The choice between a one phase vs 2 phase lock in amplifier boils down entirely to phase stability in your signal path. Choose a 2-phase (dual-phase) LIA for 95% of modern research applications—including impedance spectroscopy, optical pump-probe setups, and cryogenic material sweeps—because it calculates signal magnitude independent of phase drift. Choose a 1-phase (single-phase) LIA only when your reference and signal paths are physically fixed, you are operating on a strict sub-$5,000 budget, or you specifically need to isolate the pure in-phase baseband component without quadrature mixing.

Lock-in amplifiers (LIAs) extract microvolt or nanoamp signals buried under 100 dB of noise using phase-sensitive detection (homodyne detection). But whether you need one or two internal detection channels dictates your measurement accuracy, your hardware cost, and whether your data survives a temperature sweep. Here is the bench-level breakdown of how these architectures differ and where they fail.

The Single Physical Difference Driving the Divide

The entire divergence between single-phase and dual-phase architectures stems from one hardware addition: a second phase-sensitive detector (PSD) operating at a 90° phase shift (quadrature) relative to the first.

In a 1-phase LIA, the incoming signal is multiplied by a reference sine wave and passed through a low-pass filter. This yields the in-phase component, X. The math is straightforward: X = Vsig cos(θ), where θ is the phase difference between your signal and the reference. If θ is perfectly zero, X equals your true signal amplitude.

In a 2-phase LIA, the hardware duplicates this process. A second multiplier feeds the signal against a reference wave shifted by exactly 90°. This yields the quadrature component, Y = Vsig sin(θ). The instrument’s internal DSP then calculates the true magnitude R = √(X² + Y²) and the phase angle θ = arctan(Y/X).

Why does this matter? Because cos(θ) is highly sensitive to drift. If your phase shifts by just 10° due to thermal expansion in a cable, a 1-phase LIA reading will drop by 1.5%. A 30° shift causes a 13.4% error. A 2-phase LIA ignores this entirely; because cos²(θ) + sin²(θ) = 1, the calculated magnitude R remains perfectly stable regardless of phase drift.

Hardware Specifications and Bench Realities

Modern LIAs achieve this using high-speed ADCs and FPGA-based digital multipliers rather than analog analog-multiplier ICs. Below is a data-dense comparison of standard benchtop architectures, highlighting the real-world trade-offs between 1-phase DSP, standard 2-phase DSP, and high-frequency RF implementations.

Table 1: LIA Architecture and Specification Comparison
Specification 1-Phase DSP (e.g., SRS SR810) 2-Phase DSP (e.g., SRS SR830) 2-Phase RF (e.g., Zurich HF2LI)
PSD Architecture Single Digital Multiplier Dual Simultaneous Multipliers Dual High-Speed FPGA Multipliers
Frequency Range 1 mHz to 102 kHz 1 mHz to 102 kHz DC to 50 MHz
Dynamic Reserve 100 dB 100 dB 120 dB
Phase Resolution 0.01° (Manual tuning required) 0.001° (Auto-calculated) 0.0001° (Auto-calculated)
Baseband Outputs X only X, Y, R, θ X, Y, R, θ (Demodulator 1 & 2)
Approx. Price (2026) ~$4,500 ~$6,800 ~$16,000+

As seen in the Stanford Research SR830 specifications, the jump from 1-phase to 2-phase in the sub-100 kHz regime adds roughly $2,300 to the bill of materials. This cost covers the secondary ADC channel, the expanded FPGA logic blocks, and the more complex analog front-end routing required to maintain phase orthogonality.

Where the Two Architectures Are NOT Interchangeable

If you are simply chopping a laser beam with a mechanical chopper and feeding the chopper's TTL output directly into the LIA's reference input via a 1-foot BNC cable, a 1-phase LIA will work fine. The phase relationship is fixed. However, there are three specific scenarios where a 1-phase LIA will physically fail to deliver accurate data, making 2-phase mandatory.

1. Complex Impedance Spectroscopy

When measuring a component's impedance (Z = R + jX), you must separate the real (resistive) and imaginary (reactive/capacitive) parts. The resistive part maps to the in-phase channel (X), while the reactance maps to the quadrature channel (Y). A 1-phase LIA cannot measure complex impedance without mechanically sweeping the reference phase and taking sequential readings—a process far too slow for frequency sweeps and entirely useless for dynamic electrochemical cells.

2. Cryogenic Temperature Sweeps

Coaxial cables use PTFE (Teflon) dielectrics. The dielectric constant of PTFE changes significantly as it cools from 300 K down to 4 K. This alters the propagation velocity of the signal, introducing continuous phase drift. Standard RG-316 cable introduces roughly 1 ns of delay per foot. If your signal cable and reference cable experience different thermal gradients inside a cryostat, the phase θ will drift wildly. A 1-phase LIA will record this phase drift as a false amplitude drop-off, ruining your material characterization data. A 2-phase LIA's R output ignores the drift entirely.

3. High-Frequency RF Measurements

At 10 MHz, a mere 1-foot difference in cable length between your signal and reference paths results in a 12° phase shift. At 50 MHz, that same 1-foot mismatch shifts the phase by 60°, which would attenuate a 1-phase X reading by 50%. While Zurich Instruments notes in their lock-in detection principles that modern RF LIAs handle this via digital delay lines, you still fundamentally need the dual-channel architecture to resolve the vector accurately at high frequencies.

Cost, DSP Overhead, and Availability

Given the massive advantages of 2-phase detection, why do manufacturers still build and sell 1-phase units like the SRS SR810? The answer lies in DSP overhead and specific demodulation needs.

Digitizing a signal at 16-bit or 18-bit resolution at high sampling rates requires expensive, low-noise ADCs. Running two of these ADCs in perfect synchronization, alongside the FPGA fabric required to execute two simultaneous multiply-accumulate (MAC) operations and a CORDIC algorithm for the arctangent calculation, increases the PCB complexity and thermal noise floor. For educational labs, basic optical chopping experiments, or simple AM-radio baseband demodulation where phase is irrelevant, the 1-phase architecture provides the necessary signal-to-noise ratio at a 30% discount.

Furthermore, in some ultra-high-frequency analog lock-ins (above 100 MHz), generating a precise 90° analog phase shift across a wide bandwidth is notoriously difficult due to the limitations of analog Hilbert transformers or RC-CR networks. In those niche RF domains, engineers sometimes revert to single-phase detection or use sequential dual-phase (measuring X, then physically shifting the reference 90° and measuring Y in a second sweep).

Decision Matrix: Choose A When / Choose B When

Use this framework to specify the right instrument for your next grant proposal or bench setup.

  • Choose 1-phase when:
    • You are performing basic optical absorption/emission measurements with a mechanical chopper and short, matched BNC cables.
    • You are strictly demodulating an AM envelope where the absolute phase angle of the carrier is irrelevant.
    • Your budget is hard-capped below $5,000 and you do not need to measure complex impedance.
    • You are building a custom FPGA-based LIA and need to minimize logic block usage and ADC channel count.
  • Choose 2-phase when:
    • You are performing electrochemical impedance spectroscopy (EIS) or measuring capacitor ESR.
    • Your signal path involves long cable runs, optical fibers, or cryogenic environments where thermal phase drift is guaranteed.
    • You need to track the physical phase shift of a phenomenon (e.g., measuring the phase lag in a PLL or a mechanical resonator).
    • You are using an auto-phase routine at the start of an experiment and want the data to remain accurate even if the system drifts over a 12-hour sweep.

Ultimately, the 2-phase lock-in amplifier is the default standard for modern precision measurement. The ability to decouple amplitude from phase drift saves hours of manual recalibration and prevents catastrophic data loss during long-term environmental sweeps. Reserve the 1-phase architecture for fixed, low-frequency, budget-constrained educational setups.