`
`APL_MAS_ITC_00015738
`
`104
`
`Design of pulse oximeters
`
`process it, is dependent on the clock rate. The clock controls the duty cycle. Duty
`cycle is defined as the fraction of time the output is high compared to the total
`time. When designing such subsystems we have to examine the duty cycle.
`8. 3.4. 1 Clock generator and timer circuit. A 555 timer can be used to generate a
`n-minute timer, but isn't accurate enough for this kind of application. For more
`precise timing we usually use a signal derived from a crystal-controlled
`oscillator. This clock is stable but is too high in frequency to drive a processor
`interrupt input directly. Therefore, it is divided with an external counter device
`to an appropriate frequency for the interrupt input. Usually such a system
`contains counter devices such as the Intel 8253 or 8254, which can be
`programmed with instructions to divide an input frequency by any desired
`number.
`The big advantage of using these devices is that you can load a count into
`them, and start them and stop them with instructions in a software program.
`Sometimes addition of a wait state may be needed along with this device to
`compensate for the delay due to the decoders and buffers on board.
`We usually reset the circuit using simple resistors and capacitors, which are
`held low during power-on. This maintains the logic at a known state, while the
`crystal oscillator and the power supplies stabilize.
`The timer circuit could control the following units on the subsystem:
`
`1. Set the baud rate of the UART communication network.
`2. Generate interrupts for controlling the display circuit, as these are usually
`multiplexed to avoid use of high current.
`3. Audio frequency generator, for alarms.
`4. Clock frequency for the notch filter, used to suppress the power line noise.
`5. Synchronous circuit operation for pattern generator.
`8. 3.4.2 Watchdog timer circuit. This is a kind of fail- safe timer circuit, which
`turns the oximeter off if the microprocessor fails.
`A counter controls the input to a D flip-flop, which is tied to a shutdown
`signal in the power supply. The counter is reset using a control signal from the
`microprocessor and a latch. Using some current-limiting protection, this signal is
`ac-coupled to the reset input of the counter. Therefore if the counter is not reset
`before the counter output goes high, the flip-flop gets set and the power supply is
`turned off.
`8. 3. 4. 3 UART. Within a MBS , data arc transferred in parallel, because that is the
`fastest way to do it. Data are sent either swichronously or as.wich,-mu,usb,. A
`UART (Universal Asynchronous Receiver Transmitter), is a device which can be
`programmed to do asynchronous communication.
`
`8.3.5 Pattern generator
`
`This is a multipurpose section which is primarily used to generate timing patterns
`used for synchronous detection gating, LED control, and fur synchronizing the
`power supply. The heart of this system is the EPROM (Erasable Program mable
`ROM). Here preprogrammed bit patterns are stored and are cycled out through
`the counter, and are tapped off using certain address lines. Using the address the
`bit pattern is sent out and latched using an octal latch. There may be an additional
`
`1
`
`1
`
`121
`
`MASIMO 2053
`Apple v. Masimo
`IPR2022-01299
`
`
`
`RX-0035.0122
`
`APL_MAS_ITC_00015739
`
`Electronic instrument control
`
`105
`
`latch used to deglitch the system, in which the last byte i s held until the counter
`increments itself to the next address and the next pattern is obtained. Various
`patterns within the EPROM are used to select the sampling speeds of the LEDs or
`the synchronous detector pulse, the calibration patterns and diagnostic timing.
`
`8.4 ANALOG PROCESSING SYSTEM (NELLCOR®)
`
`8.4.1 Analog signal.flow
`
`Signals obtained are usually weak and may have electromagnetic interference.
`These signals must be filtered and then amplified. Usually a 50 or 60 Hz low-pass
`(for example may be a 2nd order Butterworth) filter is used. The signal is then ac
`coupled to stages of amplifiers and depending on the kind of response, variable
`gain circuits can be designed. The aim here is to maximize the signal before it
`enters the detector circuit, where the IR and red signal are separated, so the
`signal-to-noise ratio (SNR) is also kept as high as possible.
`
`8.4.2 Coding resistor, temperature sensor, and prefiltering
`
`Before examining the analog signal flow path, it is necessary to mention how the
`MBS decides what compensation to use for those LEDs which do not have their
`peak wavelength at the desired value. As LEDs are manufactured in bulk and
`tested in a random fashion, the probes may not always have the LEDs with the
`desired wavelength. Therefore the MBS generates some compensation, in order
`to solve this problem. Probes must be calibrated. Pulse oximeter systems have a
`coding resistor in every probe connector. A current is provided to the probe
`which allows the MBS to determine the resistance of the coding resistor by
`measuring the voltage drop across it. Thus the particular combination of LED
`wavelengths can be determined. Following this the MBS can then make necessary
`adjustments to determine the oxygen saturation.
`As the wavelengths of the LEDs depend on the temperatures, for accurate
`measurements the effects of the temperatures must also be known, for adequate
`compensation (Cheung et al 1989). A temperature sensor may be employed,
`whose signal along with that of the coding resistor is used to select the calibration
`curves which are to be employed for compensation.
`Despite efforts to minimize ambient light interference via covers over the
`probes and sometimes red optical filters, interfering light does reach the
`photodiode. Light from the sun and the incandescent lamp are continuous. The
`fluorescent light source emits ac light. This may overload the signal produced by
`the photodiode in response to the light received.
`
`8.4.3 Preamplifier
`
`The photodiode generates a current proportional to the light incident upon it. The
`signal from the photodiode is received by a preamplifier. Figure 8.4 shows- that
`the preamplifier consists of a differential current-to-voltage amplifier and a
`single ended output amplifier. A gain determination resistor converts the current
`flowing through it into voltage. But along with the current-to-voltage conversion,
`external interference is also amplified, making the true signal difficult to extract
`
`122
`
`MASIMO 2053
`Apple v. Masimo
`IPR2022-01299
`
`
`
`RX-0035.0123
`
`APL_MAS_ITC_00015740
`
`i
`
`106
`
`Design of pulse oximeters
`
`from the resulting output. The differential amplifier produces positive and
`negative versions of the output. This dual signal is then passed via a single ended
`amplifier with unity gain, which results in a signal with twice the magnitude of
`that of the input. Due to the opposite signs of the outputs of the differential
`amplifiers, the external interference is canceled out. As the noise factor increases
`by a marginal factor the signal-to-noise ratio improves. The mixed signal is then
`fed into two sample-and-hold (S/H) circuits whose timings are controlled such
`that each circuit samples the signal input to the demodulator during the portion of
`the signal corresponding to the wavelength to which it responds.
`
`To Red LED
`
`·-CE]
`
`A Pgm
`R Per,
`DC oflset -t gain
`amplifier
`
`Preampliner
`
`Demodulator
`
`Red
`
`--+Vamp Winded H
`
`From pholodetector
`
`St-I
`
`Fdle,
`
`Mux
`
`fir-1--f-2/1
`
`To IR LED
`
`~ Infrared
`IR Pgm
`IR Pom
`DC offset -* gain
`ampliner
`
`(a)
`
`To R drv
`
`Sample,
`ambient
`
`To IR drv
`
`To R drv-- 1 V -
`7 Driver F-
`To IR dn,- 1 Circuit ~-
`
`Mux
`
`DAC
`
`1-.Vref
`From MBS
`
`Sampleand
`hold circuit
`
`prom MBS
`
`(b)
`
`Figure 8.4 The analog signal flow path along with the signal demodulator and modulator circuit
`(from Cheungetal 1989).
`
`8.4.4 Demodulator and filtering
`
`This section splits the IR and the red signals from the mixed signal from the
`photodiode. The mixed signal is demultiplexed synchronously and steered
`depending on the type of signals present. The inputs to this circuit are the
`photodiode output and the timing or control signal from the MBS. The
`microprocessor along with the information stored in the EPROM calculates the
`time period each signal component is present in the photodiode output. Switching
`at the right time results in the two components getting separated. In order to
`eliminate the high-frequency switching noise, low-pass filters are provided. To
`optimize cost, size and accuracy, switched capacitor filters are used. These filters
`
`123
`
`11 I
`
`1
`
`1
`
`L
`
`MASIMO 2053
`Apple v. Masimo
`IPR2022-01299
`
`
`
`RX-0035.0124
`
`APL_MAS_ITC_00015741
`
`Electronic instrument control
`
`107
`
`cause the two signals (red and infrared) to be identical in gain and phase
`frequency response. In order to filter out the noise generated by this switched
`capacitor a second filter follows in the cascade to filter out the switching
`frequency noise. This stage is a high roll-off stage, allowing the first stage to be
`the dominating one, resulting i Ii higher accuracy. Then using programmable DC
`01'fset eliminators and programmable gain amplifiers. the two signals are
`multiplexed along with other analog signals prior to being fed into an ADC.
`Offset amplifiers offset the signals by a small positive level. This ensures that the
`offsets caused by the chain of amplifiers do not allow the signal to be negative as
`this is the input to the ADC, and the ADC only accepts inputs from 0 to 10 V.
`Also sometimes the gain of the red or the IR channel may be greater than the
`other, and therefore the offset must be compensated accordingly.
`
`8.4.5 DC offset elimination
`
`To exploit the entire dynamic range of the ADC the two signals (red and IR) have
`to be processed further. Before discussing how this processing is done let us
`examine why this is done.
`We know that the mixed signal consists of a pulsatile and a nonpulsatile
`component. The nonpulsatile component approximates the intensity of the light
`received at the pholodiode when only the absorptive nonpulsatile component is
`present at the site (finger, earlobe, etc). This component is relatively constant
`over short periods, but clue to probe position variation and physiological changes
`this component may vary significantly over large intervals. But as we analyze
`these signals in small interval windows, this is not a major problem. Figure 8.5
`shows that this nonpulsatile component may be S_LOW, with the difference
`between S_HIGH and S_LOW being the varying pulsatile component, due to the
`arterial pulsations at the site. This pulsatile component is very small compared to
`the nonpulsatile componeni. Therefore great care must be taken when
`determining and eventually analyzing these values, as we desire the pulsatile
`component.
`
`Volts
`
`S_HIGH
`
`S_LOW
`
`,
`
`Time
`
`Figure 8.5 The nature of the signal transmission received by the photodiode circuit.
`
`Amplifying and converting to digital form the substantial nonpulsatile
`component will use up most of the resolution of the ADC. Therefore in order to
`exploit the entire dynamic range we must eliminate this component, digitize it and
`later add it back to the pulsatile component.
`For example, consider a ADC having an input range of 0 to 10 V. From
`figure 8.5 the AC may be 1% of the DC, let S_HIGH = 5.05 V and S_LOW = 5
`V. For a 12-bit ADC, the resolution of this device is almost 2 12. This means that
`
`124
`
`MASIMO 2053
`Apple v. Masimo
`IPR2022-01299
`
`
`
`RX-0035.0125
`
`APL_MAS_ITC_00015742
`
`h
`
`1
`
`1
`
`108
`
`Design of pulse oximeters
`
`the total signal is discretized into 4096 levels. Therefore from the above value of
`the pulsatile component (S_HIGH - S_LOW), we see that only 20 levels are
`utilized. Therefore if the nonpulsatile component is removed we can use all the
`4096 levels, improving the resolution of the ADC.
`Cheung er W ( 1989) discuss this concept of nonpulsatile component
`elimination and addition. The photodiode output contains both the nonpulsatile
`and the pulsatile component. The programmable subtractors (offset amplifier)
`remove a substantial 01Tset portion of the nonpulsatile component of each signal
`and the programmable gain amplifiers increase the gain of the remaining signal
`for conversion by the ADC. A digital reconstruction of the original signal is then
`produced by [he MBS, which through the use of digital feedback information
`removes the gain and adds the offset voltages back to the signal.
`Feedback from the MBS to the analog and the digital sections of the board is
`required for maintaining the values for the offset subtraction voltage, gain, and
`driver currents at levels appropriate for the ADC. Therefore for proper
`operation. the MBS must continuously analyze and respond to the offset
`subtraction voltage, gain, and driver currents.
`Figure 8.6 shows thal thresholds L I and L2 are slightly below and above the
`maximum positive and negative excursions L3 and L4 allowable for the ADC
`input and are established ariel monitored by the MBS at the ADC. When the signal
`at the input of the ADC or at the output of the ADC exceeds eilher of the
`thresholds Ll or Ll [lie LED driver currents are readjusted to increase o r
`decrease the intensity of light impinging upon the photodiode. I n this manner the
`ADC is protected from overdrives and the margins between L3, LI. and L2, L#
`helps ensure this even for rapidly varying signals. An operable voltage margin
`for the ADC exists outside the threshold, allowing the ADC to continue operating
`while the appropriate feedback does the required adjustments,
`
`.1JI11/ll,}1/!
`/>»Reset drive /
`ll,..
`
`L 3 High rail
`
`\ NXReset offsets ~
`
`Desired
`operating
`range of LEDs
`
`_~» Mid scale
`L 6
`
`=ov
`
`..... lhi ~1
`**93§6 L 5
`:*3323%
`3%99» L2
`
`/%01>Reset drive ///7
`12/4........11 // L 4 Low rail
`
`Figure 8.6 When the signal exceeds thresholds, the LED driver currents are readjusted to prevent
`overdriving the ADC (from Cheung et al 1989),
`
`When the signal for the ADC exceeds the desired operating voltage
`threshold, L5 and L6, the MBS responds by signaling the programmable
`subtractor to increase or decrease the offset voltage being subtracted.
`The instructions for the MBS program that controls this construction and
`reconstruction are stored in the erasable, programmable, read-only memory
`(EPROM).
`
`125
`
`MASIMO 2053
`Apple v. Masimo
`IPR2022-01299
`
`
`
`RX-0035.0126
`
`APL_MAS_ITC_00015743
`
`Electronic uistrument control
`
`109
`
`8.4.6 Timing diagram (Nellcor®)
`
`Figure 8.7 shows that the Nellcor pulse oximeter system uses a four state clock,
`or has a duly-cycle of 1/4, as compared to a Ohi-neda system, where the duty-
`cycle is 1/3.
`
`1
`1
`t
`Phase 2
`Phase 1
`Phase 4
`Phase 3
`IRON | IROFF i RON ' ROFF |
`
`Sync detector I
`polar ty
`
`1
`IRLED on
`(IR') I
`
`R LED on
`(FED')
`
`1
`
`Gated IR
`energy to
`channel 2 filter
`
`Gated R
`energy to
`channel 1 filer
`
`Channe12 1
`1
`CIR) gate
`1
`1
`(IRGATE)
`1
`1
`Channel 1
`1
`1
`(R) gate
`1
`(REDGATE) 1
`Typical -, Ambient I
`detector
`response
`1
`1
`1
`1
`1
`1
`1
`1
`approx.
`1-~ 170 Fs
`1
`1
`
`1
`
`1
`
`P
`1
`1
`1
`1
`1
`1
`1
`r-L_-1-L
`1
`Ambient I
`
`11
`
`1
`1
`1
`1
`
`1
`
`~ IR+ Ambienti
`
`- IR+ Ambient
`
`< RED +Ambient
`1
`1
`1
`1
`
`11
`
`1
`1
`1
`
`L._/ Ambient
`1
`
`1
`1
`1
`
`l
`
`1
`1
`1
`1
`
`11
`
`Figure 8.7 Timing diagram (reprinted with permission from Nellcor, Inc. ©Nellcor, Inc. 1989).
`Note that the typical detector response is inverted.
`
`In the first quarter the IR LED is on and in the third quarter the R LED is
`on. In the second and the fourth quarters these LEDs are off. It is during this
`period that the ambient light measurements are done. The gate pulses are the
`sampling pulses applied lo the input signal to separate out the R and the IR
`components from the input signal. The sample pulse during the OFF period of the
`respective LED is used to sample the ambient. The gradual rise or fall is due to
`the transients, which are smoothed out using low-pass filers. The ambient
`component is larger in the fourth quarter, compared to the value in the second.
`
`126
`
`MASIMO 2053
`Apple v. Masimo
`IPR2022-01299
`
`
`
`RX-0035.0127
`
`APL_MAS_ITC_00015744
`
`It
`
`ill'
`
`11
`
`1 2 1
`
`. 1
`
`1
`
`110
`
`Design of pulse oximeters
`
`Using suitable values for the gain in the programmable DC offset amplifiers we
`can eliminate this ambient component. The AC plus the DC components of the R
`and IR signals are digitized and sent to the MBS.
`
`8.4.7 LED driver circuit
`
`The need to drive both LEDs at different intensities requires analog switches that
`are used for gating the separate drive voltages. The factor that influences the
`amount of drive voltage necessary is the signal level from the photodiode and this
`value is set by the sample-and-hold section. The necessary control signals come
`from the pattern generator. The main purpose of this drive circuit is to convert
`this drive voltage to drive current.
`Figure 8.8 shows the drive voltages VIR and VR and gating signals IR' and
`R'. These signals are used to multiplex the two signals back into one, and are fed
`into a voltage-to-current (V-I) converter such that the output of this V-/
`converter is a bipolar current signal, that is used to light up only one LED at a
`time. As the two LEDs are tied in a back to back configuration, this bipolar
`current drive ensures that only one LED is on at a time.
`
`VIR
`-~ Multiplexer
`IR' -4".-
`
`VR__I-
`
`Multiplexer
`
`Voltage to
`-- current
`convener
`
`2% 1 LEDs
`
`Sensing
`resistors
`
`Compensating
`
`circuittFrom MBS
`
`Figure 8.8 LED driver circuit with the sensor resistors to monitor and control the amount of
`current into the LED (adapted from Nellcor N.200® (Nellcor ]989))
`
`The drive current requires a control to convert the specified voltage to the
`proportional drive current. Within the voltage to current converter is an error
`amplifier that compares the voltage from the current sensing resistors with the
`specified voltage. There are two bridge networks with current boosters and drive
`and steering transistors which steer current around this conversion network. The
`drive output is connected to a pair of parallel back-to-back IR/R LEDs. The
`current through the LEDs is determined by a sensing resistor and fed back to the
`error amplifier to maintain a constant current proportional to the de,;ired output
`voltage and lo be independent of the other voltages present across the bridge
`circuit. Maximum LED current at 25% duty cycle is approximately 120 mA. The
`back-to-back configuration is such thal when one LED is forward biased the
`
`127
`
`MASIMO 2053
`Apple v. Masimo
`IPR2022-01299
`
`
`
`RX-0035.0128
`
`APL_MAS_ITC_00015745
`
`Electronic instrument control
`
`111
`
`other is not. Chapter 5 describes the LED driver circuit used in the Nellcor®
`system.
`
`8.4.8 Analog processing system (Ohmeda®)
`
`The main block diagram indicated how different signals on board a pulse
`oximeter system flow, and showed the signal transfer from one major block ( for
`e.g. ECG, probe, MBS, power supply, etc) to the other. This section will
`elaborate on the analog signal flow from the photodiode output until the analog
`signal is ready to drive the LEDs to make another measurement. See figure 8.9.
`
`T
`
`Amplifier
`
`Photodiode
`
`Calibration
`~r-- signal
`
`Ambient light
`L- cancellation,
`storage capactior,
`
`Drive current 1 20mA (max)
`R
`mullplexer,
`DAC
`
`4~Ek.. RED LED
`
`Drive current 60mA (max) ,___-=Mi„- B LED
`IR
`multiplexer,
`DAC
`
`OC galn
`S.Mt rOCIntor
`
`r ~ir- r
`
`Pmbu
`ID
`
`1* Analog Alt
`m,iltlple·.u'
`
`DC
`cepa,ator
`
`DC
`
`ADC
`
`To MBS
`
`DC
`st ripper,
`high pass
`11'le,
`
`AC gain _- RED/IR __.~ Sample ~
`ard hold ~
`separator
`
`Figure 8.9 Functional block diagram of the pulse oximeter syswm showing all the main blocks
`involved in analog signal processing (adapted from Ohmeda 3740® (Ohmeda 1988)).
`
`8.4.8.1 LED drive and monitor. The probe consists of the LEDs and the
`photodiode. The currents through the LEDs are controlled by a pair of
`multiplexers and switches and digital-to-analog converters (DACs). The
`maximum drive current is 120 mA through the R LED and 60 mA through the
`IR LED. The multiplexer and the switch turn the R and the IR LED drive on and
`off. The timing signal from the MBS controls the switches. The duly cycle of this
`timing signal is approximately 1/3 (Note that the duty cycle in devices from
`Nellcor® is 1/4). Therefore the subsequent hardware and analog and digital
`signal processing is different, The notable differences are in the multiplexers and
`sample and hold circuit. In the Ohmeda version, as the duty cycle is 1/3, first the
`R, then the IR, and finally the ambient component (measured when both the R
`and IR LEDs are off), are separated. In the Nellcor version as the duty cycle is
`1/4 (see figure 8.7), the ambient component is measured twice.
`The LED drive currents are monitored by switches and capacitors when both
`the R and IR LEDs are on individually and when both of them are off.
`
`128
`
`MASIMO 2053
`Apple v. Masimo
`IPR2022-01299
`
`
`
`RX-0035.0129
`
`APL_MAS_ITC_00015746
`
`0
`
`1
`
`Design of pulse oximeters
`
`112
`8. 4. 8. 2 Calibration test signal. The signal received by the photodiode contains
`information on the AC and DC components of the pulsatile arterial blood flow
`measured by both the R and IR LEDs and also the ambient light component which
`is measured when both the R and IR LEDs are off. The calibration signal is a test
`signal injected into the signal path. The calibration signal is used to emulate the
`photodiode amplifier output which represents a known oxygen concentration and
`pulse rate of 150 to 210 beats per minute. The MBS checks the calibration of the
`oximeter by setting a test signal. This selects the calibration signal to be passed
`through the switch of the multiplexer in place of the photodiode amplifier output.
`8.4. 8. 3 Ambient light cancellation. Ambient light cancellation is done to remove
`the effects of ambient light from the photodiode signal. A capacitor and a switch
`of a multiplexer are used to first charge up this capacitor to a voltage difference
`between the input signal and ground, when the input signal contains only the
`ambient component (R and IR LEDs are off). After this phase this voltage is
`subtracted from the input signal, now containing the R and IR components.
`8. 4. 8. 4 DC gain set resistor. The DC gain of the input signal
`the
`is set under
`control of the MBS. One resistor from a resistor bank is selected and along with
`another fixed resistor is used to set the gain of the amplifier.
`8. 4. 8. 5 DC separator. This block separates the DC components of the R and IR
`signals. This section consists of multiplexers and low-pass filters. The switches,
`controlled by the MBS, allow the red or the infrared component of the signal to
`pass through the low-pass filter. A set of amplifiers amplifies this DC before it is
`sent into the analog-to-digital (ADC) converter for conversion before being fed
`into the MBS. As a result of this stage we obtain the DC components of the R and
`the IR signals.
`8. 4. 8.6 Low-pass filtering and DC stripping. A switched capacitor low-pass filter
`is used in this section. Since the DC components have been separated and
`measured previously, it is not necessary to filter during the ambient time. The R
`and IR components are low-pass filtered during the R and IR. time.
`DC stripping is used to separate the pulsatile component from the signal. The
`low-passed signal is sent via a high-pass switching filter, and depending on the R
`and IR LED times, the pulsatile or the AC components of the R and IR signals are
`generated. This stage yields the AC components of the R and the IR signals.
`8. 4. 8.7 Red/infrared separator. Multiplexers separate the red and infrared
`pulsatile signals into two independent channels. Low-pass filters are also used to
`smooth the separated signals. To compensate for the gain differences between the
`red and infrared signal paths, the gain of the infrared amplifier is adjustable by
`potentiometer.
`8. 4. 8. 8 Sample and hold circuits. Sample and hold circuits sample the red and
`infrared pulsatile signals simultaneously so that they can be measured by the
`ADC. An additional sampling signal controls the timing of the sampling of the
`pulsatile components at a rate synchronous to the power line frequency. This
`sampling frequency helps to suppress interference generated from sources
`connected to the line power.
`
`1
`
`1
`
`129
`
`MASIMO 2053
`Apple v. Masimo
`IPR2022-01299
`
`
`
`RX-0035.0130
`
`APL_MAS_ITC_00015747
`
`113
`8.4.8.9 Probe identification. This is the voltage generated by passing a known
`
`Electronic instrument control
`
`amount of current through the probe coding resistor to identify the wavelengths
`associated with the probe. This signal is digitized, compared to a lookup table in
`the MBS's memory, and the associated wavelength values are used for further
`processing.
`An analog multiplexer is used to choose one of the many inputs and feed it to
`the ADC. The MBS for the Ohmeda system is similar to the one used in Nellcor,
`but uses Zilog's Z-8002. Motion artifact elimination using the R wave (ECG
`synchronization), as seen in Nellcor N-200 is not present in this system.
`
`8.4.8.10 Timing diagram. In the Ohmeda Biox 3700® oximeter the LED on-off
`cycle is repeated at a rate of 480 Hz (figure 8.10). This cycling allows the
`oximeter to know which LED is on at any instant of time (Pologe 1987). The
`duty cycle in this system is 1/3. The red LED is on for the first 1 /3 of the cycle,
`the infrared LED is on for the second 1/3 and both LEDs are off for the third
`1/3, allowing for the ambient light measurements. This kind of measurement of
`ambient light is necessary so that it can be subtracted from the levels obtained
`when the LEDs are on.
`
`480 Hz
`
`Red on
`ti me
`
`Infrared
`
`--_-_ on time-1- -1-1
`
`LED 011
`time
`
`Photodiode
`current
`
`TOffset due to
`
`Time
`
`ambient light level
`
`Figure 8.10 Oulpul of the photodiode of the pulse oximeter system (adapted from Ohmeda
`3700® (Ohmed: 1988))
`
`8.5 ECG SECTION
`
`1
`
`Pulse oximeters use the ECG to eliminate disturbances caused by motion artifacts
`and ambient light. There is a time delay between the electrical and the mechanical
`activity of the heart. When an ECG QRS electrical complex is detected, a
`mechanical pulse will be detected at the sensor after a transit delay of about 100
`ms. This delay depends on factors such as the heart rate, the compliance of the
`
`130
`
`MASIMO 2053
`Apple v. Masimo
`IPR2022-01299
`
`
`
`RX-0035.0131
`
`APL_MAS_ITC_00015748
`
`1
`
`1
`
`114
`
`Design of pulse oximeters
`
`arteries, and the distance of the probe from the heart. The pulse oximeter
`computes this delay and stores it, and an average delay is generated after a few
`pulses. This average delay is used to establish a time window, during which the
`pulse is expected at the probe site. So if a pulse is received within this time
`window, it is treated as real and is processed. Any pulse arriving outside this
`window is simply rejected. Note that the time averaging and the time window are
`constantly updated to account for the patient's physiological changes.
`
`8.5.1 Active filters
`
`Figure 8.11 shows that the ECG signal from the patient has to pass through a
`series of filters before it is used for processing. Usually these filter stages provide
`gain, as the signal level received is very small. The most commonly used filters
`are as follows.
`
`l. A low-pass filter with a corner frequency of 40 Hz.
`2. A switched capacitor notch filter at the power line frequency. The capacitor
`switching frequency is determined by the timer pulse, which js in turn set by
`the microprocessor. The microprocessor along with its associated circuitry
`determines the power line frequency, and accordingly sets the notch
`frequency.
`3. A second 40 Hz low-pass filter may be used to filter out the transients
`generated by the capacitor switching.
`4. A high-pass filter, with a corner frequency of 0.5 Hz, is used for the lower
`end of the range. This filter has a substantial gain and has a long time
`constant. The reset condition discharges this capacitor. The most common
`situations desiring a reset are the lead fall off condition, muscle contraction
`under the electrode, or a sudden shift in the baseline of the ECG, due to the
`already high combined gain due to the front end section and the filters
`preceding this stage.
`
`When pulse oximeters are used in electromagnetic environments (MRI
`1992), such as magnetic resonance imaging (MRI), special care has to be taken, as
`EMI interferences are quite disturbing for pulse oximeters. Probes and
`connectors are shielded, using faraday cages, and additional EMI elimination
`filters are incorporated in the pulse oximeter (see chapter 11).
`
`8.5.2 Offset amplifiers
`
`Analog-to-digital converters have a specified input dynamic range for obtaining
`the maximum digitized output. Usually these are in the positive range, from 0 to
`5 or 0 to 10 V. Therefore an amplifier that can offset the analog signals to a
`value beyond 0 V and convert its peak value to 5 to ]0 V is needed. For example
`if there is a signal from -0.7 V to +6,7 V and an ADC with dynamic range of 0
`to 5 V, the offset amplifier will convert this range to 0 to 5 V. Then we can make
`use of the entire resolution of the ADC.
`
`8.5.3 Detached lead indicator
`
`ECG signals are sensed by the electrodes placed on the body and the signals are
`transferred from the site to the pulse oximeter via leads. If the electrode falls off
`
`131
`
`MASIMO 2053
`Apple v. Masimo
`IPR2022-01299
`
`
`
`RX-0035.0132
`
`APL_MAS_ITC_00015749
`
`El.ectronic instru.ment control,
`
`115
`
`from the surface of the body, the pulse oximeter' s front end display must indicate
`this. The indicator system consists of a voltage comparator, absolute value circuit
`and a latching flip flop. This stage examines the ECG signal al the input to the
`switched capacitor notch filter (60 Hz), after it has passed through the low-pass
`stage preceding it. There is a biasing resistor network that drives the ECG signal
`to either *15 V, if one or more ECG leads are detached from the patient's body.
`If the signal rails to -15 V, it is converted to positive voltage by a level shifter
`amplifier. Using this signal, the data are latched in a flip flop and the processor is
`notified that a lead has fallen off. After the processor recognizes this, it resets this
`latching flip tiop so it is now ready to sense any other fall off.
`
`ECG
`
`Notch clock
`frequency -~
`
`signal--1 Low ~ ~- Line trequency ~ Low
`
`pass
`lill nr
`40 Hz
`
`notch filter
`
`pal:
`hit-
`40 Hz
`
`Absolute
`1 4. value
`Voltage
`circuit -~ comparator -1 flop 1
`
`ECG
`signal
`
`Vrel
`
`A Buffer
`
`Poak »dotector circuit
`
`Reset
`from MBS
`
`Comparator
`
`P / S I fpmier I
`
`1,0,1 ~ ~ ECG
`
`High
`lili„f
`0 5 Hz
`
`Of Isebollage
`
`D- Lead off
`
`ADC F--
`
`Figure 8.11 ECG signal processing section along with the lead all off indicator and peak
`detector unit (adapted from Nellcor N-3000® (Nellcor 1991))
`
`8.5.4 Power line frequency sensing
`
`Electric devices usually have the main power ac signal transformed to a root
`mean square value for power line analysis. A voltage comparator is used to
`generate a signal that interrupts the microprocessor at the frequency of the ac
`power line. This signal is then used to set the notch filter at the line frequency to
`eliminate the line frequencies. Most devices have provision for a 50 Hz or 60 Hz
`line.
`
`8.5.5 ECG output
`
`This section is used to generate pulses to synchronize the processor with the R-
`wave arrival. There is a peak follower circuit that stores the peak R-wave pulse
`in a slowly decaying fashion. This is employed to ensure that the capacitor doesn't
`discharge before the next R wave arrives. An adjustable threshold is provided for
`sensing each R-wave peak. This parameter is set by the processor, which in turn
`
`132
`
`MASIMO 2053
`Apple v. Masimo
`IPR2022-01299
`
`
`
`RX-0035.0133
`
`APL_MAS_ITC_00015750
`
`116
`
`Design of pulse oximeters
`
`is influenced by many external parameters. A voltage comparator produces a
`high output whenever the ECG input exceeds the adjustable threshold determined
`by the previous R-wave peak.
`
`8.6 SIGNAL CONVERSION
`
`The signal conversion unit consists of an ADC or DAC. Signals have dc offsets
`subtracted and even amplified before processing. This enables us to extract
`signals having low modulation and riding on a high DC, and this helps improve
`the response time of the system.
`
`8.6.1 Analog-to-digital conversion technique
`
`Analog-to-digital conversion on the processor board is accomplished by uxing a
`sample-and-hold circuit. which holds a voltage until it is sampled by a routine
`written in the memory of the processor. Both software and hardware play a
`important role in the conversion.
`Figure 8.12 shows that first the processor writes to an analog multiplexer to
`select one of its several analog inputs that desire digital conversion. These signals
`could be the demultiplexed and filtered IR or the red photodiode channel signal,
`the filtered ECG waveform. or filtered voltage from the coding resistor. The
`selected signal is first latched and the analog circuitry is notified of the amplitude