A readout that never goes blind

A way of reading a quantum sensor that never goes blind and needs no feedback lock, so it keeps measuring when the platform moves.

It measures at three fixed phase offsets, 0°, 120° and 240°, and combines them. It needs no feedback loop and no real-time adjustment, and it recovers the sign of the signal over the full range.

What it is

A standard readout is blind at the top and bottom of every fringe

In an interferometer the signal is a fringe, cos φ, where φ is the phase that gravity, a magnetic field or a frequency writes onto the atoms or spins. A standard readout looks at the fringe at one phase. Its sensitivity follows sin²φ, which falls to 0 twice per fringe. These blind regions are the dead zones.

The combined sensitivity is the same at every phase: exactly half of the standard readout’s peak, which is the standard readout’s average. Click a sketch to draw it again.

Why it is needed

In the field the phase cannot be held on the steep slope

The phase keeps moving

Motion, vibration, turning and changing gravity sweep the phase across the fringe, so a standard readout keeps passing through its dead zones.

Locks lose the fringe

The usual fix is a feedback lock that holds the operating point on the slope. A lock that holds on a bench can lose the fringe on a moving platform.

Adaptive methods need control

Bayesian adaptive methods also work, but they need real-time computation and control of the measurement settings.

The trade, stated plainly

It gives up the peak sensitivity of a perfectly locked readout. On a quiet bench, a locked or two-quadrature readout is as good or better.
It does not beat the standard quantum limit under uncorrelated noise. Theory says nothing can, and it does not.
For entangled (GHZ) probes it needs three probes instead of one, which is a real hardware cost.
Phase-shifting readout is a decades-old idea in classical optics (phase-shifting interferometry, 1980s). Our contribution is making it work on field quantum sensors, inside a verified firmware and calibration stack, with data to show it.

Evidence

Published paper

Phase Multiplexing for Non-Adaptive Dead-Spot Suppression in GHZ Parity Metrology

Manan Jain, 2026, Zenodo doi:10.5281/zenodo.21390434

2.3 %peak to peak: three GHZ probes at 0, 2π/3 and 4π/3 give a combined sensitivity uniform to within this, with no feedback or Bayesian update.

11.5 %of the phase range: where a single probe falls close to zero.

Limits. Under fair accounting of time and resources it does not exceed the standard quantum limit under uncorrelated dephasing, as the no-go theorems require (Huelga et al. 1997). The cost is three times the probe preparation and readout.

Sea-state gravity gradiometer

Simulation, no hardware yet

The sign of the signal was recovered. A feedback lock would see 19 rad of vibration per shot, against a limit of about 2 rad.

Limits. At the quantum limit, navigation improved by only 11–20 % (median). At quiet sites, established readouts were as good or better: two-quadrature readout came within about 10 %, and feedback tracking surveyed a line in 140 minutes against our 227.

Status

Theory and simulation are done. The first hardware test is next.

The first hardware test runs on an NV-diamond testbed. Its pass bars are fixed in advance:

the dead-zone fraction must stay below 1 %;
a single-phase readout must show its dead zones, as a control;
the two-quadrature result is reported whichever way it goes.