Not long ago, the only accelerometers available employed either a quartz crystal or moving coil. Typical costs of these devices were $500 or more. Now, a new generation of solid-state devices is smaller and lighter than its predecessors and is less expensive as well. Some of the new accelerometers cost only about $50, but cannot handle certain kinds of applications. In fact, choosing an accelerometer sometimes can be harder than using one.

One of the more common accelerometers is called a suspended-mass type. Suspended-mass accelerometers are characterized by their small size (solid-state devices are about as big as a transistor) and ability to respond to dc (static) accelerations. Frequency response ranges from dc to about 5 kHz depending on the particular type. High-g-range units generally have higher frequency response than their low-range equivalents. A simple model of such a device is a mass suspended on the end of a cantilever beam, but actual devices are more complicated.

For instance, the new solid-state versions of this type suspend a silicon mass from four deflection beams. The mass and beams are in a cavity in the silicon structure.

Two features usually protect the beams from overflexing. In addition to protecting the accelerometer, these mechanisms also affect its measuring capabilities. The first safety device is a mechanical stop. It protects the beam from damage during peak accelerations. Second, oil provides viscous damping. Damping also increases the accelerometer's frequency response and prevents damage from frequencies near the beam's resonant frequency.

Suspended-mass devices use two different methods to determine acceleration. The first method is to measure the force (strain) induced in the beam by the mass. Here, strain gages on the beam generate a signal proportional to acceleration. A bridge circuit connects the four strain-sensing elements. The result is a device having performance similar to that of standard piezoresistive strain-sensing elements. Accelerometers of this design are usually called strain-gage bridge accelerometers, regardless of the type of strain sensing element used.

Rather than gauging deflections with strain gages, some suspended-mass devices use capacitive sensing. This produces an electrical signal proportional to displacement. An advantage of some capacitive accelerometers is that they are air damped. This makes their damping coefficient relatively insensitive to temperature, unlike fluid-damped devices.

The most rugged of all accelerometers is the piezoelectric type. A typical unit, rated 10,000 g, is housed in a cylindrical package about 1 in. long by in. in diameter. Piezoelectric transducers also offer a wider temperature range (-195 to 260°C) than any other accelerometers.

The frequency response of these devices is wide because the active sensor is small. Most units have a frequency range of at least 5 kHz. Typical resonant frequencies are over 30 kHz.

Though tough, the devices are not without their problems. One is that they do not respond to static accelerations. For instance, typical units have a minimum response in the range of 0.01 to 5 Hz. But a bigger concern is with the output.

For accurate measurements, the accelerometer and cable must be calibrated as a system. If the cable length is changed, another calibration is required.

The alternative is to measure the charge generated by an acceleration. This techniques negates cable effects, but requires a charge amplifier. Compared with other amplifiers, charge amplifiers are expensive.

Another problem associated with piezoelectric accelerometers is the need to maintain high insulation resistance and low noise in the cabling between the accelerometer and the charge amplifier. Otherwise, measurement errors will be introduced into the system. Generally, the cable must be of special low-noise type (supplied by the accelerometer maker) and must be fastened in place to reduce vibration-induced electrical noise.

As an alternative to an external amplifier, some manufacturers offer piezoelectric accelerometers with built-in amplifiers. These devices overcome the cabling and amplifier problems. They also have low output impedance. This means that connecting them with oscilloscopes or chart recorders is relatively easy. The primary disadvantage of internally amplified units it that they require a separate power supply.

Another common accelerometer is the electromechanical force-balance type. These are commonly called servo accelerometers because they use a closed-loop control system to measure acceleration.

Mechanically, the devices consist of a permanent magnet and a sensing mass attached to a moving coil. Servo accelerometers also require some internal circuitry generally consisting of a position sensor, error amplifier, and a voltage-to-current (V/I) converter.

The sensor output represents the location of the sensing mass with respect to a null or center position. An error amplifier and V/I converter change the position signal into a current which drives the moving coil.

The moving coil and permanent magnet exert a force on the mass. The force is proportional to the current in the coil. When the mass begins to move away from the null position because of an external acceleration, the position error generates a coil current, returning the mass to the null position. A voltage signal representing coil current is used as the accelerometer output signal.

One variation in the construction of the servo accelerometer uses a bobbin-shaped coil mounted on a disc. The disc supports the coil with two beams, allowing the coil to move axially. The coil is surrounded by a permanent magnet. Current forces the coil into a null position in response to external accelerations. The advantage of this structure is high reliability because the coil needs no pivot.

Servo accelerometers typically require supply voltage of ±15 Vdc. Output signals of ±5 V are typical. Because the structure is bulkier than suspended-mass accelerometers, frequency response is lower. Most units do not respond above a few hundred cycles/sec. These accelerometers usually come with mechanical stops or fluid dampers to prevent damage from accelerations beyond their range.

Frequency response is often the single most important issue in selecting accelerometers. A primary concern is whether response to static accelerations, such as gravity, is needed. If so, use a suspended-mass or servo type. Of the two, suspended-mass types are more durable.

However, most applications do not really need static response. Instead, the response must simply be close to dc. Piezoelectric accelerometers have low (0.01 to 5-Hz) frequency cutoff but do not respond to static acceleration. At the high end of the frequency response spectrum (above a few thousand Hertz), only piezoelectric devices operate properly.

Another important consideration is g-level range. This determines the maximum g load a device can measure accurately. It also indicates the maximum acceleration that a device can withstand without damage or permanent scale shift. It should be noted that the maximum acceleration encountered can be substantially higher than expected, because of incorrect mounting or loose parts. Except for servo accelerometers, other types withstand shocks over several thousand g.

Working temperatures can also influence selection. Piezoelectric accelerometers with separate charge amplifiers have the widest temperature ranges.

An accelerometer has a temperature stability specification, which influences performance. Temperature stability has two facets: thermal gain shift and thermal zero shift. Thermal gain shift tells how much temperature changes accelerometer sensitivity. This is usually expressed in percent of full scale per degree.

Thermal zero shift is the amount of accelerometer output shift caused by temperature changes with no acceleration applied. This is given in percent of full scale per degree.

Servo accelerometers generally have the best temperature stability, but specific temperature ranges must be considered. Temperature shift is usually nonlinear over typical operating ranges. As might be expected, operating the device in a narrow temperature band minimized these effects. Depending on the device, full-scale changes over the rated temperature range vary from (0.1% to 10%). Some accelerometers also have ratings for storage temperature that differ from the operating temperature limit.