Test 1 Flashcards
measurement
assigning a value to a physical quantity based on a standard
data domains
the ways that information can be encoded in an object or via a signal throughout the measurement process
error types
random, systemic (consistently lower or higher), human (bias)
what measurement represents precision
standard deviation, = random error
calibration sensitivity
slope of the detection line
analytical sensitivity
slope over standard deviation
ppm=
mg/L
ppb=
ug/L
flow of information in an instrument
stimulus (from device)> sample > response> transducer > some data domain> information processor> readout (usually a computer)
analog quantities are measured
by magnitude (electrical current, etc) over time
transducer
interconverts info from electrical to nonelectrical domains
input transducer
transducer that takes info from nonelectric to electric domains
detector
mechanical, electrical, or chemical detector to show change in temp, pressure, electricity, radiation, ion conc, *smoke detector
readout device
transducer that converts from electric to nonelectric domains
sensor
monitors a specific chemical species and changes reversibly in response to a change. has recognition>transducer>readout *glucose sensor
how to select instrumentation
consider required accuracy, amount of sample, concentration range, interference, physical and chemical properties, and number of samples to be processed
pop. mean and SD
u and sigma
sample mean and sd
x bar and s
normal distribution
1s: 68.3%, 2s: 95.45%, 3s: 99.73%
least square assumption
error in y > error in x
bias (gamma)
calculated as y= u - xbar (pop. - sample). = system error
accuracy calculation
comparing sample mean and pop. mean (bias calculation)
LOD equation
C = (3Sbl/m)
selectivity eq (general anal chem)
S = maCa + mbCb +……
selectivity coefficient (for interfering spp in sensors)
k(b,a) = mb/ma (closer to 0 is better)
dynamic range
which concs have reliable results, located btwn the LOQ and LOL
type 1 error
false positive
type 2 error
false negative
LOQ
limit of quantitation, above in measurements have 5% standard error
degree of freedom
n-2 because we need 2 of the points just to make a line
degree of freedom in sample mean
n-1 bc the mean occupies one number
Cdl
=3Sbl/m (the conc at the DL is 3 SD of the blank out)
standard deviation units
same as the mean
relative error
(Cmeasured - Creal)/Creal
thermal/Johnson noise
charge carriers in resistors, capacitators, (anything with resistance) cause random error corresponding to the bandwidth and temperature. - frequency independent
shot noise
quantized particles (whole numbers) cross a junction, cause random noise, frequency independent. can be dec with bandwidth change.
flicker noise
aka 1/f noise, inversely proportional to frequency
environmental noise
frequency dependent bc it is at specific frequencies, non-fundamental, things like EM waves from radio
S/N
signal to noise ratio is set equal to mean signal/standard deviation also = 1/RSD
non-random noise
systemic error in measurement equipment causes high or low interference
fundamental noise
noise due to uncertainty principle etc or factors of physical instrumentation
noise reducing hardware
grounding/shielding, difference/instrument amplifier, analog filter, modulation
noise reduction in measurement
take the same measurement many many times using S/N=sqrt(n)(Sx/Nx) for desired ratio
n in noise/signal
the number of measurements conducted
S/N at which signal is indiscernible
S/N =3
conjugate quantities
when one is measure precisely it is impossible to measure the other precisely ie. position and speed, creates fundamental noise
white noise
aka gaussian noise, any frequency independent noise (thermal, shot)
A/C frequency (in US)
60 Hz
chemical noise
variations in chemicals (ie batches of the same soln vary by 0.1%)
rising time
time interval it takes between 10% and 90% power of an electrical signal when activated
Poisson process
used to measure probability of something discrete (ie e- for shot noise) P= (lambda^k(e^-lambda))/k!
bandpass filter
filters out high and low ranges leaving some middle range of signal
high-pass filters
filter out long wavelengths
low-pass filters
filter out short wavelengths, often used for DC, thermal and shot noise
grounding and shielding
physically prevent electronic interference by surrounding with grounded conductive material which attracts signal to make it not be noise.
modulation
used for DC or low frequency signal, as 1/f noise is high, and then high pass filter can be used and it can be returned to og frequency with reduced noise. example: mechanic chopper
ensemble averaging
higher n to approach true value thru averaging