Tan delta test results become unstable or inconsistent when the measurement circuit, environment or instrument changes between readings: grounding and guarding problems, surface contamination, moisture and temperature shifts, electromagnetic interference, or the test set operating beyond its load or warm-up limits. Classify the instability pattern first, eliminate the measurement causes in order, and only then interpret the results as asset condition.
Define the Instability Pattern Before Changing the Setup
Record exactly how the readings behave: slow drift, random scatter, jumps between two values, or values that depend on the operator or time of day. Each pattern points to a different cause. Drift is often temperature, warm-up or surface wetting; scatter is often interference or contact; jumps are often mode, guard or connection changes; and time-of-day dependence points to environmental or load conditions in the substation.
Capture the pattern in the test record before touching anything, because the evidence is destroyed once the setup is changed. Include the time, temperature, humidity, nearby equipment state and the exact readings with their timestamps.
Plot the readings in time order if the instrument or software supports it. A visual record distinguishes drift from scatter more clearly than a list of numbers, and it becomes part of the evidence package when the result needs to be explained to an engineer or a client. The plot should show the test conditions so the pattern can be interpreted in context.
Check Grounding, Guarding and Test-Lead Placement
The most common cause of inconsistent tan delta readings is an incorrect measurement circuit. Verify that the energised, ground and guard terminals are connected to the intended points, that the guard lead is continuous, and that no floating metal, tool or temporary conductor is near the test object. A guard that is open or connected to the wrong point changes the measured circuit and can produce readings that look like a condition change.
Inspect the test leads for damaged insulation, moisture and poor contact. Long leads should follow the instrument’s routing guidance, kept away from energised conductors and laid without sharp bends that stress the screen. Repeat the test after correcting any connection issue before considering any other cause.
When a connection issue is found and corrected, retest the full measurement rather than only the affected section, because the corrected circuit changes the whole reading. Keep the pre-correction and post-correction values in the record so the trend history shows exactly where the setup was the cause.
Floating metal is a hidden circuit changer. A ladder, a temporary ground lead, a parked tool or an ungrounded metal enclosure near the test object can couple into the measurement and shift the reading. Survey the work area and either bond or remove floating conductors before the test, and record the survey as part of the setup.
Control Surface Contamination, Moisture and Temperature
Bushing and insulator surfaces are part of the external circuit, and contamination, fog, rain or condensation changes the surface leakage path. A tan delta reading taken on a wet or polluted surface can differ from a clean-surface reading by more than the internal insulation change you are looking for. Schedule measurements on clean, dry surfaces where the procedure requires, and record the surface condition.
Temperature affects dissipation factor directly, and the winding or insulation temperature changes the reading even when the insulation is unchanged. If the procedure requires temperature correction, apply the defined method; otherwise, compare readings taken under similar temperature conditions and record the temperature with every result.
Where the procedure permits, clean the surface with the approved method and allow it to dry before measuring. Compare the clean-surface reading with the earlier value to quantify the contamination effect, and record both so the trend history is not polluted by a one-time surface condition. Never clean energised equipment, and follow the site’s work and isolation rules.
Moisture control also applies to the instrument and leads: keep the set and its connectors dry, and allow condensation to clear before the test, because moisture in the measurement path changes the result just as moisture on the test object does.
Identify Electromagnetic Interference and Frequency Effects
Nearby energised equipment couples interference into the measurement circuit, which appears as unstable or offset tan delta values. Evaluate the interference current relative to the measurement current, use the instrument’s interference suppression and grounding arrangement, and consider the time of day if interference follows the substation load. Moving the leads, re-routing them, or using the instrument’s automatic interference rejection can stabilise the reading.
Frequency effects matter if the instrument measures at more than one frequency. Insulation loss varies with frequency, so a reading at 50 Hz is not directly comparable with a reading at 60 Hz or a variable-frequency diagnostic sweep. Record the test frequency in every result and compare only like-for-like frequencies.
Interference can also enter through the supply side: an unstable site supply, shared circuit loading or a failing generator can show up as measurement noise. Confirm the set’s supply is clean and within specification, and check the internal battery or mains state if the instrument reports supply warnings.
Confirm Instrument Warm-Up, Calibration and Load Capability
Give the test set the warm-up time its manual requires, confirm the supply voltage and check the calibration validity before drawing conclusions. A set that is cold, under-supplied or out of calibration produces repeatable but wrong readings, which are the most dangerous kind of inconsistency because they look stable.
Verify that the test object’s capacitance is within the set’s rated load at the selected voltage. Driving the set beyond its capacitive load rating produces distorted voltage and unstable measurements. If the set cannot support the load, the correct answer is a different set or a lower voltage that the procedure allows, not a repeated attempt that the set cannot complete reliably.
Estimate the load before the test from the nameplate capacitance or the previous test record, and compare it with the set’s rating at the planned voltage. If the estimate is close to the rating, reduce the risk by testing at the lowest voltage the procedure permits or by using a set with more output headroom.
Repeat the Test and Record Uncertainty Transparently
After correcting the measurement causes, repeat the test under controlled conditions and record the spread of readings, not just the average. The repeatability window, together with the measurement uncertainty of the instrument, defines how small a change in tan delta can be detected with confidence. A change smaller than the repeatability window is not evidence of a condition change.
| Symptom | Likely measurement cause | First check | If it persists |
|---|---|---|---|
| Slow drift during the test | Warm-up, temperature change, surface wetting | Warm-up time, temperature record, surface condition | Review temperature correction and repeat under stable conditions |
| Random scatter between readings | Interference, poor contact, lead damage | Leads, connections, routing and grounding | Use interference suppression; change measurement time |
| Jumps between two values | Guard or mode connection change | Guard terminal and mode selection | Re-verify circuit against the connection diagram |
| Time-of-day dependence | Substation load and interference | Interference current and lead routing | Schedule measurements at stable load; document conditions |
Record the corrected repeat readings, the uncertainty statement and the conditions in the final report. If the results remain unstable after all measurement causes are eliminated, state that the readings do not support a condition conclusion and escalate the measurement setup or instrument rather than the transformer.
The uncertainty statement should name the sources that were controlled and the ones that remain: the repeatability window, the temperature uncertainty, the interference level and the instrument’s stated accuracy. A transparent uncertainty record is what allows a borderline trend to be judged fairly, because it defines how much of the change could be measurement variation.
Finally, define the decision rule for the next test date in the report: the conditions that trigger a routine repeat, the conditions that trigger additional diagnostics, and the conditions that trigger escalation. Writing the rule in advance removes judgement under pressure and keeps the programme consistent across operators.
Frequently Asked Questions
Why does the reading drift during a long test?
Drift is commonly caused by instrument warm-up, temperature changes in the insulation or surface, or slowly wetting contamination. Record temperature and surface condition, allow the set to warm up fully, and repeat under stable conditions before interpreting the value.
How can we tell noise from a real condition change?
Noise typically produces scatter that changes with lead routing, grounding, interference current or time of day, while a condition change produces a repeatable shift under controlled conditions. Correct the setup, repeat the test, and compare the repeatability window before drawing any conclusion.
What should we record to make repeatability verifiable?
Record the full conditions with every reading: test mode, voltage, frequency, temperature, humidity, surface condition, lead arrangement, interference level, instrument warm-up and calibration date. A repeatable record is what allows a second engineer to reproduce the measurement and trust the trend.
For the framework that defines what tan delta testing can and cannot establish, see the tan delta testing guide. To qualify a test set with the interference suppression and load capability your site needs, review tan delta test set options and request a technical proposal.