Electrical noise in the OR Chris Thompson Senior Staff Specialist Royal Prince Alfred Hospital SYDNEY SOUTH WEST AREA HEALTH SERVICE NSW HEALTH Electrical noise in the OR
Root causes Tiny little signals with massive amplification High EMI / RFI noise levels Unshielded patient connection wires Stimulators close to sensor wires Unequal impedance patient connections Grounding issues many more
basic electrical concepts Voltage Current Resistance, Impedance Inductance and capacitance Electrical and magnetic fields Electromagnetic radiation Induced current Antenna theory Grounding Shielding Decoupling Balanced amplifiers Common mode rejection
Current - Amperes Electrons / sec flowing down conductor 1 Amp = 1 coulomb / sec = 6.242 x 10^18 electrons / sec Tesla car: 650A Starting current, car battery: 100A 50w 12V halogen: 4.2A 50w 240V light bulb: 0.2A Brain 0.000 001 A if that
Voltage - Volts Potential difference between two points Electrical field intensity Drives electrons down a conductor Static electricity > 10,000 s of volts Car battery 14V Cell membrane potential 100mV Surface ECG 5mV Surface EEG 50uV AEP s 2-5uV
AC vs DC DC = Direct current (or voltage) constant, or changes slowly e.g. car battery, solar cell output AC = Alternating current (or voltage) changes rapidly over time e.g. current to loudspeakers EEG Radio carrier signals Mains power
RMS vs Peak to Peak 240V AC has the same power as 240V DC i.e. a light globe would be equally bright. Frequency : Hz cycles/sec
Power vs Amplitude of the EEG Power = Volts * Amps i.e. Voltage squared Resistors get four times as hot when voltage doubles. EEG power spectrograms: a) emphasise peaks b) hide low amplitude changes
Resistance - Ohms Ω Resistance to electron flow - Ohms Plastics, air: 10^14 ohms Dry skin: 10,000-10,000,000Ω ECG dot, prepped skin: 300-10,000Ω Needle electrodes: 600-2,000Ω Copper wire per metre: 0.001 Ohm For AC voltages, use Impedance.
Internal Resistance A new car battery can deliver a lot of current and keep it s voltage up. It has LOW internal resistance A very old weak car battery cannot deliver much current without its voltage dropping greatly It has HIGH internal resistance
Measuring biomedical voltages All measuring devices draw *some* current Drawing current from the source drops the apparent voltage being measured Most obvious when the voltage source has high internal resistance (high output impedance), e.g. brain High input impedance devices draw the least current Input impedance of EEG machines must be VERY high e.g. >>10MOhm
Impedance With AC signals, resistance varies with frequency Impedance is the equivalent of DC resistance at a specified frequency Capacitative impedance falls as frequency increases
Skin Impedance Frequency dependent Gel Electrodes unprepared skin Falls as frequency increases At low frequency, skin impedance is largely resistive At intermediate frequencies, it is capacitively dominated and drops rapidly with increasing frequency. At high frequencies, it becomes resistively dominated and stays low. 1cm^2 gel electrode 10k-1M skin impedance at 1Hz (dc almost) 1000 times less at 100kHz IEEE Transactions on Biomedical Engineering 53 35(8):649-51 September 1988
Needle Impedance Also falls with increasing frequency Usually measured at 10kHz Typical values 0.5-2 kω Stainless needles, different types Kalvøy, H., Tronstad, C., Nordbotten, B. et al. Ann Biomed Eng (2010) 38: 2371. doi:10.1007/s10439-010-9989-2
Capacitance unit: Farad A one Farad capacitor stores 1 coulomb of electrons at 1Volt on parallel conductive plates infinitely high DC resistance (blocks DC) frequency dependent and lower resistance for AC exists between any conductive materials in proximity - capacitative coupling Farad from Michael Faraday, 1791-1867, English physicist
Capacitative impedance falls as frequency increases
Capacitors are widely used to short circuit high frequency noise to ground Simple RC Low Pass filter Time constant High frequencies attenuated above cutoff frequency Fc = 1 / 2πRC τ = RC
Low pass filter results red = input blue = output High frequency noise much less Overall amplitude less Less likely to hit a rejection threshold Waveform in pass band more distinct BUT phase shift delays output
High Pass filters block low frequencies e.g. to prevent baseline wander RC HIGH pass filter τ = RC blocks LOW frequencies below cutoff frequency Fc = 1 / 2πRC
Effect of a high pass filter on baseline wander ECG Same applies to raw EEG waveforms e.g. removes surgical hand movement which may shift the baseline so much as to hit the artefact rejection threshold *but* makes delta activity less obvious
Noise sources Mostly 50Hz hum from mains wiring Random switching of power supplies, fluoro lights etc Mobile phones contacting towers Sparks, pulses from diathermy Stimulator pulses Muscle tone, surgical movement, tools All appear on signal wires as electrical noise Amplified 1000-100,000 times
Grounding issues AC current flows through earth wires all the time If grounding is imperfect, AC voltages appear on the ground and in shielding Modern patient circuits are floating type The patient usually is *not* grounded So what happens on the mains side shouldn t matter BUT isolation of the patient circuit is not perfect Small swings in ground voltage >> EEG voltages Always check ground wires if intractable 50Hz noise
Electromagnetic Waves In-phase, oscillating electrical and magnetic fields Propagate through a vacuum Once emitted, are independent of the source (unlike static fields) e.g. Radio waves, 240V wiring, electrical machines, diathermy, impedance checkers
Electromagnetic Pulse Pulse or burst of electric, magnetic or electromagnetic energy Typically electromagnetic Natural: lightning, meteor impacts Man-made: Nuclear bomb (induces10kv/m) Any switching on/off of electricity Switchmode power supplies Spark plugs, brushed electric motors Nerve stimulator pulses
Wavelength Speed of light / frequency Mains 50Hz: 6000km Diathermy 1MHz: 300m 300 x 10^6 m/s f FM Radio 100MHz: 3m Bluetooth 2.4Ghz: 12cm
Australian frequency allocation
EMI / RFI = electromagnetic interference Free ended wires make great ANTENNAS ½ to ¼ wavelength antennas resonate with incoming EM waves Induced voltage / current on each end of any unshielded wire ¼ wave antenna: FM radio 0.75m BT 3cm EMP is picked up by any length wire
Capacitative coupling Two wires close together are weakly capacitatively coupled Electric fields on one wire can be capacitatively transferred to the other; = noise The patient is capacitatively coupled to wires, shielding in the walls, the operating table, etc. Primarily at 50Hz. Result is very high common mode 50Hz noise signal. Farad from Michael Faraday, 1791-1867, English physicist
Figure 6: Electromagnetic interference (EMI) ways (for the capacitance values [9]). Arrows show the interference currents. (I) Voltage due to magnetic field to electrode cable loop is illustrated. (II) Displacement current on subject head due to electrical field causes voltage drop across electrodes. (III) Displacement current on subject body due to electrical field causes voltage drop across electrodes. (IV) Additionally, this current causes voltage between measurement electrode and amplifier common pin [24].
Inductive coupling Changing magnetic field flux induces current. Basic principle behind transformers and generators Two wires close together are weakly inductively coupled for AC signals Amount of induced current/voltage depends on - magnitude of current in source - proximity to the magnetic field
What happens to induced current in patient monitoring wires? Partly shorted out in the patient Patient presents a low resistance pathway to ground via electrodes The lower the electrode impedance, the more the noise will he attenuated Remainder appears as a false signal - i.e. noise on EEG and ECG monitors
Ways of reducing noise Eliminate bad sources of EMI / RFI Equal impedance @ patient connection Don t run signal wires in parallel with noise wires Shortest possible exposed wire length Equal length signal wires Twist common input pairs Shield all low level signal wires, cables, preamp Front end filtering (steep high and low pass) EP signal averaging High quality amplifiers (high CMMR, high impedance, low noise)
Common mode rejection Only the DIFFERENCE gets amplified Rejection of anything COMMON to both wires CMMR = common mode rejection ratio typically > 100dB greatly improved by filtering out unwanted noise before and after amplification
Common mode rejection limitations Impedance mismatch allows noise to masquerade as signal (noise goes more easily down the lower resistance patient connection, instead of equally down both) Massive noise levels can overwhelm CMRR Massive noise levels can overwhelm the input amplifier
Summary Eliminate bad sources of EMI / RFI Equal impedance @ patient connection Twisted pair, equal length, short signal wires Not parallel with noise wires Shortest possible exposed wire length Shield all low level signal wires, cables, preamp Correct filtering (steep high and low pass @ optimal values) Estimate the noise frequency - look for causes 50Hz noise look for ground problems (cable/power point)