2026-05-30
The translinear principle, formalized by Barrie Gilbert in 1975, is one of analog design's most elegant tricks. It exploits the exponential I-V relationship of bipolar junction transistors to perform multiplication, division, and square roots — operations that would otherwise require complex op-amp networks — using nothing but matched transistors and bias currents.
The foundation is the Ebers-Moll equation: a BJT's collector current is IC = IS · exp(VBE/VT), where VT ≈ 26 mV at room temperature. Rearranging: VBE = VT · ln(IC/IS). The base-emitter voltage is logarithmic in collector current.
The translinear loop principle: in any closed loop containing an equal number of clockwise and counterclockwise base-emitter junctions, the product of clockwise collector currents equals the product of counterclockwise collector currents. The VT · ln() terms sum around the loop, and exponentiating both sides converts sums of logs into products.
Example — Geometric Mean Circuit: Build a loop with four matched NPN transistors Q1–Q4 where Q1 and Q2 face one direction and Q3 and Q4 face the other. If I1 flows through Q1, I2 through Q2, and an output current IOUT flows through both Q3 and Q4, the loop equation gives:
I1 · I2 = IOUT2, therefore IOUT = √(I1 · I2)
A pure square-root operation, performed by physics, not math. No op-amps, no resistors in the signal path.
Real-world application: The AD834 four-quadrant multiplier (used in RF mixers, modulators, and RMS detectors up to 500 MHz) uses a translinear core. So does the AD538 — a "computational amplifier" that performs VOUT = VY · (VZ/VX)m for arbitrary exponent m, used in linearizing thermocouples and computing true-power readings.
Critical design constraints:
Rule of thumb: If your analog computation can be expressed as a product or ratio of positive quantities, a translinear loop will likely beat any op-amp-based solution in bandwidth, dynamic range, and parts count — provided you have matched transistors on a shared substrate.
