaporates). This effect is not a factor for new sticks.
3. Offset error due to the electronic circuitry interchangeability, this error is
negligible over 10% fuel moisture.