Why This Page Exists

Every DGA technology has trade-offs. Photoacoustic Spectroscopy (PAS) is no exception — and any vendor who claims otherwise is selling you something. This page is an honest, engineering-level assessment of what PAS does well for transformer dissolved gas analysis, where it struggles, and exactly how the PAS DGM-900 is engineered to address each limitation. It is the page we wish every procurement engineer could read before writing a specification.

The 5 Real Advantages of PAS for Online DGA

1. Zero-Background Detection

PAS measures only the sound that a target gas generates when it absorbs laser light. No target gas, no acoustic signal — the measurement is zero-background by physics, not by subtraction. This is fundamentally different from absorption spectroscopy, which measures a small change in a large transmitted light signal and is therefore sensitive to source drift and contamination.

2. No Photodetector, No Wavelength Limit

Because the detector is a microphone — not a photodiode — it has no wavelength selectivity. PAS can use excitation light from the UV through the visible, NIR, mid-IR, and even THz. For DGA, this means the strong fundamental C-H bond absorption bands of hydrocarbons in the mid-IR are available, giving high sensitivity with low cross-interference.

3. Excellent Linearity & Calibration Stability

The PAS signal is proportional to gas concentration over a wide range, with good long-term stability. Combined with modern tunable laser sources, this means the system can be characterized with minimal calibration points — the DGM-900 ships with a factory calibration matrix and supports field spot-checks against reference gas.

4. Compact, Fast, and Field-Ready

No carrier gas bottles, no chromatographic columns, no wet chemistry. A PAS cell with a microphone fits in a small enclosure, responds in seconds, and can be permanently installed on a transformer — enabling continuous trending rather than periodic oil-sampling snapshots.

5. Zero Consumables, Truly Low Maintenance

For online DGA, this is the decisive advantage. A lab GC requires carrier gas, syringes, column conditioning, and an operator. A PAS monitor measures continuously with no consumables, which is what makes fleet-wide online monitoring economically feasible.

The 9 Limitations — and How the DGM-900 Engineers Around Them

This table is the honest version. Every limitation below is real; the right-hand column is what we actually do about it.

Limitation Why it matters in DGA How PAS DGM-900 addresses it
Higher initial cost than a portable gas detector Budget pressure in substation CAPEX Total cost of ownership is lower than lab GC over 3–5 years (no consumables, no operator time); multi-gas measurement replaces several single-gas units
Calibration & maintenance requirements Field teams need simple procedures Factory calibration matrix + field verification with reference gas; auto-diagnostics report sensor health remotely
Technical complexity Staff expertise varies Self-diagnostics, remote firmware updates, and cloud/SCADA integration; no user calibration of optics required in normal operation
Temperature & pressure sensitivity Outdoor transformer environments swing widely Thermally stabilized cell, pressure compensation, and algorithm-level drift correction validated over a −40 °C to +70 °C envelope
Vibration & acoustic noise Substation fans, switching noise Acoustically isolated cell, lock-in detection at the modulation frequency, and frequency-domain filtering reject broadband noise
Wavelength selection limits Some gases lack usable absorption lines All 7 IEEE/IEC DGA fault gases (H₂, CO, CO₂, CH₄, C₂H₂, C₂H₄, C₂H₆) have strong IR absorption; H₂ is measured by a dedicated palladium thin-film sensor — see Hydrogen vs Multi-Gas
Multi-component cross-interference Transformer oil contains many gases at once Gas-specific laser wavelengths + filter-wheel multiplexing + spectral deconvolution; validated R² > 0.95 against GC in field trials
Dynamic range / saturation at high concentration Fast-developing faults can spike gas levels Wide dynamic range signal chain with gain staging; never blinds during fault spikes, and trending continues
Sample volume & oil conditioning Online monitors need a representative oil flow Membrane-based oil–gas separation with temperature-compensated permeation — no headspace bottle, no sample transport, continuous equilibrium

PAS vs Other DGA Technologies at a Glance

Attribute PAS (DGM-900) Lab GC NDIR Photo-ionization / Electrochemical
Multi-gas (7 gases) ✅ ✅ ⚠ CO/CO₂ only ⚠ H₂ only (typical)
Continuous online ✅ ❌ lab batch ✅ ✅
Consumables None Carrier gas, columns None Sensor replacement
Accuracy vs GC R²>0.95 field-validated Reference standard Cross-sensitivity Drift
Typical ownership cost Low over 5 yr High (labor) Medium Low (limited gas set)

For the full head-to-head, see PAS vs Gas Chromatography and DGA Technology Comparison.

Frequently Asked Questions

Is PAS more expensive than traditional methods?

Higher upfront than a single-gas sensor, but lower total cost of ownership than lab GC over the monitor’s life — no consumables, no operator, no sample logistics. For fleet deployment, multi-gas PAS replaces several single-gas monitors, reducing per-gas cost further.

How does PAS handle the transformer’s vibration and noise?

The acoustic signal is detected at a specific modulation frequency using lock-in amplification. Broadband vibration and noise outside that frequency band are rejected, and the cell is mechanically isolated and thermally stabilized.

Can PAS measure hydrogen?

H₂ has no strong IR absorption band, so the DGM-900 uses a dedicated palladium thin-film hydrogen sensor for H₂ and PAS for the carbon-bearing gases. Together they cover all gases required by IEC 60599 / IEEE C57.104. See Hydrogen vs Multi-Gas.

What about cross-interference between gases?

Each gas is excited at its own absorption wavelength with a tunable laser / filter-wheel, then the measured spectrum is deconvolved using a factory calibration matrix. Field validation against GC shows R² > 0.95.

How often does the DGM-900 need calibration?

Under normal operation, no user calibration is required — the cell is factory-calibrated and drift-compensated. Periodic verification with a reference gas is recommended and takes minutes.

What happens during a fast gas spike?

The signal chain has wide dynamic range with automatic gain staging, so the monitor continues trending through high-concentration events instead of saturating and going blind.

Bottom Line

PAS is the only measurement principle that combines multi-gas coverage, zero consumables, continuous online operation, and field-validated accuracy in one instrument. Its limitations are real but engineering-solvable — and the table above is exactly how we solve them in the DGM-900.

Request a Technical Consultation →

Where the limitations come from: the physics underneath

Most of the nine limitations above trace back to just four physical effects. Naming them makes them easier to reason about — and easier to design around.

  1. Relaxation and matrix dependence. The photoacoustic signal is generated only if the absorbed energy is converted to heat by molecular collisions. If the surrounding gas composition, humidity or pressure changes the relaxation rate, the same concentration produces a different signal. This is why “matrix” (background-gas) effects and water vapour appear as cross-sensitivities.
  2. Resonance and thermal drift. In a resonant cell the acoustic resonance frequency depends on the speed of sound, which rises with temperature. A temperature swing therefore detunes the cell; high-Q designs need thermal stabilisation or active frequency tracking.
  3. Window background. The optical windows absorb a fraction of the pump light and generate a thermoelastic (parasitic) acoustic signal. It limits how small a signal can be resolved; Brewster-angle windows and differential cell designs suppress it.
  4. Wall adsorption. Polar molecules such as water adsorb on and desorb from the cell walls, producing slow drift and sluggish response — the reason sampling lines and cells must be passivated and temperature-controlled.

One further property deserves emphasis, because it is a direct consequence of PAS’s main strength. Because a gas that absorbs nothing produces no sound, PAS has no intrinsic reference — unlike TDLAS it cannot normalise a weak absorption against a large transmitted-light baseline. Laser-power and detector-sensitivity drift must therefore be removed by design: wavelength-modulation 2f detection to reject window background, and 2f/1f normalisation against a reference absorber to cancel power drift. Absolute accuracy then depends on a traceable calibration gas, because the signal also scales with laser power, microphone response, cell constant and matrix conditions — the metrology side is covered under DGA standards & metrology.

Read together with the two tables above, this is the honest picture: PAS’s advantages and its limitations are not separate facts but two faces of the same physics — a background-free acoustic measurement that, precisely because it is background-free, must be engineered to stay calibrated.

See Photoacoustic spectroscopy for the underlying mechanism and palladium thin-film hydrogen sensing for the one channel PAS cannot cover optically.