Access Official Technical PDF

Share Summary:LinkedInEmail

A lock-in amplifier can measure a small AC signal even when noise and interference are much larger, provided the signal’s reference frequency is known. The Ithaco Model 3921 applies this approach across a 10 Hz to 15 kHz operating range, using a phase-sensitive detector (PSD) and output filtering to separate the in-phase signal from unrelated components. Understanding how its reference, sensitivity, phase and time-constant controls work helps engineers configure measurements and interpret their results.

How the 3921 lock-in amplifier extracts a signal

The instrument compares the incoming signal with an external reference that is coherent with the signal of interest. Its PSD effectively multiplies the signal by a reference-derived waveform. A component at the reference frequency is converted to a DC level at the detector output, while other frequencies appear as varying components. A low-pass filter then attenuates those variations, leaving a reading associated with the signal at the reference frequency.

This method does not eliminate every source of error. The reference must remain locked, the input must be configured appropriately, and the output filter must suit the signal and measurement timing. In particular, the reference phase affects the measured amplitude: a phase mismatch reduces the reading, and a 180-degree change reverses its polarity.

Harmonic rejection without signal-channel filtering

The 3921’s PSD was designed to avoid a common drawback of conventional switching detectors: responses to odd harmonics of the reference. Rather than relying on a simple square-wave multiplication, its detector uses incremental switching to approximate sinusoidal multiplication. The manual specifies harmonic rejection typically greater than 55 dB through the 14th harmonic.

This detector design allows the instrument to operate without signal-channel predetection filters, which can otherwise require tuning and may be sensitive to reference-frequency drift. It is especially relevant when a measurement signal contains a strong fundamental component alongside harmonic content. The published rejection figure describes the detector’s typical performance; it should not be treated as a guarantee that every real-world signal or setup will produce the same result.

Set sensitivity with input level and interference in mind

The 3921’s sensitivity range switch selects an input range in a 1-10-100 sequence. A separate ten-turn MULT control adjusts post-detector gain from 1.00 to 11.00. Together, these controls determine the input level corresponding to full-scale output. With MULT at 1.00, the selected sensitivity setting corresponds to full scale; increasing MULT reduces overall sensitivity.

Setting gain is also a trade-off between reading resolution and tolerance of unwanted input. The manual identifies greater dynamic reserve at a MULT setting of 1.00, while higher settings can improve stability and measurement accuracy at the cost of interference tolerance. A practical setup process is to:

  • Begin with a conservative sensitivity setting and a low MULT value.
  • Increase sensitivity until the wanted signal produces a useful meter or output reading.
  • Watch the separate input and output overload indicators. Input overload indicates excessive drive to the AC input amplifier; output overload indicates that the post-detector output stage is saturated.
  • If input overload occurs despite a small meter reading, reduce the gain or investigate interference rather than assuming the wanted signal is too large.

The instrument provides a 10 V DC full-scale output for external measurement equipment. Its analog meter indicates output magnitude, while polarity lamps show whether the DC output is positive or negative.

Choose phase and bandwidth deliberately

Use autophase or manual adjustment

In automatic mode, the 3921 uses an electronic servo to adjust reference phase for a maximum signal output. The AUTO lamp indicates that the system has detected successful phasing; the 0°/180° control can reverse the reference when needed. Autophase can simplify setup and can continue to track phase changes while engaged. However, the lamp has a signal-to-noise threshold. If it does not illuminate, that does not by itself prove the measurement is unusable, but automatic adjustment may be slow or unreliable at low signal levels.

In manual mode, the phase control provides a continuous adjustment, with additional controls for 90-degree shift and 180-degree reversal. A useful technique described in the manual is to adjust for a minimum output, then switch the reference by 90 degrees to obtain the maximum response. This can make phase setting easier when noise or a long time constant makes it difficult to identify a maximum directly.

Balance smoothing against response time

The TIME CONSTANT control sets the output low-pass filter, with settings from 10 milliseconds to 30 seconds and a nominal 6 dB-per-octave roll-off. Longer settings reduce output fluctuations but increase settling time: the manual notes that the output takes roughly four to five time constants to settle. The optional POST FILTER adds a fixed one-second, two-section filter with a steeper 12 dB-per-octave response. It can help suppress low-frequency ripple, although additional filtering further slows the response.

Input grounding is part of measurement setup

The input can operate in FLOAT or GND mode. FLOAT provides a quasi-differential input intended to help avoid ground-loop interference when the signal source has a ground return. GND connects the input configuration for a source without a ground reference. The manual recommends coaxial cable and grounding its shield at one point to reduce unwanted loops.

When using FLOAT, observe the specified common-mode limit: the voltage between the input BNC shell and instrument cabinet must not exceed ±10 V peak. The signal center conductor has separate stated limits relative to the BNC shell. Grounding should be checked across the source, instrument chassis and power connection, particularly if changing FLOAT/GND does not resolve unexpected readings.

Practical implications for engineers

The 3921 illustrates the key decisions behind narrowband signal measurement: establish a stable reference, select sensitivity without overloading the input, align phase, and choose filtering that balances noise rejection with response time. Its harmonic-rejecting PSD reduces the need for predetection filtering, but does not remove the need to verify grounding, reference lock and overload status. For engineers and integrators, treating these controls as a coordinated measurement setup—not independent adjustments—is central to obtaining a meaningful output.

Access Official Technical PDF

About The Author