GENERAL MEASUREMENT: PHYSICS -->| |->SAMPLE -->| unwanted -->| |->ELECTRONICS ->| physics unwanted->| |->DATA--> CAKE signals unwanted--->| data Physics that leads to cake {data} = signal Unwanted {physics+signals+data} = noise + interference signal Figure of merit: ------------------ noise+interference (frequency) ---------------------------------------------------------------------- NOISE & INTERFERENCE IN A MEASUREMENT 1 --------------------------------------- NOISE: Looks like : random-like "continuous" spectrum Origin : nature-made Lowest level: set by physics and electronics INTERFERENCE: Looks like : specific spectral components Origin : man-made Lowest level: zero GOOD SETUP : interference < noise-floor ---------------------------------------------------------------------- NOISE & INTERFERENCE IN A MEASUREMENT 2 ------------------------------------- 1- CHECK FOR INTERFERENCE: Use analog scope, trigger mode = normal (no trigger = no light) turn trigger level knob see a repetitive signal ? (vary timebase) --> Interference 2- CHECK FOR NOISE: a calculate total RMS noise from sample + electronics for a given bandwidth b use a filter to limit the bandwidth, connect to scope estimate top-top value, RMS-value = (peak-peak)/6 c If b < 2*a then setup = ok ---------------------------------------------------------------------- SIGNAL BANDWIDTH versus MEASUREMENT BANDWIDTH --------------------------------------------- Example: Ibias -> Rsample -> wires -> Voltage amplifier some values: Cwires : 300 pF Rsample: 100 kohm Rnoise : 50 nV/sqrtHz Signal bandwidth : dc..5 kHz However: As the output signal drops >5kHz also the noise level drops The usefull MEASUREMENT bandwidth can also be larger. example extending the signal bandwidth: Using a dedicated amplifier design with a differentiating response from 5kHz up to 5MHz: (ampnoise = 0.5 nV/sqrtHz) Signal bandwidth : dc..5MHz noise <500kHz : 50 nV/sqrtHz noise 1MHz : 100 nV/sqrtHz noise 5MHz : 500 nV/sqrtHz ---------------------------------------------------------------------- CONCLUSIONS -------------- -Only the ratio between the signal and the combined noise and interference in the used bandwidth should be considered, not the absolute levels. -Noise should be discriminated from interference to be able to improve and qualify the measurement. -Noise is an instrumentation and sample problem. Interference is a construction and environment problem. -A wellbuild measurement setup is limited by noise -Signal bandwidth is an overrated parameter as it has no direct relation to the measurement bandwidth.