Measuring electrical phenomena presents a number of challenges for biomedical signal measurement systems. The tiny, microvolt-level electrical pulses that signal a firing neuron or a muscle response are often obscured by high-amplitude noise or accompanied by significant dc potentials. Quite often, the signal of interest is a small, transient pulse that occurs intermittently or only once. In some applications, minute chemical and catalytic changes occur over a matter of several minutes or even hours, making it critical that important experimental events can be captured in a single acquisition.
A useful tool for making low-level stimulus and response measurements in biomedical research environments is a digital phosphor oscilloscope (DPO) combined with a differential amplifier to improve signal quality. This article discusses the basic techniques for using these tools to obtain accurate biomedical measurements.
To distinguish low-level signals and transients from the surrounding noise, biomedical measurement systems must provide not only wide dynamic range and high-quality signal conditioning, but also long record lengths for long duration event capture, extensive storage capacity, and single-shot acquisition capabilities with adequate time resolution. Further, proper documentation and reporting of results is a critical step in the bioscience research process. This means that researchers and biomedical engineers need the ability to store not only acquired waveforms, but also waveform measurement results and test setups. As such, the measurement system must provide comprehensive data storage and retrieval, analysis, and documentation capabilities.
A DPO and a low-noise differential amplifier or preamplifier provide a powerful, portable, and affordable measurement system for capturing and analyzing low-amplitude biomedical phenomena in the presence of noise. This solution allows engineers and researchers to pick up extremely small amplitude signals, such as neurons firing during an experiment or a muscle fiber response to external stimulus. It can also be used to capture the fine nuances of microvolt-level electrophysiological signals, including single-shot signals, in the presence of much larger common-mode noise signals.
A major challenge in measuring low-level signals is dealing with unwanted noise. When measuring signals in the microvolt range, noise can often be thousands of times greater in amplitude than the signal of interest. Noise in the biomedical environment can be divided into two categories: noise inherent in the signal and noise caused by the external environment. Inherently noisy response signals are usually caused by a noisy stimulus signal or some other source of noise within the test and measurement apparatus itself. External noise is generated outside the test and measurement equipment by sources such as florescent lights, stray electric or magnetic fields, and poor shielding or grounding.
If the desired signal is inherently noisy, the noise will be amplified along with the signal of interest. Selective filtering can be employed to eliminate the noise. For instance, low-pass filters in the preamplifier can be used to eliminate high-frequency noise from the signal. In most cases, this noise filtering technique will not alter the essential character of the signal of interest. In cases of extreme noise, sharp cut-off notch filters or signal averaging may be required to extract the desired signal.
Another approach is to use averaging to increase the oscilloscope’s effective vertical resolution. In this mode, real-time digital filters are applied to the oscilloscope’s digitizer output prior to writing the acquired signal to memory. This can significantly reduce high-frequency noise on lower-frequency signals. In many oscilloscopes, the averaging algorithm is implemented with fixed-point math, which retains the full bandwidth of the signal but can’t be used for single-shot acquisitions. For medical applications, the solution is a boxcar averaging technique that calculates and displays the average of all sequential sample values in each sample interval. This mode does not rely on the presence of a stable trigger and, therefore, can be used on single-shot or nonrepetitive events.
Noise that enters the measurement from the external laboratory environment and affects both the signal and the reference equally is called common-mode noise. Touching a finger to an oscilloscope probe, for example, will likely result in a large 50 or 60 Hz signal on the oscilloscope’s display. This is common-mode noise that a body, acting as an antenna, picks up from the environment. Biological specimens can pick up these same undesirable signals. Some of these common-mode signals can be eliminated by removing noise-generating devices, such as fluorescent lights, from the laboratory. Surrounding the lab setup with a grounded electrical mesh will also help to eliminate common-mode noise. Even with these precautions, however, some common mode noise may still be present. This remaining common-mode noise is chiefly due to the inability to ground the biological specimen adequately. The solution to this problem is the use of a high-performance differential amplifier.
A properly balanced differential amplifier can amplify very small signal differences, while attenuating or decreasing the amplitude of common-mode noise. The ability of a differential amplifier to reject noise is called its common-mode rejection ratio (CMRR). Higher-end differential amplifiers offer a CMRR of 100,000:1, allowing capture of small signals in the microvolt range (5–10 µV) when high-amplitude common-mode noise is present.
Differential amplifiers have two inputs, both of which are designed to be connected to the specimen. A differential amplifier is shown in Figure 1. Neither of these inputs are grounded; in other words, the amplifier floats above ground potential. A ground electrode is sometimes connected to the specimen to reference it to the measurement system. When the two differential inputs are connected to the specimen and the impedance at the two connections are reasonably well matched to the amplifier’s impedance, the amplifier sees only the true difference signal.
Figure 1. This is a typical differential amplifier, in this case a Tektronix ADA400A Differential Preamplifier.
The standard oscilloscope differential amplifier input impedance at the frequencies encountered in biophysical research is 1 MΩ each side to ground, or 2 MΩ across the differential inputs. This impedance level occurs because the input impedance of the oscilloscope is typically set to 1 MΩ and 1× passive voltage probes are used for minimum attenuation, resulting in a 1-MΩ input resistance of each input to ground. These values produce a net 500-kΩ impedance to ground for common-mode signals.
The effectiveness of differential amplifier for common-mode rejection is illustrated in Figures 2a and 2b. Figure 2a shows unfiltered monitor signals. Note, the signal on channel one contains both a simulated cardiac signal, similar to what would be seen on an ordinary heart monitor, as well as a large 60 Hz sinusoidal noise trace. The signal on channel two contains the same 60 Hz sine wave but without the cardiac signal. Because the sine wave noise is significantly larger than the cardiac signal, it is difficult to view the anomalies in the at-rest area following the main beat. After connecting the differential amplifier to the composite signal, the large common-mode signal is removed between the two inputs, and the resulting differential is displayed as shown in Figure 2b.
Figure 2a. This shows large scale 60 Hz sinusoidal noise with (yellow Channel 1) and without (blue Channel 2) small scale cardiac output signal.
Figure 2b. Here is a small scale cardiac signal after that differential amplifier was set to 100x gain. Note, 617.3 μV pk-pk vs. composite signal of sinusoidal noise and cardiac output in Figure 2a with peak-to-peak voltage of 1.015 V.
In another example, the differential amplifier is used to measure a 100-µV signal from a specimen that produces a 0.5-V common-mode signal. In this experiment, the specimen interface impedances were low and matched. Using a vertical scaling on the oscilloscope of 50 µV/div, the resulting display shows the amplitude of the signal of interest occupying two vertical divisions of the screen, while the common-mode noise takes up only 0.1 division. The 100,000:1 CMRR of the differential preamplifier causes the common-mode noise to be attenuated from 0.5 V to 0.5 µV, essentially eliminating it from the measurement.
This example illustrates the usefulness of a differential amplifier for measurements when the source impedances are low and well matched. In practice, however, it may not be possible to control the source impedances. In such situations, the CMRR of the differential amplifier will be degraded.
Figure 3 offers an example of a situation where specimen interface impedances of 2 kΩ and 0.5 kΩ were created when the research procedure was unable to establish good control between the electrodes and the specimen.
Figure 3. Here is a differential amplifier connection with unbalanced source impedances.
If the specimen interface creates a high and possibly different impedance between the electrode pairs, as in Figure 3, the measured signal will not truly represent the signal at the specimen interface. Also, the voltage dividers will be different, causing the CMRR to degrade according to the following formula:
CMRR = |(R3 or R4)/(R1-R2)| = 1 MΩ / (2 kΩ- 0.5 kΩ) = 666:1
When the CMRR is degraded like this, the displayed common-mode noise is much greater. If, in the example given above, the CMRR of the differential amplifier is
degraded to 666:1, the amplitude of the common-mode noise will occupy the equivalent of 15 vertical divisions on the oscilloscope’s display (which extends beyond the top and bottom of the screen). Even with the high gain for the differential signal, the 15-division display of the common-mode noise will make the two-division response signal unreadable.
The solution to the problem of degraded CMRR is to raise the input impedance of the differential amplifier. If the differential amplifier had an essentially infinite input impedance, the circuit in Figure 3 would look like the circuit in Figure 4. In this case, there is essentially no voltage divider action due to the mismatched interface impedances, and the full CMRR can be very nearly attained.
Figure 4. This illustrations shows a differential amplifier with increased input impedances.
This is accomplished by disconnecting the internal 1-MΩ resistors in the differential amplifier (usually using jumpers), thus presenting an essentially infinite impedance to the source. Also, the gate current, generally less than 25 picoAmps of the field effect transistor (FET) at the amplifier input, must have an external path to instrument ground. This path is usually provided by the specimen or the signal source itself. On rare occasion, the source can be purely capacitive and some conductance must be added, either in the amplifier itself or at the source, to instrument ground. (Because the gate current is very low, this path can be resistive.)
Electrode potentials exist whenever metallic electrodes interface with the specimen via an electrolyte. Differences between electrode-pair contact potentials produce an offset potential, typically in the range of hundreds of millivolts, which appear as a dc voltage source in series with the desired signal. The nominal dc-coupled amplifier load of 2 MΩ will tend to discharge these batteries, but residual offset may displace the desired signal off-screen, especially at high sensitivities. Here are a few suggestions for cancelling the effects of this offset potential:
Also, ac coupling cannot be used in the high-impedance mode described.
It is desirable not to have noise signals to contend with in the first place. Eliminating noise sources such as fluorescent lighting or constructing a grounded mesh around the test setup are good first steps, but there are also others that can be taken.
Leakage of currents to ground through the specimen can be avoided by connecting stimulus pulse generators through stimulus isolators, which present the stimulus pulse across a discrete area. Stimulators with one lead grounded could produce large ground currents through the specimen. If these currents flow through the response pick-off point, the resulting potential drop will show as an unwanted signal. If a grounded stimulator is used, the grounded electrode should always be placed between the signal electrode and the measurement electrodes, as shown in Figure 5.
Figure 5. The is the set up for differential measurements with one grounded input signal.
An extension of this principle can be applied when making stimulus-response measurements on an excised nerve of a biological specimen. As shown in Figure 6, a grounded electrode could be placed across the nerve between the stimulus isolator and the recording electrodes to effectively bypass surface currents to ground. The recording electrodes will then see the conducted action potential with very little stimulus artifact.
Note that ground in this context refers to circuit ground, typically located at the differential amplifier. The triangular ground circuit symbol is used to signify circuit ground. Safety ground or earth ground is denoted by a rake ground symbol.
Often, multiple pieces of line-operated equipment are used to perform biomedical experiments. How this equipment is connected can greatly affect the level of noise generated in the measurement system. The voltage of third wire ground connection, for example, at various wall outlets may not be at exactly the same ground potential, or at the same level between outlets. If two or more pieces of equipment are connected together via coaxial cables, it is possible for circulating line currents to flow in the outer braid. This ground loop can inject line ripple into the inputs of susceptible devices, such as amplifiers. To avoid these problems, safety grounds should be solid and all equipment used in a measurement should be connected to the same ground bus.
Figure 6. This is the set up for differential measurements with floating input signals.
Any cable, shielded or otherwise, can pick up induced currents if they pass close to power transformers, line cords, or other ac-current-carrying leads. Care has to be taken to route single-ended signal leads away from such sources, and paired differential leads are often twisted together to cancel out induced currents. Ideally, differential amplifier circuitry is placed at the probe end where it is as close as possible to the specimen being tested. Probes that interface with the animal or specimen should be shielded and grounded at the equipment end. Never ground both ends of signal leads, as this immediately sets up a ground loop. Figure 7 shows the correct grounding technique. Using this test setup, the differential amplifier eliminates the effects of ground loops while keeping the oscilloscope safely grounded. In the United States, OSHA warns that floating test equipment above ground can be hazardous and increases chances of electric shock. To be safe, never float instruments by disabling the safety ground connection.
Figure 7. Here is the proper grounding technique for differential measurements
By paying careful attention to the grounding of the equipment, isolation of the signal generators, and shielding of the probes and leads, it is possible to obtain refined biomedical measurements without complicated and expensive test equipment setups, preconditioning equipment, and external filters. Using an oscilloscope and a differential amplifier, engineers and researchers can obtain complete biomedical measurements. Such a test system delivers precise signal conditioning, outstanding acquisition confidence, comprehensive on-board signal processing end analysis, and accurate results storage and report generation capabilities, making it versatile enough to solve a variety of complex measurement problems in the areas of manufacturing test, bioscience research, power electronics or power supply design, and electronic product service end repair.