TEV vs ultrasound for switchgear partial-discharge surveys

What does a switchgear partial-discharge survey tell you?

Use transient earth voltage (TEV) and ultrasound together for switchgear partial-discharge surveys: TEV detects fast electrical pulses on grounded metal enclosures, while ultrasound detects discharge sound through an available air path or a suitable contact sensor. EA Technology's UltraTEV Plus² manual V13.05.00 uses pulse magnitude, repetition rate and pattern evidence to interpret activity; a single dB reading does not establish switchgear condition. Record background noise, sensor position, phase reference, operating state, temperature and humidity so each finding can be repeated and assigned a follow-up action. [S1]

Partial discharge is localized electrical discharge that does not completely bridge the insulation between electrodes. Detection establishes signal evidence; localization identifies its origin; diagnosis identifies the defect; the operating decision assigns the response. Keep those four conclusions separate in the report. [S1]

A satisfactory infrared inspection does not clear an insulation-discharge finding: resistive heating and PD are different condition evidence. Place the survey within the medium-voltage switchgear maintenance program, with a defined route from finding to diagnostic work.

Define the asset and operating state before choosing the sensor

Mark bus, breaker, cable and instrument-transformer compartments on photographs and the electrical one-line. Record manufacturer, model, insulation medium, nominal and available measured operating voltage, breaker position, accessible surfaces and ownership boundary. Define the surveyed substation by asset IDs, including inaccessible compartments; an open breaker alone does not establish isolation throughout the lineup. [S4, S14]

Where do TEV and ultrasound work best?

Select the sensor from the path between the suspected source and the accessible measurement point. Internal electromagnetic emissions can escape enclosure discontinuities and couple onto exterior metal; airborne ultrasound needs an air passage, while a contact probe needs effective structural coupling. This table is Anvilfield’s method synthesis. [S1, S8]

Why the two methods can disagree

A low TEV result can coexist with surface-discharge sound, while loud acoustic activity can come from vibration or leakage. EA’s UTP2 manual V13.05.00, §4.3.5, excludes contact-probe use where another layer separates the probe from the tank containing the suspected source. Mark that compartment unassessed by contact ultrasound; an outer cabinet reading cannot clear the inner tank. [S1, S4]

Identify the insulation around the component separately from the breaker interruption medium. A vacuum breaker can occupy an air-insulated lineup; its nameplate does not establish external acoustic access to the sealed interrupter. High-frequency current transformer (HFCT) and ultra-high-frequency (UHF) follow-up likewise require suitable earth-conductor or sensor arrangements, not simply an available instrument. [S1, S8]

What are the TEV partial-discharge limits?

Use the instrument's magnitude, pulse rate and pattern interpretation together with equipment condition and failure consequence. EA Technology's UltraTEV Plus² manual V13.05.00 specifies a 10 dBmV minimum pulse-count level, so a zero counter below that level does not establish the absence of partial discharge. The numerical boundaries below describe instruments and measurement frameworks, not universal switchgear acceptance bands. [S1]

Keep dBmV, dBµV, dB SPL and pC separate

TEV dBmV references 1 mV; UTP2 ultrasonic dBµV describes its transducer chain; acoustic SPL uses a sound-pressure reference. Apparent charge in pC belongs to a calibrated electrical measuring arrangement. Do not subtract ultrasound from TEV or convert an enclosure TEV reading directly into apparent charge. [S1, S7, S9]

Where manufacturers and standards differ

EA's narrowband airborne sensor and FLIR's selectable acoustic filters describe different systems; preserve each configuration on repeat visits. FLIR's acoustic severity assessment uses distance, voltage and component type, whereas UTP2 interpretation combines TEV magnitude and repetition rate. Neither manufacturer's severity label transfers directly to the other. [S1, S9, S10]

The UTP2 manual defines its short-term TEV Severity index as magnitude in millivolts multiplied by pulses per cycle, not displayed dBmV multiplied by pulse count. Retain both inputs separately. Its adjustable alarm thresholds also mean that a green display records a configured response, not apparatus acceptance. [S1, §§3.9 and 3.12.2]

How do you distinguish PD from electrical and acoustic noise?

Use phase-resolved partial-discharge (PRPD) plots with localization and noise checks: interference synchronized to the supply can also appear phase-related. EA’s UTP2 manual V13.05.00, §4.3.1, describes discharge audio as crackling; compare that sound with the stored phase plot, background and source location. The instrument shifts ultrasound into the audible range for headphones; retain the original audio files for review. EA identifies radio transmissions, overhead insulators and transformer bushings as external electromagnetic sources; safely accessible, unrelated metalwork provides a useful background comparison. The matrix below is Anvilfield’s diagnostic synthesis. [S1, S2]

What distinguishes surface, internal and floating-metal discharge?

Name the suspected mechanism separately from confidence and action priority. FLIR groups surface/internal discharge in one PRPD class, so its camera label alone cannot identify the defective component. NETA World’s Fall 2020 account explains that poor-contact and floating-metal patterns can resemble each other, move in phase and mask weaker sources. [S9, S13]

Verify the phase reference before naming the defect

EA’s 2021 phase-reference note shows two captures at the same 11 kV cable box: an unlocked capture resembles noise, while a locked capture reveals an internal-void-type pattern. Save reference source, frequency, lock state and capture duration; restore lock before repeating a smeared pattern. The UTP2 manual states that ultrasonic classification is unavailable without phase lock, so a missing classifier is not a negative PD result. [S3; S1, §3.10.3]

How do you make PD measurements repeatable?

EA Technology's UltraTEV Plus² manual V13.05.00 requires ultrasonic comparisons at the same position with the same sensor type. Anvilfield adds three captures per point and route-end function checks in the protocol below. [S1, p. 62]

Voltage supplies electrical stress; current records operating state and influences temperature. Keep both where available, without importing a thermography minimum-load percentage into PD screening. Label comparisons matched conditions, changed conditions or unassessed. A humid-condition increase still needs action, while its cause requires comparison within the recorded environmental conditions. [S4, S6]

The UTP2 manual V13.05.00 retains bracketed historical temperature/RH readings after sensor removal. Let live readings settle and record their timestamp and live/historical status; a value carried from the previous panel is not a fresh environmental measurement. [S1, §3.6]

Distinguish repeatability from accuracy

Instrument accuracy excludes changes in geometry, background and coupling. Compare replicate spread before interpreting a small rise, and check range-end readings for clipping before calling a flat maximum stable. New localization or pattern evidence can justify action even when amplitude barely changes. [S1, S3; Anvilfield interpretation]

Can you survey energized switchgear with the doors closed?

Yes, where approved external measurements and installed signal paths support the method. EA’s UTP2 manual V13.05.00 permits ground-potential use only, requires earthed metalwork and gives the instrument, leads and accessories no inherent shock protection. It prohibits altering earth connections or disturbing plant mechanically, electrically or by applying heat during measurement. [S1, p. ii]

Keep doors, barriers, interlocks and bonds intact during external screening. Follow the site task assessment and arc-flash information; PD amplitude determines neither PPE nor approach distance. Refer bond defects to the grounding test scope without changing the bond for a reading.

Keep diagnostic access separate from routine screening

For utilization installations covered by OSHA Subpart S, 29 CFR 1910.333(a)(1) requires deenergizing exposed live parts before work on or near them unless the employer demonstrates increased hazards or infeasibility from design or operating limitations. Plan intrusive inspection under the electrical isolation procedure, with source isolation, stored-energy control, lockout/tagout and qualified-person voltage verification, including backfeed and induced voltage. OSHA 1910.333(b)(2)(iv)(B) requires checking the voltage tester immediately before and after verification on circuits over 600 V nominal; a TEV detector is not that voltage tester. OSHA 1910.331 defines the scope and exclusions for generation, transmission and distribution work. [S14]

CRYSOUND’s open-panel localization applied an external HV source after supply isolation: the diagnostic circuit was energized during testing. Treat it as a separate controlled HV test. EA’s manual prohibits measurements during nearby electrical storms or immediately after circuit energization; it gives no fixed waiting time. Its preferred clearance of more than 30 cm from perpendicular metalwork concerns sensor geometry, not personnel safety clearance. [S12, S1]

Which PD findings need action, and what makes a trend credible?

A positive result needs evidence review and an assigned action, even without a long trend. Anvilfield’s register below separates diagnostic confidence, suspected mechanism and consequence; the authorized owner/operator controls switching. Prioritize a localized finding by the insulation involved, defect evidence, available redundancy, fault exposure and outage consequence. A quiet trend on a critical bus still needs an assigned diagnostic response.

A PD amplitude trend is comparable only when sensor, position, acquisition settings, operating state and environmental conditions are recorded together; changed conditions require a separate interpretation. This Anvilfield rule follows the acquisition and monitoring evidence in S1, S3, S4 and S6.

Published repair evidence versus a continuing deterioration trend

CRYSOUND reports 30 dB TEV and 26 dBµV ultrasound before localization of a loose field deflector associated with an undersized O-ring, followed by 5 dB TEV and noise-level ultrasound after correction. Anvilfield calculates a 25 dB TEV difference from those readings. Retain those values as a repair comparison, with environmental matching recorded as unknown. [S12]

NETA World’s Fall 2020 discussion explains that PD amplitude can rise or fall with operating conditions and defect development; amplitude is not a remaining-life clock. EA’s Trinity Mirror case provides physical closure evidence instead: contact ultrasound localized activity, inspection found CT surface treeing, and replacement followed by retesting eliminated the source. [S13, S5]

Worked example: low TEV with positive ultrasound evidence

EA Technology’s May 15, 2018 survey used an UltraTEV Plus² and airborne ultrasound at indoor Substation 1, with 31 °C and 27% RH recorded. F1A was open while F1B carried the load. The table retains the source’s amplitude bins and ultrasonic “dB” notation. [S4]

Turn the observed result into a checkable work order

Anvilfield groups the two positive point records into one F1A feeder finding, preserving front-breaker and rear-cable evidence separately. Assign the maintenance lead a repeat-survey task and planned isolated inspection or offline testing, with the original environmental record attached. Localize before assigning two separate defects: the front and rear points can receive emissions from one source.

Anvilfield estimates an August 15 to November 15, 2018 retest window by adding the report’s three to six calendar months to May 15. Use that window on the F1A work order, with earlier review if the evidence worsens. [S4]

How often should a PD survey be repeated?

IEEE DEIS’s July 2023 webinar abstract describes routine survey programs operating every few months or annually; EA’s surface-discharge case specifies three to six months, and its service page describes one to two weeks of temporary monitoring. Use these published practices to select a condition-specific task, with an accountable reviewer and due date. [S11, S4, S6]

Specify who reviews monitored records, who receives alarms and who authorizes the operating response. Installing sensors does not resolve an open finding; the monitoring task must answer whether activity follows plant operation, environment or a persistent localized source. [S6; Anvilfield workflow]

Which follow-up tests confirm a suspected source?

Choose follow-up by the unresolved question: source direction, acoustic location, cable path, controlled electrical response or visible insulation damage. Anvilfield’s register assigns test objectives and closure evidence.

Controlled offline PD tests can measure inception and extinction voltages under the applied voltage sequence. Retain voltage, waveform, frequency, calibration and background, and keep these results separate from the operating survey series. NETA World’s Fall 2020 article explains this online/offline distinction; IEC 60270:2025 supplies the charge-based measurement framework. [S13, S7]

Insulation-resistance testing answers a different insulation question, while thermography documents surface-temperature evidence. Neither alone clears a localized PD finding. Combine their results by asset and defect, without substituting one method’s severity scale for another.

What the closeout record must retain

Keep the original raw traces, audio, point photographs, instrument settings, checks, background, operating conditions and coverage exclusions. Add the actual defect, corrected component, test basis and repeat result. “Source eliminated at repeat test” is a useful closure statement tied to evidence and scope. [S1, S4, S5]

How much does a switchgear PD survey cost, and how long does it take?

Anvilfield estimates $2,000 in direct cost for a 12-panel survey allowing two technicians four field hours, four total reporting person-hours, $300 in instrument allocation and $200 in travel. Increasing field time to six elapsed hours raises that estimate to $2,500. The selected planning inputs are shown below.

Build the time allowance from measurement points

Count marked points, methods and repeat captures before assigning crew time. For 12 panels with three points each, two methods and three captures per method, the plan contains 216 records. At the ten-second acquisition described by EA’s phase-reference note, raw recording totals 2,160 seconds, or 36 minutes. [S3; Anvilfield estimate]

Method note: Anvilfield’s November 5, 2026 estimate multiplies field and reporting person-hours by a selected $125 hourly rate, then adds instrument and travel allowances. Four field hours cover preparation, access, background checks, movement, placement and flagged-point repeats; four reporting hours cover evidence review and the action register. The rate is an estimate input, not a measured market average.

Price controlled HV testing, switching services, repairs, outage losses and overnight travel separately. Overhead, profit and taxes also sit outside this direct-cost model. Accessible points and diagnostic follow-up drive the scope more accurately than panel count alone.

How do you scope a repeatable switchgear survey program?

Specify a complete point map, complementary method coverage, background checks, raw evidence, condition labels and a dated findings register before counting panels. Obtain equipment records, maintenance history and operating restrictions from the owner. State which compartments each method covers and which remain unassessed.

Use the predictive-maintenance program to connect baselines, criticality and work orders, while retaining PD-specific comparability. The preventive-maintenance program guide supplies cross-trade route and closure practices. Keep thermal, PD and outage-test findings together by asset without averaging their incompatible severity scores.

Scope by site, then verify the indoor equipment inventory

Group sites geographically to plan travel between facilities. Contractors building a regional survey route use PorchRocket's Asset Search to locate visible substations and power sites with the owner of record attached to each find.

Facility records establish the indoor switchgear inventory; field testing establishes its measured condition. Confirm the utility/customer operating boundary against the one-line, since property ownership alone does not define permission to survey every asset.

Field checklist: from first reading to closed finding

Use this Anvilfield checklist to complete the survey record. Every positive or unresolved finding needs a responsible role, a due date and closure evidence; inaccessible scope receives “not assessed,” never an invented zero.

Sources and how to cite this page

Anvilfield. Editorial review date: November 5, 2026. Sources checked: November 6, 2026. The IEC references below are public publisher scopes; the summaries do not reproduce paywalled clauses.

1. S1. EA Technology, UltraTEV Plus² Operating Manual, model UTP2, 2388-MANUL-V13.05.00, June 17, 2025.
2. S2. EA Technology Australia, PD FAQ: Background Noise, February 9, 2023.
3. S3. EA Technology, Importance of accurate phase reference lock when using the UltraTEV Plus2, v1, September 17, 2021.
4. S4. EA Technology, Partial Discharge Survey Examples of MV & HV Assets, report June 8, 2018, survey May 15, 2018.
5. S5. EA Technology, UltraTEV Plus2 confirms and classifies Partial Discharge on Trinity Mirror HV Network, undated case study; initial survey February 2016.
6. S6. EA Technology Americas, Partial Discharge Surveys, undated live service page.
7. S7. IEC 60270:2025, edition 4.0, published June 5, 2025, public publisher scope.
8. S8. IEC TS 62478:2016, edition 1.0, published August 29, 2016, public publisher scope.
9. S9. FLIR Si2 series, Partial discharge detection, T810637-AG-en-US, live manual.
10. S10. FLIR Si2a, Locating sound sources, T810671-AA-en-US live manual.
11. S11. IEEE Dielectrics and Electrical Insulation Society, Improving Network Reliability with Partial Discharge Monitoring, July 2023 webinar announcement/abstract for August 8, 2023.
12. S12. CRYSOUND, Floating Partial Discharge in MV Switchgear: Detection and Repair Case Study, undated page.
13. S13. NETA World, Partial Discharge Secrets, Tips, and Tricks, Fall 2020, EA Technology-hosted publisher reprint.
14. S14. OSHA, 29 CFR 1910.331, Scope, and 1910.333, Selection and use of work practices, public regulatory text checked November 6, 2026.

Update log: November 5, 2026, final article review added layered-enclosure exclusions, clarified energized diagnostic access and OSHA isolation requirements, preserved suspected-source confidence, and removed repeated explanations.

How to cite this page: Anvilfield, TEV vs ultrasound for switchgear partial-discharge surveys, 2026, https://anvilfield.com/field-guides/electrical/switchgear-partial-discharge-survey-interpretation/

Related tools

FAQ

People also ask

• Why does medium-voltage switchgear need maintenance if it runs fine?
• What does an infrared inspection find?
• What is predictive maintenance?