Insulation Test Fixtures: Measure Fixture Leakage Before the Product

Measurement purpose

Characterize fixture leakage during an insulation measurement and determine whether it can be separated from the intended specimen current.

Specimens and conditions

Populated fixture state

Identify electrode spacing, ceramic support, fired glaze boundaries and contact locations in the actual installed geometry.

Blank representation

Document whether the blank is an empty fixture, geometric dummy or other controlled configuration and which fields or paths it does not reproduce.

Equipment and records required

• Low-current measurement chain: Preserve voltage, polarity, guarding, range, timing and cable position across blank and product observations.
• Environmental record: Track local temperature, humidity, conditioning and handling through the entire bracketing sequence.

Method sequence

1. Acquire the first blank

   Measure the identified fixture state under the required electrical and environmental sequence.

   Record: Blank current, voltage and timing.

2. Acquire the specimen

   Install the identified specimen without unrecorded changes to cables, guards or support conditions.

   Record: Total current and physical arrangement.

3. Acquire the closing blank

   Repeat the original blank condition and compare its movement with the correction allowance.

   Record: Closing blank, drift assessment and correction decision.

Decision and uncertainty

Correction is permitted only when the unwanted current is comparably represented and the residual uncertainty supports the required result.

Installing a specimen can change electric fields, surface paths and fixture coupling; interpolation between blank readings does not prove those changes are absent.

The method owner approves blank equivalence and corrections; the product owner controls which terminal-to-terminal leakage paths belong to acceptance.

Traceable outputs

Draw the blank and populated circuits separately

Mark the current path through the specimen and the unwanted paths through supports, connectors and cables. Then draw the blank condition as a second circuit. Removing the ceramic may also remove a surface contact, change electrode spacing or alter the electric field near a support.

A good blank comparison states these differences openly. An empty fixture is useful evidence about the empty fixture, not automatic proof of the current it would draw when populated. A geometric dummy may better reproduce some conditions, but its own surface and bulk conduction must also be considered.

Bracket the product reading with comparable blanks

Acquire a blank before the specimen, measure the specimen and then repeat the original blank arrangement. Preserve the same voltage history and observation time for all three. Record actual timestamps, because fixture movement or environmental drift can occur while the specimen is loaded.

The closing blank tests whether the initial blank remained representative over the sequence. If the two differ materially, a single fixed correction is not supported. Linear interpolation may be a useful model only when the drift behavior and timing justify it; two endpoints alone cannot rule out a jump during the product reading.

Subtract current rather than resistance

For parallel ohmic paths at the same voltage, currents and conductances add. If a valid total measurement corresponds to 80 gigaohms and a comparable fixture blank to 400 gigaohms, the intended conductance is 1/80 minus 1/400 in inverse gigaohms. The resulting intended resistance is 100 gigaohms.

Subtracting 80 from 400, or subtracting 400 from 80, has no corresponding parallel-circuit meaning. At an assumed 10 volts, the same example is total current 0.125 nanoampere minus blank current 0.025 nanoampere, leaving 0.100 nanoampere. The values are illustrative and not fixture or substrate specifications.

Compare blank uncertainty with the residual signal

A correction can improve the nominal result while making its relative uncertainty much worse. Suppose the residual specimen current is 0.100 nanoampere and the uncertainty associated with the blank contribution is 0.005 nanoampere. That contribution alone is five percent of the residual current, before adding the populated measurement uncertainty.

As the total and blank currents approach each other, subtraction becomes increasingly sensitive to their uncertainties and covariance. A residual near zero should lead to a justified bound or an unresolved result, not a large precise resistance obtained by dividing through a barely positive number.

Decide whether the blank supports a correction

Use the physical comparison and the bracketing history together. A correction is a measurement-model decision, not a cosmetic way to move resistance above a limit.

Do not guard away the leakage the product must withstand

A driven guard can divert selected unwanted currents, but the placement determines which paths are removed. If the requirement concerns leakage between the finished assembly terminals, diverting part of its surface current may change the measurand instead of improving the same measurement.

Maintain separate names for a guarded material-oriented result and an unguarded assembly result where both are needed. Follow the instrument's connection and safety instructions because a driven guard need not be at ground potential. A shield and a guard have different functions and should not be interchanged by appearance.

Control handling and environmental history

Humidity, contamination and residual cleaning films can change both fixture and specimen currents. Record who handled the contact surfaces, the permitted cleaning state and the time between conditioning and measurement. Do not clean only the specimen while leaving an unexplained fixture contribution in place.

A deliberate cleaning comparison should preserve before-and-after records and the identity of the treated part. It can help locate a contamination contribution, but it creates a changed specimen state. The cleaned result does not automatically replace the original acceptance result unless the production and retest procedure authorizes that disposition.

Deliver the correction model with the measured result

Retain raw blank and populated currents, timestamps, actual voltage and the exact correction formula. State whether interpolation was used and why. Include the fixture drawing or photographs needed to reconstruct electrode and guard placement, together with the uncertainty of the resulting specimen quantity.

If the fixture correction consumes too much of the required margin, improving insulation, routing or the measurement boundary may be more useful than collecting more decimal places. For customer review, identify the minimum insulation requirement and the installed product paths so the test can distinguish fixture limitations from genuine circuit behavior.