Flagship: Noise-Stationary Locking
Quantum-grade measurement stability for interferometric optical sensors — the first and most developed module, detailed throughout this site.
Advanced Supervisory Modules · measurement-aware interferometry
Lock the error bar, not just the fringe.
Conventional locks keep an instrument optically stable. PQSensing builds Advanced Supervisory Modules — a firmware performance layer that asks a harder question: is the instrument parked where it produces the most usable measurement information? Our flagship module brings this to quantum-grade optical sensing, and the family extends across inertial, timing, and quantum-communications instruments.
Supervisory performance layer.
Flagship: Noise-Stationary Locking.
Runs on standard lab instruments.
The platform
One supervisory idea, a family of measurement modules.
Across precision instruments — interferometric optical sensors, inertial and fiber-optic gyros, frequency references and optical clocks, quantum links — the limiting factor is rarely whether the device is locked. It is whether it is held at the operating point where its numbers are actually best. PQSensing packages that discipline as a family of Advanced Supervisory Modules: a firmware layer that runs alongside your existing lock and keeps the instrument parked where the measurement is cleanest.
An extensible family
Further modules target inertial sensing, optical clocks and frequency references, and quantum communications — each tuned to the figure of merit that matters for that instrument class.
Runs on instruments you have
Delivered as firmware for standard FPGA-based lab instruments, including Liquid Instruments Moku and Red Pitaya — deployed alongside the locks already on the box.
Seeking collaboration
Help validate a new noise-reduction layer for interferometers.
PQSensing is looking for a university lab, metrology partner, instrument company, or technical sponsor to help test Noise-Stationary Locking on a classical tabletop interferometer.
The first experiment is deliberately modest: compare a conventional PDH, power, or PID-style lock against a normalized noise-objective lock under controlled vibration, thermal drift, and phase/path-length disturbance.
What we need: access to an optical bench or Mach-Zehnder/cavity test setup, photodetector readout, a PZT or phase actuator, DAQ/controller support, and technical guidance on a defensible validation protocol.
What partners get: a focused applied optics/control project, possible student-useful data, early visibility into Kansas-origin photonics IP, and a clear sponsored-research or prototype path if the traces are interesting.
Why this exists
A locked interferometer is not always an optimized interferometer.
Most precision interferometers are locked to an optical condition: a cavity resonance, a fringe point, a reflected-power null, or a transmitted-power peak. That is necessary, but it is not always enough.
The customer does not buy a pretty fringe. The customer buys a smaller error bar: displacement, phase, strain, rotation, acceleration, refractive index, or timing measured with less uncertainty and less wasted integration time.
NSL + DPI is for instruments that are optically locked but not necessarily measurement-optimized. The optical lock says stable. The noise floor says otherwise.
The buyer pain
“My interferometer is locked, but my measurement is still noisy.”
That is the use case. The lock is holding the optics, but vibration, thermal drift, laser-frequency noise, platform motion, speckle, detuning jitter, or quadrature-angle error still corrupt the measurement band.
More averaging stops helping.
Integration time should shrink the error bar. When the operating point drifts or the noise becomes nonstationary, longer averaging can collect drift instead of information.
Quantum advantage fades in the field.
Squeezed-light benefit depends on alignment, loss, and phase stability. A squeezed source is not enough if the readout drifts away from the low-variance quadrature.
The dark port is underused.
A nominally dark port can be more than an empty input. DPI turns it into a controlled channel for calibration, coding, ranging, interference rejection, and squeezed-light injection.
What we are doing
We add a measurement-performance layer to interferometric sensors.
NSL + DPI is intended to augment established optical locking methods such as PDH, not pick a needless fight with them. PDH keeps the cavity or interferometer captured. NSL trims the operating point toward the quietest useful measurement condition. DPI gives the dark port a controlled job.
Keep the optical lock.
Use PDH, bias locking, or another conventional method to maintain optical capture and keep the instrument in range.
Add Noise-Stationary Locking.
Compute normalized measurement-band noise and servo the system toward a point where small detuning errors no longer cause first-order noise degradation.
Activate the dark port.
Inject a controlled auxiliary field, phase code, pilot signal, or squeezed vacuum through the dark mode and recover useful signatures at the output.
Interactive demonstration
Watch the two locks disagree.
A power lock parks at the fringe extremum. A noise-stationary lock parks where the slope of Vnorm(δ) = Nband(δ) / P(δ)k goes to zero. Whenever real-world noise has structure that isn't perfectly aligned with the fringe — which it rarely does — those two operating points are not the same. Drag the controls.
Power lock parks at
δ = 0.00
Vnorm there = —
NSL parks at
δ = 0.00
Vnorm there = —
NSL advantage
— dB
in this regime
P(δ) — optical transmission
Nband(δ) — measurement-band noise
Vnorm(δ) — the NSL objective
Pedagogical model, not bench data. Analytic curves illustrate the slope-null thesis. The validation sprint replaces these with measured curves from a tabletop Mach–Zehnder.
Why optimize variance?
Because the sensor’s product is an estimate with an uncertainty.
For a photon-limited phase measurement, the useful relationship is:
(Δφ)2 ≈ Vnorm / ( ṄT )
Lower and more stable Vnorm means the same photon flux and integration time produce a smaller phase-error variance. That can mean better precision in the same time, the same precision faster, or the same result with less optical power.
How we intend to do it
Measure the noise that matters, normalize it, find its slope, and lock there.
Acquire
Read the detector stream y(t) from a photodiode, balanced detector, homodyne receiver, or related interferometric readout.
Compute
Estimate band-limited noise Nband, compute an optical power metric P, and form the normalized objective.
Dither
Apply a small operating-point dither to δ outside the protected measurement band to probe the objective’s slope.
Demodulate
Synchronously demodulate Vnorm against the dither reference to estimate dVnorm/dδ.
Servo
Drive the actuator toward dVnorm/dδ ≈ 0 while preserving a power floor and using mislock recovery logic.
Integrate
Hold the quietest useful operating point so photon integration continues to reduce the measurement error bar.
Controller objective
Vnorm(δ) = Nband(δ) / P(δ)k + power-floor penalty
Slope-null condition
dVnorm/dδ → 0
Phase-Controlled Dark-Mode Injection
DPI gives the dark port a job.
In a conventional interferometer, the dark input is often treated as unused or as a passive vacuum input. DPI uses that port intentionally. A known phase waveform g(t), pilot signal, code, chirp, coherent auxiliary field, or squeezed vacuum can be injected and recovered through synchronous detection, correlation, matched filtering, or homodyne readout.
- Calibration and system identification
- Code-division signatures and multiplexing
- Interference rejection and matched-filter recovery
- Ranging or time-delay signatures
- Squeezed-light injection and quadrature control
What can this help measure?
Weak phase-based signals that were too noisy, too slow, or too unstable to recover reliably.
If the physical quantity can be converted into an optical phase, path-length, quadrature, or time-delay change, NSL + DPI may help move it from “buried in noise” to “recoverable with usable integration time.”
Displacement and vibration
Mirror motion, MEMS motion, wafer-stage position, precision machine vibration, optomechanical displacement.
Strain and deformation
Fiber strain, structural deformation, composite stress, shape sensing, pressure-vessel or pipeline strain.
Acceleration, force, pressure, acoustics
Proof-mass motion, cantilever force, membrane pressure, acoustic diaphragm motion, hydrophone-style sensing.
Rotation and inertial drift
Sagnac-style rotation, fiber gyro readout, GPS-denied navigation support, platform stabilization.
Refractive-index and chemical change
Trace gas, fluid contamination, biosensing, lab-on-chip refractive-index shifts, molecular binding effects.
Quantum-enhanced weak phase signals
Sub-shot-noise phase shifts, squeezed-light readout, low-variance quadrature tracking, dark-port quantum resources.
Timing, ranging, and coded signatures
Dark-port code recovery, matched-filter timing, correlation peaks, multiplexed instrument identification.
The practical claim
Not magic. Not “measure anything.” The claim is sharper: preserve the measurement condition so small interferometric signals can be recovered sooner and with less noise-floor wandering.
Validation plan
First prove it on a classical tabletop Mach-Zehnder.
The first milestone is intentionally lean. No squeezed-light source is required to validate the core control principle. Build a classical MZI, sweep the operating point, map power and Vnorm, inject controlled detuning jitter, and compare a conventional power lock against NSL.
- Map P(δ), Nband(δ), and Vnorm(δ)
- Show whether the noise-stationary point differs from the power extremum
- Inject repeatable jitter and compare locked time traces
- Demonstrate dither/demod convergence toward dVnorm/dδ ≈ 0
- Package raw data, plots, and reproducible analysis
Prototype metrics
Evidence that makes partner conversations rational.
These are validation targets, not guaranteed product specifications. The point is to produce defensible traces that prove or kill the control thesis.
MetricTargetEvidence output
Setpoint separationVnorm stationary point differs from the power extremum in at least one injected-noise regime.δ sweep: P(δ), Nband(δ), Vnorm(δ)
Jitter-induced noise reductionFirst-demo target: ≥6 dB vs power lock; stretch target: 10–20 dB if bench stability supports it.Locked traces under controlled jitter
Slope-null convergence|dVnorm/dδ| driven toward zero while maintaining P above a useful floor.Dither/demod error and actuator command
Reacquisition≥5 successful reacquisitions after intentional mislock or disturbance trials.State log: SEARCH → CAPTURE → BIAS_LOCK → NOISE_LOCK
DPI optional testRecover known g(t) by correlation/matched filtering with a visible peak over uncorrelated disturbance.Correlation peak, lag estimate, processing-gain plot
Market wedge
Start where the lock is stable but the reading is not.
Metrology OEMs: cavities, mode cleaners, MZI/Michelson platforms, displacement and precision phase sensors.
Defense and navigation: inertial sensing, gyro readout, platform vibration, GPS-denied measurement support.
Photonics and integrated sensing: MZI/PIC control firmware, dark-port coding, correlation signatures, interference rejection.
Quantum-sensing labs and partners: squeezed-light injection, quadrature alignment, homodyne readout, practical quantum advantage.
Audio overview
Listen to the metrology thesis.
A concise spoken overview of the PQSensing approach, the tabletop validation sprint, and why noise-stationary control matters for interferometric metrology.
Immediate ask
$10K minimum useful prototype/data sprint.
The first funding unit is not a full quantum hardware build. It is a lean classical MZI validation package: optical bench or partner lab access, detector/readout path, actuator, DSP, jitter injection, and reproducible data comparing conventional locking against NSL.
Optics + mechanics: breadboard, mirrors, beamsplitters, mounts, alignment hardware.
Readout + control: photodiode or balanced detector path, DAQ/interface, PZT/phase actuator, driver.
DSP + analysis: real-time Vnorm pipeline, dither/demod loop, reproducible notebooks.
Output: raw logs, plots, updated deck, technical package for companies, SBIR/STTR, investors, and lab partners.