Transcript Slide 1
(Analog) Data Acquisition D. Gordon E. Robertson, PhD, FCSB Issues • • • • • • What type of signal? (AC or DC) How many signals (channels) will be needed? Characteristics of signal (range, frequency) How does the A/D converter record the data? Is signal conditioning necessary? (amplification, filtering) How will the data be saved for use by other software? What Kind of Signal? • Use AC-coupling to remove biases from AC signals such as EMGs, EEGs or ECGs. • Use DC-coupling for all other signals • Piezoelectric signals, such as from accelerometers or force transducers, may require recording of bias levels for zeroing EMGs are AC Forces are DC How Many Channels? NI 32-channel cDAQ A/D • A/D converters sample multiple signals one channel at a time. • Typically 4, 8, 16, 32 or 64 channels are possible • Note, the more channels that are sampled the lower the maximum sampling rate • A/D converters have maximum rates (e.g., 100 kHz) that must be divided among the number of sampled channels • Most systems sample all channels at the same rate. Some allow different rates for each channel. – 1 Kistler force platform: 8 channels AMTI force platform – 1 AMTI force platform: 6 channels – 1 3D head accelerometers: 9 channels – 1 joint elgon: 1–3 channels Biometrics 2-channel electrogoniometer Characteristics of Signal • Voltage range: – 0 to +5 V, 0 to +10 V – +/–1.5 V, +/–5 V, +/–10 V • What is the signal’s frequency spectrum? – forces: DC to 10 Hz – audio: 20 Hz to 20 kHz – EMG: 20 to 500 Hz What Signal Conditioning is Done? • Amplification or attenuation to fit dynamic range of A/D converter – 1000x for EMGs – force transducers 1/100, 1/10, 1/1 Honeywell bridge amp • Analog filtering – band-pass filtering of EMGs Bortec 8-channel EMG, gains from 1x to 15000x How does the A/D converter Record the Data? • number of data bits: • • • – 8 bits gives 0 to 255 – 12 bits gives 0 to 4095 – 16 bits gives 0 to 65 536 – 24 bits gives 0 to over 1 million are the data stored in ones or twos complement or without sign bit bit formats: Intel, DEC, Sun software must be able to convert the integer data to floating point in volts or other measurement unit Relationship between Integer and Real Data • resolution = voltage range/bits • for 12 bit, +/–10 V; resolution = 20/4096 = 0.0048828 V/bit 8 bit 12 bit 16 bit 255 4095 216–1 128 2048 215 0 0 0 +/–12 bit 2047 input V actual +10 V +5000 N 0 0V 0N –2048 –10 V –5000 N How Will the Data be Saved? • integer format is the most compact and fastest • integer format must be converted to floating point by other software (e.g., BioProc2/3, MATLAB, commercial software) • real format (machine language) requires less processing but more memory • ASCII requires greatest amount of memory but is easiest to read by other software (Excel, QuattroPro, SPSS) and may reduce accuracy due to rounding • public format: .C3D, .WAV, .EDF, .WK1 • spreadsheet: .WK1, .XLS, .QPW