Developing a Digital Blood Pressure Meter
by CS Chua and Slew Mun Hin, Motorola Sensor Product Division
Blood pressure meters are among the most frequently used instruments in the medical industry. It is essential for healthcare professionals to use the most accurate and efficient meters for their applications. In order to provide better solutions, many of today's engineers are designing digital systems to monitor a patient's blood pressure.
Engineers developing a digital blood pressure system can choose many design solutions. But one of the most effective is developing a system which employs an integrated pressure sensor, analog signaling-conditioning circuitry, microcontroller hardware/software, and a liquid crystal display (LCD). Through this design, engineers can develop systems which can read cuff pressure (CP) and can extract pulses for analysis and determination of systolic and diastolic pressure.
Oscillometric Method
When developing a digital blood pressure system, many engineers produce devices that employ the oscillometric method, which is used by the majority of non-invasive devices. With this method, a limb and its vasculate are compressed by an encircling, inflatable compression cuff. The blood pressure readings for systolic and diastolic values are then read at the parameter identification point.
The oscillometric method is simply a measurement of the amplitude of the pressure change in the cuff as the cuff is inflated from above the systolic pressure. The amplitude suddenly grows larger as the pulse breaks through the occlusion. This is very close to systolic pressure.
As the cuff pressure is further reduced, the pulsation increase in amplitude reaches a maximum and then diminishes readily. The index of the diastolic pressure is then taken where the rapid transition begins. Therefore, the systolic blood pressure and the diastolic blood pressure are obtained by identifying the region where there is a rapid increase then decrease in the amplitude of the pulses.
In the digital blood pressure system, the cuff pressure is sensed by an integrated pressure sensor, such as the X-ducer developed by Motorola (Phoenix, Ariz.). The output of the sensor is then split into two paths which are used for two different purposes. One path is used as the cuff pressure while the other path is further processed by a circuit. Since the pressure sensor is signal-conditioned by an internal operation amplifier, the cuff pressure can be directly interfaced with an analog-to-digital converter (ADC) for digitization. The other path will filter and amplify the raw CP signal to extract an amplifier version of the CP oscillations, which are caused by the expansion of the subject's arm each time pressure in the arm increases during cardiac systole.
The output of the sensor consists of two signals, with the approximate 1 Hz oscillation signal riding on the CP signal. As a result, a two-pole, high-pass filter is needed to block the CP signal before the amplification of the oscillation signal. If the CP signal is not properly attenuated, the baseline of the oscillation will not be constant and the amplitude of each oscillation will not have the same reference for comparison (Fig. 1).
The filter consists of two RC networks which determine two cut-off frequencies. These two poles are carefully chosen to ensure that the oscillation signal is not distorted or lost. The two cut-off frequencies can be approximated using the equations:
Figure 2 shows the frequency response of the filter. (This plot does not include the gain of the amplifier.)
The oscillation signal varies from person to person. But, in general, it varies from less than 1 to 3 mmHg. From the transfer function of the pressure sensor, this translates into a voltage output of 12 to 36 mV. Since the filter gives an attenuation of 10 dB to the 1 Hz signal, the oscillation signal becomes 3.8 to 11.4 mV. Experiments indicate that the amplification factor of the amplifier is chosen to be 150 so that the amplified oscillation signal is within the output limit of the amplifier (Figs. 3 and 3b)
In the digital meter design, the pressure sensor and the amplifier are linked to ports on the microcontroller. This port is an input to the on-chip, 8-b ADC. The pressure sensor provides a signal output to the microprocessor of approximately +0.2 VDC at 0 mmHg to +4.7 VDC at 75 mmHg of applied pressure. The amplifier, on the other hand, offers a signal from +0.005 to +3.5 VDC. In order to maximize resolution, separate voltage references should be provided for the ADC instead of using a +5 VDC supply. In this example, the input range of the ADC is set to approximately 0 to +3.8 VDC. This compresses the range of the ADC around 0 to 300 mmHg to maximize the resolution.
The digital meter's voltage divider is connected to the +5 VDC supply powering the system. The output of the pressure sensor is ratiometric to the voltage applied to it. The pressure sensor and the voltage divider are connected to a common supply, yielding a system which is ratiometric. By nature of the ratiometric system, variation in the voltage of the power supplied to the system will have no effect on system accuracy.
The LCD is directly driven by from input/output (I/O) ports on the microcontroller. The operation of an LCD required that the data and backplane pins must be driven by an alternating signal, This function is provided through a software routine which toggles the data and the backplane at a 30 Hz approximate rate.
Other than the LCD, there are two more I/O devices which are connected to the plasma length converter of the microcontroller-a buzzer and a light emitting diode (LED). The buzzer, which is connected to the pulse length converter, can produce two different frequencies-122 Hz and 1.953 kHz. Thus, when the microcontroller encounters a certain error due to improper inflation of the cuff, a low frequency tone is alarmed. One the contrary, when a measurement is successful, a high frequency tone will be heard.
The microcontroller section of the system requires certain support hardware in order to function. For example, a 4 MHz crystal provides the external portion of the oscillator function for clocking the microcontroller and provides a stable base for time-based functions.
Power Up
At system power up of the digital blood pressure meter, the user needs to manually pump the cuff to approximately 30 or 160 mmHg above the previous systolic blood pressure. During the pumping of the inflation bulb, the microcontroller ignores the signal at the output of the amplifier. After the subroutine TAKE senses a decrease in CP for a continuous duration of more than 0.75 s, the microcontroller assumes that the user is no longer pumping the bulb and starts to analyze the oscillation signal (Fig. 4).
In Figure 4, the threshold level of a valid pulse is set to +1.75 VDC in order to eliminate noise or spike. As soon as the amplitude of a pulse is identified, the microcontroller will ignore the signal for 450 ms in order to prevent any false identification caused by the presence of premature overshoot provided by oscillation. As a result, the algorithm can only detect pulse rate which is less than 133 beats/minute.
The amplitudes of all the pulses detected are then stored in the random access memory (RAM) for further analysis. If the microcontroller senses a non-typical oscillation envelope shape, an error message is output to the LCD. When this occurs, the user will have to exhaust all pressure in the cuff before re-pumping the CP for the next higher value. An algorithm within the system ensures that the user exhausts all of the air present in the cuff before allowing any re-pumping. If this does not happen, the venous blood trapped in the distal arm may affect the next measurement.
A Final Note
When choosing a microcontroller for the digital blood pressure system, engineers should look for a few key characteristics. The microcontroller should be equipped with 2 kB of on-chip read only memory (ROM) space, 150 B of RAM space, a 2-channel ADC, a 16 b free running counter timer, an LCD driver, 32 B of EEPROM, and power saving modes.
C.S. Chua and Slew Mun Hin, Sensor Application Engineering, Motorola, Inc., Sensor Products Division, P.O. Box 20912, Phoenix, AZ 85036, Tel: (602) 303-5454, Fax: (602) 244-4201.