feature

Pre-trimmed to ±0.25% maximum four-quadrant error (AD534L); all inputs (X, Y, and Z) differential high impedance [(X1-X2)(Y1-Y2)/10 V] + Z2 transfer function; scale factor adjustable , provides up to × 100 gain; low noise design: 90μV rms, 10 Hz to 10 kHz; low cost overall construction; excellent long-term stability.

application

High-quality analog signal processing; differential ratio and percentage calculations; algebraic and trigonometric synthesis; wideband, high peak rms-to-dc conversion; precision voltage-controlled oscillators and filters; available in chip form.

General Instructions

The AD534 is a monolithic laser-trimmed four-quadrant multiplier voltage divider with an accuracy specification previously found only in expensive hybrid or modular products. The maximum multiplication error of ±0.25% is guaranteed without any external trimming of the AD534 L. The excellent power supply rejection, low temperature coefficient and long-term stability of on-chip thin film resistors and buried Zener references maintain accuracy even under adverse usage conditions. This is the first multiplier to offer fully differential, high impedance operation on all inputs, including the Z input, a feature that greatly increases its flexibility and ease of use. The scale factor is predefined to a standard value of 10.00 V; it can be reduced to values as low as 3 V with external resistors.

The wide range of applications and the availability of multiple grades make this multiplier the first choice for all new designs. The AD53J (±1% max error), AD534 K (±0.5% max), and AD534 L (±0.25% max) are specified for operation over the 0°C to +70°C temperature range. The AD534S (±1% max) and AD534 T (±0.5% max) are specified over the extended temperature range, 55°C to +125°C. All grades are available in sealed TO-100 metal cans and SBDIP packages. AD534K, AD534S and AD534T chips are also available.

Absolute Maximum Ratings

Stresses listed above the Absolute Maximum Ratings may cause permanent damage to the device. This is a stress rating only; functional operation of the device under the conditions described in the operating section of this specification or any other conditions above is not implied. Exposure to absolute maximum rating conditions for extended periods may affect device reliability.

Thermal resistance

θ is specified for the worst case, that is, a device soldered in a circuit board for a surface mount package.

Typical performance characteristics

Typically ±VS=±15 V dc at 25°C unless otherwise noted.

Function description

Figure 1 shows the functional block diagram of the AD534. The input is converted to differential current by three identical voltage-to-current converters, each zero-offset trimmed. The product of the X and Y currents is generated by a multiplying unit using Gilbert's translinear technique. The on-chip embedded Zener provides a highly stable reference that is laser trimmed to provide an overall scale factor of 10V. The difference between XY/SF and Z is then applied to a high gain output amplifier. This allows a variety of closed-loop configurations and greatly reduces nonlinearity due to the input amplifier, a major source of distortion in early designs.

The effectiveness of the new scheme can be judged from the fact that under typical multiplier conditions, when X is full scale (±10v), the nonlinearity of the Y input is ±0.005% of FS. Even at its worst point (which occurs when X=±6.4v), the nonlinearity is typically only ±0.05% of FS. On the other hand, the nonlinearity of the signal applied to the X input is almost entirely determined by the multiplier elements and is parabolic. This error is the main factor in determining the overall accuracy of the device and is therefore closely related to the device class.

The generalized transfer function of the AD534 is given by:

Where: A is the open-loop gain of the output amplifier, usually 70 dB DC. X1, Y1, Z1, X2, Y2, and Z2 are the input voltages (full scale = ±SF, peak = ±1.25 square feet). SF is the scale factor, pre-set to 10.00 V, but can be stepped down to 3 volts by the user voltage.

In most cases, the open loop gain can be considered as infinite and the SF is 10v. Then, the operation performed by the AD534 can be described by the following formula:

The user can adjust the value of SF between 10V and 3V by connecting an external resistor in series with the potentiometer between SF and -V.

200f-e95b-4363-bf47-ed1e90b0b8ad.png" title="1.3.png" />

Due to equipment tolerances, a potentiometer should be used to vary R by ±25%. Bias current, noise, and drift can be significantly reduced by reducing SF. This has the overall effect of increasing signal gain without increasing noise. Note that the peak input signal is always limited to 1.25 SF (ie, SF = 4 V ± 5 V), so the overall transfer function shows a maximum gain of 1.25. However, since the dynamic range of the input is now fully utilized, using a lower scale factor can improve performance when the input signal is small. Using this option will not affect bandwidth.

Typically, the supply voltage is assumed to be ±15 V. However, satisfactory operation can be reduced to ±8V (see Figure 7). Since all inputs are kept constant at a peak input capability of ±1.25 SF, some feedback attenuation is required to achieve output voltage fluctuations in excess of ±12 V when higher supply voltages are used.

Provides low noise gain

The AD534 was the first general-purpose multiplier capable of providing gains of up to ×100, often without the need for a separate instrumentation amplifier to preprocess the input. The AD534 can be used very effectively as a variable gain differential input amplifier with high common-mode rejection. The gain option is available for all modes and simplifies the implementation of many function fitting algorithms, such as those used to generate sine and tangent. The AD534's inherently low noise: 90 µV rms (depending on gain), which is 10 times lower than previous monolithic multipliers, enhances the usefulness of this feature. Drift and feedthrough are also greatly reduced compared to earlier designs.

Operation as a multiplier

Figure 15 shows the basic connection for multiplication. Note that the circuit meets all specifications without trimming.

To reduce AC feedthrough to a minimum (as in a suppressed carrier modulator), apply an external trim voltage (the desired ±30 mV range) to the X or Y input (see Figure 3). Figure 10 shows a typical AC feedthrough in this regulation mode. Note that the Y input is 10 times lower than the X input and should be used in applications where zero suppression is important.

The high-impedance Z terminal of the AD534 can be used to sum additional signals into the output. In this mode, the output amplifier behaves as a voltage follower with a small-signal bandwidth of 1 MHz and a slew rate of 20 V/µs. This terminal should always be referenced to the ground point of the drive system, especially in remote situations. Likewise, the differential inputs should be referenced to their respective ground potentials to achieve the full accuracy of the AD534.

As shown in Figure 16, using a feedback attenuator can achieve lower scale voltages without reducing the input signal range. In this example, the scale is such that V=(x-x)(y-y), so that the circuit can exhibit a maximum gain of 10. This connection results in a reduction in bandwidth to about 80 kHz without the peak capacitor C = 200 pF. Additionally, the output bias voltage is increased by a factor of 10, requiring external adjustment in some applications. Adjust by connecting a 4.7 MΩ resistor between Z and a potentiometer slider connected to the power supply to provide a trim range of ± 300 mV at the output.

Feedback attenuation also preserves the ability to add a signal to the output. Signals can be applied to the high impedance Z terminals, where they are amplified by +10, or to the common ground connection, where they are amplified by +1. The input signal can also be applied to the lower end of the 10 kΩ resistor, resulting in a gain of -9. Other values of the feedback ratio, up to ×100, can be used for a combination of multiplication and gain.

Occasionally, it may be necessary to convert the output to a current into a load with no specified impedance or DC level. For example, the function of multiplication is sometimes followed by integration; if the output is in the form of a current, a simple capacitor provides the function of integration. Figure 17 shows how this can be achieved. This method can also be applied to square, division and square root modes with appropriate terminal selection. This technique is used in the voltage-controlled low-pass filter and differential input voltage-to-frequency converter shown in the application information section.

square operation

The squaring operation is implemented in the same way as a multiplier, except that the X and Y inputs are used in parallel. Differential inputs can be used to determine output polarity (positive for X=Y and X=Y, negative if one of the inputs is opposite). The accuracy in square mode is usually 2 times better than in multiply mode, and the maximum error occurs when the output value is small when the input voltage is below 1V.

If the application relies on precise operation of an input consistently less than ±3 V, it is recommended to use a reduced value of SF as described in the functional description section. Alternatively, a feedback attenuator can be used to increase the output level. This will be used in the squared difference application to compensate for the 2 loss factors involved in generating the sum term (see Figure 20).

The squared difference function is also used as the basis for the new rms-to-dc converter shown in Figure 27. The averaging filter is a true integrator, the loop seeks to have its input zero. For this to happen, (V) - (V) = 0v (for signals with periods well below the average time constant). Therefore, V must be equal to the rms value of V. The absolute accuracy of this technique is very high; at intermediate frequencies and near full-scale signals, it is almost entirely determined by the ratio of the resistors in the inverting amplifier. The multiplier scaling voltage affects only the open loop gain. The data shown is typical performance that can be achieved with the AD534K, but even with the AD534J, this technique can provide better than 1% accuracy over a wide frequency range, even with crest factors in excess of 10.

Actions as delimiters

Figure 18 shows the connections required for division. Unlike earlier products, the AD534 provides differential operation on both the numerator and denominator, allowing the generation of ratios of two floating variables. Further flexibility is due to the result of summing the high impedance input to Y. As with all dividers that use a multiplier in the feedback loop, the bandwidth is proportional to the size of the denominator, as shown in Figure 14.

The AD534K and AD534L are accurate enough to maintain 1% error over the 10 V to 1 V denominator range without additional trimming. This range can be extended to 100-1, reducing the X offset by simply applying an external offset voltage (the desired range is ±3.5 mV max) to the unused X input (see Figure 3). To trim, apply a +100 mV to +V ramp at 100 Hz on X and Z (if X is used for offset adjustment; otherwise, reverse signal polarity) and adjust trim voltage to minimize output change Since the output voltage is close to 10 volts, it should be AC coupled in this adjustment. The increase in noise level and the reduction in bandwidth enable operation at a ratio well beyond 100:1.

As with the multiplier connection, the overall gain can be introduced by inserting a simple attenuator between the output and Y. This option and the differential ratio function of the AD534 are used in the percentage computer application shown in Figure 24. This configuration produces an output proportional to the percent deviation of one variable (A) relative to a reference variable (B), scaled at 1% per volt.

square root operation

The operation of the AD534 in square root mode is shown in Figure 19. The diode prevents a latch-up state, which can happen if the input changes polarity momentarily. As shown, the output is always positive; it can be changed to a negative output by reversing the diode direction and swapping the X input. Since the signal inputs are differential, all combinations of input and output polarities are possible, but operation is limited to one quadrant associated with each input combination.

In contrast to earlier devices, the AD534 cannot tolerate capacitive loads in square root mode and is stable to at least 1000 pF with all loads. For critical applications, small adjustments to the Z input offset (see Figure 3) can improve accuracy for inputs below 1v.

unprecedented flexibility

Precise calibration and differential Z inputs provide flexibility found in other currently available multipliers. Standard multiply, divide, square, and square root (MDSSR) functions are easy to implement, while removing earlier design restrictions on specific input/output polarities. Signals can be summarized into outputs, with or without gain, in either positive or negative sense. Many new models based on implicit function synthesis are possible, often requiring only external passive components. If desired, the output can be in the form of a current for ease of integration, etc.

application information

The versatility of the AD534 allows creative designers to implement circuits such as wattmeters, frequency multipliers, and automatic gain controls.

Dimensions