`
`APL_MAS_ITC_00015691
`
`Light-emitting diodes and their control
`
`57
`
`5.].J Description, materials, and operation
`
`An LED is an optoelectronic semiconductor which produces light by
`electroluminescenc·e (D.A.TA. Handbook 1992). LEDs are characterized by high
`light emitting efficiency compared to other methods of light emission such as
`cathode, high-temperature, and photoluminescence. The electroluminescence
`occurs by the injection and recombination of minority carriers in the forward-
`biased 1 ) - 1 1 junction . Most LEDs are made from III-V ,
`II-VI ,
`and IV
`semiconductors, with the most common materials being gallium arsenide
`phosphide (GaAsP), gallium phosphide (GaP), and gallium arsenide (GaAs).
`GaAsP and GaP LEDs emit light in the visible spectrum (approximately 380 to
`78() nm), while GaAs is used in infrared LEDs. Another material not as
`commonly used to make LEDs which can produce light in both the visible and IR
`regions of the spectrum is gallium aluminum arsenide, GaAIAs.
`Figure 5.] shows the light emission mechanism of an LED. When an
`electron gains enough energy to cross the forbidden energy gap Eg, it enters the
`conduction band. When an electron in this conduction band returns to the lower
`energy level of the valence band, the electron releases energy in the form of a
`photon of light. The wavelength of light emitted from an LED is determined by
`(5.1)
`g
`E = Iicil,
`I where Eg is the forbidden bandwidth in electron volts, h is Planck's constant
`(6.626 x 10-34 J s), c is the speed of light in a vacuum (3.00 >: 108 m/s), and A is
`the wavelength of the emitted photon. The value of Eg, which is a physical
`property of the LED material(s), determines the wavelength of emitted photons
`and is directly related to the forward voltage of an LED (see section 5.2.1).
`
`1
`
`5.1.2 Bandwidth considerations
`
`Another factor considered in the use of LEDs in pulse oximetry is the emission
`spectrum of the LED. Because of the steep slope of the deoxyhemoglobin (Hb)
`extinction curve at 660 nm, it is extremely important that the red LEDs used in
`pulse oximeter probes emit a very narrow range of wavelengths centered at the
`desired 660 nm in order to minimize error in the Sp02 reading which is the
`pulse oximeter's estimation of arterial oxygen saturation (New and Corenman
`1987, 1988). The width of the wavelength range of the IR LED is not as
`important for accuracy due to the relative flatness of both the Hb and HbO2
`(oxyhemoglobin) extinction curves at 940 nm. LEDs again perform very well for
`this requirement. Typical LEDs have a spectral bandwidth in the range of 60 nm
`to less than 20 nm, with visible LEDs usually having smaller bandwidths of
`approximately 25 nm and IR LEDs typically having larger bandwidths near 50
`nm.
`
`5.2 LIGHT-EMITTING DIODE SPECIFICATIONS
`
`Before discussing the specifications of LEDs available on the market, the
`performance desirable for LEDs in pulse oximetry will be given. The two
`predominant factors are the radiated power (or light output) and the size of the
`LEDs.
`
`74
`
`MASIMO 2053
`Apple v. Masimo
`IPR2022-01300
`
`
`
`RX-0035.0075
`
`APL_MAS_ITC_00015692
`
`58
`
`Design of pulse oximeters
`
`electron
`
`conduction band
`
`light
`emission
`
`E
`
`g
`
`00®
`-z valence band
`)poof-
`hole -*-
`
`p-n junction
`
`p-type region
`
`n-type region
`
`-lili+
`
`Figure 5.1 The light emission mechanism of an LED. Electrons gain energy moving to the
`conduction band. They emit light when dropping to the va[ence band.
`
`The radiated power of an LED is measured in milliwatts. The typical
`radiated power of both the red and [R LEDs used in pulse oximetry is 1 inW at
`20 mA dc. Brighter LEDs are available, but generally the radiated power does
`not exceed 10 mW,
`Modern manufacturing techniques have shrunk LEDs to sizes smaller than a
`millimeter in length or diameter, while remaining bright enough to be used in
`devices such as pulse oximeters, LED size is not an obstacle in the design of pulse
`oximeter probes.
`
`5.2.1 Forward voltage
`
`The forward voltage is defined as the potential drop across the p-,1 junction of
`the diode from anode to cathode. While ordinary silicon diode forward voltages
`are near 0.7 V, LEDs forward voltages can range from 0.9 to 2.5 V typically.
`Equation (5. ] } shows that an inverse relationship exists between a material's
`forbidden energy gap Eg and the wavelength of emitted photons. In addition, the
`forward voltage of an LED is directly related to Eg. Therefore, an LED with a
`relatively small forward voltage has a small Eg and a long emitted wavelength
`(e.g. in the infrared region). Conversely, an LED with a relatively large forward
`voltage has a large Eg ariel a short emitted wavelength le.g, in the blue-green
`region).
`
`5.2.2 Forward current
`
`The forward current is defined as the currenl flowing through the LED in the
`direction from anode to cathode. With sufficient current, an LED will emit light.
`A very important property of LEDs is that radiated power, to a first
`
`75
`
`:41
`
`t
`
`11
`
`MASIMO 2053
`Apple v. Masimo
`IPR2022-01300
`
`
`
`RX-0035.0076
`
`APL_MAS_ITC_00015693
`
`Light-emitting diodes and their control
`
`59
`
`approximation, varies linearly with forward current over the range of-~ cui-rent
`found i n pulse oximeters. Typical values for forward current have a large range,
`from 2 to 50 mA. Figure 5.2 shows the relationship between current and voltage
`for a typical 660 nm LED.
`
`Forward
`current
`(mA)
`
`60
`50
`40
`30
`20
`10 '
`
`1
`
`1.0 1.5 2.0 2.5 3.0
`Forward voltage (V)
`
`Figure 5.2 Forward current-voltage characteristic for a typical 660 nin GaP LED.
`
`5.2.3 Power dissipation
`
`Another consideration for LEDs used in pulse oximetry is power consumption.
`While the vast majority of pulse oximeters are used in a stationary environment
`where power is readily available from the nearest wall outlet, some are portable
`units used in a variety of emergency medical situations. These portable units may
`need to function for an extended period of time without a power supply recharge.
`Il is therefore essential that LED power consumption be minimized while still
`providing adequate radiated power for pulse oxime[ry.
`The maximum power dissipation rating for an LED can be defined as the
`largest amount of power that can be dissipated while still remaining within safe
`operating conditions. This power is a function of three paramelers: ambient
`temperature, rated maximum junction temperature, and the increase in junction
`temperature above ambient per unit of power dissipation for the given LEI)'s
`package and mounting configuration. The latter of these parameters is defined as
`the thermal resisiance of the device, and is very important
`in reliable system
`design. The worst-case value for [liermal resistance, that with no heat sink, can be
`calculated from
`(5.2)
`where RTH is the thermal resistance, 71 is the junction temperature, TA is tile
`ambient temperature, and PD is the rated power dissipation of the LED. Another
`method for calculating thermal resistance is to use the negative reciprocal of the
`slope of the forward current versus ambient temperature graph, figure 5 3. This
`value is in units of °C/mA, which can be converted to the thermal res,8tance in
`¤C/W by multiplying the denominator by the LED forward voltage (D.A.T.A.
`
`RTH = (TJ - TA)/PD °C/W,
`
`76
`
`MASIMO 2053
`Apple v. Masimo
`IPR2022-01300
`
`
`
`RX-0035.0077
`
`APL_MAS_ITC_00015694
`
`60
`
`Design of pulse oximeters
`
`Handbook 1992). Because the skin is the primary sink for LED heat in pulse
`oximetry, the design engineer must consider power dissipation in order to
`prevent possible burns to the patient's skin.
`Typical LEDs are 2 to 10% efficient, meaning that the majority of power
`dissipated by an LED becomes heat. The optical power absorbed by the tissue also
`becomes heat. As with forward current, a broad range of power ratings is
`available, typically from 20 to 300 mW. An interesting fact to note is that the
`typical IR LED, with its lower forward voltage (see section 5,2.1), required a
`greater forward current to dissipate the same optical power as a typical red LED.
`This is because red photons contain more energy than infrared photons.
`
`5.2.4 Reverse breakdown voltage
`
`As with all diodes, under reverse bias virtually no current will flow across the p-
`n junction until the reverse breakdown voltage has been reached. Above that
`voltage, large currents flow and damage the diode, unless a resistor limits the
`current. Most LEDs have a fairly small value for this specification, usually in the
`range of 3 to 5 V. This specification is important in pulse oximetry due to the
`arrangement of the LEDs in a probe. To minimize the number of wires in each
`probe (and hence cost), the LEDs are wired in a parallel arrangement with
`polarities reversed. This means that while one LED is ON, the other LED is
`under reverse bias. The typical LED has a reverse breakdown voltage that is
`larger than the forward voltage of most LEDs, minimizing the difficulty in
`dealing with this specification.
`
`5.2.5 Reverse current
`
`In an ideal diode, no current tlowN in the reverse direction when the p-n junction
`is reverse-biased. in reality, a minute amount of current actually does tlow in the
`reverse direction. In LEDs, this current typically ranges from 0.01 to 10 FLA.
`Since this current is extremely small compared to the forward current of the
`LED wired in parallel, this shunt current has a negligible effect.
`
`5.2.6 Operating temperature
`
`Pulse oximeters are usually used in a stable medical environment at room
`temperature. However, emergency situations may arise in which a pulse oximeter
`has to operate under extreme temperatures. Fortunately, LEDs are extremely
`rugged devices with a basic specified range of operating temperature from -40 to
`85 °C. Many LEDs with an even larger operating temperature range are
`available.
`Most LED parameters are specified at a given temperature. In addition,
`information is given for how some of these parameters vary over a given
`temperature range (see section 5.5). The most important of these parameters is
`maximum forward current versus temperature, which determines the thermal
`resistance of the LED (see section 5.2.3). Figure 5.3 shows the relationship
`between maximum forward current and temperature for a typical high-power
`660 nm red LED.
`
`77
`
`MASIMO 2053
`Apple v. Masimo
`IPR2022-01300
`
`
`
`RX-0035.0078
`
`APL_MAS_ITC_00015695
`
`Light-emitting diodes an.d their control
`
`61
`
`Forward
`Current
`(mA)
`
`60
`50
`40
`30
`20
`10
`0
`
`0 20 40 60 80 100
`Ambient Temperature (°C)
`
`Figure 5.3 Maximum forward current versus temperature for a typical high-power red LED.
`
`5.2.7 Switching times
`
`Switching time is the time required for an LED to switch from its ON state to its
`OFF state or vice versa. Most LEDs have a switching time in the low hundreds of
`nanoseconds. In the application of pulse oximetry, this is much faster than
`required because of the extremely low frequency of the arterial pulsatile
`waveform (-1 Hz). For reasons which will be explained in chapter 8, in most
`pulse oximeters LED switching cycles occur at a rate of 480 Hz, much more
`slowly than the maximum switching capabilities of LEDs.
`
`5.2.8 Beam angle
`Beam angle is defined as the angular measure of radiated power measured on an
`axis from half-power point to half-power point. It is simply a measure of how
`focused the emitted light is. In LEDs on the market today, beam angles can range
`from a few degrees to a maximum of 180°. In pulse oximetry, the beam angle
`only needs to be narrow enough to ensure that the maximal light output enters the
`tissue. The scattering of light occurring in the tissue serves to ensure that the light
`spreads over the entire sensor area.
`
`5.2.9 Pulse capability
`Pulse capability is defined as the maximum allowable pulse current as a function
`of duty cycle and frequency. This parameter is important in pulse oximeters for
`two main reasons. The first reason is that, as discussed in chapter 8, LEDs are
`pulsed in pulse oximeters. The second reason is that the small LEDs used by some
`manufacturers in pulse oximeter probes may not be able to tolerate enough
`sustained current to sufficiently excite the photodiode. Since the allowable pulse
`current is always substantially higher than the maximum sustained current,
`smaller LEDs can be used than could be if the LEDs were constantly on. For
`
`1
`
`78
`
`MASIMO 2053
`Apple v. Masimo
`IPR2022-01300
`
`
`
`RX-0035.0079
`
`APL_MAS_ITC_00015696
`
`62
`
`Design of pulse oximeters
`
`example, the LEDs in Criticare pulse oxinieters have a duty cycle near 5%, while
`in Nellcor devices it is 25%. Figure 5.4 shows the pulse capability of a typical
`660 nni LED.
`
`IJ
`
`Ll•/11
`
`U.ID(
`
`1 ' J·- .0.1 10'' 10.1 10' , .0
`-4
`
`Figure 5.4 Pulse capability of a typical high-power 660 nm LED. The maximal pulse current is a
`function of the duty cycle d and frequency (from Siemens 1993).
`
`5.2.10 Cost
`
`Chapter 7 states that a disposable probe for use in pulse oximetry has some
`advantages over reusable probes, such as convenience and guaranteed sterility.
`With the widespread use of disposable probes, cost is the prohibitive fack,r in
`their manufacture. The cost of the two LEDs used in each probe is therefore
`important for the purpose of minimizing the overall expense of each probe.
`Today, both red and IR LEDs can be purchased in bulk for just a few cents each,
`making then-1 a minor factor in the overall cost of a probe. (Allied Electronics,
`Inc. 1995, Digi-Key Corporation 1995). However, testing each LED to find its
`peak wavelength, as discussed in the following section, does increase the overall
`cost to the manufacturer.
`
`5.3 MEASURING AND IDENTIFYING LED WAVELENGTHS
`
`Chapter 4 notes that the choice of 660 and 940 nm for the light wavelengths was
`not arbitrary with respect to optical considerations. Because of the steep slope of
`the Hb extinction curve at 660 nm, it is important that the red LEDs used in pulse
`oximeter probes have a peak wavelength of exactly 660 nm in order to minimize
`error in the SPO.2 reading (see chapter 11). Error in the peak wavelength of the
`IR LED is not as important for accuracy due to the relative tlatness of both the
`lib and HbO2 extinction curves at 940 nin. An alternative to having LEDs with
`precise peak wavelengths of 660 and 940 nm is to have the pulse oximeter itself
`somehow compensate for any deviation from those nominal values. This section
`discusses these concerns.
`As is the case with all mass nianufacturing processes. imperfections occur in
`each lot of LEDs produced. For pulse oximetry, the most important of these is
`peak wavelength shift. Peak wavelength is defined as the wavelength at which the
`radiated power of the device is maximum. Although bulk LEDs theoretically all
`have the same peak wavelength, figure 5.5 shows that the actual peak wavelength
`of any LED may vary from the rated value by as much as 15 nm (Pologe 1987).
`
`79
`
`MASIMO 2053
`Apple v. Masimo
`IPR2022-01300
`
`
`
`RX-0035.0080
`
`APL_MAS_ITC_00015697
`
`being able to use only LEDs with peak wavelengths of exactly 66() and 94() nm.f 15'm-7
`
`610 620 630 640 650 660 670 680 690 700 710
`WAVELENGTH (nm)
`
`Figure 5.5 Center wavelength variation of LED, of the same type from tile saine lot (from
`Pologe 1987).
`
`While many methods exist to solve this problem, only the one most
`commonly used in pulse oximeters will be explained here.
`The first step in the process for the probe manufacturer is to test each
`individual LED to find its exact peak wavelength. This is done by testing each
`individual LED with a spectrophotometer to experimentally determine the
`wavelength of light at which the LED has its highest power output. The LEDs are
`then separated into a certain number of groups, with each group having a small,
`distinct range of wavelengths, for example 660 to 661 nm.
`Knowing the center wavelengths for a particular LED pair allows the proper
`set of calibration curves, specific to that wavelength combination, to be chosen
`from the entire family of curves that exist. This is most often done by developing
`a two-dimensional matrix with, for example, the red LED wavelength values in
`the heading row and the IR LED wavelength values in the heading column. Each
`matrix location then identifies the appropriate set of calibration curves for the
`given pair of LEDs.
`The final problem to be solved is to have the pulse oximeter somehow
`interrogate each new probe to find out which calibration curve must be used to
`accurately determine arterial oxygen saturation. The most common technique is
`to include in the probe connector a coding resistor with a specific value. Each
`unique resistor value represents to the pulse oximeter those pairings of LED
`wavelengths that correspond to one calibration curve. The microprocessor simply
`sends a current through the resistor and measures the voltage drop across it, in
`effect finding the value of the coding resistor. By finding this voltage value in a
`lookup table, the microprocessor can indirectly determine the proper calibration
`curve to be used for that probe (New and Corenman 1987,1988).
`Chapter 8 provides details of how the pulse oximeter performs this
`interrogation of each probe.
`Kastle et al ( 1997) describes how Hewlett-Packard avoided using a coding
`resistor by selecting red LEDs within ai l nm variation of wavelength. A later
`
`80
`
`Light-emitting diodes and their control
`
`63
`
`In order to solve this problem, pulse oximeters can compensate for a number of
`different LED peak wavelengths. This technique has the advantage of lowering
`cost by allowing probe manufacturers to buy and use LEDs in bulk instead of
`
`INTENSITY
`
`MASIMO 2053
`Apple v. Masimo
`IPR2022-01300
`
`
`
`RX-0035.0081
`
`APL_MAS_ITC_00015698
`
`1
`
`1
`
`1
`
`1
`
`64
`
`Design of pulse oximeters
`
`sensor used new high-efficiency AlGaAs red LEDs to achieve a four-fold
`increase in intensity, with corresponding lowered heat dissipation. They also note
`that the red LED may emit an undesired secondary emission peak (<4% of
`maximum intensity) at about 800 to 850 nm, which may interfere with the IR
`LED. They place the LED in an integrating sphere to diffuse the light for
`wavelength measurement by an optical spectrometer having a wavelength
`resolution of 0.2 nm.
`
`5.4 LED DRIVER CIRCUIT
`
`This section presents an overview of the operation of a specific LED driver
`circuit used in many pulse oximeters. Greater detail about the microprocessor
`control, signal processing, and other hardware or software concerns can be found
`in chapters 8 and 9.
`Figure 5.6 shows the LED driver circuit. This circuit, and the LED driver
`circuits in many of the pulse oximeters on the market today, provide up to 50 mA
`of pulse current to each LED. The microprocessor automatically alters the
`amount of current supplied to the LEDs according to the absorption of the tissue
`on which the pulse oximeter is being used. Factors such as skin pigmentation, skin
`thickness, and optical path length, among many others, determine the absorption
`of the tissue. The microprocessor first determines if the photodiode is receiving a
`proper amount of light, enough to adequately excite but not saturate the
`photodiode. The microprocessor then supplies voltage feedback to the LED
`driver circuit, which allows current to the LEDs to be adjusted as needed. No
`complex calculations are necessary to determine current adjustments, as radiated
`power varies nearly linearly with drive current over the range of current utilized
`in pulse oximetry.
`The microprocessor controls how much current is provided to each LED by
`dynamically adjusting the reference voltage seen at the driver amplifier, U3A. U 1
`supplies the reference voltage, which is switched selectively for the red and IR
`LEDs using U4A and U4B. The microprocessor changes the reference voltage
`for the red or IR LEDs by changing the data supplied to Ul, which is a
`multiplying DAC, before the voltage is switched to the amplifier. The
`microprocessor attempts to achieve and keep the optimal drive current without
`clipping the transducer signal.
`The control signals REDLED/ and IRLED/ come from the microprocessor
`and control the switches U4A and U4B along with the transistor network that
`drives the LEDs. The LEDs are never on at the same time, although during part
`of the LED switching cycle they are both off to allow the photodiode to detect
`ambient light.
`When REDLED/ is low, IRLED/ is high and U4A is closed, placing the red
`reference voltage at pin 3 of U3A. Since REDLED/ is low, transistor Q5 is off,
`which allows Q3 to turn on and conduct current from the LED through Rl, the
`sense resistor. Also with REDLED/ low, Q2 turns on, allowing current to flow
`from the positive supply to the red LED anode, turning on the red LED. Since
`IRLED/ is high, Ql is off, with no current conduction, and Q6 is on, pulling the
`base of Q4 to ground, which keeps Q4 off. The current path in this case is from
`VCC through Q2, the red LED, Q3, and Rl, the sense resistor. The voltage drop
`across R 1 is fed back to U3A, which compares it to the reference voltage and
`changes the drive current of Q3 accordingly.
`
`81
`
`MASIMO 2053
`Apple v. Masimo
`IPR2022-01300
`
`
`
`RX-0035.0082
`
`APL_MAS_ITC_00015699
`
`Light-emitting diodes and their control
`
`In
`
`1-i . ft .it
`
`1-'
`
`-4.
`
`....1
`
`. . 0,
`
`i
`
`rn
`
`..
`
`im:- 1
`
`•~_I-• I
`
`42.
`
`'4.
`
`1 11 -1.:r'
`
`[-In- 2 -=L--4.'St
`717
`.2.
`
`3 - I tul.-i
`
`rT
`
`.r....1 Y
`
`1*44-1
`
`1
`
`.4.
`
`A
`
`;;~Ci,
`.... 1.1
`r-71
`
`09 9 R 8 8 Villumi
`
`1
`
`Figure 5.6 LED driver circuit. (From Protocol. 1994. Propaq® 100-Series Monitors Sch,·mictics
`& Drawings Set. Schematic 00950 , 6 of 7 . Section 2E, Protocol Systems, inc .. Beaverton. OR).
`
`82
`
`MASIMO 2053
`Apple v. Masimo
`IPR2022-01300
`
`
`
`RX-0035.0083
`
`APL_MAS_ITC_00015700
`
`66
`
`Design of pulse oximeters
`
`When REDLED# is high, IRLED# is low and the reference voltage is applied
`to U3A through U4B. Having REDLED/ high causes Q2 to be off and Q5 to be
`on, turning off Q3. Having IRLED/ low turns Q1 on, allowing current to flow to
`the IR LED anode, turning on the IR LED. Q6 is also turned off, which allows
`the base of Q4 to be pulled up in voltage by U3A until Q4 conducts. The current
`path in this case is from VCC through Ql, the IR LED, Q4, and Rl. The voltage
`drop across R 1 is again fed back to U3A, which compares it to the reference
`voltage and changes the drive current of Q4 accordingly.
`In the case when both the IRLED/ and REDLED/ control signals are high,
`both switches, U4A and U4B, are open and all of the drive transistors are off.
`The resistors R2, R4, and R3 form a voltage divider network that makes the
`reference input of U3A slightly negative with respect to ground. Because of this,
`U3A drives its output negative. However, D1 will not allow U3A's output to drop
`below approximately -0.6 V so that the drive transistors Q3 and Q4 can be
`turned on quickly when needed (Protocol 1994, pp 2ES-6).
`
`5.5 LED PEAK WAVELENGTH SHIFT WITH TEMPERATU RE
`As discussed in section 5,3, pulse oximeter probe manufacturers could use any of
`a number of methods to compensate for LED peak wavelengths which vary from
`the nominal values of 660 and 940 nm, with the method of choice being the use of
`a coding resistor to indicate to the microprocessor which set of calibration curves
`to use for a given probe.
`However, the peak wavelength of an LED can shift during operation due to a
`change of the p-n junction temperature. It is more difficult to account for this
`wavelength shift when determining which set of calibration curves to use. The
`effect that LED peak wavelength shift due to temperature has upon S~O2 will be
`discussed in detail in chapter 11. Lastly, two methods of minimizing tne negative
`effects of temperature changes upon Sp02 will be discussed.
`5.5.1 p-n junction heating
`
`Equation (5.1) shows that the wavelength of emitted light in an LED depends on
`the forbidden energy gap Eg. In turn, Eg is dependent upon temperature (Varshni
`1967, Panish and Casey 1969). In GaAs, GaP, and most other common
`semiconductor materials, Eg decreases as temperature increases. Therefore, the
`peak wavelength of an LED should increase as the p-n junction temperature
`increases. Typically, the peak wavelength will increase by 0.35 to 0.6 nm/'C
`(Miller and Kaminow 1988).
`The main factor affecting the p-n junction temperature of the two LEDs is
`drive current, which causes ohmic heating at the p-n junction. Although the
`LEDs are sequentially pulsed with a duty cycle of 2 to 50% depending on the
`make of the oximeter (Reynolds et al 1991), the resulting average current (duty
`cycle multiplied by drive current) is still sufficient to substantially heat the p-n
`junction, as power dissipated is directly proportional to the drive current I
`according to the equation P = VI.
`
`5.5.2 Studies
`
`de Kock et al (1991) tested the effect upon peak wavelength of driving a red and
`IR LED at 10% and 100% of the rated maximum drive current. The nominal
`
`1
`
`83
`
`MASIMO 2053
`Apple v. Masimo
`IPR2022-01300
`
`
`
`RX-0035.0084
`
`APL_MAS_ITC_00015701
`
`Light-emitting diodes and their amtrol
`
`67
`
`wavelengths of the tested LEDs were 660 and 950 nm, with 30 min allotted 1 o r
`thermal equilibrium to be reached. At 380 Hz with a 25% duty cycle, they found
`that the increased drive current increased the center wavelength of the red LED
`by 8 nm. while the center wavelength of the IR LED did not shift at all. Figures
`5.7 and 5.8 show their results.
`
`035=
`
`2 2 0/ i
`
`C• ~~'
`
`ro
`1 A & 9 4. 5 - h.'
`
`000 600
`
`do
`Wo, elength ( rn 3
`Figure 5.7 Normalized red LED spectra at low and high forward current (from de Kock et al
`1991)
`
`720
`
`~ 0 04 i 22-/ t
`10021 1
`1 ~"r
`
`10. POW" f
`
`V
`
`9,
`
`0 04 1
`
`f ° °j j
`
`7.
`0 0 02 1
`
`tool ]
`
`40
`
`Figure 5 . 8 Normalized IR LED spectra at
`1991).
`
`920
`9€C
`Wo·,elenath i
`low and high forward current (from de Kock et Cd
`
`'360
`
`The same group studied the effect of ambient temperature upon LED peak
`wavelength in 1991. As in the study mentioned above, the red and IR wavelengths
`were 660 and 950 nm. The spectrum of each LED was measured using a
`spectrophotometer at 2 nm intervals at ambient temperatures ranging from 0 to
`50 °C in 10 °C steps. Ten minutes was given for thermal equilibrium to be
`established at each temperature step. The group found that over this range of
`ambient temperatures, the red LED had an increase of 5.5 nm in its peak
`wavelength, while the IR LED had an increase of 7.8 nm. In addition, no
`significant change was found in the spectral bandwidth of either LED over the
`temperature range. The measured bandwidths were found to be approximately 25
`and 55 nm for the red and IR LEDs, respectively. Figures 5.9 and 5.10 show
`these results.
`Klistle et al (1997) list a wavelength shift of about 0.12 nm/K but note that
`the red and IR LEDs tend to track temperatures, which compensates for errors m
`the ratio R.
`
`=1 -
`
`84
`
`MASIMO 2053
`Apple v. Masimo
`IPR2022-01300
`
`
`
`RX-0035.0085
`
`APL_MAS_ITC_00015702
`
`68
`
`Design of pulse oximeters
`
`5.5/m
`
`ond
`widlh
`nm
`
`0.08
`0.07
`> 0.06
`
`E 0.05
`C
`U 0.04
`2 0.03
`Z 0.02
`
`600
`
`620
`
`0.01 - A0
`
`660
`640
`Wavelength (nm)
`Figure 5.9 Shift in emission spectrum of a red LED as ambient temperature is increased from ()
`to 50°Cin 10°Cintervals (from Reynolds etal 1991).
`
`680
`
`700
`
`0.040 -
`
`0.035 -
`
`2 0.030
`
`7.8 nm
`
`,/f I
`
`121Il
`
`Normalized intensi
`
`//// Bant,1.,ath 14
`- 55 •m
`
`0'c
`
`50 C
`
`0.025
`0.020
`0.015
`0.010
`0.005 .1
`0
`880 900 920 940 960 980 1000
`Wavelength (nml
`
`Figure 5.10 Shift in emission spectrum of an IR LED as ambient temperature is increased froin ()
`to 50 °C in 10 °C intervals (from Reynolds et at 1991)
`
`5.5.3 Two methods to compensate for LED temperature changes
`
`As expected, a shift in LED peak wavelength due to a change in temperature can
`cause erroneous Sp02 readings. A full discussion of this problem will be given in
`chapter 11.
`One way to compensate for LED temperature changes is to have a
`temperature sensor built into the probe along with the LEDN and photodiode
`(Cheung et at 1993 1. Temperature information is fed back to the microprocessor.
`which then estimates how much the peak wavelength of each LED has changed
`from its rated value (which the microprocessor determined from the probe's
`coding resistor). The microprocessor then chooses the set of calibration curves ti>
`match the new set of LED wavelengths. One inherent problem with this method is
`that the temperature-peak wavelength relationship given as a specification by the
`manufacturer will not be exactly the same for each individual LED, making the
`microprocessor's calculation of new LED peak wavelengths polenlially
`inaccurate. Another problem is the difference between the sensed temperature
`and the actual temperature of the p-,1.junctions of the LEDs. 1 f the two LEDs are
`being driven with different currents, as is normally the case, they will probably
`be at different temperatures. The temperature sensor will read al best an average
`
`85
`
`1
`
`1
`
`1
`
`J
`
`1
`
`1
`
`MASIMO 2053
`Apple v. Masimo
`IPR2022-01300
`
`
`
`RX-0035.0086
`
`APL_MAS_ITC_00015703
`
`Light-emitting diodes and their control
`
`69
`
`of the two LED temperatures, and at worst an average of the two LED
`temperatures along with the skin and ambient temperatures. In addition, the
`sensor and additional wires needed will add cost to the probes, making a cost-
`benefit analysis of this method necessary before its inclusion in a pulse oximeter
`design.
`A second, similar method to compensate for LED temperature changes is to
`measure the LED drive current directly. The microprocessor would then use that
`drive current value to calculate the estimated temperature change, and from that,
`calculate the estimated peak wavelength shift. This method eliminates the second
`problem listed above for the temperature sensor solution, but still leaves the
`problem of variations in the relationship between temperature and peak
`wavelength among individual LEDs. Another advantage of this method is that no
`extra wires or other components need to be added to a probe, making this the less
`expensive of the two methods discussed here.
`
`5.6 PREVENTION OF BURNS IN PULSE OXIMETRY
`
`In order to prevent burns on a patient's skin due to LED heat, the Food and Drug
`Administration now requires that the contact region between the skin and the
`oximeter probe not exceed 41 °C. Given an average body temperature of 37 °C, a
`pulse oximeter system should be designed to yield a maximum temperature rise
`of 4 'C at the skin-probe contact region, which is the primary dissipator of the
`LED heat. The relevant LED specification is thermal resistance, discussed in
`section 5.2.3. In pulse oximetry, the therma] resistance of each LED is on the
`order of a standard LED mounted in a PC board, which is a specification given in
`LED product catalogs. As previously mentioned, many pulse oximeters on the
`market have a maximal LED pulse current of 50 mA. This is sufficiently small to
`prevent dangerous LED heating, while still providing adequate light to the
`photodiode.
`Mills and Ralph (1992) tested the heating of six pulse oximeter probes over a
`span of 3 h. The probes were placed in an incubator kept at a constant
`temperature of 36.9 to 37 °C. The working temperatures of the probes were
`quite similar, with a range of 39.1 to 39.7 °C over the entire 3 h. One of the
`probes was monitored for 24 h, and during that time its temperature remained
`constant within a range of *0.1 °C.
`The conditions of this test, however, did not do anything to simulate the
`reaction of skin to heating of a few degrees for several hours. To ensure that no
`burning occurs, the probe's point of application should be inspected often. In
`addition, the position of the probe on the patient should be changed regularly,
`especially if the probe application area suffers from low perfusion, which limits
`the skin's ability to dissipate heat.
`
`5.7 LED PACKAGING
`
`Most LED packages are made of resin, offering superior mechanical strength and
`the ability to withstand vibration and shock. In some of today's pulse oximeter
`probes, the two LEDs can be found in one package, which has the distinct
`advantage of keeping costs down. In the Nellcor SCP-10 reusable and Oxisensor
`II D-25 disposable pulse oximeter probes, the two LEDs come encased in a
`transparent rectangular solid with approximate dimensions of 5 mm long by 4
`
`86
`
`MASIMO 2053
`Apple v. Masimo
`IPR2022-01300
`
`
`
`RX-0035.0087
`
`APL_MAS_ITC_00015704
`
`70
`
`Design of pulse oximeters
`
`mm wide by 2 min thick. The LEDs themselves are flat squares with sides ot
`approximately 0.25 mm.
`Other probes have discrete LEDs inside, with the LEDs lying side by side
`and a mirror to reflect light at a 900 angle to the tissue. Some probes even have
`three or four LEDs in them to increase light output. The details of these and
`many other probes will be discussed in Chapter 7.
`There is no wrong choice for LED packaging as long as the LEDs are small
`yet powerful enough to perform the task at hand. However, there is definit