I have been reading NIST information about spectral reflectance and uncertainty with a spectrophotometer with an integrating sphere
http://www.nist.gov/calibrations/upload/sp250-48.pdf
Also, I have found it with an application to colorimetry
http://www.cormusa.org/uploads/4_-_Colorimetry_Nadal_CORM2011.pdf
The issue here is that in order to calculate the total uncertainty I don't only need repetition of the test (with accounts to TYPE A uncertainty) but also the other sources of uncertainty such as signal to noise ration, signal uncertainty and so on. The problem is that I do not know how to get this values.
According to some texts I have reviewed, the computation of uncertainty is just like Ms. Nadal described in that link I present. But gathering those variables for the total uncertainty budget is what I have problems with.
Also the instrument gives me the 100% line, 0% line and the relative spectral reflectance of my sample. I am following the guidelines of ASTM E903.
I think you need to read document of ASTM E2554-13 in this link
http://www.astm.org/Standards/E2554.htm
Basically, this ASTM standard can bring your illustration about "Standard Practice for Estimating and Monitoring the Uncertainty of Test Results of a Test Method Using Control Chart Techniques". This document provides one way for a laboratory to develop data-based Type A estimates of uncertainty as referred to STM Standards.
Fundamentally, the information about ASTM E2554-13 are the standard measurement Lab based on Control Charts - Measurement Uncertainty - Precision - Precision And Bias - Statistical Analysis - Test Method Evaluation to Measurment Analysis Techniques.
I guess, if this session possible to get your additional information about detecting the uncertainty value from Control Chart Techniques.
Best regards
Mr. La Ode Husein Z Toresano thanks for your answer. I already checked the ASTM standard you cite. However, I am not only interested in Type A uncertainty but the whole budget. The budget is detailed in Ms. Nadal presentation.
You are after the Type B values. These are the systematic offsets and device or measurement uncertainties.
The S/N can be estimated by acquiring a spectrum with a signal and its background. Take the RMS fluctuation of the background as a representation of the noise level. A good exercise for the spectrometer is to measure the S/N as a function of integration time. You should find, the S/N goes as the inverse of the square root of integration time. Count for four times long to get twice the S/N.
The signal uncertainty can be estimated in the first step by the level of A/D conversion of your detector. A 2 bit convert can read from 0 to 22 - 1 = 3, and its relative uncertainty is 1/4 = 25%. An 8 bit converter by comparison can read to 255 with a relative precision of 0.3%. We can argue points of whether to divide the uncertainty by 2 or not as "fine details for further investigation".
Another estimate of signal uncertainty is just the RMS noise level itself during operation. Alternatively, a much lower value is the RMS noise level of the "dark count" of the spectrometer.
The A/D conversion uncertainty is generally well below the RMS noise level on instruments with A/D converters greater than 8 bit. One way to tell whether you need to consider A/D conversion over RMS is to look for "staircase" steps in your noise signal. The "digitization problem" becomes apparent for example on 8 bit converter systems where the RMS noise is ± 1%. A spectrum of a flat background at the full range of noise might show about 3-4 "staircase steps" (1%/0.3% ~ 3 - 4).
The systematic offset uncertainties are determined by a calibration of the instrument response given a well-known input. You may get a linear or non-linear calibration equation. The uncertainties on your fitting coefficients in a plot of measured response versus expected response are the systematic offsets.
This sounds like a great exercise for a focused lab project to define your instrument. I hope I have given some pointers to further searching even without exact references. Good luck and have fun!
- Badges
- Science topic
- Similar topics
- Engineering
- Mechanical Engineering