Precision Pulse Oximeter LED Current Sinks Using the ADA4505-2 Op Amp, the ADR1581 Shunt Reference, and the ADG1636 Switches
A pulse oximeter is a noninvasive medical device used for continuously measuring the percentage of hemoglobin (Hb) saturated with oxygen and the pulse rate of a patient. Hemog-lobin that is carrying oxygen (oxyhemoglobin) absorbs light in the infrared (IR) region of the spectrum; hemoglobin that is not carrying oxygen (deoxyhemoglobin) absorbs visible red (R) light. In pulse oximetry, a clip containing two LEDs (sometimes more, depending on the complexity of the measurement algorithm) and the light sensor (photodiode) is placed on the finger or earlobe of the patient. One LED emits red light (wavelengths of 600 nm to 700 nm), and the other emits light in the near IR (wavelengths 800 nm to 900 nm) region. The clip is connected by a cable to a processor unit. The LEDs are rapidly and sequentially excited by two current sources (one for each LED) whose dc levels depend on the LED being driven, based on manufacturer requirements. The detector is synchronized to capture the light from each LED as it is transmitted through the tissue.
Low power, precision current sources (if the current flows into the load) or current sinks (if the current flows out of the load) used in pulse oximeter designs are required to deliver a few decades of milliamps (hundreds of milliamps for legacy products). The active elements in these circuits are a low power precision operational amplifier, a precision shunt voltage reference, and a MOSFET or a bipolar transistor. To save power an analog switch can be added to power off the current source/sink when it is in the standby mode. If ultraprecision design is required, then an ultraprecision series voltage reference can be utilized instead of the shunt voltage reference.
An excellent low power, low cost precision amplifier to use in this medical application is the dual 10 μA rail-to-rail zero crossover distortion ADA4505-2. A good ultralow power, low cost precision shunt voltage reference to accompany the amplifier is the 1.25 V ADR1581 (A grade). An excellent choice for the analog switch is the 1 Ω on-resistance ADG1636 dual SPDT switch.
The maximum quiescent currents for the devices are as follows when these devices are operated from a power supply rail of 5 V and working under the industrial temperature range of –40°C to +85°C: one-half the ADA4505-2, 15 μA; ADR1581, 70 μA; ADG1636, 1 μA. These numbers add up to a total of 86 μA current consumption per circuit, which is suitable for portable and battery-powered instruments.
Current sources (sinks) implemented using the low power, low cost ADA4505-2 amplifier; the micropower, low-cost ADR1581 shunt voltage reference; and the ultralow power ADG1636 SPDT analog switch are characterized by being precise, low power, cost effective, flexible, and small in PCB footprint.
To maximize battery life, the current sinks are turned on only when needed. One half of the ADG1636 SPDT analog switch is used to connect/disconnect the 1.25 V voltage reference to/from each current circuit.
When the current sinks are driving their respective LEDs, the ADR1581 (A grade) 1.25 V voltage reference is buffered by one-half the ADA4505-2. The IRLMS2002 N-channel MOSFET is connected as a source follower and is inside the op amp feedback loop. This forces the voltage across the current setting resistor (121 Ω or 82.5 Ω) to exactly 1.25 V. This in turn sets the current in the current sources to 10.3 mA or 15.2 mA. In essence, the ADA4505-2 is acting as both a voltage reference buffer and current switch control.
The equation that sets each current sink value is
where VREF is the 1.25 V voltage reference, RS is the 121 Ω or 82.5 Ω current sink resistors, VOS and IB are the ADA4505-2’s offset voltage and bias current, respectively.
If we ignore the VOS and IB of the amplifier in order to simplify the circuit analysis, the ISINK is the 10.3 mA or 15.2 mA current through the red or infrared LED.
The current sinks are turned-off by disconnecting the voltage reference from its current sink resistor and connecting this resistor to ground.
When the current sinks are on, each is on only a certain amount of time, and not at the same time. This time is set by the duty cycle of the waveform driving its corresponding current sink (pins IN1 and IN2 of the ADG1636 switches). These waveforms are pulses with duty cycle of approximately 25% and a typical period of 1 ms (1 kHz). This means that each current sink is on during 250 μs in a 1 ms period. A typical timing of these red and infrared current sinks is shown in Figure 2.
Figure 2. Pulse Oximeter Red and Infrared Current Sources Typical Timing
With this timing, the total current consumption of both current sinks can be calculated by the equation
This gives a total of less than 6.5 mA out of the 5 V supply needed to obtain a pulse oximetry reading. It is worth pointing out that the 86 μA is the previously calculated current consumption of the ADA4505-2, ADR1581, and ADG1636 combination, and this is only 1.3% of the total 6.5 mA needed. Therefore, these three active elements add a negligible load to the battery.
The VOS error term in the equation should not only account for the amplifier’s offset voltage but for all the errors caused by the op amp’s non-ideal behavior. Each of these errors should be mathematically accounted for as an additional VOS reflected back to the input of the op amp. By using a precision amplifier as the ADA4505-2, the total sum of all these errors is negligible compared to the op amp’s inherent offset voltage.
The ADA4505-2’s VOS is 3 mV maximum which represents an error of 0.24% in the sink current. By the same reasoning, the IB of 2 pA maximum in the ADA4505-2 will give an error that can be considered as zero, when compared to the 10.3 mA and 15.2 mA required out the red and infrared current sinks.
If to this VOS error of 0.24% we add the 0.8% maximum initial accuracy of the ADR1581 and the 0.1% sink resistor tolerance (chosen for this design), then the total worst case error is 1.14%, and the uncertainty analysis error is 0.83%. Now, if we take into account in this uncertainty analysis the Gaussian distribution curves of the errors from each of the contributing elements (voltage reference, sink resistor, and op amp), the less pessimistic expected error is 0.28% (see Holman, J. P., Experimental Methods for Engineers, McGraw-Hill, Fourth Edition, 1984).
The error allowable in the current sink depends on the degree of accuracy needed in the pulse oximeter reading. In a red and infrared LED, the accuracy of the current driving it is directly proportional to the accuracy of its radiant flux (radiant output power). This radiant flux is a measure of how much power is contained in the light being emitted. Therefore, high LED current accuracy means LED radiant flux high accuracy prediction. A typical radiant flux vs. forward current curve is shown in Figure 3 for the Hamamatsu L5276 , L5586, and L6286 infrared LEDs.
Figure 3. Radiant Flux vs. Forward Current for Hamamatsu Subminiature Infrared LED: L5276, L5586, L6286
Regarding the other elements in the design, the 22 pF capacitors are used to improve stabilization (in-loop compensation) of the ADA4505-2 amplifiers. (See Analog Dialogue, “Ask the Applications Engineer—25.” Analog Devices.) The 1 kΩ feedback resistors are used to provide some current limiting into the amplifiers’ inverting pin. The 22 Ω resistors in series with the ADA4505-2 outputs serve two purposes: one is to prevent possible oscillation when driving the N-MOSFET’s input capacitance (Ciss); the second is to dampen some of the transient response of the N-MOSFETs when they turn on and off. These resistors and capacitances may need some further optimization in the specific application.
Table 1 shows the calculated (ideal) and measured values on both red and infrared current sinks for the design in Figure 1 at ambient temperature.
Table 1. Ideal and Measured Pulse Oximeter Red and Infrared Current Sources
|Current Sink||Ideal Value (mA)
||Measured Value (mA)
The use of the IRLMS202 N-channel MOSFET allows this design to be used for currents up to hundreds of milliamps (special attention has to be placed in making sure its safe operating area is not violated). When the current levels are in the order of decades of milliamps, the reliable and cost-effective BSS138 N-channel MOSFET can be utilized instead.
If the accuracy and temperature drift of the total design has stringent requirements, use a more accurate and lower temperature coefficient drift voltage reference, such as the high precision series voltage reference ADR127 or the higher accuracy B grade of the ADR1581, ADR1581 (B grade); a tighter tolerance and low temperature drift should also be selected for the current sink resistor; and a very low VOS precision amplifier should be chosen, for instance the auto-zero (zero-drift) AD8629 (2.7 V to 5 V) or the OP07D (8 V to 36 V).
If high precision current sinks are not required, and depending on the total design tolerance requirements, then an all-purpose, cost-effective op amp can be used. Good examples of these device are the AD8515 (1.8 V to 5 V), AD8542 (2.7 V to 5.5 V), AD8529 (2.7 V to 12 V), AD8566 (4.5 V to 16 V), and OP275 (9 V to 44 V).
|ADA4505-2||10 µA, RRIO, Zero Input Crossover Distortion Dual Op Amp||
|ADG1636||1 Ω Typical On Resistance, ±5 V, +12 V, +5 V, and +3.3 V Dual SPDT Switches||
|ADR1581||1.25 V Micropower, Precision Shunt Voltage Reference||