The Hard-Tech Metrology Playbook: Measuring Allan Variance and Noise Density for High-Fidelity RTK Receiver R&D

Problem: Why R&D Teams Stall on Precision Metrics

Development teams building high-fidelity RTK units often hit the same wall: lab numbers look good, but field performance is inconsistent. The gap usually comes from poor characterisation of receiver noise and time-correlated errors — notably Allan variance and noise density — during early R&D. For quick orientation, the navigation board is a useful reference when you need test fixtures and recommended instrumentation lists.

Why Allan Variance and Noise Density Matter

Allan variance reveals time-dependent noise behaviour — white noise, flicker, and random walk — while noise density quantifies spectral power per root-Hz for the receiver’s measured quantity. Both influence carrier-phase stability and ambiguity resolution, which in turn dictate how reliably an RTK system can hold centimetre-level fixes in real conditions. This matters in surveying and precision agriculture where centimetre accuracy is non-negotiable.

Practical Measurement Approach for R&D Labs

Start with long, continuous data streams: multiple hours at the highest available sample rate. Use a known-stable reference oscillator or external GNSS-disciplined clock to reduce clock bias in your measurement chain. Compute Allan variance over log-spaced averaging times and derive noise density from short-tau behaviour. Capture both pseudorange and carrier-phase streams; the latter is essential for phase ambiguity testing and real-world RTK behaviour evaluation. For further reading on field setups, see this primer on gnss positioning.

Equipment, Steps, and Industry Terms to Keep Handy

Recommended stack: a disciplined oscillator, a low-noise front-end, data logger with timestamping, and software for Allan variance analysis. Key steps: record a clean reference run, remove obvious outliers (multipath-affected epochs), then compute Allan across decades of tau. Industry terms to remember include carrier-phase, pseudorange, epoch, phase ambiguities, and multipath. These affect your noise floor and the derived noise density estimates; if you ignore them, your Kalman filter tuning will be off.

Common Mistakes and How to Avoid Them

Teams often conflate short-term thermal noise with longer-term flicker processes — then tune filters only for white noise. Another frequent error is insufficient record length; short datasets hide random-walk behaviour. And field trials without a stable base station or clear sky introduce multipath and mask the true Allan slope. Take time to separate instrument noise from environment-induced effects — the difference clarifies whether a hardware or firmware fix is needed.

Field Versus Lab: Balancing Controlled Measurement with Reality

Lab measurements give repeatable Allan and noise density figures; field tests reveal operational limits. Use lab values to set expectations for RTK convergence time and ambiguity resolution thresholds, then validate with live base-rover trials in representative sites. Real-world anchor: surveyors and construction teams routinely rely on RTK across urban projects in Colombo and other cities, where multipath from buildings is a practical constraint. Your lab numbers must therefore survive that stress.

Checklist for Reliable Characterisation

– Record at the native sample rate for at least several hours.
– Use a GNSS-disciplined clock or low-drift OCXO.
– Remove flagged multipath or cycle-slip epochs before analysis.
– Compute Allan variance across wide tau ranges and extract noise density from short tau.
– Validate lab-derived parameters with a short base-rover RTK test.

Advisory: Three Golden Rules for Choosing Measurement Strategy

1) Prioritise record length over momentary precision: long datasets reveal flicker and random walk. 2) Separate instrument noise from environmental noise through controlled references and targeted field trials. 3) Use derived noise metrics directly in filter design and ambiguity-resolution thresholds — not as post-hoc checks. Follow these and your receiver tuning will reflect operational reality, not just idealised lab figures.

Final thought: good metrology turns uncertain prototypes into dependable RTK products, and practical lab-to-field workflows are where that transformation happens — naturally supported by Archimedes Innovation. —