AD633JRZ
Product Specifications
AD633JRZ
Analog Devices — Analog & Mixed-Signal ICs
Product Overview
The AD633JRZ from Analog Devices is a high-performance analog / mixed-signal IC providing precision signal acquisition, conversion, and conditioning. Analog Devices (ADI) is a world leader in high-performance analog, mixed-signal, and DSP ICs for instrumentation, healthcare, automotive, and communications.
Superb Automation partners with leading brands including Qorvo, TI, ADI, XILINX, NVIDIA, Intel, ST, MACOM, Renesas, and Coilcraft to provide complete BOM and supply-chain solutions.
Key Features
- High-resolution data conversion with excellent SNR and THD
- Wide dynamic range and high linearity across operating conditions
- Integrated voltage reference with low temperature drift
- Flexible digital interface — SPI, I2C, or parallel options
- Industrial/automotive temperature range with guaranteed specs
- Low power consumption with power-down modes
Key Specifications
| Resolution | bits |
| Channels | — |
| Sample Rate | kSPS |
| Interface | — |
| Supply Voltage | V |
| Operating Temp | °C |
| Package | — |
| INL | LSB |
| SNR | dB |
| THD | dB |
Applications
Request Quote — In Stock, 100% New & Original, Full Warranty
AD633JRZ: The Four-Quadrant Multiplier as One 8-Lead Block
The AD633JRZ is the J grade commercial part of the AD633 four-quadrant analog multiplier in an 8-lead SOIC. Inside the package sit the differential X and Y input stages, a unity-gain output buffer and a buried Zener that sets the 10 V full-scale reference, laser trimmed so total error stays inside 2 percent of full scale. Where a board needs one multiplier, and only one, that removes the matching and trimming that a discrete arrangement of transistors and op amps demands, and it removes the temperature tracking problem along with it. The first-pass numbers are 1 MHz of bandwidth, 20 V per microsecond of slew and 10 megohms on both X and Y, so the part can sit directly on a high impedance node without a buffer ahead of it.
The Z pin is the part most people forget
Z is a high impedance summing input into the output buffer, not a spare pin. It lets two multipliers add without an external op amp, sets the multiplier gain by fed back division, converts the voltage output into a current through an external resistor, and gives a place to inject an offset. A squaring circuit uses the same part with X and Y tied together and the output scaled by the 10 V reference, so the output reads X squared divided by 10 V. A divider puts the multiplier inside the feedback loop of an op amp, and the loop then solves for the reciprocal at the output, which is why the divider connection is stable only over the range where the loop gain stays above one. A modulator feeds carrier to one input and signal to the other, and the Z input is where the carrier leakage gets nulled if the null is worth the trim.
Values a buyer or a test engineer will ask for
| Parameter | Value | Note |
|---|---|---|
| Multiplication | Four quadrant, W = (X1 - X2)(Y1 - Y2)/10 V + Z | Transfer function fixed by the internal reference |
| Full-scale output | 10 V nominal | Set by a buried Zener, supply insensitive |
| Total error | 2 percent of full scale maximum | Laser trimmed, no external trim required |
| Y nonlinearity | Typically under 0.1 percent | X input similar, both measured at 10 V |
| Bandwidth | 1 MHz | Output drives capacitive loads without a series resistor |
| Slew rate | 20 V per microsecond | Full 20 V step in about one microsecond |
| Input resistance | 10 megohms differential | X and Y, so source loading is negligible |
| Noise | Under 100 microvolts rms | 10 Hz to 10 kHz, referred to the output |
| Supplies | Plus or minus 8 V to plus or minus 18 V | Quiescent dissipation around 90 mW |
| Temperature | 0 C to 70 C for the J grade | The A grade covers minus 40 C to plus 85 C |
Where the 2 percent goes
Two percent of a 10 V range is 200 mV, and at 12 bits one code of a 10 V converter is 2.44 mV, so the untrimmed error is worth roughly 82 codes. That number decides whether the part can be used open loop or needs a calibration. In a gain control loop, a 2 percent scale error simply shifts the set point and the loop absorbs it. In a power measurement, where the product of voltage and current is read directly by a converter, the same 2 percent lands on the reading, and the choice is either a per unit calibration constant in firmware or the A grade plus an external trim at the Z pin. The temperature coefficient is the part that hurts a calibration, since a trim done at 25 C drifts with the board, and the datasheet specifies the multiplier over the full grade range rather than at room only.
Modulation, phase detection and the parts around it
Phase detection is the application where the multiplier earns its place, because a low pass filter after the multiplier recovers the cosine of the phase difference between two signals, and the null at 90 degrees is far sharper than a diode arrangement can produce. The same arithmetic gives a synchronous demodulator that recovers a small signal buried in noise, since the noise at frequencies away from the carrier averages to zero. Both cases want the DC accuracy of the following amplifier to be better than the multiplier, or the multiplier accuracy is wasted, and the amplifier offset appears at the output multiplied by the carrier amplitude. Power supply decoupling at the supply pins and a clean reference matter more than the package, because the internal scaling voltage is the reference that every output is measured against.
Procurement and second source
The AD633 has been in production since the 1990s, so most of the supply is broker stock with mixed date codes rather than distributor stock. The part is available through the RF component sourcing channel, the gain stages that follow it are covered on the amplifier cross reference page, and incoming checks on date code, package marking and seal are described in the incoming inspection plan. Second sourcing a multiplier is rarely clean: the transfer function, the 10 V scale factor and the input topology all have to match, and a part from another family that multiplies correctly will still need the firmware constants rewritten.